1 2017-12-12 Julia Koval <julia.koval@intel.com>
3 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
4 (PTA_CANNONLAKE): Remove PTA_CLWB.
6 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
10 (vcondv2div2df): Likewise.
11 (vconduv2dfv2di): Likewise.
13 2017-12-12 Jakub Jelinek <jakub@redhat.com>
15 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
16 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
17 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
18 ATTR_MATHFN_FPROUNDING_ERRNO.
20 2017-12-12 Richard Biener <rguenther@suse.de>
22 PR tree-optimization/83385
23 * graphite-scop-detection.c (get_order, order): Remove.
24 (bb_to_rpo): New global.
26 (build_scops): Sort pbbs in RPO order.
28 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
30 * combine.c (simplify_set): Do not transform subregs to zero_extends
31 if the destination mode is a vector mode.
33 2017-12-12 Jakub Jelinek <jakub@redhat.com>
35 PR tree-optimization/83359
36 * tree-cfg.h (fold_loop_internal_call): Declare.
37 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
38 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
39 (find_loop_dist_alias): New function.
40 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
41 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
44 PR tree-optimization/80631
45 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
46 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
47 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
48 hardcoding zero as the value if COND_EXPR is never true. For
49 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
50 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
51 hardcoding MAX_EXPR as the reduction operation.
52 (is_nonwrapping_integer_induction): Allow negative step.
53 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
54 vect_create_epilog_for_reduction, if no value is suitable, don't
55 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
57 2017-12-12 Richard Biener <rguenther@suse.de>
59 PR tree-optimization/81889
60 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
61 range info from the non-wrapping IV instead of just the range
64 2017-12-12 Julia Koval <julia.koval@intel.com>
66 * config.gcc: Add vaesintrin.h.
67 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
69 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
70 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
72 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
73 * config/i386/immintrin.h: Include vaesintrin.h.
74 * config/i386/sse.md (vaesdec_<mode>): New pattern.
75 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
76 _mm_aesdec_epi128): New intrinsics.
78 2017-12-12 Julia Koval <julia.koval@intel.com>
80 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
81 OPTION_MASK_ISA_VAES_UNSET): New.
82 (ix86_handle_option): Handle -mvaes.
83 * config/i386/cpuid.h: Define bit_VAES.
84 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
85 * config/i386/i386-c.c (__VAES__): New.
86 * config/i386/i386.c (ix86_target_string): Add -mvaes.
87 (ix86_valid_target_attribute_inner_p): Ditto.
88 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
89 * config/i386/i386.opt: Add -mvaes.
90 * doc/invoke.texi: Ditto.
92 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
94 * debug.h (gcc_debug_hooks): Add inline_entry.
95 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
96 * debug.c (do_nothing_debug_hooks): Likewise.
97 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
98 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
99 (dwarf2_lineno_debug_hooks): Likewise.
101 * common.opt (gstatement-frontiers): New, setting
102 debug_nonbind_markers_p.
103 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
104 * toplev.c (process_options): Autodetect value for debug statement
106 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
107 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
109 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
110 markers. Integrate source bind into debug stmt expand loop.
111 (pass_expand::execute): Check debug marker limit. Avoid deep
112 TER and expand debug locations for debug bind insns only.
113 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
115 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
116 * final.c (reemit_insn_block_notes): Take current block from
117 nonbind markers. Declare note where it's first set.
118 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
119 begin stmt markers if enabled.
120 (notice_source_line): Handle nonbind markers. Fail if their
121 location is unknown or that of builtins.
122 (rest_of_handle_final): Convert begin stmt markers to notes if
123 var-tracking didn't run.
124 (rest_of_clean_state): Skip begin stmt markers.
125 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
127 * function.c (allocate_struct_function): Set begin_stmt_markers.
128 * function.h (struct function): Add debug_marker_count counter
129 and debug_nonbind_markers flag.
130 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
131 * gimple-low.c (lower_function_body): Adjust
132 debug_nonbind_markers.
133 (lower_stmt): Drop or skip gimple debug stmts.
134 (lower_try_catch): Skip debug stmts.
135 * gimple.c (gimple_build_debug_begin_stmt): New.
136 (gimple_copy): Increment debug_marker_count if copying one.
137 * gimple.h (gimple_build_debug_begin_stmt): Declare.
138 * gimplify.c (rexpr_location): New.
139 (rexpr_has_location): New.
140 (warn_switch_unreachable_r): Handle gimple debug stmts.
141 (shortcut_cond_r): Call expr_location.
143 (find_goto_label): New.
144 (shortcut_cond_expr): Call expr_has_location, expr_location, and
146 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
148 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
149 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
150 (LANG_HOOKS_INITIALIZER): ... this.
151 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
152 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
154 (update_ebb_live_info): Skip debug insn markers.
155 * lra.c (debug_insn_static_data): Rename to...
156 (debug_bind_static_data): ... this.
157 (debug_marker_static_data): New.
158 (lra_set_insn_recog_data): Select one of the above depending
160 (lra_update_isn_regno_info): Don't assume debug insns have
162 (push_insns): Skip debug insns.
163 * lto-streamer-in.c (input_function): Drop debug stmts
164 depending on active options. Adjust debug_nonbind_markers.
165 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
166 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
167 begin stmt marker notes.
168 (print_insn): Likewise.
169 * recog.c (extract_insn): Recognize rtl for debug markers.
170 * rtl.def (DEBUG_MARKER): New.
171 * tree-inline.c: Include params.h.
172 (remap_gimple_stmt): Handle nonbind markers.
173 (maybe_move_debug_stmts_to_successors): Likewise.
174 (copy_debug_stmt): Likewise.
175 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
176 markers regardless of no side effects.
177 (tsi_link_before): Don't update container's side effects when adding
179 (tsi_link_after): Likewise.
180 (expr_first): Skip begin stmt markers.
181 (expr_last): Likewise.
182 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
183 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
184 Disregard nonbind markers.
185 * tree.c (make_node_stat): Don't set side effects for begin stmt
187 (build1_stat): Likewise.
188 * tree.def (DEBUG_BEGIN_STMT): New.
189 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
190 * var-tracking.c (delete_debug_insns): Renamed to...
191 (delete_vta_debug_insns): ... this.
192 (reemit_marker_as_note): New.
193 (vt_initialize): Reemit markers.
194 (delete_vta_debug_insns): Likewise.
195 (vt_debug_insns_local): Reemit or delete markers.
196 (variable_tracking_main_1): Likewise.
197 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
198 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
199 (gimple_debug_nonbind_marker_p): New.
200 (gimple_build_debug_bind): Adjust.
201 (gimple_build_debug_begin_stmt): New.
202 * doc/invoke.texi (max-debug-marker-count): New param.
203 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
204 (debug_parameter_ref, debug_marker): New.
205 (NOTE_INSN_BEGIN_STMT): New.
206 (DEBUG_INSN): Describe begin stmt markers.
208 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
209 without debug insns, we wouldn't, but clean up debug insns
210 after a control flow insn nevertheless.
212 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
213 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
214 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
215 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
216 (rtl_tidy_fallthru_edge): Likewise.
217 (rtl_verify_fallthru): Likewise.
218 (rtl_verify_bb_layout): Likewise.
219 (skip_insns_after_block): Likewise.
220 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
221 * dwarf2out.c: Include print-rtl.h.
222 (dwarf2out_next_real_insn): New.
223 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
224 Dump debug binds in asm comments.
225 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
226 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
227 callers to use gsi_start_nondebug_bb instead.
228 (gsi_after_labels): Skip gimple debug stmts.
229 (gsi_start_nondebug): New.
230 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
231 (proper_loop_form_for_interchange): Adjust.
232 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
233 * gimple.h (gimple_seq_last_nondebug_stmt): New.
234 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
235 (collect_fallthrough_labels): Likewise.
236 (should_warn_for_implicit_fallthrough): Likewise.
237 (warn_implicit_fallthrough_r): Likewise.
238 (expand_FALLTHROUGH_r): Likewise.
239 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
240 (graphite_copy_stmts_from_block): Skip nonbind markers.
241 * haifa-sched.c (sched_extend_bb): Skip debug insns.
242 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
243 * jump.c (clean_barriers): Skip debug insns.
244 * omp-expand.c (expand_parallel_call): Skip debug insns.
245 (expand_task_call): Likewise.
246 (remove_exit_barrier): Likewise.
247 (expand_omp_taskreg): Likewise.
248 (expand_omp_for_init_counts): Likewise.
249 (expand_omp_for_generic): Likewise.
250 (expand_omp_for_static_nochunk): Likewise.
251 (expand_omp_for_static_chunk): Likewise.
252 (expand_omp_simd): Likewise.
253 (expand_omp_taskloop_for_outer): Likewise.
254 (expand_omp_taskloop_for_inner): Likewise.
255 (expand_oacc_for): Likewise.
256 (expand_omp_sections): Likewise.
257 (expand_omp_single): Likewise.
258 (expand_omp_synch): Likewise.
259 (expand_omp_atomic_load): Likewise.
260 (expand_omp_atomic_store): Likewise.
261 (expand_omp_atomic_fetch_op): Likewise.
262 (expand_omp_atomic_pipeline): Likewise.
263 (expand_omp_atomic_mutex): Likewise.
264 (expand_omp_target): Likewise.
265 (grid_expand_omp_for_loop): Likewise.
266 (grid_expand_target_grid_body): Likewise.
267 (build_omp_regions_1): Likewise.
268 * omp-low.c (check_combined_parallel): Skip debug stmts.
269 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
270 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
271 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
273 * tree-cfg.c (make_blobs_1): Skip debug stmts.
274 (make_edges): Likewise.
275 (cleanup_dead_labels): Likewise.
276 (gimple_can_merge_blocks_p): Likewise.
277 (stmt_starts_bb_p): Likewise.
278 (gimple_block_label): Likewise.
279 (gimple_redirect_edge_and_branch): Likewise.
280 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
282 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
284 * tree-pretty-print (print_declaration): Omit initializer in slim
286 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
288 (eliminate_unnecessary_stmts): Stabilize block removal order.
289 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
290 * var-tracking.c (get_first_insn): New.
291 (vt_emit_notes): Call it.
292 (vt_initialize): Walk any insns before the first BB.
293 (delete_debug_insns): Likewise.
295 * gimple.h (enum gimple_debug_subcode): Add
296 GIMPLE_DEBUG_BEGIN_STMT.
297 (gimple_debug_begin_stmt_p): New.
298 (gimple_debug_nonbind_marker_p): New.
299 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
300 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
301 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
302 * insn-notes.def (BEGIN_STMT): New.
303 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
304 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
305 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
306 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
307 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
308 (INSN_DEBUG_MARKER_KIND): New.
309 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
310 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
311 (INSN_VAR_LOCATION_PTR): New.
312 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
314 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
315 deps only if debug bind insns are enabled.
316 (pass_expand::execute): Avoid deep TER and expand
317 debug locations for debug bind insns only.
318 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
319 debug stmts special handling down to debug bind stmts.
320 * combine.c (try_combine): Narrow debug insns special handling
321 down to debug bind insns.
322 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
323 Narrow debug insns preexisting special handling down to debug
325 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
326 handling down to debug bind insns.
327 * function.c (instantiate_virtual_regs): Skip debug markers,
328 adjust handling of debug binds.
329 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
330 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
331 * haifa-sched.c (schedule_insn): Narrow special handling of debug
332 insns to debug bind insns.
333 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
334 special handling of debug stmts to debug bind stmts.
335 * ipa-split.c (split_function): Likewise.
336 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
337 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
339 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
340 * regrename.c (build_def_use): Likewise.
341 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
342 (pass_cprop_hardreg): Narrow special casing of debug insns to
344 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
345 * reload1.c (reload): Likewise.
346 * sese.c (sese_insert_phis_for_liveouts): Narrow special
347 casing of debug stmts to debug bind stmts.
348 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
349 * ssa-iterators.h (num_imm_uses): Likewise.
350 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
351 debug stmts to debug bind stmts.
352 * tree-inline.c (tree_function_versioning): Narrow special casing
353 of debug stmts to debug bind stmts.
354 * tree-loop-distribution.c (generate_loops_for_partition):
355 Narrow special casing of debug stmts to debug bind stmts.
356 * tree-sra.c (analyze_access_subtree): Narrow special casing
357 of debug stmts to debug bind stmts.
358 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
359 stmts to debug bind stmts.
360 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
361 casing of debug stmts to debug bind stmts.
362 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
363 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
364 casing of debug stmts to debug bind stmts.
365 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
367 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
368 debug stmts to debug bind stmts.
369 (gimple_replace_ssa_lhs): Likewise.
370 (insert_debug_temp_for_var_def): Likewise.
371 (insert_debug_temps_for_defs): Likewise.
372 (reset_debug_uses): Likewise.
373 * tree-ssanames.c (release_ssa_name_fn): Likewise.
374 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
375 (adjust_debug_stmts): Likewise.
376 (adjust_phi_and_debug_stmts): Likewise.
377 (vect_do_peeling): Likewise.
378 * tree-vect-loop.c (vect_transform_loop): Likewise.
379 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
380 * var-tracking.c (adjust_mems): Narrow special casing of debug
381 insns to debug bind insns.
382 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
383 (compute_bb_dataflow, vt_find_locations): Likewise.
384 (vt_expand_loc, emit_notes_for_changes): Likewise.
385 (vt_init_cfa_base): Likewise.
386 (vt_emit_notes): Likewise.
387 (vt_initialize): Likewise.
388 (vt_finalize): Likewise.
390 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
391 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
392 (next_nonnote_nondebug_insn_bb): New.
393 (prev_nonnote_nondebug_insn_bb): New.
394 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
395 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
396 (prev_nonnote_nondebug_insn_bb): Declare.
397 (next_nonnote_nondebug_insn_bb): Declare.
398 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
399 * cfgrtl.c (get_last_bb_insn): Likewise.
400 * lra.c (push_insns): Likewise.
402 2017-12-11 David Malcolm <dmalcolm@redhat.com>
405 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
407 * selftest.c (selftest::test_runner::test_runner): New ctor.
408 (selftest::test_runner::~test_runner): New dtor.
409 * selftest.h (class selftest::test_runner): New class.
411 2017-12-11 Carl Love <cel@us.ibm.com>
413 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
414 vec_extract_fp32_from_shortl]): Add #defines.
415 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
416 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
417 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
418 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
419 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
420 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
421 * doc/extend.texi: Add documentation for the added builtins.
423 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
425 PR rtl-optimization/80693
426 PR rtl-optimization/81019
427 PR rtl-optimization/81020
428 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
429 are not mentioned in i3. Place the REG_UNUSED note on i2,
430 possibly modified to REG_DEAD, if it did not originate in i3.
432 2017-12-11 Jakub Jelinek <jakub@redhat.com>
434 * recog.c (store_data_bypass_p_1): New function.
435 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
436 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
438 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
440 PR rtl-optimization/83361
441 * ifcvt.c (if_convert): Call fixup_partitions.
443 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
445 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
446 early folding of splat_u{8,16,32}.
448 2017-12-11 Jakub Jelinek <jakub@redhat.com>
450 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
451 output_operand_lossage first argument with capital letter.
452 (aarch64_override_options): Don't start error and sorry first argument
455 2017-12-11 Andi Kleen <ak@linux.intel.com>
457 PR gcov-profile/83355
458 * auto-profile.c (string_table::get_index_by_decl): Don't
459 recurse when abstract origin points to itself.
461 2017-12-11 Bin Cheng <bin.cheng@arm.com>
463 PR tree-optimization/83320
464 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
465 (prune_datarefs_not_in_loop): Ditto.
467 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
469 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
471 2017-12-10 Jakub Jelinek <jakub@redhat.com>
473 PR tree-optimization/83337
474 * gimple-loop-interchange.cc (compute_access_stride): Handle
475 bitfield DRs properly.
477 2017-12-09 Jakub Jelinek <jakub@redhat.com>
479 PR tree-optimization/83338
480 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
481 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
482 vector of unsigned integers to vector of signed integers.
484 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
486 PR rtl-optimization/83317
487 * lra-constraints.c (process_address_1): Add insn code check.
489 2017-12-08 Michael Matz <matz@suse.de>
491 Fix PR tree-optimization/83323
492 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
493 head-controlled loops and loop BBs.
494 * common.opt (funroll-and-jam): Remove, instead ...
495 (floop-unroll-and-jam): ... reuse this option.
496 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
497 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
498 (-floop-unroll-and-jam): ... this option.
500 2017-12-08 Jakub Jelinek <jakub@redhat.com>
502 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
503 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
504 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
506 2017-12-08 Richard Biener <rguenther@suse.de>
509 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
510 handle accesses outside of zero-sized vars.
512 2017-12-08 Martin Jambor <mjambor@suse.cz>
514 PR tree-optimization/83141
515 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
516 test for MEM_REFs implicitely changing types with padding. Remove
518 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
520 2017-12-08 Jakub Jelinek <jakub@redhat.com>
522 * config/arc/arc.c (arc_attribute_table): Add exclusions to
524 * config/avr/avr.c (avr_attribute_table): Likewise.
525 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
526 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
527 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
528 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
529 of struct attribute_spec.
530 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
532 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
535 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
537 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
540 * profile-count.c (profile_count::from_gcov_type): Move from
541 profile-count.h; handle overflow.
542 * profile-count.h (profile_count::from_gcov_type): Move offline.
544 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
546 PR rtl-optimization/83304
547 * combine.c (move_deaths): If we do not know where a register died,
550 2017-12-08 Richard Biener <rguenther@suse.de>
552 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
553 Provide -fopt-info-loop feedback when we interchange in a nest.
555 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
557 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
558 for armv6 ARM CPU IDs.
560 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
562 * common/config/arm/arm-common.c: Include <algorithm>.
563 (INCLUDE_VECTOR): Define.
564 (compare_opt_names): New function.
565 (arm_rewrite_selected_arch): Only strip out extensions that can be
566 expressed through -mfpu. Sort the remaining extensions
569 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
571 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
572 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
573 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
574 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
576 2017-06-08 Tristan Gingold <gindold@adacore.com>
579 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
580 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
582 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
584 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
585 of struct attribute_spec.
587 2017-12-08 Julia Koval <julia.koval@intel.com>
589 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
590 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
591 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
592 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
593 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
594 _mm_maskz_dpwssds_epi32): Ditto.
596 2017-12-08 Richard Biener <rguenther@suse.de>
598 PR tree-optimization/81303
599 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
600 conditions try to create a comparison vector type matching
601 the data vector type.
602 (vectorizable_condition): Adjust.
603 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
604 Leave invariant conditions alone in case we can vectorize those.
606 2017-12-08 Julia Koval <julia.koval@intel.com>
608 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
609 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
610 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
611 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
612 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
614 2017-12-08 Julia Koval <julia.koval@intel.com>
616 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
617 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
618 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
619 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
620 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
621 _mm_maskz_dpbusds_epi32): New intrinsics.
623 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
625 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
627 (ld<bh>io_signed): Likewise.
628 (st<bhw>io): Likewise.
629 * config/nios2/predicates.md (ldstio_memory_operand): Allow
630 SMALL_INT12 constant integer operand.
632 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
634 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
636 (nios2_large_constant_allowed): ...this. Adjust uses.
637 (nios2_plus_symbolic_constant_p): Rename to...
638 (nios2_plus_large_constant_p): ...this. Adjust uses.
639 (nios2_legitimate_address_p): Correct CONST_INT handling.
640 (nios2_symbolic_memory_operand_p): Rename to...
641 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
642 (nios2_large_constant_p): Check for large constant integers too.
643 (nios2_split_large_constant): Handle constant integers.
644 (nios2_split_symbolic_memory_operand): Rename to...
645 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
646 (nios2_legitimize_constant_address): Handle constant integers.
647 (r0rel_constant_p): Handle small constant integers.
648 (nios2_print_operand_address): Handle r0-relative integer addresses.
649 * config/nios2/nios2-protos.h: Adjust for renamed functions.
650 * config/nios2/nios2.md: Adjust for renamed functions.
652 2017-12-07 Andrew Waterman <andrew@sifive.com>
654 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
655 (TARGET_HAVE_SRODATA_SECTION): New define.
656 (riscv_select_section): New function.
658 2017-12-08 Joseph Myers <joseph@codesourcery.com>
659 Alexander Monakov <amonakov@ispras.ru>
660 Jakub Jelinek <jakub@redhat.com>
663 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
665 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
667 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
668 a bool success value. Don't call output_operand_lossage here.
669 (aarch64_print_ldpstp_address): Return a bool success value.
670 (aarch64_print_operand_address): Call output_addr_const if
671 aarch64_print_address_internal fails.
672 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
673 'y'; call output_operand_lossage instead. Call output_operand_lossage
674 if aarch64_print_ldpstp_address fails.
676 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
678 * tree-vector-builder.h
679 (tree_vector_builder::binary_encoded_nelts): Declare.
680 * tree-vector-builder.c
681 (tree_vector_builder::binary_encoded_nelts): New function.
682 * fold-const.c (negate_expr_p): Likewise.
683 (operand_equal_p, fold_checksum_tree): Likewise.
684 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
685 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
686 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
687 (uniform_vector_p): Likewise.
688 * varasm.c (const_hash_1, compare_constant): Likewise.
689 * tree-ssa-ccp.c: Include tree-vector-builder.h.
690 (valid_lattice_transition): Operate directly on the VECTOR_CST
692 * ipa-icf.c: Include tree-vector-builder.h.
693 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
694 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
696 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
698 * tree.c (build_vector): Delete.
699 * tree.h (build_vector): Make static and move into the self-testing
702 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
704 * vector-builder.h (vector_builder::derived): New const overload.
705 (vector_builder::elt): New function.
706 * tree-vector-builder.h (tree_vector_builder::type): New function.
707 (tree_vector_builder::apply_step): Declare.
708 * tree-vector-builder.c (tree_vector_builder::apply_step): New
710 * gimple-fold.h (tree_vector_builder): Declare.
711 (gimple_build_vector): Take a tree_vector_builder instead of a
712 type and vector of elements.
713 * gimple-fold.c (gimple_build_vector): Likewise.
714 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
716 (get_initial_defs_for_reduction): Likewise.
717 (vectorizable_induction): Likewise.
719 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
721 * tree-vector-builder.h
722 (tree_vector_builder::new_binary_operation): Declare.
723 * tree-vector-builder.c
724 (tree_vector_builder::new_binary_operation): New function.
725 * fold-const.c (fold_relational_const): Use it.
726 (const_binop): Likewise. Check that both input vectors have
727 the same number of elements, thus excluding things like WIDEN_SUM.
728 Check whether it is possible to operate directly on the encodings
731 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
733 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
734 new_unary_operation, operating only on the encoded elements.
735 (const_unop): Likewise.
736 (exact_inverse): Likewise.
737 (distributes_over_addition_p): New function.
738 (const_binop): Use tree_vector_builder and new_unary_operation
739 for combinations of VECTOR_CST and INTEGER_CST. Operate only
740 on the encoded elements unless the encoding is strided and the
741 operation does not distribute over addition.
742 (fold_convert_const): Use tree_vector_builder and
743 new_unary_operation. Operate only on the encoded elements
744 for truncating integer conversions, or for non-stepped encodings.
746 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
748 * config/sparc/sparc.c: Include tree-vector-builder.h.
749 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
750 * expmed.c: Include tree-vector-builder.h.
751 (make_tree): Use tree_vector_builder instead of build_vector.
752 * fold-const.c: Include tree-vector-builder.h.
753 (const_binop): Use tree_vector_builder instead of build_vector.
754 (const_unop): Likewise.
755 (native_interpret_vector): Likewise.
756 (fold_vec_perm): Likewise.
757 (fold_ternary_loc): Likewise.
758 * gimple-fold.c: Include tree-vector-builder.h.
759 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
761 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
762 (simplify_vector_constructor): Use tree_vector_builder instead
764 * tree-vect-generic.c: Include tree-vector-builder.h.
765 (add_rshift): Use tree_vector_builder instead of build_vector.
766 (expand_vector_divmod): Likewise.
767 (optimize_vector_constructor): Likewise.
768 * tree-vect-loop.c: Include tree-vector-builder.h.
769 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
770 of build_vector. Explicitly use a stepped encoding for
772 * tree-vect-slp.c: Include tree-vector-builder.h.
773 (vect_get_constant_vectors): Use tree_vector_builder instead
775 (vect_transform_slp_perm_load): Likewise.
776 (vect_schedule_slp_instance): Likewise.
777 * tree-vect-stmts.c: Include tree-vector-builder.h.
778 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
779 (vect_gen_perm_mask_any): Likewise.
780 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
781 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
783 (build_vector_from_val): Likewise. Explicitly use a duplicate
786 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
788 * doc/generic.texi (VECTOR_CST): Describe new representation of
790 * vector-builder.h: New file.
791 * tree-vector-builder.h: Likewise.
792 * tree-vector-builder.c: Likewise.
793 * Makefile.in (OBJS): Add tree-vector-builder.o.
794 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
795 * tree-core.h (tree_base): Add a vector_cst field to the u union.
796 (tree_vector): Change the number of elements to
797 vector_cst_encoded_nelts.
798 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
799 (VECTOR_CST_ELTS): Delete.
800 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
801 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
802 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
803 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
804 (VECTOR_CST_ENCODED_ELT): Likewise.
805 (vector_cst_encoded_nelts): New function.
806 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
807 VECTOR_CST_NELTS_PER_PATTERN as arguments.
808 (vector_cst_int_elt, vector_cst_elt): Declare.
809 * tree.c: Include tree-vector-builder.h.
810 (tree_code_size): Abort if passed VECTOR_CST.
811 (tree_size): Update for new VECTOR_CST layout.
812 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
813 VECTOR_CST_NELTS_PER_PATTERN as arguments.
814 (build_vector): Use tree_vector_builder.
815 (vector_cst_int_elt, vector_cst_elt): New functions.
816 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
817 encoded elements and then create the vector in the canonical form.
818 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
819 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
820 (tree_c_tests): Call test_vector_cst_patterns.
821 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
823 (hash_tree): Likewise.
824 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
825 (streamer_write_tree_header): Likewise.
826 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
827 (streamer_alloc_tree): Likewise. Update call to make_vector.
828 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
830 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
832 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
833 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
834 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
836 2017-12-07 Bin Cheng <bin.cheng@arm.com>
837 Richard Biener <rguenther@suse.de>
839 PR tree-optimization/81303
840 * Makefile.in (gimple-loop-interchange.o): New object file.
841 * common.opt (floop-interchange): Reuse the option from graphite.
842 * doc/invoke.texi (-floop-interchange): Ditto. New document for
843 -floop-interchange and mention it for -O3.
844 * opts.c (default_options_table): Enable -floop-interchange at -O3.
845 * gimple-loop-interchange.cc: New file.
846 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
847 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
848 * passes.def (pass_linterchange): New pass.
849 * timevar.def (TV_LINTERCHANGE): New time var.
850 * tree-pass.h (make_pass_linterchange): New declaration.
851 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
852 interchange. Record IV before/after increment in new parameters.
853 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
854 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
856 (check_reduction_path): ...New function here.
857 * tree-vectorizer.h (check_reduction_path): New declaration.
859 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
862 PR rtl-optimization/80818
863 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
864 always early clobbered.
865 * lra-lives.c (process_bb_lives): Check input hard regs for early
866 clobbered non-operand hard reg.
868 2017-12-07 Jakub Jelinek <jakub@redhat.com>
871 * tree-cfg.c (verify_gimple_assign_binary): Don't require
872 types_compatible_p, just that TYPE_MODE is the same.
874 2017-12-07 Martin Sebor <msebor@redhat.com>
877 * attribs.c (empty_attribute_table): Initialize new member of
878 struct attribute_spec.
879 (decl_attributes): Add argument. Handle mutually exclusive
880 combinations of attributes.
881 (selftests::test_attribute_exclusions): New function.
882 (selftests::attribute_c_tests): Ditto.
883 * attribs.h (decl_attributes): Add default argument.
884 * selftest.h (attribute_c_tests): Declare.
885 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
886 * tree-core.h (attribute_spec::exclusions, exclude): New type and
888 * doc/extend.texi (Common Function Attributes): Update const and pure.
889 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
890 of struct attribute_spec.
891 * config/arc/arc.c (arc_attribute_table): Same.
892 * config/arm/arm.c (arm_attribute_table): Same.
893 * config/avr/avr.c ( avr_attribute_table): Same.
894 * config/bfin/bfin.c (bfin_attribute_table): Same.
895 * config/cr16/cr16.c (cr16_attribute_table): Same.
896 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
897 * config/h8300/h8300.c (h8300_attribute_table): Same.
898 * config/i386/i386.c (ix86_attribute_table): Same.
899 * config/ia64/ia64.c (ia64_attribute_table): Same.
900 * config/m32c/m32c.c (m32c_attribute_table): Same.
901 * config/m32r/m32r.c (m32r_attribute_table): Same.
902 * config/m68k/m68k.c (m68k_attribute_table): Same.
903 * config/mcore/mcore.c (mcore_attribute_table): Same.
904 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
905 * config/mips/mips.c (mips_attribute_table): Same.
906 * config/msp430/msp430.c (msp430_attribute_table): Same.
907 * config/nds32/nds32.c (nds32_attribute_table): Same.
908 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
909 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
910 * config/rl78/rl78.c (rl78__attribute_table): Same.
911 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
912 * onfig/rx/rx.c (rx_attribute_table): Same.
913 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
914 * config/sh/sh.c (sh_attribute_table): Same.
915 * config/sparc/sparc.c (sparc_attribute_table): Same.
916 * config/spu/spu.c (spu_attribute_table): Same.
917 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
918 * config/v850/v850.c (v850_attribute_table): Same.
919 * config/visium/visium.c (visium_attribute_table): Same.
921 2017-12-07 Tamar Christina <tamar.christina@arm.com>
924 * config/arm/arm.c (INCLUDE_STRING): Define.
925 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
926 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
929 2017-12-07 Michael Matz <matz@suse.de>
931 Add unroll and jam pass
933 * gimple-loop-jam.c: New file.
934 * Makefile.in (OBJS): Add gimple-loop-jam.o.
935 * common.opt (funroll-and-jam): New option.
936 * opts.c (default_options_table): Add unroll-and-jam at -O3.
937 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
938 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
939 * passes.def: Add pass_loop_jam.
940 * timevar.def (TV_LOOP_JAM): Add.
941 * tree-pass.h (make_pass_loop_jam): Declare.
942 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
943 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
944 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
945 to flow_loop_tree_node_add.
946 (duplicate_subloops, copy_loops_to): Append to sibling list.
947 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
948 * doc/invoke.texi (-funroll-and-jam): Document new option.
949 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
951 2017-12-07 Richard Biener <rguenther@suse.de>
953 PR tree-optimization/83296
954 PR tree-optimization/67769
955 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
956 flow sensitive info in an unrelated BB.
957 (value_replacement): Use reset_flow_sensitive_info.
958 (minmax_replacement): Reset flow sensitive info on the def
959 we move. Do not reset flow sensitive info in the whole BB
961 (abs_replacement): Likewise.
963 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
966 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
967 rs6000_cpu to the given -mcpu=, or to the default processor.
969 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
971 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
972 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
973 instead of rs6000_cpu_attr.
974 (rs6000_adjust_cost): Ditto.
975 (is_microcoded_insn): Ditto.
976 (rs6000_adjust_priority): Ditto.
977 (rs6000_issue_rate): Ditto.
978 (rs6000_use_sched_lookahead): Ditto.
979 (rs6000_use_sched_lookahead_guard): Ditto.
980 (rs6000_sched_reorder): Ditto.
981 (force_new_group): Ditto.
982 * config/rs6000/rs6000.md (cpu attribute): Ditto.
983 (group_ending_nop): Ditto.
985 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
987 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
988 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
989 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
990 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
992 (rs6000_reassociation_width): Ditto.
993 (rs6000_emit_epilogue): Ditto.
994 (rs6000_adjust_cost): Ditto.
995 (is_microcoded_insn): Ditto.
996 (is_cracked_insn): Ditto.
997 (rs6000_adjust_priority): Ditto.
998 (rs6000_sched_reorder): Ditto.
999 (rs6000_sched_reorder2): Ditto.
1000 (insn_must_be_first_in_group): Ditto.
1001 (insn_must_be_last_in_group): Ditto.
1002 (rs6000_register_move_cost): Ditto.
1003 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1006 2017-12-07 Julia Koval <julia.koval@intel.com>
1008 * config.gcc: Add vaesintrin.h.
1009 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1011 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1012 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1014 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1015 * config/i386/immintrin.h: Include vaesintrin.h.
1016 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1017 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1018 _mm_aesdec_epi128): New intrinsics.
1020 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1022 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1023 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1024 c-family/c-spellcheck.cc.
1025 (best_macro_match::best_macro_match): Likewise.
1027 (struct edit_distance_traits<cpp_hashnode *>): Move to
1028 c-family/c-spellcheck.h.
1029 (class best_macro_match): Likewise.
1031 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1033 PR tree-optimization/83293
1034 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1035 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1036 might insert into empty bb.
1039 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1040 simplify for plus with :c added, and pointer_plus without that.
1041 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1042 with undefined overflow and the conversion is not widening,
1043 perform negation in utype and only convert to type afterwards.
1044 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1045 simplify for plus with :c added, and pointer_plus without that.
1046 If type is integral with undefined overflow and the conversion is
1047 not widening, perform minus in utype and only convert to type
1048 afterwards. Move the last pointer_diff_expr simplify into the
1051 2017-12-06 Martin Sebor <msebor@redhat.com>
1053 PR tree-optimization/82646
1054 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1055 strncpy, not maxlen.
1057 2017-12-06 Martin Sebor <msebor@redhat.com>
1059 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1062 PR tree-optimization/83075
1063 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1064 strncat/strncpy don't change length of source string.
1066 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1069 2017-11-29 Martin Aberg <maberg@gaisler.com>
1071 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1072 to prevent b2bst errata sequence.
1073 (sqrtdf2_fix): Likewise.
1075 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1077 PR tree-optimization/81945
1078 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1079 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1080 to dest_cfun has orig_loop_num set, either remap it to the new
1081 loop number if the loop got moved too, or clear it.
1083 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1085 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1087 (thunderx2t99_nothing): New insn reservation.
1088 (thunderx2t99_mrs): New insn reservation.
1089 (thunderx2t99_multiple): New insn reservation.
1090 (thunderx2t99_alu_basi): Add bfx to reservation.
1091 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1093 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1096 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1098 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1100 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1101 tree-ssa-pre.c::remove_dead_inserted_code.
1102 * tree-ssa-dce.h: New file.
1103 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1104 (remove_dead_inserted_code): Move and rename to function
1105 tree-ssa-dce.c::simple_dce_from_worklist.
1106 (pass_pre::execute): Update use.
1108 2017-12-05 Richard Biener <rguenther@suse.de>
1110 PR tree-optimization/83277
1111 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1112 to code-gen liveout vars.
1114 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1116 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1117 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1118 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1121 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1123 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1125 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1126 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1129 2017-12-05 Richard Biener <rguenther@suse.de>
1131 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1132 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1133 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1134 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1136 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1138 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1140 2017-12-05 Martin Liska <mliska@suse.cz>
1141 Jakub Jelinek <jakub@redhat.com>
1143 * doc/invoke.texi: Document the options.
1144 * flag-types.h (enum sanitize_code): Add
1145 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1146 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1147 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1148 * opts.c: Define new sanitizer options.
1149 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1150 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1152 2017-12-05 Julia Koval <julia.koval@intel.com>
1154 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1155 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1156 (ix86_handle_option): Handle -mavx512vnni.
1157 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1158 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1159 * config/i386/i386-c (__AVX512VNNI__): New.
1160 * config/i386/i386.c (ix86_target_string): Handle new option.
1161 (ix86_valid_target_attribute_inner_p): Handle new option.
1162 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1163 * config/i386/i386.opt (mavx512vnni): New option.
1165 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1168 * athlon.md: Disable for generic.
1169 * haswell.md: Enable for generic.
1170 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1171 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1173 (ix86_adjust_cost): Move generic to haswell path.
1175 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1177 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1178 instead of MEM_P in a couple more places. Fix formatting issues.
1180 2017-12-04 Jim Wilson <jimw@sifive.com>
1182 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1183 instead of GP_REG_LAST-1.
1184 (riscv_adjust_libcall_cfi_prologue): Likewise.
1185 (riscv_adjust_libcall_cri_epilogue): Likewise.
1186 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1189 2017-12-04 Luis Machado <luis.machado@linaro.org>
1191 * ipa-pure-const.c (check_decl): Add missing newline.
1192 (state_from_flags): Likewise.
1194 2017-12-04 Jeff Law <law@redhat.com>
1196 PR tree-optimizatin/78496
1197 * gimple-ssa-evrp-analyze.h
1198 (evrp_range_analyzer::get_vr_values): Simplify.
1199 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1200 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1201 and gimple-ssa-evrp-analyze.h.
1202 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1203 (simplify_stmt_for_jump_threading): Copy a blob of code from
1204 tree-vrp.c to use ranges to simplify statements.
1205 (dom_opt_dom_walker::before_dom_children): Call
1206 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1207 (dom_opt_dom_walker::after_dom_children): Similarly for
1208 evrp_range_analyzer::leave.
1209 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1212 * gimple-ssa-evrp-analyze.c
1213 (evrp_range_analyzer::extract_range_from_stmt): Always use
1214 vr_values::update_value_range so preexisting range info is
1215 medged with new range info, even if the new range is VR_VARYING.
1217 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1219 * combine.c: Adjust comment.
1220 (use_crosses_set_p): Delete.
1221 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1222 (try_combine): Ditto.
1224 2017-12-04 Richard Biener <rguenther@suse.de>
1226 PR tree-optimization/83255
1227 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1228 Re-add zero-iteration check.
1230 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1232 PR rtl-optimization/83245
1233 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1234 hard registers as earlyclobber, also if not in an asm.
1236 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1240 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1243 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1244 rs6000_cpu based on cpu_index, not tune_index.
1246 2017-12-04 Richard Biener <rguenther@suse.de>
1248 PR tree-optimization/83238
1249 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1250 code match comment, rejecting invalid SESE regions.
1252 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1254 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1255 require base operand is a REG_POINTER prior to reload on targets
1256 with non-equivalent space registers.
1258 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1260 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1261 (update_specialized_profile): Fix updating of counts.
1262 (perhaps_add_new_callers): Likewise.
1264 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1267 * x86-tune.def: Remove obsolette FIXMEs.
1268 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1269 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1270 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1272 (X86_TUNE_PAD_RETURNS): Disable for generic.
1274 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1276 PR tree-optimization/83170
1277 PR tree-optimization/83241
1278 * gimple-ssa-store-merging.c
1279 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1280 gimple_vuse (ins_stmt) in case it has changed.
1281 (imm_store_chain_info::output_merged_store): Likewise.
1283 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1287 * tree-cfg.c (pass_warn_function_return::execute): Handle
1288 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1289 with BUILTINS_LOCATION.
1293 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1294 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1295 instead of DECL_MODE.
1297 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1298 last argument to rtx pointer.
1299 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1300 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1301 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1302 depending on the chosen ISAs.
1303 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1304 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1305 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1307 * config/i386/sse.md (mov<mode>_internal): Likewise.
1308 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1310 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1312 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1314 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1316 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1317 parameters from prototype.
1318 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1319 parameters. Don't print a comment.
1320 (emit_fusion_gpr_load): Adjust.
1321 (emit_fusion_load_store): Adjust.
1322 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1323 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1324 comment on the second line.
1326 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1329 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1330 rs6000_cpu based on cpu_index, not tune_index.
1332 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1334 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1335 which_alternative instead of which_alternative + 1.
1336 (output_asm_insn): Print an extra tab if the template is short.
1338 2017-12-01 Jim Wilson <jimw@sifive.com>
1340 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1342 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1343 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1345 * doc/tm.texi: Regenerate.
1347 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1350 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1351 whether we can allocate pseudos before trying to fix an address.
1352 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1353 memory address is indexed or indirect.
1354 (floatuns_<mode>si2_hw2): Likewise.
1356 2017-12-01 Jason Merrill <jason@redhat.com>
1358 * Makefile.in (TAGS): Add c-family/*.cc.
1360 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1362 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1363 (call_value_insn): Likewise.
1364 (sibcall_insn): Likewise.
1365 (sibcall_value_insn): Likewise.
1366 (movsi_aarch64): Likewise.
1367 (movdi_aarch64): Likewise.
1368 (add_losym_): Likewise.
1369 (ldr_got_small_): Likewise.
1370 (ldr_got_small_sidi): Likewise.
1371 (ldr_got_small_28k_): Likewise.
1372 (ldr_got_small_28k_sidi): Likewise.
1373 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1374 Move output_addr_const to symbolic case. Add error check.
1376 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1378 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1379 * config/s390/s390.md: Disable MVC merging peephole if it would
1380 disable operand forwarding.
1381 (new peephole2): Split MVCs if it would turn them into up to 2
1384 2017-12-01 Richard Biener <rguenther@suse.de>
1386 PR tree-optimization/83232
1387 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1388 detection of same access. Instead of breaking the group here
1389 do not consider the duplicate. Add comment explaining real fix.
1391 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1393 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1395 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1397 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1399 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1401 * function.h (struct function): Remove cilk_frame_decl,
1402 is_cilk_function and calls_cilk_spawn fields.
1403 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1405 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1406 cilk_elemental field.
1407 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1408 * target.def: Adjust comment.
1409 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1410 Don't test cilk_elemental.
1412 PR tree-optimization/83233
1413 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1414 bswap_stat name for the struct.
1417 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1418 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1419 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1420 SWITCH_BREAK_LABEL_P set on the label.
1421 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1422 added for default case if it was missing and not all cases covered.
1423 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1424 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1426 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1427 to avoid -Wimplicit-fallthrough warning.
1428 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1429 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1431 PR tree-optimization/83221
1432 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1434 (init_reassoc): Formatting fix.
1437 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1438 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1442 * tree-cfg.c: Include asan.h.
1443 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1445 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1447 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1449 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1451 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1452 vpcmpeqd instruction.
1454 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1456 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1457 registers type for 128bit mode.
1459 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1461 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1463 * gengtype-state.c (read_state_pair): Likewise.
1464 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1465 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1466 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1467 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1468 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1469 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1470 * builtins.c (fold_builtin_3): Likewise.
1471 * graphite-scop-detection.c
1472 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1473 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1476 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1479 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1480 parameter. Return true if ignore_decl_rtl_set_p is true and other
1481 conditions are satisfied.
1482 * asan.h (asan_protect_global): Add new parameter.
1483 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1484 to asan_protect_global calls.
1486 2017-11-30 Jim Wilson <jimw@sifive.com>
1488 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1489 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1490 -mstrict-align, add info on default value. Delete redundant lines for
1491 -mabi. Add missing -mexplicit-relocs docs.
1493 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1495 * config/arc/arc.md (trap): New pattern.
1497 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1499 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1500 instruction to end into a delay slot.
1501 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1502 can be placed into a delay slot against reg_note.
1504 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1506 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1507 labels number of usages.
1509 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1511 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1513 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1515 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1518 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1519 multiplication by power of 2 constant into two shifts + comparison.
1521 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1524 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1526 2017-11-30 Richard Biener <rguenther@suse.de>
1528 PR tree-optimization/83202
1529 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1530 allow_peel argument and guard peeling.
1531 (canonicalize_loop_induction_variables): Likewise.
1532 (canonicalize_induction_variables): Pass false.
1533 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1534 peeling from cunrolli.
1536 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1538 * combine.c (try_combine): Print a message to dump file whenever
1539 I0, I1, or I2 cannot be combined into I3.
1541 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1543 PR rtl-optimization/83156
1544 PR rtl-optimization/82621
1545 * combine.c (try_combine): Don't split an I2 if one of the dests is
1546 set again before I3. Allow unused dests.
1548 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1550 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1552 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1554 PR rtl-optimization/80818
1555 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1556 recursively. Use insn code for clobber.
1557 (lra_set_insn_recog_data): Pass the new arg to
1558 collect_non_operand_hard_regs.
1559 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1561 (lra_update_insn_regno_info): Pass insn to
1562 add_regs_to_insn_regno_info.
1564 2017-11-29 Jim Wilson <jimw@sifive.com>
1565 Andrew Waterman <andrew@sifive.com>
1567 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1568 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1569 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1570 use SINGLE_SHIFT_COST.
1571 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1572 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1574 2017-11-29 Julia Koval <julia.koval@intel.com>
1576 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1577 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1578 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1579 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1580 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1581 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1582 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1583 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1584 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1585 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1586 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1587 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1588 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1589 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1590 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1591 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1592 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1593 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1594 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1595 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1596 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1597 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1598 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1599 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1600 __builtin_ia32_vpshldv_v2di_mask,
1601 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1602 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1603 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1605 2017-11-29 Julia Koval <julia.koval@intel.com>
1607 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1608 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1609 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1610 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1611 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1612 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1613 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1614 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1615 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1616 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1617 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1618 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1619 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1620 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1621 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1622 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1623 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1624 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1625 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1626 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1627 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1629 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1631 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1632 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1635 2017-11-29 Martin Aberg <maberg@gaisler.com>
1637 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1638 to prevent b2bst errata sequence.
1639 (sqrtdf2_fix): Likewise.
1641 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1643 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1644 (div_sqrt_insn_p): New function.
1645 (sparc_do_work_around_errata): Insert NOP instructions to
1646 prevent sequences that could trigger the TN-0013 errata for
1647 certain LEON3 processors.
1648 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1649 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1650 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1651 (in_branch_delay): Prevent div and sqrt in delay slot if
1653 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1655 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1657 * config/sparc/sparc.c (atomic_insn_p): New function.
1658 (sparc_do_work_around_errata): Insert NOP instructions to
1659 prevent sequences that could trigger the TN-0010 errata for
1661 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1662 instruction referable in atomic_insns_p.
1664 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1666 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1667 (atomic_compare_and_swap_leon3_1): Likewise.
1670 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1672 * config/sparc/sparc.c (fpop_insn_p): New function.
1673 (sparc_do_work_around_errata): Insert NOP instructions to
1674 prevent sequences that could trigger the TN-0012 errata for
1676 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1677 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1678 (in_branch_annul_delay): Prevent floating-point instructions
1679 in delay slot of annulled integer branch.
1681 2017-11-29 Richard Biener <rguenther@suse.de>
1683 PR tree-optimization/83202
1684 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1686 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1687 nodes vectorized to the same stmts multiple times.
1688 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1689 (vect_analyze_slp_instance): Adjust.
1690 (scalar_stmts_to_slp_tree_map_t): New typedef.
1691 (vect_schedule_slp_instance): Add a map recording the SLP node
1692 representing the vectorized stmts for a set of scalar stmts.
1693 Avoid code-generating redundancies.
1694 (vect_schedule_slp): Allocate map and pass it down.
1696 2017-11-29 Nathan Sidwell <nathan@acm.org>
1699 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1702 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1704 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1705 cases for printing LDP/STP memory addresses.
1706 (aarch64_print_address_internal): Renamed from
1707 aarch64_print_operand_address, added parameter, add Pmode check.
1708 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1709 (aarch64_print_operand_address): Indirect to
1710 aarch64_print_address_internal.
1711 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1712 'y' operand output specifier.
1714 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1717 * tree.c (build_simple_mem_ref_loc): Handle
1718 get_addr_base_and_unit_offset returning a MEM_REF.
1721 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1724 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1727 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1729 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1731 (layout::move_to_column): Likewise.
1733 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1736 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1737 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1739 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1740 Martin Jambor <mjambor@suse.cz>
1743 * tree.h (expr_type_first_operand_type_p): Declare
1744 * tree.c (expr_type_first_operand_type_p): New function.
1745 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1746 (ipa_value_from_jfunc): Adjust declaration.
1747 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1748 Use it as result type for arithmetics, unless it is NULL in which case
1749 be more conservative.
1750 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1751 ipa_get_jf_pass_through_result.
1752 (propagate_vals_across_pass_through): Likewise.
1753 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1754 is to propagate_vals_across_pass_through.
1755 (propagate_constants_across_call): Pass PARM_TYPE to
1756 propagate_scalar_across_jump_function.
1757 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1758 ipa_value_from_jfunc.
1759 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1760 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1761 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1762 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1763 target_type, pass it to ipa_value_from_jfunc.
1764 (update_indirect_edges_after_inlining): Pass parameter type to
1765 try_make_edge_direct_simple_call.
1767 2017-11-28 Jeff Law <law@redhat.com>
1769 * gimple-ssa-evrp-analyze.c
1770 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1771 refine ranges if scev_initialized_p returns true.
1772 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1774 2017-11-28 Julia Koval <julia.koval@intel.com>
1776 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1777 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1778 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1779 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1780 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1781 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1782 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1783 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1784 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1785 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1786 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1787 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1788 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1789 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1790 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1791 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1792 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1793 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1794 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1795 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1796 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1797 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1799 2017-11-28 Julia Koval <julia.koval@intel.com>
1801 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1802 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1803 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1804 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1805 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1806 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1807 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1808 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1809 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1810 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1811 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1812 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1813 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1814 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1815 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1816 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1817 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1818 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1819 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1820 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1821 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1822 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1823 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1824 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1825 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1826 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1827 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1828 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1829 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1830 __builtin_ia32_vpshld_v2di_mask): New builtins.
1831 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1832 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1834 2017-11-28 Richard Biener <rguenther@suse.de>
1836 PR tree-optimization/80776
1837 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1839 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1841 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1842 If the incoming edge is an effective fallthru because the other
1843 edge only reaches a __builtin_unreachable () then record ranges
1844 derived from the controlling condition in SSA info.
1845 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1846 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1848 2017-11-28 Olivier Hainque <hainque@adacore.com>
1850 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1851 on mingw build hosts.
1853 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1856 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1858 2017-11-28 Richard Biener <rguenther@suse.de>
1861 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1862 copies generated from memcpy use a character array as reference
1865 2017-11-28 Julia Koval <julia.koval@intel.com>
1866 Sebastian Peryt <sebastian.peryt@intel.com>
1868 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1869 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1870 cilk-common.o, cilk.h, cilk-common.c): Remove.
1872 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1873 * builtins.c (is_builtin_name): Remove cilkplus condition.
1874 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1875 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1876 cilk-builtins.def, cilkplus.def): Remove.
1877 * cif-code.def (CILK_SPAWN): Remove.
1878 * cilk-builtins.def: Delete.
1879 * cilk-common.c: Ditto.
1881 * cilkplus.def: Ditto.
1882 * config/darwin.h (fcilkplus): Delete.
1883 * cppbuiltin.c: Ditto.
1884 * doc/extend.texi: Remove cilkplus doc.
1885 * doc/generic.texi: Ditto.
1886 * doc/invoke.texi: Ditto.
1887 * doc/passes.texi: Ditto.
1888 * gcc.c (fcilkplus): Remove.
1889 * gengtype.c (cilk.h): Remove.
1890 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1892 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1894 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1896 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1897 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1898 cilkplus conditions.
1899 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1900 inline_read_section): Ditto.
1901 * ipa-inline-analysis.c (cilk.h): Remove.
1902 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1903 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1904 append_linker_options): Remove condition for fcilkplus.
1905 * lto/lto-lang.c (cilk.h): Remove.
1906 (lto_init): Remove condition for fcilkplus.
1907 * omp-expand.c (expand_cilk_for_call): Delete.
1908 (expand_omp_taskreg, expand_omp_for_static_chunk,
1909 expand_omp_for): Remove cilkplus
1911 (expand_cilk_for): Delete.
1912 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1913 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1914 execute_lower_omp, diagnose_sb_0): Ditto.
1915 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1916 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1917 * tree-nested.c: Ditto.
1918 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1919 (dump_generic_node): Ditto.
1920 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1921 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
1923 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
1925 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1927 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
1928 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
1929 for vector mode and !TARGET_SIMD.
1931 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1933 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
1935 * tree.h (SWITCH_LABELS): Remove.
1936 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
1937 assert SWITCH_BODY is non-NULL.
1938 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
1940 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
1942 PR tree-optimization/80788
1943 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
1944 has TREE_OVERFLOW set, call drop_tree_overflow.
1946 2017-11-28 Richard Biener <rguenther@suse.de>
1948 PR tree-optimization/83158
1949 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
1951 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
1954 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
1955 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
1957 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
1959 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
1960 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
1962 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1965 * varasm.c (compare_constant): Take the mode of the constants into
1966 account when comparing floating point constants.
1968 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
1970 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
1971 from explicit instantiation of debug_helper.
1972 * vec.h (DEFINE_DEBUG_VEC): Ditto.
1974 2017-11-27 Richard Biener <rguenther@suse.de>
1976 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
1979 2017-11-27 Richard Biener <rguenther@suse.de>
1981 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
1982 (type_hash_canon): Also clear int_cst_hash_table entry for
1984 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
1986 2017-11-27 Tamar Christina <tamar.christina@arm.com>
1988 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1989 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1990 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1992 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1994 * hash-map.h (gt_cleare_cache): Avoid UB.
1996 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
1998 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
1999 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2000 (decide_unroll_runtime_iterations): Likewise.
2001 (decide_unroll_stupid): Likewise.
2003 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2006 * config/i386/i386.md: Add a loop with incssp.
2008 2017-11-27 Martin Jambor <mjambor@suse.cz>
2010 PR tree-optimization/81248
2011 * tree-sra.c (splice_param_accesses): Remove size check.
2012 (decide_one_param_reduction): Fix size check.
2013 * gimple-pretty-print.c (dump_profile): Silence warning.
2014 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2016 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2019 * dbxout.c (lastlineno): New variable.
2020 (dbx_debug_hooks): Use dbxout_switch_text_section as
2021 switch_text_section debug hook.
2022 (dbxout_function_end): Switch to current_function_section
2023 rather than function_section. If crtl->has_bb_partition,
2024 output just one N_FUN, depending on in_cold_section_p.
2025 (dbxout_source_line): Remember last lineno in lastlineno.
2026 (dbxout_switch_text_section): New function.
2027 (dbxout_function_decl): Adjust dbxout_block caller.
2028 (dbx_block_with_cold_children): New function.
2029 (dbxout_block): Return true if any LBRAC/RBRAC have been
2030 emitted. Use dbx_block_with_cold_children at depth == 0
2031 in second partition. Add PARENT_BLOCKNUM argument, pass
2032 it optionally adjusted to children. Output LBRAC/RBRAC
2033 around recursive call only if the block is in the current
2034 partition, if not and anything was output, emit empty
2036 * final.c (final_scan_insn): Compute cold_function_name
2037 before calling switch_text_section debug hook. Call
2038 that hook even if dwarf2out_do_frame if not emitting
2042 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2043 TREE_READONLY decls.
2045 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2047 PR rtl-optimization/82488
2048 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2050 2017-11-26 Julia Koval <julia.koval@intel.com>
2052 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2054 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2057 2017-11-26 Julia Koval <julia.koval@intel.com>
2059 * config/i386/driver-i386.c (host_detect_local_cpu):
2060 Detect skylake-avx512.
2062 2017-11-26 Julia Koval <julia.koval@intel.com>
2064 * config.gcc: Add -march=cannonlake.
2065 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2066 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2067 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2068 (PTA_CANNONLAKE): New.
2069 (processor_target_table): Add cannonlake.
2070 (ix86_option_override_internal): Ditto.
2071 (fold_builtin_cpu): Ditto.
2072 (get_builtin_code_for_version): Handle cannonlake.
2073 (M_INTEL_COREI7_CANNONLAKE): New.
2074 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2075 * doc/invoke.texi: Add -march=cannonlake.
2077 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2079 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2080 (try_init_one_plugin): Alternative implementation for MinGW.
2081 * Makefile.in (plugin_implib): New.
2082 (gengtype-lex.c): Fix broken AIX workaround.
2083 * configure: Regenerate.
2084 * doc/plugins.texi: Document support for MinGW.
2086 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2088 PR rtl-optimization/81553
2089 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2090 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2091 is a shift where C1 has different mode than the whole shift, use C1's
2092 mode for MULT rather than the shift's mode.
2095 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2096 builtins not enabled in the currently selected ISA.
2098 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2100 PR tree-optimization/71026
2101 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2102 (insert_reciprocals): Change to insert reciprocals before a division
2103 by a square and to insert the square of a reciprocal.
2104 (execute_cse_reciprocals_1): Change to consider division by a square.
2105 (register_division_in): Add importance parameter.
2107 2017-11-24 Richard Biener <rguenther@suse.de>
2109 PR tree-optimization/82402
2110 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2111 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2113 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2115 * match.pd (0-ptr): New transformation.
2117 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2120 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2121 when skipping checking.
2123 2017-11-24 Richard Biener <rguenther@suse.de>
2125 PR tree-optimization/83128
2126 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2127 (vn_reference_lookup_3): Likewise.
2129 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2132 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2133 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2136 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2137 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2138 check for the latter with gimple_call_builtin_p. Do not handle
2139 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2141 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2143 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2146 2017-11-23 Julia Koval <julia.koval@intel.com>
2148 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2149 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2150 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2151 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2152 _mm512_maskz_expandloadu_epi16): New intrinsics.
2153 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2154 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2155 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2156 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2157 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2158 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2159 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2160 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2161 _mm256_maskz_expandloadu_epi8): New intrinsics.
2162 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2163 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2164 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2165 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2166 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2167 * config/i386/sse.md (VI248_VLBW): New iterator.
2168 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2170 2017-11-23 Julia Koval <julia.koval@intel.com>
2172 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2173 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2174 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2175 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2176 _mm512_mask_compressstoreu_epi16): New.
2177 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2178 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2179 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2180 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2181 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2182 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2183 _mm256_mask_compressstoreu_epi8): New.
2184 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2185 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2186 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2187 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2188 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2189 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2190 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2191 __builtin_ia32_compresshi128_mask,
2192 __builtin_ia32_compressstoreuqi512_mask,
2193 __builtin_ia32_compressstoreuhi512_mask,
2194 __builtin_ia32_compressstoreuqi256_mask,
2195 __builtin_ia32_compressstoreuqi128_mask,
2196 __builtin_ia32_compressstoreuhi256_mask,
2197 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2198 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2200 (ix86_expand_special_args_builtin): Handle new types.
2201 (s4fma_expand): Handle new builtin array.
2202 * config/i386/immintrin.h: Include new headers.
2203 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2204 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2206 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2208 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2209 fixes. Declare temp and g variables at the top in order to avoid
2210 {} in most of the cases.
2212 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2214 * match.pd (ptr-0): New transformation.
2216 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2218 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2219 (struct addr_mode_cost_table): New.
2220 (struct tune_params): Add field addr_mode_costs.
2221 * config/arm/arm.c (generic_addr_mode_costs): New.
2222 (arm_slowmul_tune): Initialise addr_mode_costs field.
2223 (arm_fastmul_tune): Likewise.
2224 (arm_strongarm_tune): Likewise.
2225 (arm_xscale_tune): Likewise.
2226 (arm_9e_tune): Likewise.
2227 (arm_marvell_pj4_tune): Likewise.
2228 (arm_v6t2_tune): Likewise.
2229 (arm_cortex_tune): Likewise.
2230 (arm_cortex_a8_tune): Likewise.
2231 (arm_cortex_a7_tune): Likewise.
2232 (arm_cortex_a15_tune): Likewise.
2233 (arm_cortex_a35_tune): Likewise.
2234 (arm_cortex_a53_tune): Likewise.
2235 (arm_cortex_a57_tune): Likewise.
2236 (arm_exynosm1_tune): Likewise.
2237 (arm_xgene1_tune): Likewise.
2238 (arm_cortex_a5_tune): Likewise.
2239 (arm_cortex_a9_tune): Likewise.
2240 (arm_cortex_a12_tune): Likewise.
2241 (arm_cortex_a73_tune): Likewise.
2242 (arm_v7m_tune): Likewise.
2243 (arm_cortex_m7_tune): Likewise.
2244 (arm_v6m_tune): Likewise.
2245 (arm_fa726te_tune): Likewise.
2246 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2248 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2250 * config/arm/arm.c (arm_mem_costs): New function.
2251 (arm_rtx_costs_internal): Use arm_mem_costs.
2253 2017-11-23 Mark Wielaard <mark@klomp.org>
2255 * dwarf2out.c (init_sections_and_labels): Use generation to create
2256 unique ranges_section_label and ranges_base_label. Return generation.
2257 (output_rnglists): Add generation argument. Use generation to create
2258 unique ranges labels.
2259 (dwarf2out_finish): Get generation from init_sections_and_labels
2260 and pass generation to output_rnglists.
2262 2017-11-23 Mike Stump <mikestump@comcast.net>
2263 Eric Botcazou <ebotcazou@adacore.com>
2265 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2266 * cfgloop.h (struct loop): Add unroll field.
2267 * function.h (struct function): Add has_unroll bitfield.
2268 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2269 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2270 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2271 (pass_rtl_unroll_loops::gate): Likewise.
2272 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2273 for which loop->unroll==1.
2274 (decide_unroll_constant_iterations): Use note for consistency and
2275 take loop->unroll into account. Return early if loop->unroll is set.
2276 Fix thinko in existing test.
2277 (decide_unroll_runtime_iterations): Use note for consistency and
2278 take loop->unroll into account.
2279 (decide_unroll_stupid): Likewise.
2280 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2281 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2282 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2284 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2285 (print_loop): Print loop->unroll if set.
2286 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2287 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2288 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2290 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2291 loop->unroll is set and smaller than the trip count. Otherwise bypass
2292 entirely the heuristics if loop->unroll is set. Remove dead note.
2293 Fix off-by-one bug in other note.
2294 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2295 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2297 (tree_unroll_loops_completely): Make static.
2298 (pass_complete_unroll::execute): Use correct type for variable.
2299 (pass_complete_unrolli::execute): Fix formatting.
2300 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2302 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2304 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2305 enable when TARGET_PREFER_AVX128 is set.
2307 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2309 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2311 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2312 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2314 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2316 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2317 frequencies. Fix estimation of aggregate parameters.
2319 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2321 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2324 2017-11-23 Tom de Vries <tom@codesourcery.com>
2326 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2328 2017-11-23 Tom de Vries <tom@codesourcery.com>
2330 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2333 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2336 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2337 sibcall_value_pcrel_fdpic): Use local variable instead of
2339 (calli_tbr_rel): Add missing operand 2.
2340 (call_valuei_tbr_rel): Add missing operand 3.
2342 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2345 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2346 bitpos/bitsize covering the whole destination, use store_expr only if
2347 the complex mode is the same. Otherwise, use expand_normal and if
2348 it returns CONCAT, subreg each part separately instead of trying to
2349 subreg the whole result.
2351 2017-11-23 Richard Biener <rguenther@suse.de>
2353 PR tree-optimization/23094
2354 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2355 come here from walking over backedges in the first iteration.
2356 (vn_reference_lookup_3): Skip clobbers that store the same value.
2358 2017-11-23 Richard Biener <rguenther@suse.de>
2360 PR tree-optimization/81403
2361 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2362 a block we need a leader relative to.
2363 (phi_translate_1): For nary processing require a leader from
2364 get_representative_for given we run expression simplification
2365 using match-and-simplify. Remove previous fix.
2367 2017-11-22 Jeff Law <law@redhat.com>
2369 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2370 Use new method allocate_value_range rather than accessing the
2371 vrp_value_range_pool data member directly.
2372 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2373 to use extract_range_from_stmt method to avoid need for
2374 extract_range_from_assignment method.
2375 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2376 method rather than setting values_propgated data member directly.
2377 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2378 and values propagated data members and extract_range_from_assignment
2379 method. Reorder private data members to conform to standards.
2380 Add new methods set_lattice_propagation_complete and
2381 allocate_value_range.
2383 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2385 PR rtl-optimization/83030
2386 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2387 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2388 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2389 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2390 CROSSING_JUMP_P flag is already set before setting it.
2391 * cfgrtl.c (fixup_partition_crossing): Likewise.
2392 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2395 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2397 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2398 Handle the case where both arguments are using gen_const_vec_series.
2400 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2403 * pretty-print.c (pp_format): Move quoting implementation to
2404 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2405 to pass address of "quote" local.
2406 (pp_begin_quote): New function.
2407 (pp_end_quote): New function.
2408 * pretty-print.h (printer_fn): Convert penultimate param from bool
2410 (pp_begin_quote): New decl.
2411 (pp_end_quote): New decl.
2412 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2413 param from bool to bool *.
2414 * tree-diagnostic.h (default_tree_printer): Likewise.
2416 2017-11-22 Jeff Law <law@redhat.com>
2418 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2419 Initialize vr_values.
2420 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2421 vr_values via vr_values class instance rather than delegators.
2422 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2423 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2424 (evrp_range_analyzer::push_value_range): Likewise.
2425 (evrp_range_analyzer::pop_value_range): Likewise.
2426 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2427 most delegators. Those remaining are exposed as public interfaces.
2428 Make vr_values a pointer and private.
2429 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2431 (evrp_range_analyzer::get_vr_value): New method.
2432 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2433 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2434 (class evrp_dom_walker): Attach evrp_folder class, initialize
2435 it in the ctor. Remove temporary delegators.
2436 (evrp_dom_walker::before_dom_children): Call methods in attached
2437 evrp_range_analyzer class via class instance pointer. Use
2438 free value_range_constant_singleton to remove need for
2439 op_with_constant_singleton_value delegator method. Do not
2440 create a vrp_prop class instance for every call! Narrow
2441 scope of a couple variables.
2442 (evrp_dom_walker::cleanup): Call methods in attached
2443 evrp_range_analyzer class via class instance pointer.
2444 * vr-values.h (class vr_values): Privatize many methods and
2447 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2449 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2450 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2451 * expr.c (expand_expr_real_2): Likewise.
2452 * fold-const.c (const_unop): Likewise.
2453 * optabs-tree.c (optab_for_tree_code): Likewise.
2454 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2455 * tree-inline.c (estimate_operator_cost): Likewise.
2456 * tree-pretty-print.c (dump_generic_node): Likewise.
2457 (op_code_prio): Likewise.
2458 (op_symbol_code): Likewise.
2459 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2460 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2461 * internal-fn.c (direct_internal_fn_optab): New function.
2462 (direct_internal_fn_array, direct_internal_fn_supported_p
2463 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2464 * fold-const-call.c (fold_const_reduction): New function.
2465 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2467 * tree-vect-loop.c: Include internal-fn.h.
2468 (reduction_code_for_scalar_code): Rename to...
2469 (reduction_fn_for_scalar_code): ...this and return an internal
2471 (vect_model_reduction_cost): Take an internal_fn rather than
2473 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2475 (vectorizable_reduction): Use internal functions rather than tree
2476 codes for the reduction operation. Update calls to the functions
2478 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2479 Use calls to internal functions rather than REDUC tree codes.
2480 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2482 2017-11-22 Olivier Hainque <hainque@adacore.com>
2484 * config/vxworks.c (vxworks_override_options): Pick default
2485 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2486 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2487 DWARF_GNAT_ENCODINGS_DEFAULT.
2488 * config/vxworksae.h: Likewise.
2490 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2492 PR tree-optimization/83104
2493 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2496 2017-11-22 Marek Polacek <polacek@redhat.com>
2497 H.J. Lu <hongjiu.lu@intel.com>
2498 Jason Merrill <jason@redhat.com>
2503 * calls.c (initialize_argument_information): Call
2504 warn_parameter_passing_abi target hook.
2505 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2506 argument onto stack.
2507 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2508 * common.opt: Update -fabi-version description.
2509 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2510 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2512 (ix86_is_empty_record): New function.
2513 (ix86_warn_parameter_passing_abi): New function.
2514 (TARGET_EMPTY_RECORD_P): Redefine.
2515 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2516 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2517 * doc/tm.texi: Regenerated.
2518 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2519 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2520 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2521 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2522 instead of int_size_in_bytes.
2523 * expr.c (copy_blkmode_to_reg): Likewise.
2524 * function.c (aggregate_value_p): Return 0 for empty types.
2525 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2527 (locate_and_pad_parm): Call arg size_in_bytes instead
2529 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2530 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2531 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2533 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2534 (std_gimplify_va_arg_expr): Skip empty records. Call
2535 arg_size_in_bytes instead size_in_bytes.
2536 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2537 * tree-core.h (tree_type_common): Add empty_flag.
2538 (tree_decl_common): Update comments.
2539 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2541 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2542 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2544 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2545 * tree.c (default_is_empty_type): New function.
2546 (default_is_empty_record): New function.
2547 (arg_int_size_in_bytes): New function.
2548 (arg_size_in_bytes): New function.
2549 (get_ultimate_context): New function.
2550 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2551 TRANSLATION_UNIT_WARN_EMPTY_P.
2552 (default_is_empty_record, arg_int_size_in_bytes,
2553 arg_size_in_bytes, get_ultimate_context): Declare.
2555 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2557 * config/arm/arm.c (cmse_clear_registers): New function.
2558 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2559 code by call to cmse_clear_registers.
2560 (cmse_nonsecure_entry_clear_before_return): Likewise.
2562 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2564 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2565 (vdot_s32, vdotq_s32): New.
2566 (vdot_lane_u32, vdotq_lane_u32): New.
2567 (vdot_lane_s32, vdotq_lane_s32): New.
2570 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2573 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2574 for unsigned values with fewer HWIs than the precision.
2575 (test_overflow): New function.
2576 (wide_int_cc_tests): Call it.
2578 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2580 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2581 has the same mode class as Pmode.
2583 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2585 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2586 padding_bits_to_clear_ptr.
2587 (cmse_nonsecure_entry_clear_before_return): Likewise.
2589 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2591 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2592 auto_sbitap instead of integer bitfield to control register needing
2595 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2597 PR tree-optimization/83044
2598 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2599 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2600 that need the upper bound. Subtract offset from
2601 get_addr_base_and_unit_offset only if positive and subtract it
2602 before division by eltsize rather than after it.
2605 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2606 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2607 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2611 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2612 expand_binop with *mul_widen_optab, make sure at least one of the
2613 operands doesn't have VOIDmode.
2616 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2618 PR rtl-optimization/82044
2619 PR tree-optimization/82042
2620 * dse.c (record_store): Check for overflow.
2621 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2622 add_wild_read instead of clear_rhs_from_active_local_stores on
2623 overflow and log it into dump_file.
2625 2017-11-22 Richard Biener <rguenther@suse.de>
2627 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2628 fake edges to exit when looking for a place to insert.
2629 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2631 (insert_into_preds_of_block): Commit edge insertion immediately,
2632 assert that doesn't require new BBs.
2633 (fini_pre): Release expressions.
2634 (pass_pre::execute): Shuffle things around a bit, if the fn
2635 is too large do not compute AVAIL either as this is really the
2638 2017-11-22 Richard Biener <rguenther@suse.de>
2640 PR tree-optimization/83089
2641 * tree-if-conv.c (pass_if_conversion::execute): If anything
2642 changed reset SCEV and free the number of iteration estimates.
2644 2017-11-21 Martin Sebor <msebor@redhat.com>
2646 PR tree-optimization/82945
2647 * calls.h (warn_nonstring_bound): Remove unused function.
2649 2017-11-21 Martin Sebor <msebor@redhat.com>
2651 PR tree-optimization/82945
2652 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2653 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2654 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2656 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2657 * calls.h (get_attr_nonstring_decl): Declare new function.
2658 * doc/extend.texi (attribute nonstring): Update.
2659 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2660 get_attr_nonstring_decl and handle it.
2661 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2662 detection of nul-termination.
2663 (strlen_to_stridx): Change to a pointer.
2664 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2665 (pass_strlen::execute): Same.
2667 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2669 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2670 for the new option -mprefer-vector-width=[none|128|256|512].
2671 * config/i386/i386.c (ix86_target_string): remove old style options
2672 -mprefer-avx256 and make -mprefer-avx128 as alias.
2673 (ix86_option_override_internal): Apply defaults for the
2674 -mprefer-vector-width=[128|256] option.
2675 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2676 Implement macros to work with -mprefer-vector-width=.
2677 * config/i386/i386.opt: Implemented option
2678 -mprefer-vector-width=[none|128|256|512].
2679 * doc/invoke.texi: Documentation for
2680 -mprefer-vector-width=[none|128|256|512].
2682 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2684 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2686 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2688 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2690 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2692 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2694 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2695 * expr.c (expand_expr_real_2): Likewise.
2696 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2697 fold_binary_loc): Likewise.
2698 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2699 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2700 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2701 MINUS_EXPR transformations.
2702 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2703 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2704 * tree-inline.c (estimate_operator_cost): Likewise.
2705 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2706 op_symbol_code): Likewise.
2707 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2708 * vr-values.c (extract_range_from_binary_expr): Likewise.
2709 * varasm.c (initializer_constant_valid_p_1): Likewise.
2710 * tree.def: New tree code POINTER_DIFF_EXPR.
2712 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2714 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2715 integer suffix to movbe mnemonic.
2716 (*bswaphi2_movbe): Ditto.
2717 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2719 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2722 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2723 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2724 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2726 2017-11-21 Martin Liska <mliska@suse.cz>
2728 * tree-inline.c (expand_call_inline): Remove not needed
2731 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2732 Jakub Jelinek <jakub@redhat.com>
2735 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2736 Remove static keyword from f variable.
2738 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2740 PR tree-optimization/83086
2741 * gimple-ssa-store-merging.c
2742 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2743 rather than n.base_addr.
2745 2017-11-21 Martin Liska <mliska@suse.cz>
2747 PR rtl-optimization/82044
2748 PR tree-optimization/82042
2749 * dse.c (check_mem_read_rtx): Check for overflow.
2751 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2753 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2756 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2758 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2759 and its corresponding call arg location note.
2761 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2762 Andrew Burgess <andrew.burgess@embecosm.com>
2764 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2766 (arc_return_slot_offset): Likewise.
2767 (arc_eh_return_address_location): New declaration.
2768 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2769 (MUST_SAVE_REGISTER): Add exception handler case.
2770 (MUST_SAVE_RETURN_ADDR): Likewise.
2771 (arc_frame_pointer_required): Likewise.
2772 (arc_frame_pointer_needed): New function.
2773 (arc_compute_frame_size): Changed.
2774 (arc_expand_prologue): Likewise.
2775 (arc_expand_epilogue): Likewise.
2776 (arc_initial_elimination_offset): Likewise.
2777 (arc_return_slot_offset): Delete.
2778 (arc_eh_return_address_location): New function.
2779 (arc_builtin_setjmp_frame_value): Likewise.
2780 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2781 (EH_RETURN_STACKADJ_RTX): Define.
2782 (EH_RETURN_HANDLER_RTX): Likewise.
2783 * config/arc/arc.md (eh_return): Delete.
2785 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2787 * print-tree.h (debug_vec_tree): Remove prototype.
2788 * gdbinit.in (pvt): Remove macro.
2790 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2792 PR tree-optimization/83047
2793 * gimple-ssa-store-merging.c
2794 (imm_store_chain_info::output_merged_store): If the loads with the
2795 same vuse are in different basic blocks, for load_gsi pick a load
2796 location that is dominated by the other loads.
2799 * config/i386/i386.c (ix86_memmodel_check): Start
2800 -Winvalid-memory-model diagnostics with lowercase letter.
2803 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2804 set in_cold_section_p to the partition containing loc_list->first.
2805 When seeing loc_list->last_before_switch node, update secname and
2806 perform range_across_switch second partition handling only after that.
2809 * run-rtl-passes.c: Include debug.h.
2810 (run_rtl_passes): Call debug_hooks->assembly_start.
2811 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2815 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2816 OPTAB_DIRECT in calls to expand_simple_binop.
2818 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2821 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2823 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2826 * spellcheck.h (best_match::blithely_get_best_candidate): New
2829 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2831 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2832 with lower case letter.
2834 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2836 * config/i386/i386.md (bswaphi2): New expander.
2837 (*bswaphi2_movbe): New insn pattern.
2838 (bswaphi -> rorhi pepehole2): New peephole pattern.
2840 2017-11-20 Jeff Law <law@redhat.com>
2842 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2843 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2844 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2845 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2846 gimple-ssa-evrp-analyze.h.
2848 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2849 set BB_VISITED here.
2850 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2852 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2854 * tree-predcom.c: Add general comment on Store-Store chains.
2855 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2857 (get_chain_last_ref_at): Rename into...
2858 (get_chain_last_write_at): ...this.
2859 (get_chain_last_write_before_load): New function.
2860 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2861 CT_STORE_STORE when write reference is added.
2862 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2863 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2864 (initialize_root_vars_store_elim_1): Ditto.
2865 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2866 definition is created.
2867 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2868 chain by replacing it with dominant stored value.
2870 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2872 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2874 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2876 * vr-values.c (extract_range_from_binary_expr): Use a full range
2879 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2881 * config/arm/arm.md (R4_REGNUM): Define constant.
2882 (nonsecure_call_internal): Remove r4 clobber.
2883 (nonsecure_call_value_internal): Likewise.
2884 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2885 clobber and resequence match_operands.
2886 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2887 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2888 (nonsecure_call_value_reg_thumb2): Likewise.
2890 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2892 PR tree-optimization/78821
2893 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2894 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2895 to the address of the base rather than the base itself.
2896 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2897 (find_bswap_or_nop_finalize): New function.
2898 (find_bswap_or_nop): Use it.
2899 (bswap_replace): Return a tree rather than bool, change first
2900 argument from gimple * to gimple_stmt_iterator, allow inserting
2901 into an empty sequence, allow ins_stmt to be NULL - then emit
2902 all stmts into gsi. Fix up MEM_REF address gimplification.
2903 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2904 (struct store_immediate_info): Add N and INS_STMT non-static
2906 (store_immediate_info::store_immediate_info): Initialize them
2907 from newly added ctor args.
2908 (merged_store_group::apply_stores): Formatting fixes. Sort by
2910 (stmts_may_clobber_ref_p): For stores call also
2911 refs_anti_dependent_p.
2912 (gather_bswap_load_refs): New function.
2913 (imm_store_chain_info::try_coalesce_bswap): New method.
2914 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2915 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2916 (imm_store_chain_info::output_merged_store): Fail if number of
2917 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2918 and NOP_EXPR rhs_code.
2919 (pass_store_merging::process_store): Compute n and ins_stmt, if
2920 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
2921 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
2923 (pass_store_merging::execute): Calculate dominators.
2925 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
2926 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
2927 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
2928 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2929 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
2930 class pass_optimize_bswap, bswap_replace,
2931 pass_optimize_bswap::execute): Moved to ...
2932 * gimple-ssa-store-merging.c: ... this file.
2933 Include optabs-tree.h.
2934 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
2935 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2936 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
2937 anonymous namespace, remove static keywords.
2938 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
2939 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
2941 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
2944 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
2946 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
2948 * vec.h (debug_helper): New function.
2949 (DEFINE_DEBUG_VEC): New macro.
2950 * hash-set.h (debug_helper): New function.
2951 (DEFINE_DEBUG_HASH_SET): New macro.
2952 * cfg.c (debug_slim (edge)): New function.
2953 Call DEFINE_DEBUG_VEC for edges.
2954 Call DEFINE_DEBUG_HASH_SET for edges.
2955 * cfghooks.c (debug_slim (basic_block)): New function.
2956 Call DEFINE_DEBUG_VEC for basic blocks.
2957 Call DEFINE_DEBUG_HASH_SET for basic blocks.
2958 * print-tree.c (debug_slim): New function to handle trees.
2959 Call DEFINE_DEBUG_VEC for trees.
2960 Call DEFINE_DEBUG_HASH_SET for trees.
2961 (debug (vec<tree, va_gc>) &): Remove.
2962 (debug (<vec<tree, va_gc>) *): Remove.
2963 * print-rtl.c (debug_slim): New function to handle const_rtx.
2964 Call DEFINE_DEBUG_VEC for rtx_def.
2965 Call DEFINE_DEBUG_VEC for rtx_insn.
2966 Call DEFINE_DEBUG_HASH_SET for rtx_def.
2967 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
2968 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
2969 (debug (vec<rtx_insn *> *ptr): Remove.
2970 (debug_insn_vector): Remove.
2971 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
2973 2017-11-20 Tom de Vries <tom@codesourcery.com>
2975 PR rtl-optimization/82020
2976 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
2977 IF_THEN_ELSE condition.
2979 2017-11-19 Jeff Law <law@redhat.com>
2981 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2982 of degenerates resulting from ignoring an edge.
2984 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2987 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
2989 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2992 * profile-count.c (profile_count::to_sreal_scale): Fix return value
2993 for uninitialied counts.
2995 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2998 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3000 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3003 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3005 2017-11-19 Tom de Vries <tom@codesourcery.com>
3007 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3010 2017-11-19 Tom de Vries <tom@codesourcery.com>
3012 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3015 2017-11-19 Tom de Vries <tom@codesourcery.com>
3017 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3020 2017-11-19 Tom de Vries <tom@codesourcery.com>
3022 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3024 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3026 2017-11-19 Tom de Vries <tom@codesourcery.com>
3028 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3030 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3031 semicolon after MCORE_EXPORT_NAME call.
3033 2017-11-19 Tom de Vries <tom@codesourcery.com>
3036 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3037 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3039 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3041 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3042 (cgraph_edge::clone): Cleanup updating of profile.
3043 * ipa-cp.c (update_profiling_info): Likewise.
3044 * ipa-inline-transform.c (inline_transform): Likewise.
3045 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3046 * ipa-split.c (execute_split_functions): Do not split when function
3048 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3049 * profile-count.c (profile_count::dump): Add global0.
3050 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3052 (profile_count::to_sreal_scale): Likewise.
3053 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3054 (profile_count::combine_with_ipa_count): New function.
3055 * profile-count.h (profile_guessed_global0adjusted): New.
3056 (profile_count::adjusted_zero): New.
3057 (profile_count::global0adjusted): New.
3058 (profile_count::combine_with_ipa_count): New.
3059 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3060 correct profile of return block of split functions.
3061 (copy_cfg_body): Remove unused profile_count.
3062 (copy_body): Likewise.
3063 (expand_call_inline): Update.
3064 (tree_function_versioning): Update.
3066 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3068 * hash-set.h (hash_set::empty): New.
3069 * tree-ssa-threadbackward.h: Delete.
3070 * tree-ssa-threadbackward.c (class thread_jumps): New.
3071 Move max_threaded_paths into class.
3072 (fsm_find_thread_path): Remove arguments that are now in class.
3073 (profitable_jump_thread_path): Rename to...
3074 (thread_jumps::profitable_jump_thread_path): ...this.
3075 (convert_and_register_jump_thread_path): Rename to...
3076 (thread_jumps::convert_and_register_current_path): ...this.
3077 (check_subpath_and_update_thread_path): Rename to...
3078 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3079 (register_jump_thread_path_if_profitable): Rename to...
3080 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3081 (handle_phi): Rename to...
3082 (thread_jumps::handle_phi): ...this.
3083 (handle_assignment): Rename to...
3084 (thread_jumps::handle_assignment): ...this.
3085 (fsm_find_control_statement_thread_paths): Rename to...
3086 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3087 (find_jump_threads_backwards): Rename to...
3088 (thread_jumps::find_jump_threads_backwards): ...this.
3089 Initialize path local data.
3090 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3091 from within thread_jumps class.
3092 (pass_early_thread_jumps::execute): Same.
3094 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3096 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3098 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3100 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3101 to canonical location.
3103 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3106 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3108 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3110 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3112 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3113 rcrt1.o%s/grcrt1.o%s for -static-pie.
3115 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3117 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3118 ix86_shift_rotate_cost): Break out from ...
3119 (ix86_rtx_costs): ... here.
3120 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3123 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3125 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3126 when entry block was promoted unlikely.
3127 (estimate_bb_frequencies): Increase frequency scale.
3128 * profile-count.h (profile_count): Export precision info.
3130 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3132 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3133 disturbin profile of entry block.
3135 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3137 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3140 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3142 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3143 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3145 2017-11-17 Jeff Law <law@redhat.com>
3147 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3148 from evrp_dom_walker class. Various methods moved into new class.
3149 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3150 (evrp_range_analyzer::enter): New method.
3151 (evrp_range_analyzer::leave): New method.
3152 (evrp_dom_walker): Remove delegators no longer needed by this class.
3153 Replace vr_values data member with evrp_range_analyzer
3155 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3156 method extracted from evrp_dom_walker::before_dom_children.
3157 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3158 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3160 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3161 Add private copy constructor and move assignment operators.
3162 Privatize methods and class data where trivially possible.
3163 (evrp_dom_walker::cleanup): New function, extracted from
3164 execute_early_vrp. Simplify access to class data.
3166 * vr-values.h (get_output_for_vrp): Prototype.
3167 * vr-values.c (get_output_for_vrp): New function extracted from
3168 vrp_visit_assignment_or_call and extract_range_from_stmt.
3169 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3171 2017-11-17 Luis Machado <luis.machado@linaro.org>
3173 * config/aarch64/aarch64.c
3174 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3175 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3176 tune_params::AUTOPREFETCHER_WEAK.
3178 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3181 * config/arm/arm.c (arm_valid_target_attribute_rec):
3182 Parse "arch=" and "+<ext>".
3183 (arm_valid_target_attribute_tree): Re-init global options.
3184 (arm_option_override): Make non-static.
3185 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3186 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3187 (__ARM_FEATURE_CRC32): Support undef.
3188 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3189 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3191 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3193 * gdbinit.in (break-on-diagnostic): New command.
3195 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3197 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3198 return to indirect jump for EH return if control-flow
3199 protection is enabled. Change explicit 'false' argument in
3200 pro_epilogue_adjust_stack with a value of
3202 * config/i386/i386.md (simple_return_indirect_internal):
3203 Remove SImode restriction to support 64-bit.
3205 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3207 * combine.c (added_notes_insn): New.
3208 (try_combine): Handle added_notes_insn like added_links_insn.
3209 Rewrite return value code.
3210 (distribute_notes): Set added_notes_insn to the earliest insn we added
3213 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3215 PR rtl-optimization/82621
3216 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3217 dest of one of those SETs is unused.
3219 2017-11-17 Richard Biener <rguenther@suse.de>
3222 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3223 * tree-pretty-print.c (dump_generic_node): Handle
3224 annot_expr_parallel_kind.
3225 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3226 * gimplify.c (gimple_boolify): Likewise.
3228 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3230 * config.gcc (extra_headers): Add cet.h for x86 targets.
3231 * config/i386/cet.h: New file.
3232 * doc/install.texi: Add --enable-cet/--disable-cet.
3234 2017-11-17 Richard Biener <rguenther@suse.de>
3236 PR tree-optimization/83017
3237 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3238 (gen_parallel_loop): Properly count iterations.
3239 (parallelize_loops): Handle loop->can_be_parallel independent
3240 of flag_loop_parallelize_all. Make static profitability test match
3242 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3243 * invoke.texi (parloops-min-per-thread): Document.
3245 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3247 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3248 upstreaming review comments.
3250 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3252 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3253 with fast unaligned access.
3254 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3256 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3258 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3259 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3261 2017-11-17 Richard Biener <rguenther@suse.de>
3263 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3264 folding of references.
3266 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3269 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3270 of replacing call to strncmp with corresponding call to strcmp when
3273 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3275 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3276 option prefer-avx256 for skylake-avx512 configuration.
3277 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3278 (get_builtin_code_for_version): Ditto.
3280 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3281 Monk Chiang <sh.chiang04@gmail.com>
3283 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3284 (FIXED_REGISTERS): Reserve more register numbers.
3285 (CALL_USED_REGISTERS): Likewise.
3286 (REG_ALLOC_ORDER): Likewise.
3287 (REG_CLASS_CONTENTS): Likewise.
3288 (REGISTER_NAMES): Likewise.
3290 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3291 Kito Cheng <kito.cheng@gmail.com>
3293 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3295 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3297 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3299 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3300 parameter list for vec_splats.
3302 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3304 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3306 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3308 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3311 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3312 with the required version of automake 1.11.6.
3314 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3316 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3318 (DU_C2_3_power9): Correct reservation combinations.
3319 (FP_DIV_power9, VEC_DIV_power9): New.
3320 (power9-alu): Split out rotate/shift...
3321 (power9-rot): ...to here, correct dispatch resource.
3322 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3324 (power9-fp): Correct latency.
3325 (power9-sdiv): Add div/sqrt resource.
3326 (power9-ddiv): Correct latency, add div/sqrt resource.
3327 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3328 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3330 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3332 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3334 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3335 switch statement mapping KF built-ins to TF built-ins if we don't
3336 have the proper ISA 3.0 assembler support.
3338 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3340 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3341 (lower_emutls_function_body): Do not compute it.
3343 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3345 * ipa-split.c (split_bb_info): Turn time to sreal.
3346 (split_point): Likewise.
3347 (dump_split_point): Likewise.
3348 (fine_split_points): Likewise.
3349 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3351 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3353 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3356 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3358 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3360 (expensive_function_p): Remove useless assert.
3361 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3363 2017-11-16 Martin Sebor <msebor@redhat.com>
3365 PR tree-optimization/82588
3366 PR tree-optimization/82583
3367 * tree-vrp.c (check_array_ref): Handle flexible array members,
3368 string literals, and inner indices.
3369 (search_for_addr_array): Add detail to diagnostics.
3371 2017-11-16 Nathan Sidwell <nathan@acm.org>
3375 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3376 (SET_DECL_ASSEMBLER_NAME): Forward to
3377 overwrite_decl_assembler_name.
3378 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3379 (overwrite_decl_assembler_name): Declare.
3380 * tree.c (overwrite_decl_assembler_name): New.
3381 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3382 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3383 (LANG_HOOKS_INITIALIZER): Add it.
3384 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3385 * langhooks.c (lhd_set_decl_assembler_name): Use
3386 SET_DECL_ASSEMBLER_NAME.
3387 (lhd_overwrite_decl_assembler_name): Default implementation.
3389 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3390 Jackson Woodruff <jackson.woodruff@arm.com>
3392 PR tree-optimization/71026
3393 * match.pd: Canonicalize constant multiplies in division.
3395 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3397 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3398 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3399 Remove OPT_fomit_frame_pointer entry.
3400 * common/config/arc/arc-common.c: Likewise.
3401 * common/config/arm/arm-common.c: Likewise.
3402 * common/config/avr/avr-common.c: Likewise.
3403 * common/config/c6x/c6x-common.c: Likewise.
3404 * common/config/cr16/cr16-common.c: Likewise.
3405 * common/config/cris/cris-common.c: Likewise.
3406 * common/config/epiphany/epiphany-common.c: Likewise.
3407 * common/config/fr30/fr30-common.c: Likewise.
3408 * common/config/frv/frv-common.c: Likewise.
3409 * common/config/ia64/ia64-common.c: Likewise.
3410 * common/config/iq2000/iq2000-common.c: Likewise.
3411 * common/config/lm32/lm32-common.c: Likewise.
3412 * common/config/m32r/m32r-common.c: Likewise.
3413 * common/config/mcore/mcore-common.c: Likewise.
3414 * common/config/microblaze/microblaze-common.c: Likewise.
3415 * common/config/mips/mips-common.c: Likewise.
3416 * common/config/mmix/mmix-common.c: Likewise.
3417 * common/config/mn10300/mn10300-common.c: Likewise.
3418 * common/config/nios2/nios2-common.c: Likewise.
3419 * common/config/pa/pa-common.c: Likewise.
3420 * common/config/pdp11/pdp11-common.c: Likewise.
3421 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3422 * common/config/riscv/riscv-common.c: Likewise.
3423 * common/config/rs6000/rs6000-common.c: Likewise.
3424 * common/config/rx/rx-common.c: Likewise.
3425 * common/config/s390/s390-common.c: Likewise.
3426 * common/config/sh/sh-common.c: Likewise.
3427 * common/config/sparc/sparc-common.c: Likewise.
3428 * common/config/tilegx/tilegx-common.c: Likewise.
3429 * common/config/tilepro/tilepro-common.c: Likewise.
3430 * common/config/v850/v850-common.c: Likewise.
3431 * common/config/visium/visium-common.c: Likewise.
3432 * common/config/xstormy16/xstormy16-common.c: Likewise.
3433 * common/config/xtensa/xtensa-common.c: Likewise.
3434 * invoke.texi (-fomit-frame-pointer): Update documentation.
3436 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3438 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3440 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3442 * cfg.c (scale_bbs_frequencies_int,
3443 cale_bbs_frequencies_gcov_type): Remove.
3444 * cfg.h (scale_bbs_frequencies_int,
3445 cale_bbs_frequencies_gcov_type): Remove.
3447 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3449 * tree-ssa-loop-manip.c
3450 (scale_dominated_blocks_in_loop): Update to profile counts.
3451 (tree_transform_and_unroll_loop): Likewise.
3453 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3455 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3456 scale_bbs_frequencies_int.
3458 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3460 * final.c (compute_alignments): Use counts rather than frequencies.
3462 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3464 * cfgloopanal.c: Include sreal.h
3465 (average_num_loop_insns): Use counts and sreal for accounting.
3467 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3469 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3472 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3473 Kito Cheng <kito.cheng@gmail.com>
3475 * config/nds32/constraints.md: Provide more constraints.
3476 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3477 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3478 support constraints usage.
3480 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3482 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3484 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3485 Kito Cheng <kito.cheng@gmail.com>
3487 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3488 * config/nds32/nds32.opt: Refine the layout.
3489 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3490 TARGET_EXT_STRING): Support new options.
3491 * config/nds32/nds32.h: Likewise.
3492 * config/nds32/nds32.md: Likewise.
3493 * config/nds32/nds32-predicates.c: Likewise.
3494 * config/nds32/constraints.md: Likewise.
3495 * common/config/nds32/nds32-common.c: Likewise.
3497 2017-11-16 Julia Koval <julia.koval@intel.com>
3500 * config/i386/gfniintrin.h: Add sse check.
3501 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3503 2017-11-16 Julia Koval <julia.koval@intel.com>
3505 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3506 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3507 (ix86_handle_option): Handle -mavx512vbmi2.
3508 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3509 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3510 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3511 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3512 (ix86_valid_target_attribute_inner_p): Ditto.
3513 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3514 * config/i386/i386.opt (mavx512vbmi2): New option.
3515 * doc/invoke.texi: Add new option.
3517 2017-11-16 Julia Koval <julia.koval@intel.com>
3519 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3520 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3521 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3522 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3523 _mm512_gf2p8mul_epi8): New intrinsics.
3524 * config/i386/i386-builtin-types.def
3525 (V64QI_FTYPE_V64QI_V64QI): New type.
3526 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3527 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3528 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3529 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3530 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3531 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3533 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3535 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3536 explicitly as a stream of bytes.
3538 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3540 * config/rs6000/altivec.h (vec_xst_be): New #define.
3541 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3542 and externalize from *altivec_vperm_<mode>_internal.
3543 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3545 (XL_BE_V8HI): Likewise.
3546 (XL_BE_V4SI): Likewise.
3547 (XL_BE_V4SI): Likewise.
3548 (XL_BE_V2DI): Likewise.
3549 (XL_BE_V4SF): Likewise.
3550 (XL_BE_V2DF): Likewise.
3551 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3552 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3553 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3554 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3555 VSX_BUILTIN_VEC_XST_BE.
3556 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3557 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3559 (altivec_init_builtins): Replace conditional calls to def_builtin
3560 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3561 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3562 with unconditional calls. Remove calls to def_builtin for
3563 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3564 __builtin_vec_xst_be.
3565 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3566 to define_expand, and add alternate RTL generation for P8.
3567 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3568 vsx_ld_elemrev_v8hi.
3569 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3570 add alternate RTL generation for P8.
3571 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3572 vsx_ld_elemrev_v16qi.
3573 (vsx_st_elemrev_v8hi): Convert define_insn
3574 to define_expand, and add alternate RTL generation for P8.
3575 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3576 vsx_st_elemrev_v8hi.
3577 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3578 add alternate RTL generation for P8.
3579 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3580 vsx_st_elemrev_v16qi.
3582 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3585 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3586 TARGET_AVX512ER check.
3587 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3588 neither -mzeroupper nor -mno-zeroupper is used and
3589 TARGET_EMIT_VZEROUPPER is set.
3590 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3591 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3593 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3595 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3596 folding of vector compares.
3597 (fold_build_vec_cmp): New helper function.
3598 (fold_compare_helper): New helper function.
3599 (builtin_function_type): Add compare builtins to the list of functions
3600 having unsigned arguments. Cosmetic updates to comment indentation.
3601 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3602 the not+eq combination.
3604 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3606 PR tree-optimization/82726
3607 PR tree-optimization/70754
3608 * tree-predcom.c (order_drefs_by_pos): New function.
3609 (combine_chains): Move code setting has_max_use_after to...
3610 (try_combine_chains): ...here. New parameter. Sort combined chains
3611 according to position information.
3612 (tree_predictive_commoning_loop): Update call to above function.
3613 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3615 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3617 PR tree-optimization/82726
3619 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3621 PR tree-optimization/70754
3622 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3623 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3624 combined stmt before it if not NULL.
3625 (combine_chains): Process refs reversely and compute dominance point
3629 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3631 PR tree-optimization/79663
3632 * tree-predcom.c (combine_chains): Process refs in reverse order
3633 only for ZERO length chains, and add explaining comment.
3635 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3637 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3638 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3639 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3641 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3643 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3645 2017-11-15 Martin Liska <mliska@suse.cz>
3647 * tree-cfg.c (pass_warn_function_return::execute):
3648 Compare warn_return_type for greater than zero.
3650 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3654 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3655 to return true on Xeon and not on Xeon Phi.
3656 (ix86_check_avx256_register): Changed to ...
3657 (ix86_check_avx_upper_register): ... this. Add extra check for
3658 VALID_AVX512F_REG_OR_XI_MODE.
3659 (ix86_avx_u128_mode_needed): Changed
3660 ix86_check_avx256_register to ix86_check_avx_upper_register.
3661 (ix86_check_avx256_stores): Changed to ...
3662 (ix86_check_avx_upper_stores): ... this. Changed
3663 ix86_check_avx256_register to ix86_check_avx_upper_register.
3664 (ix86_avx_u128_mode_after): Changed
3665 avx_reg256_found to avx_upper_reg_found. Changed
3666 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3667 (ix86_avx_u128_mode_entry): Changed
3668 ix86_check_avx256_register to ix86_check_avx_upper_register.
3669 (ix86_avx_u128_mode_exit): Ditto.
3670 * config/i386/i386.h: (host_detect_local_cpu): New define.
3672 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3674 * config/arm/xgene1.md (xgene1): Split into automatons
3675 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3676 (xgene1_f_load): Adjust reservations and/or types.
3677 (xgene1_f_store): Likewise.
3678 (xgene1_load_pair): Likewise.
3679 (xgene1_store_pair): Likewise.
3680 (xgene1_fp_load1): Likewise.
3681 (xgene1_load1): Likewise.
3682 (xgene1_store1): Likewise.
3683 (xgene1_move): Likewise.
3684 (xgene1_alu): Likewise.
3685 (xgene1_simd): Likewise.
3686 (xgene1_bfm): Likewise.
3687 (xgene1_neon_load1): Likewise.
3688 (xgene1_neon_store1): Likewise.
3689 (xgene1_neon_logic): Likewise.
3690 (xgene1_neon_st1): Likewise.
3691 (xgene1_neon_ld1r): Likewise.
3692 (xgene1_alu_cond): Added.
3693 (xgene1_shift_reg): Likwise.
3694 (xgene1_bfx): Likewise.
3695 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3697 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3700 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3702 (can_widen_mult_without_libcall): New function.
3703 (expand_mul_overflow): If only checking unsigned mul overflow,
3704 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3705 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3706 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3707 (expand_DIVMOD): Formatting fix.
3708 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3709 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3710 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3712 PR tree-optimization/82977
3713 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3714 constructed temporary to strlen_to_stridx.put.
3716 2017-11-15 Martin Liska <mliska@suse.cz>
3718 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3719 * configure: Regenerate.
3721 2017-11-15 Martin Liska <mliska@suse.cz>
3724 * config/sh/sh-mem.cc: Use proper probability for
3727 2017-11-14 Jeff Law <law@redhat.com>
3729 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3730 the red zone for stack_clash_protection_final_dynamic_probe targets
3731 when the total dynamic stack size is zero bytes.
3733 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3734 blocks is post order.
3736 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3738 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3739 * final.c (rest_of_clean_state): Set it for the
3740 -fcompare-debug dump.
3741 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3742 class when TDF_COMPARE_DEBUG is set.
3744 * dwarf2out.c (gen_producer_string): Discard
3747 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3750 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3751 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3752 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3753 (__TGMATH_REAL_2_3): Remove macros.
3754 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3755 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3757 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3759 (remquo): Define using __TGMATH_REAL_3.
3761 2017-11-14 Jeff Law <law@redhat.com>
3763 * vr-values.c: New file with contents extracted from tree-vrp.c.
3764 * Makefile.in (OBJS): Add vr-values.o
3765 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3766 (set_value_range, set_and_canonicalize_value_range): Likewise.
3767 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3768 (value_range_constant_singleton, symbolic_range_p): Likewise.
3769 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3770 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3771 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3772 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3773 (find_case_label_range, find_case_label_index): Likewise.
3774 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3775 (range_int_cst_singleton_p, value_inside_range): Likewise.
3776 (get_single_symbol): Likewise.
3777 (switch_update): Move structure definition here.
3778 (to_remove_edges, to_update_switch_stmts): Provide externs.
3779 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3780 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3781 (vrp_val_is_min, set_value_range): Likewise.
3782 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3783 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3784 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3785 (range_is_nonnull, range_int_cst_p): Likewwise.
3786 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3787 (get_single_symbol, operand_less_p): Likewise
3788 (compare_values_warnv, compare_values): Likewise.
3789 (value_inside_range, value_range_constant_singleton): Likewise.
3790 (zero_nonzero_bitgs_from_vr): Likewise.
3791 (extract_range_from_binary_expr_1): Likewise.
3792 (overflow_comparison_p): Likewise.
3793 (to_remove_edges, to_update_switch_stmts): Likewise.
3794 (find_case_label-index, find_case_label_range): Likewise.
3795 (switch_update, set_value_range_to_nonnegative): Remove.
3796 (set_value_range_to_truthvalue): Likewise.
3797 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3798 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3799 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3800 (find_case_label_ranges, test_for_singularity): Likewise.
3801 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3802 (x_vr_values): Move to its remaining use site.
3804 2017-11-10 Jeff Law <law@redhat.com>
3806 * vr-values.h (VR_INITIALIZER): Move #define here.
3807 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3808 * Makefile.in (OBJS): Add tree-evrp.o
3809 * tree-vrp.h (assert_info): Move structure definition here.
3810 (set_value_range_to_varying): Prototype.
3811 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3812 (infer_value_range, register_edge_assert_for): Likewise.
3813 (stmt_interesting_for_vrp): Likewise.
3814 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3815 (set_value_range_to_varying): No longer static.
3816 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3817 (infer_value_range, register_edge_assert_for): Likewise.
3819 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3821 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3822 generate the XXBRD instruction.
3824 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3825 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3826 as being compatible if -mabi=ieeelongdouble.
3827 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3828 to setup float128 built-ins with hardware support.
3829 (BU_FLOAT128_HW_2): Likewise.
3830 (BU_FLOAT128_HW_3): Likewise.
3831 (BU_FLOAT128_HW_VSX_1): Likewise.
3832 (BU_FLOAT128_HW_VSX_2): Likewise.
3833 (scalar_extract_expq): Change float128 built-in functions to
3834 accommodate having both KFmode and TFmode functions. Use the
3835 KFmode variant as the default.
3836 (scalar_extract_sigq): Likewise.
3837 (scalar_test_neg_qp): Likewise.
3838 (scalar_insert_exp_q): Likewise.
3839 (scalar_insert_exp_qp): Likewise.
3840 (scalar_test_data_class_qp): Likewise.
3841 (sqrtf128_round_to_odd): Delete processing the round to odd
3842 built-in functions as special built-in functions, and define them
3843 as float128 built-ins. Use the KFmode variant as the default.
3844 (truncf128_round_to_odd): Likewise.
3845 (addf128_round_to_odd): Likewise.
3846 (subf128_round_to_odd): Likewise.
3847 (mulf128_round_to_odd): Likewise.
3848 (divf128_round_to_odd): Likewise.
3849 (fmaf128_round_to_odd): Likewise.
3850 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3851 support for KFmode and TFmode xststdcqp calls.
3852 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3853 point, switch the built-in handlers for the get/set float128
3854 exponent, get float128 mantissa, float128 test built-ins, and the
3855 float128 round to odd built-in functions. Eliminate creating the
3856 float128 round to odd built-in functions as special built-ins.
3857 (rs6000_init_builtins): Eliminate special creation of the float128
3858 round to odd built-in functions.
3859 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3860 function insns to support both TFmode and KFmode variants.
3861 (xsxsigqp_<mode>): Likewise.
3862 (xsiexpqpf_<mode>): Likewise.
3863 (xsiexpqp_<mode>): Likewise.
3864 (xststdcqp_<mode>): Likewise.
3865 (xststdcnegqp_<mode>): Likewise.
3866 (xststdcqp_<mode>): Likewise.
3868 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3870 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3871 unused path_in_freq_ptr parameter.
3872 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3874 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3876 * ipa-inline.c (edge_badness): Dump sreal frequency.
3877 (compute_inlined_call_time): Match natural implementaiton ...
3878 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3879 forgotten division by CGRAPH_FREQ_BASE.
3881 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3883 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3884 Solaris 11. Update comment.
3885 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3887 * config/sol2.h (STARTFILE_SPEC): Likewise.
3888 * configure: Regenerate.
3890 2017-11-14 Carl Love <cel@us.ibm.com>
3892 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3893 le_ and be_ prefixes to swap* variables. Remove
3894 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3897 2017-11-14 Jason Merrill <jason@redhat.com>
3899 Support GTY((cache)) on hash_map.
3900 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3901 (ggc_cache_remove): Override it instead of ggc_mx.
3902 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3903 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3904 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3905 (simple_cache_map_traits): Override maybe_mx.
3906 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3907 (hash_map): Friend gt_cleare_cache.
3908 (gt_cleare_cache): New.
3909 * tree.h (tree_cache_traits): New hash_map traits class.
3910 (tree_cache_map): New typedef.
3912 2017-11-14 Richard Biener <rguenther@suse.de>
3914 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3915 paramter and handling.
3916 (cleanup_control_flow_bb): Likewise.
3917 (cleanup_control_flow_pre): New helper performing a DFS walk
3918 to call cleanup_control_flow_bb in PRE order.
3919 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3920 via cleanup_control_flow_pre.
3922 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
3924 * config/aarch64/aarch64-simd.md
3925 (aarch64_simd_bsl<mode>_internal): Remove DImode.
3926 (*aarch64_simd_bsl<mode>_alt): Likewise.
3927 (aarch64_simd_bsldi_internal): New.
3928 (aarch64_simd_bsldi_alt): Likewise.
3930 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3932 * tracer.c (better_p): Do not compare frequencies.
3933 * reg-stack.c (better_edge): Likewise.
3934 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
3937 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3939 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
3940 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
3941 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
3942 * lto-streamer-in.c (input_function): Use update_max_bb_count.
3943 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
3944 * predict.c (maybe_hot_frequency_p): Inline to ...
3945 (maybe_hot_count_p): ... here; rewrite to counts.
3946 (counts_to_freqs): Rename to ...
3947 (update_max_bb_count): ... this one.
3948 (expensive_function_p): Use counts.
3949 (estimate_bb_frequencies): Update.
3950 (rebuild_frequencies): Update.
3951 * predict.h (counts_to_freqs): Rename to ...
3952 (update_max_bb_count): ... this one.
3953 * profile.c (compute_branch_probabilities): Add debug info
3954 * tree-inline.c (expand_call_inline): Update debug info.
3955 (optimize_inline_calls): Use update_max_bb_count..
3956 (tree_function_versioning): Use update_max_bb_count..
3957 * value-prof.c (gimple_value_profile_transformations):
3958 Do not use update_max_bb_count.
3960 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3962 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
3963 always use frequencies.
3965 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3967 * bb-reorder.c: Remove frequencies from comments.
3968 (better_edge_p): Use profile counts.
3969 (find_traces): Dump profile counts.
3970 (rotate_loop): Use profile counts.
3971 (find_traces_1_round): Likewise.
3972 (connect_better_edge_p): Use counts instead of probabilities for
3974 (copy_bb_p): Drop early check for non-0 frequency.
3975 (sanitize_hot_paths): Update comments.
3977 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3979 * ipa-split.c (struct split_point): Add count.
3980 (consider_split): Do not compute incoming frequency; compute incoming
3981 count and store it to split_point.
3982 (split_function): Set count of the call to split part correctly.
3984 2017-11-13 Carl Love <cel@us.ibm.com>
3986 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
3988 2017-11-13 Tom Tromey <tom@tromey.com>
3990 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
3992 2017-11-13 Carl Love <cel@us.ibm.com>
3994 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3995 Add support for builtins:
3996 unsigned int vec_first_{,miss}_match_{,or_eos}index,
3997 vector {un,}signed {char,int,short},
3998 vector {un,}signed {char,int,short}) arguments.
3999 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4000 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4001 Add BU_P9V_AV_2 expansions for the builtins.
4002 * config/rs6000/altivec.h (vec_first_match_index,
4003 vec_first_mismatch_index, vec_first_match_or_eos_index,
4004 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4005 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4006 new extern declaration.
4007 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4008 * config/rs6000/vsx.md (first_match_index_<mode>,
4009 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4010 first_mismatch_or_eos_index_<mode>): Add define expand.
4011 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4012 * doc/extend.texi: Update the built-in documenation file for the new
4015 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4017 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4018 into the min/max operations for _Float<N> and _Float<N>X types.
4020 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4023 * dwarf2out.c (do_eh_frame): New static variable.
4024 (dwarf2out_begin_prologue): Set it.
4025 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4027 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4029 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4031 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4033 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4036 * debug.h (dwarf2out_do_eh_frame): Declare.
4037 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4038 (dwarf2out_do_frame): Use it.
4039 (dwarf2out_do_cfi_asm): Likewise.
4040 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4041 (dwarf2out_assembly_start): Likewise.
4042 (dwarf2out_begin_prologue): Fix comment.
4043 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4044 if the target needs either debug or unwind DWARF2 info.
4045 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4046 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4047 -ftrapping-math, -ftrapv and -fwrapv.
4049 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4051 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4052 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4053 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4054 (estimate_edge_size_and_time): Likewise.
4055 (ipa_merge_fn_summary_after_inlining): Likewise.
4056 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4057 (compute_uninlined_call_time): Use sreal_frequency.
4058 (compute_inlined_call_time): Likewise.
4059 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4060 * profile-count.c: Include sreal.h.
4061 (profile_count::to_sreal_scale): New.
4062 * profile-count.h: Forward declare sreal.
4063 (profile_count::to_sreal_scale): Declare.
4065 2017-11-13 Nathan Sidwell <nathan@acm.org>
4067 * diagnostic.c (maybe_line_and_column): New.
4068 (diagnostic_get_location_text): Use it.
4069 (diagnostic_report_current_module): Likewise.
4070 (test_diagnostic_get_location_text): Add tests.
4072 2017-11-13 Luis Machado <luis.machado@linaro.org>
4074 * doc/md.texi (Specifying processor pipeline description): Fix
4075 incorrect latency for the div instruction example.
4077 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4079 PR tree-optimization/78821
4080 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4081 that bit_not_p is the same.
4082 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4083 (split_group): Count precisely bit_not_p bits in each statement.
4084 (invert_op): New function.
4085 (imm_store_chain_info::output_merged_store): Use invert_op to
4086 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4087 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4089 2017-11-13 Martin Liska <mliska@suse.cz>
4091 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4092 (struct source_info): Likewise.
4093 (add_branch_counts): Likewise.
4094 (add_line_counts): Likewise.
4095 (function_summary): Likewise.
4096 (output_intermediate_line): Likewise.
4097 (generate_results): Likewise.
4099 2017-11-13 Martin Liska <mliska@suse.cz>
4101 * gcov.c (struct block_info): Remove typedef for block_t.
4102 (struct line_info): Likewise.
4103 (line_info::has_block): Likewise.
4104 (EXIT_BLOCK): Likewise.
4105 (unblock): Likewise.
4106 (circuit): Likewise.
4107 (get_cycles_count): Likewise.
4108 (process_file): Likewise.
4109 (read_graph_file): Likewise.
4110 (solve_flow_graph): Likewise.
4111 (find_exception_blocks): Likewise.
4112 (add_line_counts): Likewise.
4113 (accumulate_line_info): Likewise.
4114 (output_line_details): Likewise.
4116 2017-11-13 Martin Liska <mliska@suse.cz>
4118 * gcov.c (struct arc_info): Remove typedef for arc_t.
4119 (struct line_info): Likewise.
4120 (add_branch_counts): Likewise.
4121 (output_branch_count): Likewise.
4122 (function_info::~function_info): Likewise.
4123 (circuit): Likewise.
4124 (output_intermediate_line): Likewise.
4125 (read_graph_file): Likewise.
4126 (solve_flow_graph): Likewise.
4127 (find_exception_blocks): Likewise.
4128 (add_line_counts): Likewise.
4129 (accumulate_line_info): Likewise.
4130 (output_line_details): Likewise.
4131 (output_function_details): Likewise.
4133 2017-11-13 Martin Liska <mliska@suse.cz>
4135 * gcov.c (struct function_info): Remove typedef for function_t.
4136 (struct source_info): Likewise.
4137 (source_info::get_functions_at_location): Likewise.
4138 (solve_flow_graph): Likewise.
4139 (find_exception_blocks): Likewise.
4140 (add_line_counts): Likewise.
4141 (output_intermediate_file): Likewise.
4142 (process_file): Likewise.
4143 (generate_results): Likewise.
4144 (release_structures): Likewise.
4145 (read_graph_file): Likewise.
4146 (read_count_file): Likewise.
4147 (accumulate_line_counts): Likewise.
4148 (output_lines): Likewise.
4150 2017-11-13 Martin Liska <mliska@suse.cz>
4152 * gcov.c (function_info::function_info): Remove num_counts
4153 and add vector<gcov_type>.
4154 (function_info::~function_info): Use the vector.
4155 (process_file): Likewise.
4156 (read_graph_file): Likewise.
4157 (read_count_file): Likewise.
4158 (solve_flow_graph): Likewise.
4160 2017-11-13 Martin Liska <mliska@suse.cz>
4162 * gcov.c (function_info::is_artificial): New function.
4163 (process_file): Erase all artificial early.
4164 (generate_results): Skip as all artificial are already
4167 2017-11-13 Martin Liska <mliska@suse.cz>
4169 * gcov.c (read_graph_file): Store to global vector of functions.
4170 (read_count_file): Iterate the vector.
4171 (process_file): Likewise.
4172 (generate_results): Likewise.
4173 (release_structures): Likewise.
4175 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4177 PR tree-optimization/82954
4178 * gimple-ssa-store-merging.c
4179 (imm_store_chain_info::coalesce_immediate_stores): If
4180 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4182 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4184 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4185 Upddate call to ENDIAN_LANE_N.
4186 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4187 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4188 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4189 and use aarch64_endian_lane_rtx.
4190 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4192 2017-11-12 Tom de Vries <tom@codesourcery.com>
4194 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4196 2017-11-12 Tom de Vries <tom@codesourcery.com>
4198 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4200 2017-11-12 Tom de Vries <tom@codesourcery.com>
4202 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4204 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4205 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4206 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4208 2017-11-11 Martin Sebor <msebor@redhat.com>
4211 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4214 * prefic.c (translate_name): Replace strncpy with memcpy to
4215 avoid -Wstringop-truncation.
4217 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4219 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4221 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4223 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4225 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4227 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4230 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4232 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4233 enable generating XXBRH if the value is in a vector register.
4234 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4235 value is in a vector register.
4236 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4237 register to register bswap64's instead of doing the GPR sequence
4238 used on previous machines.
4239 (bswapdi2_xxbrd): New insn.
4240 (bswapdi2_reg): Disallow on ISA 3.0.
4241 (register to register bswap64 splitter): Do not split the insn on
4242 ISA 3.0 systems that use XXBRD.
4244 2017-11-10 Martin Sebor <msebor@redhat.com>
4247 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4249 (find_subframework_file): Same.
4251 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4253 * auto-profile.c (afdo_indirect_call): Drop frequency.
4254 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4255 (cgraph_node::create_edge): Drop frequency argument.
4256 (cgraph_node::create_indirect_edge): Drop frequency argument.
4257 (cgraph_edge::make_speculative): Drop frequency arguments.
4258 (cgraph_edge::resolve_speculation): Do not update frequencies
4259 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4260 (cgraph_node::dump): Check consistency in IPA mode.
4261 (cgraph_edge::maybe_hot_p): Use IPA counter.
4262 (cgraph_edge::verify_count_and_frequency): Rename to ...
4263 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4264 (cgraph_node::verify_node): Update.
4265 * cgraph.h (struct cgraph_edge): Drop frequency.
4266 (cgraph_edge::frequency): New function.
4267 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4269 (cgraph_edge::rebuild_edges): Likewise.
4270 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4271 (duplicate_thunk_for_node): Do not pass frequency.
4272 (cgraph_node::create_clone): Scale only counts.
4273 (cgraph_node::create_virtual_clone): Do not pass frequency.
4274 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4275 (cgraph_node::create_version_clone): Do not pass frequency.
4276 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4277 (cgraph_node::expand_thunk): Do not pass frequency.
4278 (cgraph_node::create_wrapper): Do not pass frequency.
4279 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4281 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4282 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4283 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4284 (gather_caller_stats): Use frequency function.
4285 (ipcp_cloning_candidate_p): Use frequency function.
4286 (ipcp_propagate_stage): Use frequency function.
4287 (get_info_about_necessary_edges): Use frequency function.
4288 (update_profiling_info): Update only IPA profile.
4289 (update_specialized_profile): Use frequency functoin.
4290 (perhaps_add_new_callers): Update only IPA profile.
4291 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4292 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4293 (dump_ipa_call_summary): Use frequency function.
4294 (estimate_edge_size_and_time): Use frequency function.
4295 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4296 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4297 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4298 (update_noncloned_counts): ... ths one; scale counts only.
4299 (clone_inlined_nodes): Do not scale frequency.
4300 (inline_call): Do not pass frequency.
4301 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4302 (compute_inlined_call_time): Use IPA profile.
4303 (want_inline_small_function_p): Use IPA profile.
4304 (want_inline_self_recursive_call_p): Use IPA profile.
4305 (edge_badness): Use IPA profile.
4306 (lookup_recursive_calls): Use IPA profile.
4307 (recursive_inlining): Do not pass frequency.
4308 (resolve_noninline_speculation): Do not update frequency.
4309 (inline_small_functions): Collect max of IPA profile.
4310 (dump_overall_stats): Dump IPA porfile.
4311 (dump_inline_stats): Dump IPA porfile.
4312 (ipa_inline): Collect IPA stats.
4313 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4314 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4315 (ipa_propagate_frequency): Use frequency function.
4316 (ipa_profile): Cleanup.
4317 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4318 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4319 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4320 (input_node): Do not stream frequency.
4321 (input_edge): Do not stream frequency.
4322 (merge_profile_summaries): Scale only IPA profiles.
4323 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4324 * predict.c (drop_profile): Do not recompute frequency.
4325 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4326 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4327 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4328 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4329 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4330 * tree-inline.c (copy_bb): Do not scale frequency.
4331 (expand_call_inline): Do not scale frequency.
4332 (tree_function_versioning): Do not scale frequency.
4333 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4335 2017-11-10 Julia Koval <julia.koval@intel.com>
4337 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4338 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4339 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4340 (_mm256_maskz_gf2p8affine_epi64_epi8)
4341 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4342 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4343 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4344 (__builtin_ia32_vgf2p8affineqb_v32qi)
4345 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4346 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4348 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4352 (arm_option_override): Refactor.
4353 (arm_option_reconfigure_globals): New.
4354 (arm_options_perform_arch_sanity_checks): New.
4355 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4357 (arm_options_perform_arch_sanity_checks): Likewise
4359 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4361 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4362 (power9-qpmul): New.
4363 * rs6000/rs6000.md ("type" attr): Add qmul.
4364 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4365 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4366 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4368 2017-11-10 Martin Sebor <msebor@redhat.com>
4371 * builtins.c (compute_objsize): Handle arrays that
4372 compute_builtin_object_size likes to fail for. Make extern.
4373 * builtins.h (compute_objsize): Declare.
4374 (check_strncpy_sizes): New function.
4375 (expand_builtin_strncpy): Call check_strncpy_sizes.
4376 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4377 -Wstringop-truncation.
4378 (gimple_fold_builtin_strncat): Same.
4379 * gimple.c (gimple_build_call_from_tree): Set call location.
4380 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4381 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4382 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4383 (handle_builtin_strlen): Use strlen_to_stridx.
4384 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4386 Use strlen_to_stridx.
4387 (pass_strlen::execute): Release strlen_to_stridx.
4388 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4389 (-Wstringop-truncation): Document new option.
4391 2017-11-10 Martin Liska <mliska@suse.cz>
4393 PR gcov-profile/82702
4394 * gcov.c (main): Handle intermediate files in a different
4396 (get_gcov_intermediate_filename): New function.
4397 (output_gcov_file): Remove support of intermediate files.
4398 (generate_results): Allocate intermediate file.
4399 (release_structures): Clean-up properly fn_end.
4400 (output_intermediate_file): Start iterating with line 1.
4402 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4404 PR tree-optimization/82929
4405 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4406 ops_swapped_p non-static data member.
4407 (store_immediate_info::store_immediate_info): Clear it.
4408 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4409 ops set ops_swapped_p.
4410 (count_multiple_uses): Handle ops_swapped_p.
4412 2017-11-10 Martin Liska <mliska@suse.cz>
4414 * coverage.c (coverage_init): Stream information about
4415 support of has_unexecuted_blocks.
4416 * doc/gcov.texi: Document that.
4417 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4418 * gcov.c (read_graph_file): Likewise.
4419 (output_line_beginning): Fix a small issue with
4422 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4424 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4425 reference of trivial component.
4427 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4430 * gimple-ssa-store-merging.c
4431 (pass_store_merging::terminate_all_aliasing_chains): For
4432 gimple_store_p stmts also call refs_output_dependent_p.
4434 PR rtl-optimization/82913
4435 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4438 2017-11-09 Jeff Law <law@redhat.com>
4440 * vr-values.h: New file with vr_values class.
4441 * tree-vrp.c: Include vr-values.h
4442 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4443 data objects into the vr_values class.
4444 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4445 (get_value_range): Make it a member function within vr_values class.
4446 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4447 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4448 (op_with_constant_singleton_value_range): Likewise.
4449 (extract_range_for_var_from_comparison_expr): Likewise.
4450 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4451 (extract_range_from_binary_expr): Likewise.
4452 (extract_range_from_unary_expr): Likewise.
4453 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4454 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4455 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4456 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4457 (compare_name_with_value, compare_names): Likewise.
4458 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4459 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4460 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4461 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4462 (extract_range_from_phi_node): Likewise.
4463 (simplify_truth_ops_using_ranges): Likewise.
4464 (simplify_div_or_mod_using_ranges): Likewise.
4465 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4466 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4467 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4468 (simplify_float_conversion_using_ranges): Likewise.
4469 (simplify_internal_call_using_ranges): Likewise.
4470 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4471 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4472 poitner via x_vr_values for calls into gimple folder.
4473 (vrp_initialize_lattice): Make this the vr_values ctor.
4474 (vrp_free_lattice): Make this the vr_values dtor.
4475 (set_vr_value): New function.
4476 (class vrp_prop): Add vr_values data member. Add various member
4477 functions as well as member functions that delegate to vr_values.
4478 (check_array_ref): Make a member function within vrp_prop class.
4479 (search_for_addr_array, vrp_initialize): Likewise.
4480 (vrp_finalize): Likewise. Revamp to avoid direct access to
4481 vr_value, values_propagated, etc.
4482 (check_array_bounds): Extract vrp_prop class instance pointer from
4483 walk info structure. Use it to call member functions.
4484 (check_all_array_refs): Make a member function within vrp_prop class.
4485 Smuggle class instance pointer via walk info structure.
4486 (x_vr_values): New local static.
4487 (vrp_valueize): Use x_vr_values to get class instance.
4488 (vr_valueize_1): Likewise.
4489 (class vrp_folder): Add vr_values data member. Add various member
4490 functions as well as member functions that delegate to vr_values.
4491 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4492 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4493 class instance from vr_values. Use it to call member functions.
4494 (vrp_dom_walker): Add vr_values data member.
4495 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4496 instance via x_vr_values.
4497 (identify_jump_threads): Accept vr_values as argument. Store
4498 it into the walker structure.
4499 (evrp_dom_walker): Add vr_values class data member. Add various
4501 (evrp_dom_walker::try_find_new_range): Use vr_values data
4502 member to access the memory allocator.
4503 (evrp_dom_walker::before_dom_children): Store vr_values class
4504 instance into the vrp_folder class.
4505 (evrp_dom_walker::push_value_range): Rework to avoid direct
4506 access to num_vr_values and vr_value.
4507 (evrp_dom_walker::pop_value_range): Likewise.
4508 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4509 Use vr_values to get to dump_all_value_ranges member function.
4510 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4511 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4512 Pass vr_values class instance down to identify_jump_threads.
4513 Remove call to vrp_free_lattice.
4514 (debug_all_value_ranges): Remove.
4516 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4517 (vrp_folder): Likewise.
4519 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4520 Get it from the existing bitmap instead.
4521 (vrp_intersect_ranges_1): Likewise.
4523 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4525 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4527 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4528 set bit_not_p to it.
4529 (imm_store_chain_info::coalesce_immediate_stores): Break group
4530 if bit_not_p is different.
4531 (count_multiple_uses, split_group,
4532 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4533 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4534 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4535 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4536 to store_immediate_info ctor.
4538 2017-11-09 Jim Wilson <jimw@sifive.com>
4540 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4541 (scan_prog_file): Likewise.
4543 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4545 * bb-reorder.c (max_entry_frequency): Remove.
4546 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4547 connect_traces, push_to_next_round_p): Remove prototypes.
4548 (find_traces_1_round): Use counts only.
4549 (push_to_next_round_p): Likewise.
4550 (find_traces): Likewise.
4551 (rotate_loop): Likewise.
4552 (find_traces_1_round): Likewise.
4553 (connect_traces): Likewise.
4554 (edge_order): Likewise.
4556 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4558 * config/arm/arm.c (output_return_instruction): Add comments to
4559 indicate requirement for cmse_nonsecure_entry return to account
4560 for the size of clearing instruction output here.
4561 (thumb_exit): Likewise.
4562 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4563 return in hardfloat mode.
4565 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4567 * config/rs6000/rs6000.c (machine_function): Add a bool,
4568 "toc_is_wrapped_separately".
4569 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4570 if it wasn't explicitly set or unset, we are optimizing for speed, and
4571 doing separate shrink-wrapping.
4572 (rs6000_get_separate_components): Enable the TOC component if
4573 saving the TOC register in the prologue.
4574 (rs6000_components_for_bb): Handle the TOC component.
4575 (rs6000_emit_prologue_components): Store the TOC register where needed.
4576 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4577 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4579 2017-11-09 Martin Jambor <mjambor@suse.cz>
4581 * ipa-param-manipulation.c: New file.
4582 * ipa-param-manipulation.h: Likewise.
4583 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4584 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4585 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4586 (ipa_parm_adjustment): Likewise.
4587 (ipa_parm_adjustment_vec): Likewise.
4588 (ipa_get_vector_of_formal_parms): Moved declaration to
4589 ipa-param-manipulation.h.
4590 (ipa_get_vector_of_formal_parm_types): Likewise.
4591 (ipa_modify_formal_parameters): Likewise.
4592 (ipa_modify_call_arguments): Likewise.
4593 (ipa_combine_adjustments): Likewise.
4594 (ipa_dump_param_adjustments): Likewise.
4595 (ipa_modify_expr): Likewise.
4596 (ipa_get_adjustment_candidate): Likewise.
4597 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4598 ipa-param-manipulation.c.
4599 (ipa_get_vector_of_formal_parm_types): Likewise.
4600 (ipa_modify_formal_parameters): Likewise.
4601 (ipa_modify_call_arguments): Likewise.
4602 (ipa_modify_expr): Likewise.
4603 (get_ssa_base_param): Likewise.
4604 (ipa_get_adjustment_candidate): Likewise.
4605 (index_in_adjustments_multiple_times_p): Likewise.
4606 (ipa_combine_adjustments): Likewise.
4607 (ipa_dump_param_adjustments): Likewise.
4608 * tree-sra.c: Also include ipa-param-manipulation.h
4609 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4612 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4613 Alan Hayward <alan.hayward@arm.com>
4614 David Sherwood <david.sherwood@arm.com>
4616 * doc/sourcebuild.texi (vect_masked_store): Document.
4618 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4619 Alan Hayward <alan.hayward@arm.com>
4620 David Sherwood <david.sherwood@arm.com>
4622 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4624 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4625 Alan Hayward <alan.hayward@arm.com>
4626 David Sherwood <david.sherwood@arm.com>
4628 * doc/sourcebuild.texi (vect_variable_length): Document.
4630 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4631 Alan Hayward <alan.hayward@arm.com>
4632 David Sherwood <david.sherwood@arm.com>
4634 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4636 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4637 Alan Hayward <alan.hayward@arm.com>
4638 David Sherwood <david.sherwood@arm.com>
4640 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4642 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4643 Alan Hayward <alan.hayward@arm.com>
4644 David Sherwood <david.sherwood@arm.com>
4646 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4647 previously undocumented selectors.
4648 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4650 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4652 * doc/rtl.texi (const_vector): Say that elements can be
4653 const_wide_ints too.
4654 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4655 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4656 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4657 * optabs.c (expand_vector_broadcast): Likewise.
4659 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4660 Alan Hayward <alan.hayward@arm.com>
4661 David Sherwood <david.sherwood@arm.com>
4663 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4664 scaled index even if the unscaled address was invalid.
4665 Don't increase the complexity of using a scale in that case.
4667 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4668 Alan Hayward <alan.hayward@arm.com>
4669 David Sherwood <david.sherwood@arm.com>
4671 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4672 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4673 UNITS_PER_WORD bytes.
4674 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4675 based on REGMODE_NATURAL_SIZE of the inner mode.
4676 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4677 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4678 * expmed.c (lowpart_bit_field_p): Divide the value up into
4679 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4680 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4681 whether something is likely to occupy more than one register.
4683 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4686 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4687 profile_count::adjust_for_ipa_scaling.
4688 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4689 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4691 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4693 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4695 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4696 (split_group): Add total_orig and total_new arguments, estimate the
4697 number of statements related to the store group without store merging
4698 and with store merging.
4699 (imm_store_chain_info::output_merged_store): Adjust split_group
4700 callers, punt if estimated number of statements with store merging
4701 is not smaller than estimated number of statements without it.
4703 (handled_load): Remove has_single_use checks.
4704 (pass_store_merging::process_store): Likewise.
4706 2017-11-09 Richard Biener <rguenther@suse.de>
4708 PR tree-optimization/82902
4709 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4711 2017-11-09 Martin Liska <mliska@suse.cz>
4714 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4717 2017-11-09 Martin Liska <mliska@suse.cz>
4719 PR tree-optimization/82669
4720 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4722 2017-11-09 Martin Liska <mliska@suse.cz>
4724 PR gcov-profile/48463
4725 * coverage.c (coverage_begin_function): Output also end locus
4726 of a function and information whether the function is
4728 * gcov-dump.c (tag_function): Parse and print the information.
4729 * gcov.c (INCLUDE_MAP): Add include.
4730 (INCLUDE_SET): Likewise.
4731 (struct line_info): Move earlier in the source file because
4732 of vector<line_info> in function_info structure.
4733 (line_info::line_info): Likewise.
4734 (line_info::has_block): Likewise.
4735 (struct source_info): Add new member index.
4736 (source_info::get_functions_at_location): New function.
4737 (function_info::group_line_p): New function.
4738 (output_intermediate_line): New function.
4739 (output_intermediate_file): Use the mentioned function.
4740 (struct function_start): New.
4741 (struct function_start_pair_hash): Likewise.
4742 (process_file): Add code that identifies group functions.
4743 Assign lines either to global or function scope.
4744 (generate_results): Skip artificial functions.
4745 (find_source): Assign index for each source file.
4746 (read_graph_file): Read new flag artificial and end_line.
4747 (add_line_counts): Assign it either to global of function scope.
4748 (accumulate_line_counts): Isolate core of the function to
4749 accumulate_line_info and call it for both function and global
4751 (accumulate_line_info): New function.
4752 (output_line_beginning): Fix GNU coding style.
4753 (print_source_line): New function.
4754 (output_line_details): Likewise.
4755 (output_function_details): Likewise.
4756 (output_lines): Iterate both source (global) scope and function
4758 (struct function_line_start_cmp): New class.
4759 * doc/gcov.texi: Reflect changes in documentation.
4761 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4764 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4765 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4766 and similarly for not instead of neg.
4768 2017-11-08 Andi Kleen <ak@linux.intel.com>
4770 * config/i386/i386.opt: Add -mforce-indirect-call.
4771 * config/i386/predicates.md: Check for flag_force_indirect_call.
4772 * doc/invoke.texi: Document -mforce-indirect-call
4774 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4776 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4778 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4779 config/riscv/riscv.c (predict.h): New include.
4780 (riscv_slow_unaligned_access_p): No longer static.
4781 (riscv_block_move_straight): Add require.
4782 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4784 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4787 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4788 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4789 nonimmediate_operand predicate for operand 1 instead of
4792 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4794 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4796 * config/aarch64/constraints.md (Uml): New constraint.
4797 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4800 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4802 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4803 of two vec_duplicates into a vec_concat.
4805 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4807 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4808 Simplify vec_merge of vec_duplicate and vec_concat.
4809 * config/aarch64/constraints.md (Utq): New constraint.
4810 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4813 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4815 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4816 Simplify vec_merge of vec_duplicate and const_vector.
4817 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4819 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4820 mode iterator. Update predicate on operand 1 to
4821 handle non-const_vec constants. Delete constraints.
4822 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4824 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4826 PR tree-optimization/78821
4827 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4829 (store_operand_info::store_operand_info): Initialize it to false.
4830 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4831 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4832 store in the group, and if chain_info is non-NULL, to ignore altogether
4834 (compatible_load_p): Fail if bit_not_p does not match.
4835 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4836 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4837 (pass_store_merging::process_store): Adjust
4838 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4839 call terminate_all_aliasing_chains newly when adding a store into
4840 a chain with non-NULL chain_info.
4842 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4844 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4846 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4848 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4850 (aarch64_layout_frame): Initialise emit_frame_chain.
4851 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4852 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4854 2017-11-08 Martin Liska <mliska@suse.cz>
4856 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4857 of gimple_call_internal_p.
4859 2017-11-07 Tom de Vries <tom@codesourcery.com>
4861 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4863 2017-11-07 Tom de Vries <tom@codesourcery.com>
4865 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4868 2017-11-08 Martin Liska <mliska@suse.cz>
4871 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4873 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4875 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4877 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4880 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4881 (*cmp<mode>_ccz_1): New insn with $k alternative.
4884 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4885 RTX_COMM_COMPARE as commutative as well.
4886 (ix86_binary_operator_ok): Formatting fix.
4887 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4888 *<code><mode>3<mask_name><round_saeonly_name>,
4889 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4890 *<s>mul<mode>3_highpart<mask_name>,
4891 *vec_widen_umult_even_v16si<mask_name>,
4892 *vec_widen_umult_even_v8si<mask_name>,
4893 *vec_widen_umult_even_v4si<mask_name>,
4894 *vec_widen_smult_even_v16si<mask_name>,
4895 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4896 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4897 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4898 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4899 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4900 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4901 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4902 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4903 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4904 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4905 ix86_binary_operator_ok. Formatting fixes.
4906 (*<plusminus_insn><mode>3<mask_name><round_name>,
4907 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4910 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4912 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4913 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4914 eq and ne if TARGET_ISEL.
4915 (cmp): New code_iterator.
4916 (UNS, UNSU_, UNSIK): New code_attrs.
4917 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4918 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4920 ("eq<mode>3"): ... this.
4921 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
4923 ("ne<mode>3"): ... this.
4925 2017-11-07 Julia Koval <julia.koval@intel.com>
4928 * common/config/i386/i386-common.c
4929 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
4930 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
4931 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
4932 * config/i386/i386.opt: Ditto.
4933 * config/i386/i386.c (ix86_target_string): Ditto.
4934 (ix86_option_override_internal): Ditto.
4935 (ix86_init_mpx_builtins): Move MPX to args2.
4936 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
4937 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4938 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4939 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4940 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4941 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4942 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
4944 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
4947 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
4948 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
4949 (zero-extendsidi peephole2): Remove peephole.
4951 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4954 * stor-layout.c: Include attribs.h.
4955 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
4956 explicit lookup of "aligned" attribute.
4958 2017-11-07 Andrew Waterman <andrew@sifive.com>
4960 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
4961 (riscv_expand_block_move): Likewise.
4962 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
4964 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
4965 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
4966 * config/riscv/riscv.c (riscv_block_move_straight): New function.
4967 (riscv_adjust_block_mem): Likewise.
4968 (riscv_block_move_loop): Likewise.
4969 (riscv_expand_block_move): Likewise.
4970 * config/riscv/riscv.md (movmemsi): New pattern.
4972 2017-11-07 Michael Clark <michaeljclark@mac.com>
4974 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
4975 (MUSL_DYNAMIC_LINKER): Likewise.
4977 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
4979 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
4980 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
4983 2017-11-07 Richard Biener <rguenther@suse.de>
4985 * match.pd: Fix build.
4987 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
4988 Jackson Woodruff <jackson.woodruff@arm.com>
4990 PR tree-optimization/71026
4991 * match.pd: Canonicalize negate in division.
4993 2017-11-07 Sudakshina Das <sudi.das@arm.com>
4996 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
4998 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5000 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5001 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5003 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5005 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5006 non-scalar integral types.
5007 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5008 (-(A-B), -(~A)): New transformations.
5010 2017-11-07 Tom de Vries <tom@codesourcery.com>
5012 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5013 semicolon after "do {} while (0)".
5014 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5015 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5016 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5017 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5018 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5020 2017-11-07 Tom de Vries <tom@codesourcery.com>
5022 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5023 after "do {} while (0)".
5024 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5025 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5026 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5027 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5028 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5030 2017-11-07 Tom de Vries <tom@codesourcery.com>
5033 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5035 (arm_rtx_costs_internal): Add missing semicolon after
5036 HANDLE_NARROW_SHIFT_ARITH call.
5038 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5040 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5041 disable isel if it was not set explicitly.
5043 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5045 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5046 (add_type_attribute) likewise.
5048 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5050 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5051 of struct ix86_frame.
5052 (ix86_initial_elimination_offset): Likewise.
5053 (ix86_expand_split_stack_prologue): Likewise.
5055 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5057 * tree-vrp.h (enum value_range_type): Update stale comment.
5059 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5060 Alan Hayward <alan.hayward@arm.com>
5061 David Sherwood <david.sherwood@arm.com>
5063 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5064 (aarch64_expand_vec_perm_const): Take the number of units too.
5065 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5066 (aarch64_expand_vec_perm_const): Likewise.
5067 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5068 (vec_perm<mode>): Update accordingly.
5070 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5071 Alan Hayward <alan.hayward@arm.com>
5072 David Sherwood <david.sherwood@arm.com>
5074 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5075 Take the number of units too.
5076 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5077 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5078 but check for a vector mode before rather than after the call.
5079 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5080 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5081 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5082 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5083 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5084 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5085 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5086 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5087 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5088 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5089 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5090 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5091 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5092 (aarch64_sqdmull2_n<mode>): Update accordingly.
5094 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5095 Alan Hayward <alan.hayward@arm.com>
5096 David Sherwood <david.sherwood@arm.com>
5098 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5099 the number of units too.
5100 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5101 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5102 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5103 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5104 (vec_store_lanesxi<mode>): Update accordingly.
5106 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5107 Alan Hayward <alan.hayward@arm.com>
5108 David Sherwood <david.sherwood@arm.com>
5110 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5111 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5112 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5113 of units rather than the mode.
5114 * config/aarch64/iterators.md (nunits): New mode attribute.
5115 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5116 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5117 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5118 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5119 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5120 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5121 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5122 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5123 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5124 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5125 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5126 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5127 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5128 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5129 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5130 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5131 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5132 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5133 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5134 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5135 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5136 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5137 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5138 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5139 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5140 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5141 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5142 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5143 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5144 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5145 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5146 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5147 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5148 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5149 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5150 (aarch64_simd_vec_setv2di): Likewise.
5152 2017-11-06 Carl Love <cel@us.ibm.com>
5154 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5156 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5157 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5159 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5161 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5162 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5163 Add power 8 macro expansions.
5164 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5165 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5166 power 8 instructions. (VSX_XXBR): Add iterator.
5168 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5170 * config/arm/arm.md (predicable_short_it): Change default to "no",
5171 improve documentation, remove uses that are identical to the default.
5172 (enabled_for_depr_it): Rename to enabled_for_short_it.
5173 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5174 Remove default uses.
5175 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5176 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5177 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5178 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5180 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5183 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5184 float128 helper macros, which are no longer used after deleting
5185 the old 'q' built-in functions, and moving the round to odd
5186 built-in functions to being special built-in functions.
5187 (BU_FLOAT128_2): Likewise.
5188 (BU_FLOAT128_1_HW): Likewise.
5189 (BU_FLOAT128_2_HW): Likewise.
5190 (BU_FLOAT128_3_HW): Likewise.
5191 (FABSQ): Delete old 'q' built-in functions.
5192 (COPYSIGNQ): Likewise.
5193 (SQRTF128_ODD): Move round to odd built-in functions to be
5194 special built-in functions, so that we can handle
5195 -mabi=ieeelongdouble.
5196 (TRUNCF128_ODD): Likewise.
5197 (ADDF128_ODD): Likewise.
5198 (SUBF128_ODD): Likewise.
5199 (MULF128_ODD): Likewise.
5200 (DIVF128_ODD): Likewise.
5201 (FMAF128_ODD): Likewise.
5202 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5203 built-in names to 'f128'.
5204 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5205 old 'q' built-in functions, as the machine independent code for
5206 'f128' built-in functions handles this.
5207 (rs6000_expand_builtin): Add expansion for float128 round to odd
5208 functions, keying off on -mabi=ieeelongdouble of whether to use
5209 the KFmode or TFmode variant.
5210 (rs6000_init_builtins): Initialize the _Float128 round to odd
5212 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5213 _Float128 'q' built-in functions are now mapped into the new
5214 'f128' built-in functions.
5216 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5218 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5219 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5221 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5224 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5225 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5226 * toplev.c: Include "ipa-fnsummary.h".
5227 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5229 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5231 PR tree-optimization/82838
5232 * gimple-ssa-store-merging.c
5233 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5234 on a separate gimple_seq which is then appended to seq.
5236 2017-11-06 Jeff Law <law@redhat.com>
5239 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5240 (get_probe_interval): New functions.
5241 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5242 (ix86_adjust_stack_and_probe): Likewise.
5243 (output_adjust_stack_and_probe): Likewise.
5244 (ix86_emit_probe_stack_range): Likewise.
5245 (ix86_expand_prologue): Likewise.
5247 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5249 PR tree-optimization/82816
5250 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5251 if the modes of the two types are the same.
5252 (convert_plusminus_to_widen): Likewise.
5254 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5256 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5258 (usadv16qi): New define_expand.
5259 (usadv8hi): New define_expand.
5261 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5264 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5266 (inline_transform): Likewise.
5267 * predict.c (counts_to_freqs): Remove useless conditional.
5268 * profile-count.h (profile_count::apply_scale): Move sanity check.
5269 * tree-inline.c (copy_bb): Always scale.
5270 (copy_cfg_body): Likewise.
5272 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5275 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5277 (*cmp_ite0): Add enabled_for_depr_it attribute.
5278 (*cmp_ite1): Likewise.
5280 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5282 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5285 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5287 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5288 return the value by pointer.
5289 (extract_range_from_multiplicative_op_1): Update accordingly.
5290 Return as soon as an operation fails.
5292 2017-11-05 Tom de Vries <tom@codesourcery.com>
5295 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5296 (DEF_SANITIZER_BUILTIN): ... here.
5297 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5298 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5300 2017-11-05 Tom de Vries <tom@codesourcery.com>
5303 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5305 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5306 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5307 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5309 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5310 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5311 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5313 2017-11-05 Tom de Vries <tom@codesourcery.com>
5316 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5319 2017-11-04 Michael Clark <michaeljclark@mac.com>
5321 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5322 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5324 (*addsi3_extended): Likewise.
5325 (*addsi3_extended2): Likewise.
5326 (<optab>si3): Likewise.
5327 (<optab>di3): Likewise.
5328 (<optab><mode>3): Likewise.
5329 (<*optabe>si3_internal): Likewise.
5330 (zero_extendqi<SUPERQI:mode>2): Likewise.
5331 (*add<mode>hi3): Likewise.
5332 (*xor<mode>hi3): Likewise.
5333 (<optab>di3): Likewise.
5334 (*<optab>si3_extend): Likewise.
5335 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5336 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5337 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5339 2017-11-04 Andrew Waterman <andrew@sifive.com>
5341 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5342 TARGET_STRICT_ALIGN based upon -mtune argument.
5344 2017-11-04 Andrew Waterman <andrew@sifive.com>
5346 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5348 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5350 * config/i386/i386.c (choose_basereg): Use optional scratch
5351 register and add assertion.
5352 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5353 needed, and don't allocate stack.
5354 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5355 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5356 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5358 2017-11-03 Jeff Law <law@redhat.com>
5360 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5361 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5362 to probe at the start of a noreturn function.
5364 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5366 PR tree-optimization/78821
5367 * gimple-ssa-store-merging.c: Update the file comment.
5368 (MAX_STORE_ALIAS_CHECKS): Define.
5369 (struct store_operand_info): New type.
5370 (store_operand_info::store_operand_info): New constructor.
5371 (struct store_immediate_info): Add rhs_code and ops data members.
5372 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5373 and op1r arguments to the ctor, initialize corresponding data members.
5374 (struct merged_store_group): Add load_align_base and load_align
5376 (merged_store_group::merged_store_group): Initialize them.
5377 (merged_store_group::do_merge): Update them.
5378 (merged_store_group::apply_stores): Pick the constant for
5379 encode_tree_to_bitpos from one of the two operands, or skip
5380 encode_tree_to_bitpos if neither operand is a constant.
5381 (class pass_store_merging): Add process_store method decl. Remove
5382 bool argument from terminate_all_aliasing_chains method decl.
5383 (pass_store_merging::terminate_all_aliasing_chains): Remove
5384 var_offset_p argument and corresponding handling.
5385 (stmts_may_clobber_ref_p): New function.
5386 (compatible_load_p): New function.
5387 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5388 if there is overlap and rhs_code is not INTEGER_CST. For
5389 non-overlapping stores terminate group if rhs is not mergeable.
5390 (get_alias_type_for_stmts): Change first argument from
5391 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5392 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5393 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5395 (get_location_for_stmts): Change first argument from
5396 auto_vec<gimple *> & to vec<gimple *> &.
5397 (struct split_store): Remove orig_stmts data member, add orig_stores.
5398 (split_store::split_store): Create orig_stores rather than orig_stmts.
5399 (find_constituent_stmts): Renamed to ...
5400 (find_constituent_stores): ... this. Change second argument from
5401 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5402 to info structures rather than the statements.
5403 (split_group): Rename ALLOW_UNALIGNED argument to
5404 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5405 it. Adjust find_constituent_stores caller.
5406 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5407 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5408 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5409 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5410 (mem_valid_for_store_merging): New function.
5411 (handled_load): New function.
5412 (pass_store_merging::process_store): New method.
5413 (pass_store_merging::execute): Use process_store method. Adjust
5414 terminate_all_aliasing_chains caller.
5416 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5418 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5419 Return true for more constants, symbols and label references.
5420 (aarch64_valid_floating_const): Remove unused function.
5422 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5425 * config/aarch64/aarch64.c (aarch64_layout_frame):
5426 Undo forcing of LR at bottom of frame.
5428 2017-11-03 Jeff Law <law@redhat.com>
5431 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5432 for int_registers_saved.
5434 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5435 extracted from tree-ssa-dom.c.
5436 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5437 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5438 (record_equivalences_from_incoming_edge): Add additional argument
5439 to single_pred_edge_ignoring_loop_edges call.
5440 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5441 (uncprop_dom_walker::before_dom_children): Add additional argument
5442 to single_pred_edge_ignoring_loop_edges call.
5443 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5444 single_pred_edge_ignoring_loop_edges rather than open coding.
5445 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5447 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5449 * match.pd (-(-A)): Rewrite.
5451 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5453 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5454 (rs6000_emit_int_cmove): New declaration.
5455 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5456 (rs6000_emit_sISEL): Delete.
5457 (rs6000_emit_int_cmove): Make non-static.
5458 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5459 instead of rs6000_emit_sISEL.
5461 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5463 * asan.c (create_cond_insert_point): Maintain profile.
5464 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5466 * basic-block.h (struct basic_block_def): Remove frequency.
5467 (EDGE_FREQUENCY): Use to_frequency
5468 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5470 (find_traces): Update to use to_frequency.
5471 (find_traces_1_round): Likewise; use only IPA counts.
5472 (bb_to_key): Likewise.
5473 (connect_traces): Use IPA counts only.
5474 (copy_bb_p): Update to use to_frequency.
5475 (fix_up_crossing_landing_pad): Likewise.
5476 (sanitize_hot_paths): Likewise.
5477 * bt-load.c (basic_block_freq): Likewise.
5478 * cfg.c (init_flow): Set count_max to uninitialized.
5479 (check_bb_profile): Remove frequencies; check counts.
5480 (dump_bb_info): Do not dump frequencies.
5481 (update_bb_profile_for_threading): Update counts only.
5482 (scale_bbs_frequencies_int): Likewise.
5483 (MAX_SAFE_MULTIPLIER): Remove.
5484 (scale_bbs_frequencies_gcov_type): Update counts only.
5485 (scale_bbs_frequencies_profile_count): Update counts only.
5486 (scale_bbs_frequencies): Update counts only.
5487 * cfg.h (struct control_flow_graph): Add count-max.
5488 (update_bb_profile_for_threading): Update prototype.
5489 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5490 (find_many_sub_basic_blocks): Likewise.
5491 * cfgcleanup.c (try_forward_edges): Likewise.
5492 (try_crossjump_to_edge): Likewise.
5493 * cfgexpand.c (expand_gimple_cond): Likewise.
5494 (expand_gimple_tailcall): Likewise.
5495 (construct_init_block): Likewise.
5496 (construct_exit_block): Likewise.
5497 * cfghooks.c (verify_flow_info): Check consistency of counts.
5498 (dump_bb_for_graph): Do not dump frequencies.
5499 (split_block_1): Do not update frequencies.
5500 (split_edge): Do not update frequencies.
5501 (make_forwarder_block): Do not update frequencies.
5502 (duplicate_block): Do not update frequencies.
5503 (account_profile_record): Do not update frequencies.
5504 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5505 for global heuristics.
5506 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5507 (expected_loop_iterations_unbounded): Use counts only.
5508 * cfgloopmanip.c (scale_loop_profile): Simplify.
5509 (create_empty_loop_on_edge): Simplify
5511 (duplicate_loop_to_header_edge): Simplify
5512 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5513 (update_br_prob_note): Take care of removing note when profile
5515 (relink_block_chain): Do not dump frequency.
5516 (rtl_account_profile_record): Use to_frequency.
5517 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5518 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5519 (cgraph_update_edges_for_call_stmt_node): Likewise.
5520 (cgraph_edge::verify_count_and_frequency): Update.
5521 (cgraph_node::verify_node): Temporarily disable frequency verification.
5522 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5523 to_cgraph_frequency.
5524 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5525 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5527 (cgraph_node::expand_thunk): Update profile.
5528 * except.c (dw2_build_landing_pads): Do not update frequency.
5529 * final.c (compute_alignments): Use to_frequency.
5530 (dump_basic_block_info): Do not dump frequency.
5531 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5532 (dump_gimple_bb_header): Do not dump frequency.
5533 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5535 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5536 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5537 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5538 (init_before_recovery): Do not update frequency.
5539 (sched_create_recovery_edges): Do not update frequency.
5540 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5541 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5542 (ipa_cp_c_finalize): Set max_count to uninitialized.
5543 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5544 (param_change_prob): Use counts.
5545 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5547 * ipa-split.c (consider_split): Use to_frequency.
5548 (split_function): Use to_frequency.
5549 * ira-build.c (loop_compare_func): Likewise.
5550 (mark_loops_for_removal): Likewise.
5551 (mark_all_loops_for_removal): Likewise.
5552 * loop-doloop.c (doloop_modify): Do not update frequency.
5553 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5555 * lto-streamer-in.c (input_function): Update count_max.
5556 * omp-expand.c (expand_omp_taskreg): Update count_max.
5557 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5558 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5559 (maybe_hot_count_p): Use ipa counts only.
5560 (maybe_hot_bb_p): Simplify.
5561 (maybe_hot_edge_p): Simplify.
5562 (probably_never_executed): Do not take frequency argument.
5563 (probably_never_executed_bb_p): Do not pass frequency.
5564 (probably_never_executed_edge_p): Likewise.
5565 (combine_predictions_for_bb): Check that profile is nonzero.
5566 (propagate_freq): Do not set frequency.
5567 (drop_profile): Simplify.
5568 (counts_to_freqs): Simplify.
5569 (expensive_function_p): Use to_frequency.
5570 (propagate_unlikely_bbs_forward): Simplify.
5571 (determine_unlikely_bbs): Simplify.
5572 (estimate_bb_frequencies): Add hack to silence graphite issues.
5573 (compute_function_frequency): Use ipa counts.
5574 (pass_profile::execute): Update.
5575 (rebuild_frequencies): Use counts only.
5576 (force_edge_cold): Use counts only.
5577 * profile-count.c (profile_count::dump): Dump new count types.
5578 (profile_count::differs_from_p): Check compatiblity.
5579 (profile_count::to_frequency): New function.
5580 (profile_count::to_cgraph_frequency): New function.
5581 * profile-count.h (struct function): Declare.
5582 (enum profile_quality): Add profile_guessed_local and
5583 profile_guessed_global0.
5584 (class profile_proability): Decrease number of bits to 29;
5585 update from_reg_br_prob_note and to_reg_br_prob_note.
5586 (class profile_count: Update comment; decrease number of bits
5587 to 61. Check compatibility.
5588 (profile_count::compatible_p): New private member function.
5589 (profile_count::ipa_p): New member function.
5590 (profile_count::operator<): Handle global zero correctly.
5591 (profile_count::operator>): Handle global zero correctly.
5592 (profile_count::operator<=): Handle global zero correctly.
5593 (profile_count::operator>=): Handle global zero correctly.
5594 (profile_count::nonzero_p): New member function.
5595 (profile_count::force_nonzero): New member function.
5596 (profile_count::max): New member function.
5597 (profile_count::apply_scale): Handle IPA scalling.
5598 (profile_count::guessed_local): New member function.
5599 (profile_count::global0): New member function.
5600 (profile_count::ipa): New member function.
5601 (profile_count::to_frequency): Declare.
5602 (profile_count::to_cgraph_frequency): Declare.
5603 * profile.c (OVERLAP_BASE): Delete.
5604 (compute_frequency_overlap): Delete.
5605 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5606 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5607 * sched-ebb.c (rank): Use counts only.
5608 * shrink-wrap.c (handle_simple_exit): Use counts only.
5609 (try_shrink_wrapping): Use counts only.
5610 (place_prologue_for_one_component): Use counts only.
5611 * tracer.c (find_best_predecessor): Use to_frequency.
5612 (find_trace): Use to_frequency.
5613 (tail_duplicate): Use to_frequency.
5614 * trans-mem.c (expand_transaction): Do not update frequency.
5615 * tree-call-cdce.c: Do not update frequency.
5616 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5617 (gimple_merge_blocks): Likewise.
5618 (gimple_split_edge): Likewise.
5619 (gimple_duplicate_sese_region): Likewise.
5620 (gimple_duplicate_sese_tail): Likewise.
5621 (move_sese_region_to_fn): Likewise.
5622 (gimple_account_profile_record): Likewise.
5623 (insert_cond_bb): Likewise.
5624 * tree-complex.c (expand_complex_div_wide): Likewise.
5625 * tree-eh.c (lower_resx): Update profile.
5626 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5628 (initialize_cfun): Do not initialize frequencies
5629 (freqs_to_counts): Delete.
5630 (copy_cfg_body): Ignore count parameter.
5631 (copy_body): Update.
5632 (expand_call_inline): Update count_max.
5633 (optimize_inline_calls): Update count_max.
5634 (tree_function_versioning): Update count_max.
5635 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5636 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5638 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5639 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5640 (try_peel_loop): Likewise.
5641 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5643 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5644 (tree_transform_and_unroll_loop): Do not use frequencies
5645 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5646 Use reliable prediction only.
5647 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5648 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5649 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5650 probability scaling.
5651 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5652 not update frequency
5653 (any_remaining_duplicated_blocks): Likewise.
5654 (update_profile): Likewise.
5655 (estimated_freqs_path): Delete.
5656 (freqs_to_counts_path): Delete.
5657 (clear_counts_path): Delete.
5658 (ssa_fix_duplicate_block_edges): Likewise.
5659 (duplicate_thread_path): Likewise.
5660 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5661 * tree-tailcall.c (decrease_profile): Do not update frequency.
5662 (eliminate_tail_call): Likewise.
5663 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5664 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5665 (optimize_mask_stores): Likewise.
5666 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5667 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5668 (ubsan_expand_ptr_ifn): Update profile.
5669 * value-prof.c (gimple_ic): Simplify.
5670 * value-prof.h (gimple_ic): Update prototype.
5671 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5672 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5674 (want_inline_self_recursive_call_p): Likewise.
5675 (resolve_noninline_speculation): Only cummulate defined counts.
5676 (inline_small_functions): Use nonzero_p.
5677 (ipa_inline): Do not access freed node.
5679 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5681 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5682 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5684 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5686 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5687 non-legitimate address.
5689 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5691 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5692 (*lt0_<mode>di, *lt0_<mode>si): New.
5694 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5696 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5697 TARGET_PAIRED_FLOAT.
5699 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5700 Jim Wilson <jim.wilson@linaro.org>
5702 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5703 * config/aarch64/aarch64-tune.md: Regenerated.
5704 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5705 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5707 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5709 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5710 (arc_expand_prologue): Restore blink for millicode.
5711 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5713 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5716 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5717 gen_vec_duplicate after forcing the scalar into a register.
5719 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5721 * combine (try_combine): Print the insns input to try_combine to the
5724 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5727 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5728 Remove second argument from aarch64_process_target_attr call.
5729 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5731 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5733 (aarch64_handle_attr_arch): Remove second argument.
5734 (aarch64_handle_attr_cpu): Ditto.
5735 (aarch64_handle_attr_tune): Ditto.
5736 (aarch64_handle_attr_isa_flags): Ditto.
5737 (aarch64_process_one_target_attr): Ditto.
5738 (aarch64_process_target_attr): Ditto.
5739 (aarch64_option_valid_attribute_p): Remove second argument.
5740 on aarch64_process_target_attr call.
5742 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5744 * diagnostic.c: Include "selftest-diagnostic.h".
5745 (selftest::assert_location_text): New function.
5746 (selftest::test_diagnostic_get_location_text): New function.
5747 (selftest::diagnostic_c_tests): Call it.
5749 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5751 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5752 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5753 (class selftest::test_diagnostic_context): Move to...
5754 * selftest-diagnostic.c: New file.
5755 * selftest-diagnostic.h: New file.
5757 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5759 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5760 offset range for FT32B.
5761 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5762 * config/ft32/ft32.md: Add TARGET_NOPM.
5763 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5764 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5766 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5768 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5770 2017-11-02 Jeff Law <law@redhat.com>
5772 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5773 virtual keyword on FINAL OVERRIDE members.
5775 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5776 virtuals together. Add virtual destructor.
5777 (substitute_and_fold_engine): Similarly.
5779 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5781 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5783 2017-11-02 Richard Biener <rguenther@suse.de>
5785 PR tree-optimization/82795
5786 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5788 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5790 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5792 (gcc_GAS_CHECK_FEATURE): Remove.
5793 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5794 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5795 * configure: Regenerate.
5797 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5799 * config/arc/arc.c (hwloop_optimize): Account for empty
5802 2017-11-02 Richard Biener <rguenther@suse.de>
5805 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5806 Truncate ARRAY_REF index and element size.
5808 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5810 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5812 2017-11-01 Jeff Law <law@redhat.com>
5814 * tree-ssa-ccp.c (ccp_folder): New class derived from
5815 substitute_and_fold_engine.
5816 (ccp_folder::get_value): New member function.
5817 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5818 (ccp_fold_stmt): Remove prototype.
5819 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5820 * tree-ssa-copy.c (copy_folder): New class derived from
5821 substitute_and_fold_engine.
5822 (copy_folder::get_value): Renamed from get_value.
5823 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5824 * tree-vrp.c (vrp_folder): New class derived from
5825 substitute_and_fold_engine.
5826 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5827 (vrp_folder::get_value): New member function.
5828 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5829 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5830 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5831 provide a class interface to folder/substitute routines.
5832 (ssa_prop_fold_stmt_fn): Remove typedef.
5833 (ssa_prop_get_value_fn): Likewise.
5834 (subsitute_and_fold): Remove prototype.
5835 (replace_uses_in): Likewise.
5836 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5837 Renamed from replace_uses_in. Call the virtual member function
5838 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5839 (substitute_and_fold_dom_walker): Remove initialization of
5840 data member entries for calbacks. Add substitute_and_fold_engine
5841 member and initialize it.
5842 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5843 member functions for get_value, replace_phi_args_in c
5844 replace_uses_in, and fold_stmt calls.
5845 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5846 substitute_and_fold. Remove assert. Update ctor call.
5848 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5849 (ssa_prop_visit_phi_fn): Likewise.
5850 (class ssa_propagation_engine): New class to provide an interface
5852 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5854 (ssa_prop_visit_phi): Likewise.
5855 (ssa_propagation_engine::simulate_stmt): Moved into class.
5856 Call visit_phi/visit_stmt from the class rather than via
5857 file scoped static variables.
5858 (ssa_propagation_engine::simulate_block): Moved into class.
5859 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5860 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5861 set file scoped statics for the visit_stmt/visit_phi callbacks.
5862 * tree-complex.c (complex_propagate): New class derived from
5863 ssa_propagation_engine.
5864 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5865 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5866 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5868 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5869 ssa_propagation_engine.
5870 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5871 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5872 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5873 * tree-ssa-copy.c (copy_prop): New class derived from
5874 ssa_propagation_engine.
5875 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5876 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5877 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5878 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5879 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5880 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5881 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5883 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5885 PR rtl-optimization/82778
5886 PR rtl-optimization/82597
5887 * compare-elim.c (struct comparison): Add in_a_setter field.
5888 (find_comparison_dom_walker::before_dom_children): Remove killed
5889 bitmap and df_simulate_find_defs call, instead walk the defs.
5890 Compute last_setter and initialize in_a_setter. Merge definitions
5891 with first initialization for a few variables.
5892 (try_validate_parallel): Use insn_invalid_p instead of
5893 recog_memoized. Return insn rather than just the pattern.
5894 (try_merge_compare): Fix up comment. Don't uselessly test if
5895 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5897 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5900 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5901 Alan Hayward <alan.hayward@arm.com>
5902 David Sherwood <david.sherwood@arm.com>
5904 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5905 aarch64_hard_regno_nregs to get the number of registers
5908 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5909 Alan Hayward <alan.hayward@arm.com>
5910 David Sherwood <david.sherwood@arm.com>
5912 * config/aarch64/constraints.md (Upl): Rename to...
5914 * config/aarch64/aarch64.md
5915 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5918 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5919 Alan Hayward <alan.hayward@arm.com>
5920 David Sherwood <david.sherwood@arm.com>
5922 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
5923 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
5926 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5927 Alan Hayward <alan.hayward@arm.com>
5928 David Sherwood <david.sherwood@arm.com>
5930 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
5931 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
5932 (aarch64_evpc_dup): Generate rtl direcly, rather than using
5934 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
5935 of a single element.
5936 * config/aarch64/iterators.md: Add a comment above the permute
5937 unspecs to say that they are generated directly by
5938 aarch64_expand_vec_perm_const.
5939 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
5941 2017-11-01 Nathan Sidwell <nathan@acm.org>
5943 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
5945 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5947 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
5948 and medany code models, and describe what they do.
5950 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5952 Revert accidental duplicate:
5954 * combine.c (can_change_dest_mode): Reject changes in
5955 REGMODE_NATURAL_SIZE.
5957 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
5959 PR rtl-optimization/64682
5960 PR rtl-optimization/69567
5961 PR rtl-optimization/69737
5962 PR rtl-optimization/82683
5963 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
5964 register mentioned in the note, drop the note, unless it came from I3,
5965 in which case it should go to I3 again.
5967 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5969 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
5970 and return false if not.
5971 (clear_bytes_written_by, live_bytes_read): Update accordingly.
5973 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5975 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
5976 range is known to be empty.
5978 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5979 Alan Hayward <alan.hayward@arm.com>
5980 David Sherwood <david.sherwood@arm.com>
5982 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
5983 and CONST_VECTOR_NUNITS instead of computing the number of units from
5984 the byte sizes of the vector and element.
5985 (simplify_binary_operation_1): Likewise.
5986 (simplify_const_binary_operation): Likewise.
5987 (simplify_ternary_operation): Likewise.
5989 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5990 Alan Hayward <alan.hayward@arm.com>
5991 David Sherwood <david.sherwood@arm.com>
5993 * var-tracking.c (INT_MEM_OFFSET): Replace with...
5994 (int_mem_offset): ...this new function.
5995 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
5996 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
5997 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6000 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6001 Alan Hayward <alan.hayward@arm.com>
6002 David Sherwood <david.sherwood@arm.com>
6004 * lower-subreg.c (interesting_mode_p): New function.
6005 (compute_costs, find_decomposable_subregs, decompose_register)
6006 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6007 (resolve_clobber, dump_choices): Use it.
6009 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6010 Alan Hayward <alan.hayward@arm.com>
6011 David Sherwood <david.sherwood@arm.com>
6013 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6015 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6016 Alan Hayward <alan.hayward@arm.com>
6017 David Sherwood <david.sherwood@arm.com>
6019 * alias.c (find_base_value, find_base_term): Only process integer
6020 truncations. Check the precision rather than the size.
6022 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6023 Alan Hayward <alan.hayward@arm.com>
6024 David Sherwood <david.sherwood@arm.com>
6026 * machmode.h (is_narrower_int_mode): New function
6027 * optabs.c (expand_float, expand_fix): Use it.
6028 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6030 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6031 Alan Hayward <alan.hayward@arm.com>
6032 David Sherwood <david.sherwood@arm.com>
6034 * rtl.h (narrower_subreg_mode): New function.
6035 * ira-color.c (update_costs_from_allocno): Use it.
6037 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6038 Alan Hayward <alan.hayward@arm.com>
6039 David Sherwood <david.sherwood@arm.com>
6041 * optabs-query.h (convert_optab_p): New function, split out from...
6042 (convert_optab_handler): ...here.
6043 (widening_optab_handler): Delete.
6044 (find_widening_optab_handler): Remove permit_non_widening parameter.
6045 (find_widening_optab_handler_and_mode): Likewise. Provide an
6046 override that operates on mode class wrappers.
6047 * optabs-query.c (widening_optab_handler): Delete.
6048 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6049 parameter. Assert that the two modes are the same class and that
6050 the "from" mode is narrower than the "to" mode. Use
6051 convert_optab_handler instead of widening_optab_handler.
6052 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6053 instead of widening_optab_handler.
6054 * expr.c (expand_expr_real_2): Update calls to
6055 find_widening_optab_handler.
6056 * optabs.c (expand_widen_pattern_expr): Likewise.
6057 (expand_binop_directly): Take the insn_code as a parameter.
6058 (expand_binop): Only call find_widening_optab_handler for
6059 conversion optabs; use optab_handler otherwise. Update calls
6060 to find_widening_optab_handler and expand_binop_directly.
6061 Use convert_optab_handler instead of widening_optab_handler.
6062 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6063 find_widening_optab_handler and use scalar_mode rather than
6065 (convert_plusminus_to_widen): Likewise.
6067 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6068 Alan Hayward <alan.hayward@arm.com>
6069 David Sherwood <david.sherwood@arm.com>
6071 * machmode.h (fixed_size_mode): New class.
6072 * rtl.h (get_pool_mode): Return fixed_size_mode.
6073 * gengtype.c (main): Add fixed_size_mode.
6074 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6075 (get_raw_arg_mode): Likewise.
6076 * doc/tm.texi: Regenerate.
6077 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6078 * targhooks.c (default_get_reg_raw_mode): Likewise.
6079 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6080 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6081 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6082 (msp430_get_raw_result_mode): Likewise.
6083 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6084 * dbxout.c (dbxout_parms): Require fixed-size modes.
6085 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6086 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6087 * omp-low.c (lower_oacc_reductions): Likewise.
6088 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6089 (simplify_subreg): Update accordingly.
6090 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6091 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6092 that aren't fixed-size.
6093 (get_pool_mode): Return a fixed_size_mode.
6094 (output_constant_pool_2): Take a fixed_size_mode.
6096 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6097 Alan Hayward <alan.hayward@arm.com>
6098 David Sherwood <david.sherwood@arm.com>
6100 * doc/rtl.texi (vec_series): Document.
6101 (const): Say that the operand can be a vec_series.
6102 * rtl.def (VEC_SERIES): New rtx code.
6103 * rtl.h (const_vec_series_p_1): Declare.
6104 (const_vec_series_p): New function.
6105 * emit-rtl.h (gen_const_vec_series): Declare.
6106 (gen_vec_series): Likewise.
6107 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6108 (gen_vec_series): Likewise.
6109 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6110 * simplify-rtx.c (simplify_unary_operation): Handle negations
6112 (simplify_binary_operation_series): New function.
6113 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6114 (test_vector_ops_series): New function.
6115 (test_vector_ops): Call it.
6116 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6117 gen_const_vec_series.
6118 (altivec_lvsr): Likewise.
6119 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6121 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6122 Alan Hayward <alan.hayward@arm.com>
6123 David Sherwood <david.sherwood@arm.com>
6125 * doc/rtl.texi (const): Update description of address constants.
6126 Say that vector constants are allowed too.
6127 * common.md (E, F): Use CONSTANT_P instead of checking for
6129 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6130 checking for CONST_VECTOR.
6131 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6133 * expr.c (expand_expr_real_2): Check for vector modes instead
6134 of checking for CONST_VECTOR.
6135 * rtl.h (const_vec_p): New function.
6136 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6137 (unwrap_const_vec_duplicate): Handle them here too.
6139 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6140 David Malcolm <dmalcolm@redhat.com>
6141 Alan Hayward <alan.hayward@arm.com>
6142 David Sherwood <david.sherwood@arm.com>
6144 * rtl.h (vec_duplicate_p): New function.
6145 * selftest-rtl.c (assert_rtx_eq_at): New function.
6146 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6147 (assert_rtx_eq_at): Declare.
6148 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6149 * selftest-run-tests.c (selftest::run_tests): Call it.
6150 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6151 (simplify_unary_operation_1): Recursively handle vector duplicates.
6152 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6154 (simplify_subreg): Handle subregs of vector duplicates.
6155 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6156 (selftest::simplify_rtx_c_tests): New functions.
6158 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6159 Alan Hayward <alan.hayward@arm.com>
6160 David Sherwood <david.sherwood@arm.com>
6162 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6163 (gen_vec_duplicate): Likewise.
6164 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6166 (gen_const_vector): ...here.
6167 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6168 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6169 whose elements are all equal.
6170 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6171 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6172 (simplify_relational_operation): Likewise.
6173 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6175 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6176 (aarch64_expand_vector_init): Likewise.
6177 * config/arm/arm.c (neon_vdup_constant): Likewise.
6178 (neon_expand_vector_init): Likewise.
6179 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6180 (arm_block_set_unaligned_vect): Likewise.
6181 (arm_block_set_aligned_vect): Likewise.
6182 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6183 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6184 (expand_vec_perm_even_odd_pack): Likewise.
6185 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6186 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6187 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6188 gen_const_vec_duplicate.
6189 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6190 * config/mips/mips.c (mips_gen_const_int_vector): Use
6191 gen_const_vec_duplicate.
6192 (mips_expand_vector_init): Use CONST0_RTX.
6193 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6194 (define_split): Use gen_const_vec_duplicate.
6195 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6196 (define_split): Use gen_const_vec_duplicate.
6197 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6198 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6199 * config/spu/spu.c (spu_const): Likewise.
6201 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6202 Alan Hayward <alan.hayward@arm.com>
6203 David Sherwood <david.sherwood@arm.com>
6205 * combine.c (can_change_dest_mode): Reject changes in
6206 REGMODE_NATURAL_SIZE.
6208 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6210 * configure.ac (--enable-libssp): New.
6211 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6212 trying to determine target-specific default. Adjust indentation.
6213 * configure: Regenerated.
6214 * doc/install.texi (Configuration): Expand --disable-libssp
6217 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6219 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6222 2017-10-31 Martin Jambor <mjambor@suse.cz>
6225 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6227 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6229 * auto-profile.c (autofdo_source_profile::read): Use
6230 UNKNOWN_LOCATION rather than 0.
6231 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6232 (warning_at): ...this overload.
6233 (warning_at_rich_loc_n): Rename to...
6234 (warning_n): ...this overload.
6235 (error_at_rich_loc): Rename to...
6236 (error_at): ...this overload.
6237 (pedwarn_at_rich_loc): Rename to...
6238 (pedwarn): ...this overload.
6239 (permerror_at_rich_loc): Rename to...
6240 (permerror): ...this overload.
6241 (inform_at_rich_loc): Rename to...
6242 (inform): ...this overload.
6243 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6244 (diagnostic_n_impl_richloc): Rename to...
6245 (diagnostic_n_impl): ...this rich_location *-based decl.
6246 (inform_at_rich_loc): Rename to...
6247 (inform): ...this, and add an assertion.
6248 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6249 (warning_at_rich_loc): Rename to...
6250 (warning_at): ...this, and add an assertion.
6251 (warning_at_rich_loc_n): Rename to...
6252 (warning_n): ...this, and add an assertion.
6253 (warning_n): Update location_t-based implementation for removal of
6254 location_t-based diagnostic_n_impl.
6255 (pedwarn_at_rich_loc): Rename to...
6256 (pedwarn): ...this, and add an assertion.
6257 (permerror_at_rich_loc): Rename to...
6258 (permerror): ...this, and add an assertion.
6259 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6260 (error_at_rich_loc): Rename to...
6261 (error_at): ...this, and add an assertion.
6262 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6263 (driver::do_spec_on_infiles): Likewise.
6264 * substring-locations.c (format_warning_va): Update for renaming
6265 of inform_at_rich_loc.
6267 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6269 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6270 _Float<N>X built-in functions so that the variant without the
6271 "__builtin_" prefix is only enabled for the GNU C and Objective C
6272 languages when they are in non-strict ANSI/ISO mode.
6273 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6274 * target.def (floatn_builtin_p): Add a target hook to control
6275 whether _Float<N> and _Float<N>X built-in functions without the
6276 "__builtin_" prefix are enabled, and return true for C and
6277 Objective C in the default hook. Include langhooks.h in
6279 * targhooks.h (default_floatn_builtin_p): Likewise.
6280 * targhooks.c (default_floatn_builtin_p): Likewise.
6281 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6282 floatn_builtin_p target hook.
6283 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6285 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6286 Eric Botcazou <ebotcazou@adacore.com>
6288 PR rtl-optimization/81803
6289 * lra-constraints.c (curr_insn_transform): Also reload the whole
6290 register for a strict subreg no wider than a word if this is for
6291 a WORD_REGISTER_OPERATIONS target.
6293 2017-10-31 Jason Merrill <jason@redhat.com>
6295 * gdbinit.in: Skip over inlines from timevar.h.
6297 2017-10-31 Martin Liska <mliska@suse.cz>
6299 * doc/gcov.texi: Document new option.
6300 * gcov.c (print_usage): Likewise print it.
6301 (process_args): Support the argument.
6302 (format_count): New function.
6303 (format_gcov): Use the function.
6305 2017-10-31 Martin Liska <mliska@suse.cz>
6307 * gcov.c (struct name_map): do not use typedef.
6308 Define operator== and operator<.
6309 (name_search): Remove.
6310 (name_sort): Remove.
6311 (main): Do not allocate names.
6312 (process_file): Add vertical space.
6313 (generate_results): Use std::find.
6314 (release_structures): Do not release memory.
6315 (find_source): Use std::find.
6317 2017-10-31 Martin Liska <mliska@suse.cz>
6319 * gcov.c (struct line_info): Remove it's typedef.
6320 (line_info::line_info): Add proper ctor.
6321 (line_info::has_block): Do not use a typedef.
6322 (struct source_info): Do not use typedef.
6323 (circuit): Likewise.
6324 (get_cycles_count): Likewise.
6325 (output_intermediate_file): Iterate via vector iterator.
6326 (add_line_counts): Use std::vector methods.
6327 (accumulate_line_counts): Likewise.
6328 (output_lines): Likewise.
6330 2017-10-31 Martin Liska <mliska@suse.cz>
6332 * gcov.c (struct source_info): Remove typedef.
6333 (source_info::source_info): Add proper ctor.
6334 (accumulate_line_counts): Use struct, not it's typedef.
6335 (output_gcov_file): Likewise.
6336 (output_lines): Likewise.
6337 (main): Do not allocate an array.
6338 (output_intermediate_file): Use size of vector container.
6339 (process_file): Resize the vector.
6340 (generate_results): Do not preallocate, use newly added vector
6342 (release_structures): Do not release sources.
6343 (find_source): Use vector methods.
6344 (add_line_counts): Do not use typedef.
6346 2017-10-31 Martin Liska <mliska@suse.cz>
6348 * doc/gcov.texi: Document that.
6349 * gcov.c (add_line_counts): Mark lines with a non-executed
6351 (output_line_beginning): Handle such lines.
6352 (output_lines): Pass new argument.
6353 (output_intermediate_file): Print it in intermediate format.
6355 2017-10-31 Martin Liska <mliska@suse.cz>
6357 * color-macros.h: New file.
6358 * diagnostic-color.c: Factor out color related to macros to
6360 * doc/gcov.texi: Document -k option.
6361 * gcov.c (INCLUDE_STRING): Include string.h.
6362 (print_usage): Add -k option.
6363 (process_args): Parse it.
6364 (pad_count_string): New function.
6365 (output_line_beginning): Likewise.
6366 (DEFAULT_LINE_START): New macro.
6367 (output_lines): Support color output.
6369 2017-10-31 Martin Liska <mliska@suse.cz>
6371 PR gcov-profile/82633
6372 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6373 their interaction with GCOV infrastructure.
6374 * configure.ac: Add -fkeep-{inline,static}-functions to
6376 * configure: Regenerate.
6378 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6381 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6383 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6386 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6387 into a register if it does not fit into an immediate offset field.
6389 2017-10-31 Olivier Hainque <hainque@adacore.com>
6391 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6393 2017-10-31 Julia Koval <julia.koval@intel.com>
6395 * config.gcc: Add gfniintrin.h.
6396 * config/i386/gfniintrin.h: New.
6397 * config/i386/i386-builtin-types.def
6398 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6399 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6400 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6401 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6402 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6403 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6404 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6405 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6406 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6407 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6408 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6409 * config/i386/immintrin.h: Include gfniintrin.h.
6410 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6412 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6414 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6416 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6418 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6419 (arm_ashldi3_1bit): Remove pattern.
6420 (ashrdi3): Remove shift by 1 expansion.
6421 (arm_ashrdi3_1bit): Remove pattern.
6422 (lshrdi3): Remove shift by 1 expansion.
6423 (arm_lshrdi3_1bit): Remove pattern.
6424 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6425 cost of ashldi3 by 1.
6426 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6427 (<shift>di3_neon): Likewise.
6429 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6431 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6432 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6433 and (*aarch64_simd_mov<VQ:mode>).
6434 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6435 pattern alternative.
6436 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6437 attributes to match pattern alternative.
6439 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6441 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6442 * config/rs6000/emmintrin.h: New file.
6443 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6445 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6447 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6448 * (movdi_vfp_cortexa8): Remove pattern.
6450 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6452 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6454 (Specific, alpha*-dec-osf5.1): Remove.
6455 (Specific, mips-sgi-irix5): Remove.
6456 (Specific, mips-sgi-irix6): Remove.
6458 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6461 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6462 arguments to clear_bit_region_be.
6464 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6466 * gimplify.c: Include memmodel.h.
6468 2017-10-30 Martin Jambor <mjambor@suse.cz>
6470 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6471 condition whether loop should be executed at all.
6473 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6475 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6476 gimple folding of vec_madd() intrinsics.
6477 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6478 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6479 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6481 2017-10-30 Richard Biener <rguenther@suse.de>
6483 PR tree-optimization/82762
6485 2017-10-23 Richard Biener <rguenther@suse.de>
6487 PR tree-optimization/82129
6489 2017-08-01 Richard Biener <rguenther@suse.de>
6491 PR tree-optimization/81181
6492 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6493 (compute_antic): ... end of iteration here.
6495 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6497 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6498 -std=iso9899:2017 and -std=gnu17.
6499 * doc/standards.texi (C Language): Document C17 support.
6500 * doc/cpp.texi (Overview): Mention -std=c17.
6501 (Standard Predefined Macros): Document C11 and C17 values of
6502 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6503 * doc/extend.texi (Inline): Do not list individual options for
6504 standards newer than C99.
6505 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6507 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6510 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6512 * asan.c (asan_finish_file): Align asan globals array by shadow
6515 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6518 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6519 (struct store_immediate_info): Add bitregion_start and bitregion_end
6521 (store_immediate_info::store_immediate_info): Add brs and bre
6522 arguments and initialize bitregion_{start,end} from those.
6523 (struct merged_store_group): Add bitregion_start, bitregion_end,
6524 align_base and mask fields. Drop unnecessary struct keyword from
6525 struct store_immediate_info. Add do_merge method.
6526 (clear_bit_region_be): Use memset instead of loop storing zeros.
6527 (merged_store_group::do_merge): New method.
6528 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6529 stores as long as the surrounding bitregions have no gaps.
6530 (merged_store_group::merge_overlapping): Use do_merge.
6531 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6532 is byte aligned, rather than requiring that start and width are
6533 byte aligned. Drop unnecessary struct keyword from
6534 struct store_immediate_info. Allocate and populate also mask array.
6535 Make start of the arrays relative to bitregion_start rather than
6536 start and size them according to bitregion_{end,start} difference.
6537 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6538 struct store_immediate_info.
6539 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6540 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6541 struct keyword from struct store_immediate_info.
6542 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6543 between stores as long as the surrounding bitregions have no gaps.
6545 (struct split_store): Add orig non-static data member.
6546 (split_store::split_store): Initialize orig to false.
6547 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6548 if there is exactly a single original stmt. Change stmts argument
6549 to pointer from reference, if NULL, don't push anything to it. Add
6550 first argument, use it to optimize skipping over orig stmts that
6551 are known to be before bitpos already. Simplify.
6552 (split_group): Return unsigned int count how many stores are or
6553 would be needed rather than a bool. Add allow_unaligned argument.
6554 Change split_stores argument from reference to pointer, if NULL,
6555 only do a dry run computing how many stores would be produced.
6556 Rewritten algorithm to use both alignment and misalign if
6557 !allow_unaligned and handle bitfield stores with gaps.
6558 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6559 from bitregion_start instead of start. Compute allow_unaligned
6560 here, if true, do 2 split_group dry runs to compute which one
6561 produces fewer stores and prefer aligned if equal. Punt if
6562 new count is bigger or equal than original before emitting any
6563 statements, rather than during that. Remove no longer needed
6564 new_ssa_names tracking. Replace num_stmts with
6565 split_stores.length (). Use 32-bit stack allocated entries
6566 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6567 if possible. Handle bitfields with gaps.
6568 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6569 Compute bitregion_{start,end} for the stores and construct
6570 store_immediate_info with that. Formatting fixes.
6572 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6575 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6576 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6578 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6580 * gimplify.c: Include tm_p.h.
6582 * common.opt (gcoff): Re-add as ignored option.
6583 (gcoff1, gcoff2, gcoff3): Likewise.
6585 * Makefile.in (OBJS): Delete sdbout.o.
6586 (GTFILES): Delete $(srcdir)/sdbout.c.
6587 * debug.h: Delete sdb_debug_hooks.
6588 * final.c: Delete sdbout.h include.
6589 (final_scan_insn): Delete SDB_DEBUG check.
6590 (rest_of_clean_state): Likewise.
6591 * output.h: Delete sdb_begin_function_line.
6594 * toplev.c: Delete sdbout.h include.
6595 (process_options): Delete SDB_DEBUG check.
6596 * tree-core.h (tree_type_common): Delete pointer field of
6598 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6599 TYPE_SYMTAB_POINTER.
6600 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6601 (TYPE_SYMTAB_IS_POINTER): Delete.
6602 (TYPE_SYMTAB_IS_DIE): Renumber.
6603 * xcoffout.c: Refer to former sdbout.c file.
6604 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6606 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6607 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6608 (Debugging Options): Delete -gcoff.
6609 (-gstabs): Delete SDB reference.
6611 (-gcoff@var{level}): Delete.
6612 * doc/passes.texi (Debugging information output): Delete SDB and
6613 sdbout.c references.
6614 * doc/tm.texi: Regenerate.
6615 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6616 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6618 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6619 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6620 (SDB_DEBUGGING_INFO): Delete.
6621 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6622 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6623 * target.def (output_source_filename): Delete COFF reference.
6625 * common.opt (gcoff): Delete.
6626 (gxcoff+): Update Negative chain.
6627 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6629 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6630 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6631 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6633 (expand_function_start): Change sdb reference to past tense.
6634 (expand_function_end): Change sdb reference to past tense.
6635 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6636 * opts.c (debug_type_names): Delete coff entry.
6637 (common_handle_option): Delete OPT_gcoff case.
6638 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6640 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6641 * config/cris/cris.h: Delete SDB reference in comment.
6642 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6643 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6644 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6645 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6647 (ix86_expand_prologue): Likewise.
6648 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6649 * config/ia64/ia64.h: Likewise.
6650 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6651 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6653 * config/mmix/mmix.h: Likewise.
6654 * config/nds32/nds32.c: Likewise.
6655 * config/stormy/storym16.h: Likewise.
6656 * config/visium/visium.h: Likewise.
6657 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6659 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6661 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6662 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6663 FRAME_POINTER_REGNUM point at high end of local var area.
6665 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6667 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6668 Move comment around. Do not reset best_edge for a copiable
6669 destination if the copy would cause a partition change.
6670 (better_edge_p): Remove redundant check.
6672 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6674 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6676 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6678 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6679 for math functions that have _Float<N> and _Float<N>X variants.
6680 (mathfn_built_in_2): Add support for math functions that have
6681 _Float<N> and _Float<N>X variants.
6682 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6683 (expand_builtin_mathfn_ternary): Add support for fma with
6684 _Float<N> and _Float<N>X variants.
6685 (expand_builtin): Likewise.
6686 (fold_builtin_3): Likewise.
6687 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6688 create math function _Float<N> and _Float<N>X variants as external
6690 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6691 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6692 the __builtin_ prefix and if not strict ansi, without the prefix.
6693 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6694 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6695 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6696 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6697 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6698 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6699 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6700 function signatures for fma _Float<N> and _Float<N>X variants.
6701 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6702 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6703 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6704 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6705 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6706 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6707 * gencfn-macros.c (print_case_cfn): Add support for math functions
6708 that have _Float<N> and _Float<N>X variants.
6709 (print_define_operator_list): Likewise.
6710 (fltfn_suffixes): Likewise.
6712 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6713 for math functions that have _Float<N> and _Float<N>X variants.
6714 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6715 and _Float<N>X variants.
6716 (COPYSIGN): Likewise.
6719 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6720 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6722 (integer_valued_read_call_p): Likewise.
6723 * fold-const-call.c (fold_const_call_ss): Likewise.
6724 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6725 _Float<N> and _Float<N>X variants.
6726 (fold_const_call_ssss): Add support for fma _Float<N> and
6727 _Float<N>X variants.
6728 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6729 support for copysign and fma _Float<N> and _Float<N>X variants.
6730 (backprop::process_builtin_call_use): Likewise.
6731 * tree-call-cdce.c (can_test_argument_range); Add support for
6732 sqrt _Float<N> and _Float<N>X variants.
6733 (edom_only_function): Likewise.
6734 (get_no_error_domain): Likewise.
6735 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6736 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6737 copysign _Float<N> and _Float<N>X variants.
6738 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6739 handled by machine independent code.
6740 (FMAF128): Likewise.
6741 * doc/cpp.texi (Common Predefined Macros): Document defining
6742 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6743 fma _Float<N> and _Float<N>X variants.
6745 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6748 * config/i386/i386-modes.def (CCFPU): Remove definition.
6749 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6750 (ix86_cc_modes_compatible): Ditto.
6751 (ix86_expand_carry_flag_compare): Ditto.
6752 (ix86_expand_int_movcc): Ditto.
6753 (ix86_expand_int_addcc): Ditto.
6754 (ix86_reverse_condition): Ditto.
6755 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6756 Return true/false for unordered/ordered fp comparisons.
6757 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6758 (ix86_prepare_fp_compare_args): Update for rename.
6759 (ix86_expand_fp_compare): Update for rename. Generate unordered
6760 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6761 (ix86_expand_sse_compare_and_jump): Ditto.
6762 * config/i386/predicates.md (fcmov_comparison_operator):
6763 Remove CCFPU mode handling.
6764 (ix86_comparison_operator): Ditto.
6765 (ix86_carry_flag_operator): Ditto.
6766 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6767 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6768 (*cmpu<mode>_cc_i387): Ditto.
6769 (FPCMP): Remove mode iterator.
6770 (unord): Remove mode attribute.
6771 (unord_subst): New define_subst transformation
6772 (unord): New define_subst attribute.
6774 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6775 (*cmpi<unord>xf_i387): Ditto.
6776 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6777 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6778 using unord_subst transformation.
6779 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6780 (round_saeonly): Also handle CCFP mode.
6781 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6782 Remove UNSPEC_SAHF unspec handling.
6784 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6786 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6788 2017-10-27 Jeff Law <law@redhat.com>
6790 * gimple-ssa-sprintf.c: Include domwalk.h.
6791 (class sprintf_dom_walker): New class, derived from dom_walker.
6792 (sprintf_dom_walker::before_dom_children): New function.
6793 (struct call_info): Moved into sprintf_dom_walker class
6794 (compute_formath_length, handle_gimple_call): Likewise.
6795 (sprintf_length::execute): Call the dominator walker rather
6796 than walking the statements.
6798 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6799 gimple statement locations.
6800 (check_array_bounds): Corresponding changes. Get the statement's
6801 location directly from wi->stmt.
6803 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6806 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6808 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6810 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6811 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6813 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6816 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6817 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6818 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6819 maybe_get_pool_constant.
6820 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6823 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6825 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6826 2.26 caveat. Update gas and gld versions.
6827 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6830 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6832 * cgraph.h (set_malloc_flag): Declare.
6833 * cgraph.c (set_malloc_flag_1): New function.
6834 (set_malloc_flag): Likewise.
6835 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6836 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6838 (read_ipa_call_summary): Add support for reading is_return_callee.
6839 (write_ipa_call_summary): Stream is_return_callee.
6840 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6841 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6842 ipa-prop.h, ipa-fnsummary.h.
6843 (pure_const_names): Change to static.
6844 (malloc_state_e): Define.
6845 (malloc_state_names): Define.
6846 (funct_state_d): Add field malloc_state.
6847 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6848 (check_retval_uses): New function.
6849 (malloc_candidate_p): Likewise.
6850 (analyze_function): Add support for malloc attribute.
6851 (pure_const_write_summary): Stream malloc_state.
6852 (pure_const_read_summary): Add support for reading malloc_state.
6853 (dump_malloc_lattice): New function.
6854 (propagate_malloc): New function.
6855 (warn_function_malloc): New function.
6856 (ipa_pure_const::execute): Call propagate_malloc and
6857 ipa_free_fn_summary.
6858 (pass_local_pure_const::execute): Add support for malloc attribute.
6859 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6860 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6862 2017-10-27 Martin Liska <mliska@suse.cz>
6864 PR gcov-profile/82457
6865 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6866 for fork-like functions to be properly instrumented.
6868 2017-10-27 Richard Biener <rguenther@suse.de>
6871 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6872 info when we redirected EH.
6874 2017-10-26 Michael Collison <michael.collison@arm.com>
6876 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6878 (<optab>_trunchf<GPI:mode>2: New pattern.
6879 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6880 * config/aarch64/iterators.md (wv): New mode attribute.
6881 (vf, VF): New mode attributes.
6882 (vgp, VGP): New mode attributes.
6883 (s): Update attribute with SImode and DImode prefixes.
6885 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6887 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6888 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6889 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6890 (nios2_option_overide): Initialize it. Don't allow R0-relative
6891 addressing with PIC.
6892 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6893 (nios2_symbolic_constant_p): Likewise.
6894 (nios2_legitimate_address_p): Likewise.
6895 (nios2_r0rel_section_name_p): New.
6896 (nios2_symbol_ref_in_r0rel_data_p): New.
6897 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6898 (r0rel_constant_p): New.
6899 (nios2_print_operand_address): Handle r0rel_constant_p.
6900 (nios2_cdx_narrow_form_p): Likewise.
6901 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6902 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6903 (Nios II Options): Document -mr0rel-sec.
6905 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6907 * config/nios2/nios2.c: Include xregex.h.
6908 (nios2_gprel_sec_regex): New.
6909 (nios2_option_overide): Initialize it. Don't allow GP-relative
6910 addressing with PIC.
6911 (nios2_small_section_name_p): Check for regex match.
6912 * config/nios2/nios2.opt (mgprel-sec=): New option.
6913 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6914 (Nios II Options): Document -mgprel-sec.
6916 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6918 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6920 2017-10-26 Tom de Vries <tom@codesourcery.com>
6922 PR tree-optimization/82707
6923 * gimple.c (gimple_copy): Fix unsharing of
6924 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
6926 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
6928 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
6929 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
6930 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
6931 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
6932 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
6933 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
6934 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
6935 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
6936 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
6937 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
6938 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
6939 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
6940 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
6941 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
6942 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
6943 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
6944 _mm512_mask_cmpunord_ps_mask): New intrinsics.
6946 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6948 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
6950 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
6951 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
6952 warning to rs6000.c. Remove the Undocumented flag, since it has
6954 (-mabi=ibmlongdouble): Likewise.
6955 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
6956 already set, set the default format for long double.
6957 (rs6000_debug_reg_global): Print whether long double is IBM or
6959 (rs6000_option_override_internal): Rework setting long double
6960 format. Only warn if the user is changing the long double default
6961 and they did not use -Wno-psabi.
6962 * doc/invoke.texi (PowerPC options): Update the documentation for
6963 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
6965 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6966 Alan Hayward <alan.hayward@arm.com>
6967 David Sherwood <david.sherwood@arm.com>
6969 * rtl.h (wider_subreg_mode): New function.
6970 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
6971 rather than an unsigned int *.
6972 * ira-color.c (regno_max_ref_width): Replace with...
6973 (regno_max_ref_mode): ...this new variable.
6974 (coalesced_pseudo_reg_slot_compare): Update accordingly.
6975 Use wider_subreg_mode.
6976 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
6977 rather than an unsigned int *.
6978 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
6979 (process_alt_operands): Likewise.
6980 (invariant_p): Likewise.
6981 * lra-spills.c (assign_mem_slot): Likewise.
6982 (add_pseudo_to_slot): Likewise.
6983 * lra.c (collect_non_operand_hard_regs): Likewise.
6984 (add_regs_to_insn_regno_info): Likewise.
6985 * reload1.c (regno_max_ref_width): Replace with...
6986 (regno_max_ref_mode): ...this new variable.
6987 (reload): Update accordingly. Update call to
6988 ira_sort_regnos_for_alter_reg.
6989 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
6990 (init_eliminable_invariants): Update to use regno_max_ref_mode.
6991 (scan_paradoxical_subregs): Likewise.
6993 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6995 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
6996 (aarch64_frame): Add emit_frame_chain boolean.
6997 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6998 Move eh_return case to aarch64_layout_frame.
6999 (aarch64_layout_frame): Initialize emit_frame_chain.
7000 (aarch64_expand_prologue): Use emit_frame_chain.
7002 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7004 * config/aarch64/aarch64.c (aarch64_layout_frame):
7005 Ensure LR is always stored at the bottom of the callee-saves.
7006 Remove rarely used frame layout which saves callee-saves at top of
7007 frame, so the store of LR can be used as a valid probe in all cases.
7009 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7011 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7012 Improve unaligned TImode/TFmode base/offset split.
7014 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7015 Alan Hayward <alan.hayward@arm.com>
7016 David Sherwood <david.sherwood@arm.com>
7018 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7019 * combine.c (find_single_use_1): Likewise.
7020 (expand_field_assignment): Likewise.
7021 (move_deaths): Likewise.
7022 * lra-constraints.c (simplify_operand_subreg): Likewise.
7023 (curr_insn_transform): Likewise.
7024 * lra.c (collect_non_operand_hard_regs): Likewise.
7025 (add_regs_to_insn_regno_info): Likewise.
7026 * rtlanal.c (reg_referenced_p): Likewise.
7027 (covers_regno_no_parallel_p): Likewise.
7029 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7031 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7032 Don't print any bits outside the precision of the value.
7033 * wide-int.cc (test_printing): Add some new tests.
7035 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7037 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7038 supports -xbrace_comment option.
7039 * configure: Regenerate.
7040 * config.in: Regenerate.
7041 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7042 (ASM_CPU_SPEC): Use it.
7044 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7046 * target.def (static_rtx_alignment): New hook.
7047 * targhooks.h (default_static_rtx_alignment): Declare.
7048 * targhooks.c (default_static_rtx_alignment): New function.
7049 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7050 * doc/tm.texi: Regenerate.
7051 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7052 instead of targetm.constant_alignment. Remove call to
7054 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7055 (cris_preferred_mininum_alignment): New function, split out from...
7056 (cris_constant_alignment): ...here.
7057 (cris_static_rtx_alignment): New function.
7058 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7060 (ix86_constant_alignment): ...here.
7061 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7062 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7063 (mmix_static_rtx_alignment): New function.
7064 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7065 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7067 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7070 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7071 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7073 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7075 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7076 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7077 mark as a sign-extending load.
7078 (local_pic_load_u): Define.
7080 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7083 * fold-const.c (operand_equal_for_comparison_p): Also return true
7084 if ARG0 is a simple variant of ARG1 with narrower precision.
7085 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7087 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7089 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7091 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7092 scatter_static, scatter_per_elt.
7093 * x86-tune-costs.h: Add new cost entries.
7095 2017-10-25 Richard Biener <rguenther@suse.de>
7097 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7098 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7099 class pass_fre): Move to ...
7100 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7101 class pass_fre): ... here and adjust for statistics.
7103 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7106 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7107 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7108 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7109 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7112 2017-10-25 Richard Biener <rguenther@suse.de>
7114 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7115 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7116 eliminate_push_avail, eliminate_insert): Move inside...
7117 (class eliminate_dom_walker): ... this class in preparation
7119 (fini_eliminate): Remove by merging with ...
7120 (eliminate): ... this function. Adjust for class changes.
7121 (pass_pre::execute): Remove fini_eliminate call.
7122 (pass_fre::execute): Likewise.
7124 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7127 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7128 (VPERMI2, VPERMI2I): New mode iterators.
7129 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7130 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7132 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7133 mode iterator. Remove 3 old define_insn patterns.
7134 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7135 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7136 VPERMI2 mode iterator, remove the other two expanders.
7137 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7138 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7139 instructions, remove the other two patterns.
7140 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7141 mode iterator, remove the other two patterns.
7142 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7143 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7144 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7145 and adjust argument order accordingly.
7146 (ix86_expand_vec_perm): Adjust caller.
7147 (expand_vec_perm_1): Likewise.
7148 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7149 (expand_vec_perm_vpermt2_vpshub2): ... this.
7150 (ix86_expand_vec_perm_const_1): Adjust caller.
7151 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7154 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7155 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7156 (vec_shl_<mode>): Remove unused expander.
7157 (avx512bw_<shift_insn><mode>3): New define_insn.
7158 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7159 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7161 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7164 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7167 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7169 PR rtl-optimization/82396
7170 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7171 (autopref_multipass_init): Simplify initialization.
7172 (autopref_rank_data): Simplify sort order.
7173 * gcc/sched-int.h (autopref_multipass_data_): Remove
7174 multi_mem_insn_p, min_offset and max_offset.
7176 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7179 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7180 Check special value of flag_omit_frame_pointer.
7181 (aarch64_can_eliminate): Likewise.
7182 (aarch64_override_options_after_change_1): Simplify handling of
7183 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7185 2017-10-24 Richard Biener <rguenther@suse.de>
7187 PR tree-optimization/82697
7188 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7189 zero for conditional load and unconditional store.
7191 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7193 * doc/install.texi: Document bootstrap-cet.
7195 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7198 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7199 ENDBR instruction at function entrance if function is only
7202 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7205 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7206 patterns to better describe from which operation the CF is computed.
7207 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7208 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7209 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7210 is 0, use _0 suffixed expanders instead of emitting a comparison
7213 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7215 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7216 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7218 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7221 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7222 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7223 * loop-iv.c (iv_get_reaching_def): Likewise.
7224 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7225 variable is promoted and the partition contains undefined values.
7227 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7229 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7231 (nios2_address_cost): Define.
7232 (nios2_legitimize_address): Recognize (exp + constant) directly.
7233 (TARGET_ADDRESS_COST): Define.
7235 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7237 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7238 (nios2_symbolic_memory_operand_p): Declare.
7239 (nios2_split_large_constant): Declare.
7240 (nios2_split_symbolic_memory_operand): Declare.
7241 * config/nios2/nios2.c: Adjust includes.
7242 (nios2_symbolic_constant_allowed): New.
7243 (nios2_symbolic_constant_p): New.
7244 (nios2_plus_symbolic_constant_p): New.
7245 (nios2_valid_addr_expr_p): Recognize addresses involving
7247 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7248 (nios2_symbolic_memory_operand_p): New.
7249 (nios2_large_constant_p): New.
7250 (nios2_split_large_constant): New.
7251 (nios2_split_plus_large_constant): New.
7252 (nios2_split_symbolic_memory_operand): New.
7253 (nios2_legitimize_address): Code refactoring. Handle addresses
7254 involving symbolic constants.
7255 (nios2_emit_move_sequence): Likewise.
7256 (nios2_print_operand): Improve error output.
7257 (nios2_print_operand_address): Handle LO_SUM.
7258 (nios2_cdx_narrow_form_p): Likewise.
7259 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7260 operands involving symbolic constants.
7261 (movhi_internal, movsi_internal): Likewise.
7262 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7263 (extendhisi2, extendqi<mode>2): Likewise.
7265 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7267 * tree-pass.h (PROP_rtl_split_insns): Define.
7268 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7270 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7272 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7274 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7277 * target.def (const_not_ok_for_debug_p): Default to
7278 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7279 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7280 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7281 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7282 which targetm.const_not_ok_for_debug_p returned true.
7283 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7285 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7287 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7288 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7289 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7291 (ix86_const_not_ok_for_debug_p): New function.
7292 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7293 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7295 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7298 * system.h: Conditionally include "unique-ptr.h" if
7299 INCLUDE_UNIQUE_PTR is defined.
7300 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7301 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7303 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7305 * config/rl78/rl78.md: New define_expand "subdi3".
7307 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7310 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7311 DF_REF_INSN if DF_REF_INSN_INFO is false.
7313 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7315 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7316 xmm_move instead of sse_move.
7317 (sse_store_index): New function.
7318 (ix86_register_move_cost): Be more sensible about mismatch stall;
7319 model AVX moves correctly; make difference between sse->integer and
7321 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7322 moves; make difference between SSE and AVX.
7323 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7324 and zmm_move. Increase size of sse load and store tables;
7325 add unaligned load and store tables; add ssemmx_to_integer.
7326 * x86-tune-costs.h: Update all entries according to real
7327 move latencies from Agner Fog's manual and chip documentation.
7329 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7332 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7333 * config/i386/constraints.md (Wf): New constraint.
7334 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7335 (cmp<dwi>_doubleword): Removed.
7336 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7337 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7338 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7339 expand with cmp<dwi>_doubleword. For LTU and GEU use
7340 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7342 * common.opt (gcolumn-info): Enable by default.
7343 * doc/invoke.texi (gcolumn-info): Document new default.
7345 2017-10-23 Richard Biener <rguenther@suse.de>
7347 PR tree-optimization/82672
7348 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7349 Fold the stmt if we propagated into it.
7351 2017-10-23 Richard Biener <rguenther@suse.de>
7353 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7354 (bitmap_remove_expr_from_set): ... this. All callers call this
7355 for non-constant values.
7356 (bitmap_set_subtract): Rename to...
7357 (bitmap_set_subtract_expressions): ... this. Adjust and
7359 (bitmap_set_contains_value): Remove superfluous check.
7360 (bitmap_set_replace_value): Inline into single caller ...
7361 (bitmap_value_replace_in_set): ... here and simplify.
7362 (dependent_clean): Merge into ...
7363 (clean): ... this using an overload. Adjust.
7364 (prune_clobbered_mems): Adjust.
7365 (compute_antic_aux): Likewise.
7366 (compute_partial_antic_aux): Likewise.
7368 2017-10-23 Richard Biener <rguenther@suse.de>
7370 PR tree-optimization/82129
7372 2017-08-01 Richard Biener <rguenther@suse.de>
7374 PR tree-optimization/81181
7375 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7376 (compute_antic): ... end of iteration here.
7378 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7380 * target.def (starting_frame_offset): New hook.
7381 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7382 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7383 * doc/tm.texi.in: Regenerate.
7384 * hooks.h (hook_hwi_void_0): Declare.
7385 * hooks.c (hook_hwi_void_0): New function.
7386 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7387 STARTING_FRAME_OFFSET.
7388 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7389 * reload1.c (reload): Likewise.
7390 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7391 instead of STARTING_FRAME_OFFSET.
7392 * function.c (try_fit_stack_local): Likewise.
7393 (assign_stack_local_1): Likewise
7394 (instantiate_virtual_regs): Likewise.
7395 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7396 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7397 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7398 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7399 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7400 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7401 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7402 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7403 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7404 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7405 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7406 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7407 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7408 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7409 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7410 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7411 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7412 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7413 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7414 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7415 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7416 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7417 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7418 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7419 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7420 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7421 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7422 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7423 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7424 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7425 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7426 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7427 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7428 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7429 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7430 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7431 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7432 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7433 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7434 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7435 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7436 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7437 return a HOST_WIDE_INT.
7438 (avr_builtin_setjmp_frame_value): Use it instead of
7439 STARTING_FRAME_OFFSET.
7440 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7441 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7442 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7444 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7445 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7446 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7447 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7448 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7449 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7450 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7451 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7452 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7453 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7454 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7455 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7457 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7458 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7459 * config/mips/mips.c (mips_compute_frame_info): Refer to
7460 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7461 (mips_starting_frame_offset): New function.
7462 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7463 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7464 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7465 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7466 and return a HOST_WIDE_INT.
7467 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7468 (mmix_initial_elimination_offset): Refer to
7469 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7470 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7471 * config/pa/pa.c (pa_starting_frame_offset): New function.
7472 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7473 (pa_expand_prologue): Likewise.
7474 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7475 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7476 !FRAME_GROWS_DOWNWARD handling to...
7477 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7478 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7479 !FRAME_GROWS_DOWNWARD handling to...
7480 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7481 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7482 !FRAME_GROWS_DOWNWARD handling to...
7483 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7484 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7486 (rs6000_starting_frame_offset): New function.
7487 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7488 !FRAME_GROWS_DOWNWARD handling to...
7489 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7490 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7491 !FRAME_GROWS_DOWNWARD handling to...
7492 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7493 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7494 !FRAME_GROWS_DOWNWARD handling to...
7495 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7496 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7497 (rs6000_starting_frame_offset): New function.
7498 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7499 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7500 * config/vax/vax.c (vax_starting_frame_offset): New function.
7501 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7502 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7503 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7504 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7505 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7506 * system.h (STARTING_FRAME_OFFSET): Poison.
7508 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7510 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7511 SCALAR_TYPE_MODE instead of TYPE_MODE.
7513 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7514 Alan Hayward <alan.hayward@arm.com>
7515 David Sherwood <david.sherwood@arm.com>
7517 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7519 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7520 Alan Hayward <alan.hayward@arm.com>
7521 David Sherwood <david.sherwood@arm.com>
7523 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7525 2017-10-23 Richard Biener <rguenther@suse.de>
7527 PR tree-optimization/82129
7528 * tree-ssa-pre.c (bitmap_set_and): Remove.
7529 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7530 canonicalizing expressions in the set to those with lowest
7531 ID rather than taking that from the first edge.
7533 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7535 * combine.c (rtx_equal_for_field_assignment_p): Use
7536 byte_lowpart_offset.
7538 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7539 Alan Hayward <alan.hayward@arm.com>
7540 David Sherwood <david.sherwood@arm.com>
7542 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7543 to a SUBREG_PROMOTED_VAR.
7545 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7546 Alan Hayward <alan.hayward@arm.com>
7547 David Sherwood <david.sherwood@arm.com>
7549 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7550 (expand_debug_source_expr): Likewise.
7551 * combine.c (combine_simplify_rtx): Likewise.
7552 * cse.c (fold_rtx): Likewise.
7553 * optabs.c (expand_float): Likewise.
7554 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7555 (simplify_binary_operation_1): Likewise.
7557 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7558 Alan Hayward <alan.hayward@arm.com>
7559 David Sherwood <david.sherwood@arm.com>
7561 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7562 (record_promoted_value): Likewise.
7563 * expr.c (expand_expr_real_2): Likewise.
7564 * ree.c (update_reg_equal_equiv_notes): Likewise.
7565 (combine_set_extension): Likewise.
7566 * rtlanal.c (low_bitmask_len): Likewise.
7567 * simplify-rtx.c (neg_const_int): Likewise.
7568 (simplify_binary_operation_1): Likewise.
7570 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7571 Alan Hayward <alan.hayward@arm.com>
7572 David Sherwood <david.sherwood@arm.com>
7574 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7575 * regcprop.c (maybe_mode_change): Likewise.
7576 * reload1.c (alter_reg): Likewise.
7578 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7580 * inchash.h (inchash::hash::add_wide_int): New function.
7581 * lto-streamer-out.c (hash_tree): Use it.
7583 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7585 * inchash.h (inchash::hash::add_wide_int): Rename to...
7586 (inchash::hash::add_hwi): ...this.
7587 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7588 (polymorphic_call_target_hasher::hash): Likewise.
7589 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7590 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7591 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7592 * lto-streamer-out.c (hash_tree): Likewise.
7593 * optc-save-gen.awk: Likewise.
7594 * tree.c (add_expr): Likewise.
7596 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7599 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7600 for ordered inequality comparisons even with TARGET_IEEE_FP.
7602 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7605 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7606 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7607 Expand with cmp<dwi>_doubleword.
7609 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7611 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7612 List CET intrinsics.
7613 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7614 specific to -fcf-protection option.
7616 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7618 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7619 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7620 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7621 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7622 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7623 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7624 (extra_objs): Add cet.o for Linux/x86 targets.
7625 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7626 * config/i386/cet.c: New file.
7627 * config/i386/cetintrin.h: Likewise.
7628 * config/i386/t-cet: Likewise.
7629 * config/i386/cpuid.h (bit_SHSTK): New.
7630 (bit_IBT): Likewise.
7631 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7632 pass IBT and SHSTK bits.
7633 * config/i386/i386-builtin-types.def
7634 (VOID_FTYPE_UNSIGNED_PVOID): New.
7635 (VOID_FTYPE_UINT64_PVOID): Likewise.
7636 * config/i386/i386-builtin.def: Add CET intrinsics.
7637 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7638 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7639 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7640 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7642 * config/i386/i386.c (rest_of_insert_endbranch): New.
7643 (pass_data_insert_endbranch): Likewise.
7644 (pass_insert_endbranch): Likewise.
7645 (make_pass_insert_endbranch): Likewise.
7646 (ix86_notrack_prefixed_insn_p): Likewise.
7647 (ix86_target_string): Add -mibt, -mshstk flags.
7648 (ix86_option_override_internal): Add flag_cf_protection
7650 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7651 (ix86_print_operand): Add 'notrack' prefix output.
7652 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7653 (ix86_expand_builtin): Expand CET intrinsics.
7654 (x86_output_mi_thunk): Add 'endbranch' instruction.
7655 * config/i386/i386.h (TARGET_IBT): New.
7656 (TARGET_IBT_P): Likewise.
7657 (TARGET_SHSTK): Likewise.
7658 (TARGET_SHSTK_P): Likewise.
7659 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7660 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7661 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7662 (builtin_setjmp_setup): New pattern.
7663 (builtin_longjmp): Likewise.
7664 (rdssp<mode>): Likewise.
7665 (incssp<mode>): Likewise.
7666 (saveprevssp): Likewise.
7667 (rstorssp): Likewise.
7668 (wrss<mode>): Likewise.
7669 (wruss<mode>): Likewise.
7670 (setssbsy): Likewise.
7671 (clrssbsy): Likewise.
7672 (nop_endbr): Likewise.
7673 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7675 * config/i386/immintrin.h: Include <cetintrin.h>.
7676 * config/i386/linux-common.h
7677 (file_end_indicate_exec_stack_and_cet): New prototype.
7678 (TARGET_ASM_FILE_END): New.
7680 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7682 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7683 latencies instead of having separate table; make difference between
7684 integer and float costs.
7685 * i386.h (processor_costs): Remove scalar_stmt_cost,
7686 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7687 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7689 * x86-tune-costs.h: Remove entries which has been removed in
7690 procesor_costs from all tables; make cond_taken_branch_cost
7691 and cond_not_taken_branch_cost COST_N_INSNS based.
7693 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7695 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7697 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7699 * config/i386/i386.md (isa): Remove fma_avx512f.
7700 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7701 <avx512>_fmadd_<mode>_mask3<round_name>,
7702 <avx512>_fmsub_<mode>_mask<round_name>,
7703 <avx512>_fmsub_<mode>_mask3<round_name>,
7704 <avx512>_fnmadd_<mode>_mask<round_name>,
7705 <avx512>_fnmadd_<mode>_mask3<round_name>,
7706 <avx512>_fnmsub_<mode>_mask<round_name>,
7707 <avx512>_fnmsub_<mode>_mask3<round_name>,
7708 <avx512>_fmaddsub_<mode>_mask<round_name>,
7709 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7710 <avx512>_fmsubadd_<mode>_mask<round_name>,
7711 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7712 (*vec_widen_umult_even_v16si<mask_name>,
7713 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7714 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7716 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7718 * extend.texi: Add 'nocf_check' documentation.
7719 * gimple.texi: Add second parameter to
7720 gimple_build_call_from_tree.
7721 * invoke.texi: Add -fcf-protection documentation.
7722 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7724 2017-10-20 Richard Biener <rguenther@suse.de>
7726 PR tree-optimization/82473
7727 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7728 the largest input type.
7730 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7732 * c-attribs.c (handle_nocf_check_attribute): New function.
7733 (c_common_attribute_table): Add 'nocf_check' handling.
7734 * gimple-parser.c: Add second argument NULL to
7735 gimple_build_call_from_tree.
7736 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7737 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7739 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7740 * common.opt: Add fcf-protection flag.
7741 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7742 * flag-types.h: Add enum cf_protection_level.
7743 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7744 Add 'nocf_check' attribute propagation to gimple call.
7745 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7746 (gimple_build_call_from_tree): Update prototype.
7747 (gimple_call_nocf_check_p): New function.
7748 (gimple_call_set_nocf_check): Likewise.
7749 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7750 * ipa-icf.c: Add nocf_check attribute in statement hash.
7751 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7752 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7753 * toplev.c (process_options): Add flag_cf_protection handling.
7755 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7757 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7759 2017-10-20 Richard Biener <rguenther@suse.de>
7761 PR tree-optimization/82603
7762 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7763 remove false predicated stores.
7765 2017-10-20 Richard Biener <rguenther@suse.de>
7767 * graphite-isl-ast-to-gimple.c
7768 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7769 Remove return value and simplify, dump copied stmt after lhs
7771 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7772 Reduce dump verbosity.
7773 (gsi_insert_earliest): Likewise.
7774 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7775 * graphite.c (print_global_statistics): Adjust dumping.
7776 (print_graphite_scop_statistics): Likewise.
7777 (print_graphite_statistics): Do not dump loops here.
7778 (graphite_transform_loops): But here.
7780 2017-10-20 Nicolas Roche <roche@adacore.com>
7782 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7783 * configure: Regenerate.
7785 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7788 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7789 functions when optimizing replace GIMPLE_RETURN stmts with
7790 calls to __builtin_unreachable ().
7793 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7794 for -fsanitize=thread link of executables.
7795 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7796 link of executables.
7799 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7801 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7802 define_insns for logical vector shifts to use VI248_AVX512BW
7803 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7804 condition, useless isa and prefix attributes. Change the first
7805 2 of these define_insns to ...
7806 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7807 define_insn for avx512vl.
7808 (<shift_insn><mode>3): ... and this, new define_insn without
7809 masking for non-avx512vl.
7812 * config/i386/sse.md (*andnot<mode>3,
7813 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7814 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7815 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7816 not applied use empty suffix even for TARGET_AVX512VL.
7817 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7818 is applied, supply evex,evex or evex,evex,evex instead of just
7821 2017-10-20 Julia Koval <julia.koval@intel.com>
7823 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7824 (OPTION_MASK_ISA_GFNI_UNSET): New.
7825 (ix86_handle_option): Handle OPT_mgfni.
7826 * config/i386/cpuid.h (bit_GFNI): New.
7827 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7828 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7829 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7830 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7831 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7832 * config/i386/i386.opt: Add mgfni.
7834 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7836 * config/msp430/msp430.c (msp430_option_override): Disable
7837 -fdelete-null-pointer-checks.
7838 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7840 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7842 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7843 of x87 and SSE instructions.
7845 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7847 * asan.c (create_cond_insert_point): Do not update edge count.
7848 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7849 (afdo_propagate_circuit): Likewise.
7850 (afdo_calculate_branch_prob): Likewise.
7851 (afdo_annotate_cfg): Likewise.
7852 * basic-block.h (struct edge_def): Remove count.
7853 (edge_def::count): New accessor.
7854 * bb-reorder.c (rotate_loop): Update.
7855 (find_traces_1_round): Update.
7856 (connect_traces): Update.
7857 (sanitize_hot_paths): Update.
7858 * cfg.c (unchecked_make_edge): Update.
7859 (make_single_succ_edge): Update.
7860 (check_bb_profile): Update.
7861 (dump_edge_info): Update.
7862 (update_bb_profile_for_threading): Update.
7863 (scale_bbs_frequencies_int): Update.
7864 (scale_bbs_frequencies_gcov_type): Update.
7865 (scale_bbs_frequencies_profile_count): Update.
7866 (scale_bbs_frequencies): Update.
7867 * cfganal.c (connect_infinite_loops_to_exit): Update.
7868 * cfgbuild.c (compute_outgoing_frequencies): Update.
7869 (find_many_sub_basic_blocks): Update.
7870 * cfgcleanup.c (try_forward_edges): Update.
7871 (try_crossjump_to_edge): Update
7872 * cfgexpand.c (expand_gimple_cond): Update
7873 (expand_gimple_tailcall): Update
7874 (construct_exit_block): Update
7875 * cfghooks.c (verify_flow_info): Update
7876 (redirect_edge_succ_nodup): Update
7877 (split_edge): Update
7878 (make_forwarder_block): Update
7879 (duplicate_block): Update
7880 (account_profile_record): Update
7881 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7882 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7883 * cfgloopmanip.c (scale_loop_profile): Update.
7885 (lv_adjust_loop_entry_edge): Update.
7886 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7887 (force_nonfallthru_and_redirect): Update.
7888 (purge_dead_edges): Update.
7889 (rtl_flow_call_edges_add): Update.
7890 * cgraphunit.c (init_lowered_empty_function): Update.
7891 (cgraph_node::expand_thunk): Update.
7892 * gimple-pretty-print.c (dump_probability): Update.
7893 (dump_edge_probability): Update.
7894 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7895 * haifa-sched.c (sched_create_recovery_edges): Update.
7896 * hsa-gen.c (convert_switch_statements): Update.
7897 * ifcvt.c (dead_or_predicable): Update.
7898 * ipa-inline-transform.c (inline_transform): Update.
7899 * ipa-split.c (split_function): Update.
7900 * ipa-utils.c (ipa_merge_profiles): Update.
7901 * loop-doloop.c (add_test): Update.
7902 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7903 * lto-streamer-in.c (input_cfg): Update.
7904 (input_function): Update.
7905 * lto-streamer-out.c (output_cfg): Update.
7906 * modulo-sched.c (sms_schedule): Update.
7907 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7908 * predict.c (maybe_hot_edge_p): Update.
7909 (unlikely_executed_edge_p): Update.
7910 (probably_never_executed_edge_p): Update.
7911 (dump_prediction): Update.
7912 (drop_profile): Update.
7913 (propagate_unlikely_bbs_forward): Update.
7914 (determine_unlikely_bbs): Update.
7915 (force_edge_cold): Update.
7916 * profile.c (compute_branch_probabilities): Update.
7917 * reg-stack.c (better_edge): Update.
7918 * shrink-wrap.c (handle_simple_exit): Update.
7919 * tracer.c (better_p): Update.
7920 * trans-mem.c (expand_transaction): Update.
7921 (split_bb_make_tm_edge): Update.
7922 * tree-call-cdce.c: Update.
7923 * tree-cfg.c (gimple_find_sub_bbs): Update.
7924 (gimple_split_edge): Update.
7925 (gimple_duplicate_sese_region): Update.
7926 (gimple_duplicate_sese_tail): Update.
7927 (gimple_flow_call_edges_add): Update.
7928 (insert_cond_bb): Update.
7929 (execute_fixup_cfg): Update.
7930 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
7931 * tree-complex.c (expand_complex_div_wide): Update.
7932 * tree-eh.c (lower_resx): Update.
7933 (unsplit_eh): Update.
7934 (cleanup_empty_eh_move_lp): Update.
7935 * tree-inline.c (copy_edges_for_bb): Update.
7936 (freqs_to_counts): Update.
7937 (copy_cfg_body): Update.
7938 * tree-ssa-dce.c (remove_dead_stmt): Update.
7939 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7940 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7941 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7942 (unloop_loops): Update.
7943 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7944 * tree-ssa-loop-split.c (connect_loops): Update.
7945 (split_loop): Update.
7946 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7947 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7948 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7949 * tree-ssa-reassoc.c (branch_fixup): Update.
7950 * tree-ssa-tail-merge.c (replace_block_by): Update.
7951 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7952 (compute_path_counts): Update.
7953 (update_profile): Update.
7954 (recompute_probabilities): Update.
7955 (update_joiner_offpath_counts): Update.
7956 (estimated_freqs_path): Update.
7957 (freqs_to_counts_path): Update.
7958 (clear_counts_path): Update.
7959 (ssa_fix_duplicate_block_edges): Update.
7960 (duplicate_thread_path): Update.
7961 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7962 (case_bit_test_cmp): Update.
7963 (collect_switch_conv_info): Update.
7964 (gen_inbound_check): Update.
7965 (do_jump_if_equal): Update.
7966 (emit_cmp_and_jump_insns): Update.
7967 * tree-tailcall.c (decrease_profile): Update.
7968 (eliminate_tail_call): Update.
7969 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7970 (vect_do_peeling): Update.
7971 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7972 * ubsan.c (ubsan_expand_null_ifn): Update.
7973 (ubsan_expand_ptr_ifn): Update.
7974 * value-prof.c (gimple_divmod_fixed_value): Update.
7975 (gimple_mod_pow2): Update.
7976 (gimple_mod_subtract): Update.
7977 (gimple_ic): Update.
7978 (gimple_stringop_fixed_value): Update.
7980 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7983 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
7985 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
7987 PR rtl-optimization/82395
7988 * ira-color.c (allocno_priority_compare_func): Fix comparison step
7989 based on non_spilled_static_chain_regno_p.
7991 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7993 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
7994 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
7995 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
7997 2017-10-19 Martin Sebor <msebor@redhat.com>
7999 PR tree-optimization/82596
8000 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8002 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8004 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8005 (get_mem_refs_of_builtin_call): Likewise.
8006 * builtins.c (expand_builtin_apply): Adjust call to
8007 allocate_dynamic_stack_space.
8008 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8009 the third argument to allocate_dynamic_stack_space, otherwise -1.
8010 (expand_builtin): Deal with all alloca variants.
8011 (is_inexpensive_builtin): Likewise.
8012 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8013 * calls.c (special_function_p): Deal with all alloca variants.
8014 (initialize_argument_information): Adjust call to
8015 allocate_dynamic_stack_space.
8016 (expand_call): Likewise.
8017 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8018 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8019 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8020 use it for the stack usage computation.
8021 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8022 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8023 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8024 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8025 (in_loop_p): Remove first argument and useless check.
8026 (pass_walloca::execute): Remove useless test and adjust call to above.
8027 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8028 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8029 (gimplify_call_expr): Deal with all alloca variants.
8030 * hsa-gen.c (gen_hsa_alloca): Likewise.
8031 (gen_hsa_insns_for_call): Likewise.
8032 * ipa-pure-const.c (special_builtin_state): Likewise.
8033 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8034 * tree-object-size.c (alloc_object_size): Likewise.
8035 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8036 (call_may_clobber_ref_p_1): Likewise.
8037 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8038 (ccp_fold_stmt): Likewise.
8039 (optimize_stack_restore): Likewise.
8040 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8041 (mark_all_reaching_defs_necessary_1): Likewise.
8042 (propagate_necessity): Likewise.
8043 (eliminate_unnecessary_stmts): Likewise.
8044 * tree.c (build_common_builtin_nodes): Build
8045 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8046 (build_alloca_call_expr): New function.
8047 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8048 (CASE_BUILT_IN_ALLOCA): Likewise.
8049 (build_alloca_call_expr): Declare.
8050 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8052 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8055 * dwarf2out.c (new_die_raw): New static inline function.
8056 (new_die): Use it to create the DIE.
8057 (add_AT_external_die_ref): Likewise.
8058 (clone_die): Likewise.
8059 (clone_as_declaration): Likewise.
8060 (dwarf2out_vms_debug_main_pointer): Likewise.
8061 (base_type_die): Likewise. Remove early return for corner cases.
8062 Do not call add_pubtype on the DIE here.
8063 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8064 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8065 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8066 native order exists for base types, attach the DIE manually and call
8067 add_pubtype on it. Do not equate a reverse order DIE to the type.
8069 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8071 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8072 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8074 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8076 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8078 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8079 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8080 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8081 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8082 * opts.c (sanitizer_opts): Add builtin.
8083 * ubsan.c (instrument_builtin): New function.
8084 (pass_ubsan::execute): Call it.
8085 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8086 * doc/invoke.texi: Document -fsanitize=builtin.
8088 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8089 builtins, store max (log2 (align), 0) into uchar field instead of
8090 align into uptr field.
8091 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8092 store uchar 0 field instead of uptr 0 field.
8093 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8094 instead of passing one address of struct with 2 locations pass
8095 two addresses of structs with 1 location each.
8096 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8097 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8098 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8099 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8100 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8101 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8102 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8103 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8105 2017-10-19 Martin Liska <mliska@suse.cz>
8108 * file-find.c (remove_prefix): Remove.
8109 * file-find.h (remove_prefix): Likewise.
8110 * gcc-ar.c: Remove smartness of lookup.
8112 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8114 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8115 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8116 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8118 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8121 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8122 (setcc + and to xor + setcc): New peephole2.
8124 2017-10-19 Tom de Vries <tom@codesourcery.com>
8126 * doc/sourcebuild.texi (Test Directives, Variants of
8127 dg-require-support): Add dg-require-stack-size.
8129 2017-10-19 Martin Liska <mliska@suse.cz>
8132 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8133 that have a large alignment.
8134 (gimplify_target_expr): Likewise.
8136 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8138 PR rtl-optimization/82602
8139 * ira.c (rtx_moveable_p): Return false for volatile asm.
8141 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8144 * config/i386/i386-modes.def (CCGZ): New CC mode.
8145 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8146 * config/i386/predicates.md (ix86_comparison_operator):
8148 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8149 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8150 with double-word subtraction.
8151 (put_condition_code): Handle CCGZmode.
8153 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8155 * wide-int.cc (debug (const wide_int &)): New.
8156 (debug (const wide_int *)): New.
8157 (debug (const widest_int &)): New.
8158 (debug (const widest_int *)): New.
8160 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8163 * lra-constraints.c (curr_insn_transform): Use non-input operand
8164 instead of output one for matched reload.
8166 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8168 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8169 (tree-ssa-loop-ivopts.h): New header file.
8170 (struct builtin_info): New fields.
8171 (classify_builtin_1): Compute and record base and offset parts for
8172 memset builtin partition by calling strip_offset.
8173 (offset_cmp, fuse_memset_builtins): New functions.
8174 (finalize_partitions): Fuse adjacent memset partitions by calling
8176 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8177 Expose the interface.
8178 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8180 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8182 PR tree-optimization/82574
8183 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8184 that data reference must be executed exactly once per iteration
8185 against the outermost loop in nest.
8186 (classify_partition): Update call to above function.
8188 2017-10-18 Richard Biener <rguenther@suse.de>
8190 PR tree-optimization/82591
8191 * graphite.c (graphite_transform_loops): Move code gen message
8193 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8194 Here. Handle scop_to_isl_ast failing.
8195 (scop_to_isl_ast): Limit the number of ISL operations.
8197 2017-10-18 Richard Biener <rguenther@suse.de>
8199 * graphite-isl-ast-to-gimple.c
8200 (translate_isl_ast_to_gimple::set_rename): Simplify.
8201 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8202 (graphite_copy_stmts_from_block): ... here.
8203 (copy_bb_and_scalar_dependences): Simplify.
8204 (add_parameters_to_ivs_params): Canonicalize.
8205 (generate_entry_out_of_ssa_copies): Simplify.
8206 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8207 by passing in ISL dimension.
8208 (parameter_index_in_region_1): Rename to ...
8209 (parameter_index_in_region): ... this.
8210 (extract_affine): Adjust assert, pass down parameter index.
8211 (add_param_constraints): Use range-info when available.
8212 (build_scop_context): Adjust.
8213 * sese.c (new_sese_info): Adjust.
8214 (free_sese_info): Likewise.
8215 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8216 Remove unused typedefs.
8217 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8219 2017-10-18 Martin Liska <mliska@suse.cz>
8221 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8223 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8225 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8226 (s390_sched_init): Do not reset s390_sched_state if we entered the
8227 current basic block via a fallthru edge and all others are unlikely.
8229 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8231 * config/s390/s390.c (NUM_SIDES): New variable.
8232 (LONGRUNNING_THRESHOLD): New variable.
8233 (LATENCY_FACTOR): New variable.
8234 (s390_sched_score): Decrease score for long-running instructions on
8236 (s390_sched_variable_issue): Perform bookkeeping for long-running
8239 2017-10-18 Richard Biener <rguenther@suse.de>
8241 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8242 Simplify with removal of the parameter rename map.
8243 (set_rename): Likewise.
8244 (should_copy_to_new_region): Likewise.
8245 (graphite_copy_stmts_from_block): Likewise.
8246 (copy_bb_and_scalar_dependences): Remove initialization of
8247 unused copied_bb_map.
8249 (copy_internal_parameters): Likewise.
8250 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8251 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8252 Use INTEGRAL_TYPE_P.
8253 (parameter_index_in_region_1): Rename to ...
8254 (assign_parameter_index_in_region): ... this. Assert we have
8255 a parameter we handle.
8256 (scan_tree_for_params): Adjust.
8257 * sese.h (parameter_rename_map_t): Remove.
8258 (struct sese_info_t): Remove unused parameter_rename_map and
8259 copied_bb_map members.
8260 * sese.c (new_sese_info): Adjust.
8261 (free_sese_info): Likewise.
8263 2017-10-18 Martin Liska <mliska@suse.cz>
8266 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8267 on an abnormal edge.
8269 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8271 * doc/invoke.texi (ffunction-sections and fdata-sections):
8274 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8276 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8277 the use statement can throw internally.
8279 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8281 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8282 any RTX present on the RHS of a SET.
8283 * compare-elim.c (try_eliminate_compare): Restore comment.
8285 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8287 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8288 may be also called on tcc_type nodes.
8289 * langhooks.c (lhd_tree_size): Likewise.
8291 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8293 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8294 format_warning_at_substring.
8295 (maybe_warn): Convert source_range * param to a location_t. Pass
8296 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8297 (format_directive): Remove code to extract source_ranges and
8298 source_range * in favor of just a location_t.
8299 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8301 * substring-locations.c (format_warning_va): Convert
8302 source_range * param to a location_t.
8303 (format_warning_at_substring): Likewise.
8304 * substring-locations.h (format_warning_va): Likewise.
8305 (format_warning_at_substring): Likewise.
8307 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8309 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8311 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8312 and scatter/gather ops.
8314 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8315 vec_gather_load and vec_scatter_store.
8316 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8317 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8318 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8319 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8320 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8321 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8323 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8325 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8326 Detect FTST instruction and handle its register pops. Only pop
8327 second operand if can_pop_second_op is true.
8328 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8329 set can_pop_second_op to false in the compare_for_stack_reg call.
8331 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8332 output_fp_compare for stack register operands.
8333 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8334 instructions here. Do not emit stack register pops here. Assert
8335 that FCOMPP pops next to top stack register. Rewrite function.
8337 2017-10-17 Nathan Sidwell <nathan@acm.org>
8340 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8341 use DECL_ASSEMBLER_NAME_RAW.
8344 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8347 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8348 Wilco Dijkstra <wilco.dijkstra@arm.com>
8350 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8351 converstion to Pmode from the buf_addr.
8353 2017-10-17 Richard Biener <rguenther@suse.de>
8355 * graphite-dependences.c (scop_get_reads_and_writes): Change
8356 output parameters to references.
8358 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8360 PR 71026/tree-optimization
8361 * fold-const.c (distribute_real_division): Removed.
8362 (fold_binary_loc): Remove calls to distribute_real_divison.
8364 2017-10-17 Richard Biener <rguenther@suse.de>
8366 * graphite-scop-detection.c
8367 (scop_detection::stmt_has_simple_data_refs_p): Always use
8368 the full nest as region.
8369 (try_generate_gimple_bb): Likewise.
8370 * sese.c (scalar_evolution_in_region): Simplify now that
8371 SCEV can handle instantiation in regions.
8372 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8373 in the non-loop part of a function if requested.
8375 2017-10-17 Richard Biener <rguenther@suse.de>
8377 PR tree-optimization/82563
8378 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8380 (graphite_regenerate_ast_isl): Call it.
8381 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8383 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8385 PR tree-optimization/82549
8386 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8387 Formatting fixes. Instead of calling make_bit_field_ref with negative
8390 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8392 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8393 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8394 _mm_maskz_reduce_ss): New.
8395 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8396 __builtin_ia32_reducess_mask): Ditto..
8397 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8398 * config/i386/sse.md (reduces<mode>): Renamed to ...
8399 (reduces<mode><mask_scalar_name>): ... this.
8400 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8402 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8403 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8405 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8407 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8408 * selftest-run-tests.c (selftest::run_tests): Call
8409 selftest::unique_ptr_tests_cc_tests.
8410 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8411 * unique-ptr-tests.cc: New file.
8413 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8416 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8418 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8419 (make_hard_regno_born, make_hard_regno_dead): Update
8420 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8422 2017-10-16 Jeff Law <law@redhat.com>
8424 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8426 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8428 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8430 2017-10-16 Olivier Hainque <hainque@adacore.com>
8432 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8433 with_cpu if we were configured for an e500v2 target cpu name.
8435 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8437 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8438 * doc/invoke.texi: Document +nodsp as a valid extension for
8441 2017-10-16 Martin Liska <mliska@suse.cz>
8443 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8444 (test_set_range): Likewise.
8445 (test_range_functions): Rename to ...
8446 (test_bit_in_range): ... this.
8447 (sbitmap_c_tests): Add new test.
8449 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8451 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8453 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8454 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8456 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8458 * config/aarch64/aarch64-builtins.c
8459 (aarch64_types_quadopu_lane_qualifiers): New.
8460 (TYPES_QUADOPU_LANE): New.
8461 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8462 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8463 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8464 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8465 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8466 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8467 (Vdottype, DOTPROD): New.
8468 (sur): Add SDOT and UDOT.
8470 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8472 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8473 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8474 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8476 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8477 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8478 Enable TARGET_DOTPROD.
8479 (cortex-a75.cortex-a55): Likewise.
8480 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8482 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8484 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8485 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8487 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8489 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8490 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8491 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8492 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8493 * config/arm/types.md (neon_dot, neon_dot_q): New.
8494 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8496 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8498 * config/arm/arm.h (TARGET_DOTPROD): New.
8499 * config/arm/arm.c (arm_arch_dotprod): New.
8500 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8501 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8502 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8503 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8504 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8505 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8506 * doc/invoke.texi (armv8.2-a): Document dotprod
8508 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8510 * i386.c (ix86_vec_cost): New function.
8511 (ix86_rtx_costs): Handle vector operations better.
8512 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8513 * x86-tune-costs.h: Add new costs to all tables.
8515 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8517 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8519 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8520 divsd, sqrtss and sqrtsd
8521 * x86-tune-costs.h: Add new entries to all costs.
8522 (znver1_cost): Fix to match real instruction latencies.
8524 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8525 Michael Collison <michael.collison@arm.com>
8527 * compare-elim.c: Include emit-rtl.h.
8528 (can_merge_compare_into_arith): New function.
8529 (try_validate_parallel): Likewise.
8530 (try_merge_compare): Likewise.
8531 (try_eliminate_compare): Call the above when no previous clobber
8533 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8536 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8540 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8541 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8545 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8546 to be any operand_equal_p operands. For & (B - 1) require
8547 B to be power of 2. Recognize
8548 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8550 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8553 * optabs.c (expand_memory_blockage): Fix call of
8554 targetm.have_memory_blockage.
8556 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8559 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8560 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8561 objects to extra_objs instead of overwriting it.
8563 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8565 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8566 Use any_fp_register_operand as operand[3] predicate. Simplify
8567 equality test for operands[2] and operands[4] memory location.
8568 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8569 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8570 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8571 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8572 any_fp_register_operand as operand[1] predicate. Simplify
8573 equality test for operands[0] and operands[3] memory location.
8574 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8575 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8576 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8578 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8580 * target-insns.def: Add memory_blockage.
8581 * optabs.c (expand_memory_blockage): New function.
8582 (expand_asm_memory_barrier): Rename ...
8583 (expand_asm_memory_blockage): ... to this.
8584 (expand_mem_thread_fence): Call expand_memory_blockage
8585 instead of expand_asm_memory_barrier.
8586 (expand_mem_singnal_fence): Ditto.
8587 (expand_atomic_load): Ditto.
8588 (expand_atomic_store): Ditto.
8589 * doc/md.texi (Standard Pattern Names For Generation):
8590 Document memory_blockage instruction pattern.
8592 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8594 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8595 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8596 * config/rl78/rl78.md: New define_expand "adddi3".
8598 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8600 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8603 2017-10-13 Jeff Law <law@redhat.com>
8605 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8607 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8610 * internal-fn.c (expand_mul_overflow): If both operands have
8611 the same highpart of -1 or 0 and the topmost bit of lowpart
8612 is different, overflow is if res <= 0 rather than res < 0.
8614 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8616 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8617 TARGET_P9_VECTOR code for unaligned_load case.
8619 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8621 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8623 2017-10-13 Nathan Sidwell <nathan@acm.org>
8625 * tree-core.h (tree_contains_struct): Make bool.
8626 * tree.c (tree_contains_struct): Likewise.
8627 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8628 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8629 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8630 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8631 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8633 2017-10-13 Richard Biener <rguenther@suse.de>
8635 * graphite-isl-ast-to-gimple.c
8636 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8637 parameters and dominance check.
8638 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8639 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8640 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8641 Do not update SSA form here or do intermediate IL verification.
8642 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8643 (graphite_initialize): Remove check on the number of loops in
8644 the function and inline into graphite_transform_loops.
8645 (graphite_finalize): Inline into graphite_transform_loops.
8646 (graphite_transform_loops): Perform SSA update and IL verification
8648 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8650 2017-10-13 Richard Biener <rguenther@suse.de>
8652 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8653 graphite_expression_type_precision): Avoid global constructor
8655 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8656 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8657 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8658 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8659 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8660 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8662 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8665 * config/i386/i386.h (ix86_red_zone_size): New.
8666 * config/i386/i386.md (push peephole2s): Replace
8667 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8669 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8670 Alan Hayward <alan.hayward@arm.com>
8671 David Sherwood <david.sherwood@arm.com>
8673 * combine.c (can_change_dest_mode): Reject changes in
8674 REGMODE_NATURAL_SIZE.
8676 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8677 Alan Hayward <alan.hayward@arm.com>
8678 David Sherwood <david.sherwood@arm.com>
8680 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8681 (expand_debug_source_expr): Likewise.
8682 * combine.c (combine_simplify_rtx): Likewise.
8683 * cse.c (fold_rtx): Likewise.
8684 * fwprop.c (canonicalize_address): Likewise.
8685 * targhooks.c (default_shift_truncation_mask): Likewise.
8687 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8688 Alan Hayward <alan.hayward@arm.com>
8689 David Sherwood <david.sherwood@arm.com>
8691 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8692 (widened_mode): Likewise.
8693 (expand_unop): Likewise.
8694 * ree.c (transform_ifelse): Likewise.
8695 (merge_def_and_ext): Likewise.
8696 (combine_reaching_defs): Likewise.
8697 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8699 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8700 Alan Hayward <alan.hayward@arm.com>
8701 David Sherwood <david.sherwood@arm.com>
8703 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8704 * combine.c (gen_lowpart_for_combine): Likewise.
8705 * dwarf2out.c (rtl_for_decl_location): Likewise.
8706 * final.c (alter_subreg): Likewise.
8707 * rtlhooks.c (gen_lowpart_general): Likewise.
8708 (gen_lowpart_if_possible): Likewise.
8710 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8711 Alan Hayward <alan.hayward@arm.com>
8712 David Sherwood <david.sherwood@arm.com>
8714 * calls.c (expand_call): Use subreg_lowpart_offset.
8715 * cse.c (cse_insn): Likewise.
8716 * regcprop.c (copy_value): Likewise.
8717 (copyprop_hardreg_forward_1): Likewise.
8719 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8722 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8723 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8724 =Q constraints to +Q and into insn condition add check
8725 that operands[0] and operands[1] are equal.
8726 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8727 =Q constraints to +Q and into insn condition add check
8728 that operands[0] is equal to either operands[1] or operands[2].
8731 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8732 instead of handling MINUS_EXPR twice (once for each argument),
8733 canonicalize operand order and handle just once, use rtype where
8734 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8737 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8738 any values of __C while still being pattern recognizable as a simple
8741 2017-10-13 Richard Biener <rguenther@suse.de>
8743 PR tree-optimization/82451
8745 2017-10-02 Richard Biener <rguenther@suse.de>
8747 PR tree-optimization/82355
8748 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8749 a mapping for the enclosing loop but avoid generating one for
8751 (copy_bb_and_scalar_dependences): Remove premature codegen
8752 error on PHIs in blocks duplicated into multiple places.
8753 * graphite-scop-detection.c
8754 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8755 in the region use it as loop and nest to analyze the DR in.
8756 (try_generate_gimple_bb): Likewise.
8757 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8758 (add_loop_constraints): For blocks in a loop not in the region
8759 create a dimension with a single iteration.
8760 * sese.h (gbb_loop_at_index): Remove assert.
8762 * cfgloop.c (loop_preheader_edge): For the loop tree root
8763 return the single successor of the entry block.
8764 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8765 Reset the SCEV hashtable and niters.
8766 * graphite-scop-detection.c
8767 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8768 assert that we only have POLYNOMIAL_CHREC that vary in loops
8769 contained in the region.
8770 (scop_detection::graphite_can_represent_expr): Adjust.
8771 (scop_detection::stmt_has_simple_data_refs_p): For loops
8772 not in the region set loop to NULL. The nest is now the
8773 entry edge to the region.
8774 (try_generate_gimple_bb): Likewise.
8775 * sese.c (scalar_evolution_in_region): Adjust for
8776 instantiate_scev change.
8777 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8778 Make nest parameter the edge into the region.
8779 (create_data_ref): Likewise.
8780 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8781 entry edge into a region and adjust instantiate_scev calls.
8782 (create_data_ref): Likewise.
8783 (graphite_find_data_references_in_stmt): Likewise.
8784 (find_data_references_in_stmt): Pass the loop preheader edge
8785 from the nest argument.
8786 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8787 parameter the edge into the region.
8788 (instantiate_parameters): Use the loop preheader edge as entry.
8789 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8791 (get_instantiated_value_entry): Make instantiate_below parameter
8792 the edge into the region.
8793 (instantiate_scev_name): Likewise. Adjust dominance checks,
8794 when we cannot use loop-based instantiation instantiate by
8795 walking use-def chains.
8796 (instantiate_scev_poly): Adjust.
8797 (instantiate_scev_binary): Likewise.
8798 (instantiate_scev_convert): Likewise.
8799 (instantiate_scev_not): Likewise.
8800 (instantiate_array_ref): Remove.
8801 (instantiate_scev_3): Likewise.
8802 (instantiate_scev_2): Likewise.
8803 (instantiate_scev_1): Likewise.
8804 (instantiate_scev_r): Do not blindly handle N-operand trees.
8805 Do not instantiate array-refs. Handle all constants and invariants.
8806 (instantiate_scev): Make instantiate_below parameter
8807 the edge into the region.
8808 (resolve_mixers): Use the loop preheader edge for the region
8809 parameter to instantiate_scev_r.
8810 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8812 2017-10-13 Richard Biener <rguenther@suse.de>
8814 PR tree-optimization/82525
8815 * graphite-isl-ast-to-gimple.c
8816 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8818 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8819 Fail code generation when we cannot represent the isl integer.
8820 (binary_op_to_tree): Elide modulo operations that are no-ops
8821 in the type we code generate. Remove now superfluous code
8824 2017-10-13 Richard Biener <rguenther@suse.de>
8826 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8827 (scop_detection::harmful_loop_in_region): Remove premature
8828 IV type restriction.
8829 (scop_detection::graphite_can_represent_scev): We can handle
8830 pointer IVs just fine.
8832 2017-10-13 Alan Modra <amodra@gmail.com>
8834 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8835 "Clobbers and Scratch Registers". Add paragraph on
8836 alternative to clobbers for scratch registers and OpenBLAS
8839 2017-10-13 Alan Modra <amodra@gmail.com>
8841 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8842 example of a memory input for a string of known length. Move
8843 commentary out of table. Add a number of new examples
8844 covering array memory inputs.
8846 2017-10-12 Martin Liska <mliska@suse.cz>
8848 PR tree-optimization/82493
8849 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8850 (test_range_functions): New function.
8851 (sbitmap_c_tests): Likewise.
8852 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8853 * selftest.h (sbitmap_c_tests): New function.
8855 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8857 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8859 * config/rs6000/amo.h: Fix spacing issue.
8861 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8864 * config/i386/i386.md (*ashl<mode>3_mask_1,
8865 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8866 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8869 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8871 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8872 (profile_probability): Set max_probability
8873 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8875 * profile-count.c (profile_probability::differs_from_p): Do not
8876 rely on max_probaiblity == 10000
8878 2017-10-12 Jeff Law <law@redhat.com>
8880 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8883 2017-10-12 Martin Sebor <msebor@redhat.com>
8887 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8888 (handle_alias_pairs): Call it.
8889 * common.opt (-Wattribute-alias): New option.
8890 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8891 * doc/invoke.texi (-Wattribute-alias): Document.
8893 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8896 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8898 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8900 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8901 (make_hard_regno_born, make_hard_regno_dead): Update
8902 bb_killed_pseudos and bb_gen_pseudos.
8904 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8906 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8908 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8910 * config/alpha/alpha.c (alpha_split_conditional_move):
8911 Use std::swap instead of manually swapping.
8912 (alpha_stdarg_optimize_hook): Ditto.
8913 (alpha_canonicalize_comparison): Ditto.
8915 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8917 * tree-loop-distribution.c (struct builtin_info): New struct.
8918 (struct partition): Refactor fields into struct builtin_info.
8919 (partition_free): Free struct builtin_info.
8920 (build_size_arg_loc, build_addr_arg_loc): Delete.
8921 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8922 information from struct builtin_info.
8923 (find_single_drs): New function refactored from classify_partition.
8924 Also moved builtin validity checks to this function.
8925 (compute_access_range, alloc_builtin): New functions.
8926 (classify_builtin_st, classify_builtin_ldst): New functions.
8927 (classify_partition): Refactor code into functions find_single_drs,
8928 classify_builtin_st and classify_builtin_ldst.
8929 (distribute_loop): Don't do runtime alias check when distributing
8931 (find_seed_stmts_for_distribution): New function.
8932 (pass_loop_distribution::execute): Refactor code finding seed
8933 stmts into above function. Support distribution for the innermost
8934 two-level loop nest. Adjust dump information.
8936 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8938 * tree-loop-distribution.c: Adjust the general comment.
8939 (NUM_PARTITION_THRESHOLD): New macro.
8940 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
8941 (classify_partition): Skip builtin pattern of loop nest's inner loop.
8942 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
8943 in call to build_partition_graph.
8944 (finalize_partitions): New parameter. Make loop distribution more
8945 conservative by fusing more partitions.
8946 (distribute_loop): Don't do runtime alias check in case of loop nest
8948 (find_seed_stmts_for_distribution): New function.
8949 (prepare_perfect_loop_nest): New function.
8950 (pass_loop_distribution::execute): Refactor code finding seed stmts
8951 and loop nest into above functions. Support loop nest distribution.
8952 Adjust dump information accordingly.
8954 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8956 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
8957 and set PTYPE_SEQUENTIAL for merged partition.
8959 2017-10-12 Richard Biener <rguenther@suse.de>
8961 PR tree-optimization/69728
8963 2017-09-19 Richard Biener <rguenther@suse.de>
8965 PR tree-optimization/69728
8966 * graphite-sese-to-poly.c (schedule_error): New global.
8967 (add_loop_schedule): Handle empty domain by failing the
8969 (build_original_schedule): Handle schedule_error.
8971 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
8972 domain by returning an unchanged schedule.
8974 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8976 * genrecog.c (validate_pattern): For VEC_SELECT verify that
8977 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
8979 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
8981 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
8984 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8987 * expr.c (store_field): Don't optimize away bitsize == 0 store
8988 from CALL_EXPR with addressable return type.
8990 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8992 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
8993 * config/rs6000/rs6000.md (sel): Delete mode attribute.
8994 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
8995 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
8996 TARGET_ISEL instead of TARGET_ISEL<sel>.
8998 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9000 * config/rs6000/rs6000.c
9001 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9003 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9005 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9007 (reg_or_cint_operand): Fix comment.
9008 (reg_or_zero_operand): New predicate.
9009 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9010 * config/rs6000/rs6000.c (output_isel): Delete.
9011 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9012 instead of reg_or_cint_operand. Output instruction directly (not via
9014 (isel_unsigned_<mode>): Ditto.
9015 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9016 gpc_reg_operand. Add an instruction alternative for this. Output
9017 instruction directly.
9018 (*isel_reversed_unsigned_<mode>): Ditto.
9020 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9022 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9023 (TARGET_CANONICALIZE_COMPARISON): Define.
9025 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9028 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9029 Check whether the dest is REG before adding REG_EQUIV note.
9031 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9034 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9036 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9037 (make_hard_regno_born, make_hard_regno_dead): Update
9038 bb_killed_pseudos and bb_gen_pseudos.
9040 2017-10-11 Nathan Sidwell <nathan@acm.org>
9042 * incpath.h (enum incpath_kind): Name enum, prefix values.
9043 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9044 * incpath.c (heads, tails): Use INC_MAX.
9045 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9046 (merge_include_chains, split_quote_chain,
9047 register_include_chains): Update incpath_kind names.
9048 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9049 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9051 (add_framework_path, darwin_register_objc_includes): Likewise.
9052 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9054 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9056 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9057 Do not use float_operator operator predicate.
9058 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9059 * config/i386/predicates.md (float_operator): Remove predicate.
9061 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9063 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9064 (*jccxf_i387): Ditto.
9065 (*jcc<mode>_i387): Ditto.
9066 (*jccu<mode>_i387): Ditto.
9067 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9068 (*jcc_*_i387 splitters): Remove.
9069 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9070 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9071 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9074 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9076 * profile-count.h (slow_safe_scale_64bit): New function.
9077 (safe_scale_64bit): New inline.
9078 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9079 * profile-count.c: Include wide-int.h
9080 (slow_safe_scale_64bit): New.
9082 2017-10-11 Nathan Sidwell <nathan@acm.org>
9084 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9085 HAS_DECL_ASSEMBLER_NAME_P.
9086 * gimple-expr.c (gimple_decl_printable_name: Check
9087 HAS_DECL_ASSEMBLER_NAME_P too.
9088 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9089 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9090 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9091 * passes.c (rest_of_decl_compilation): Check
9092 HAS_DECL_ASSEMBLER_NAME_P too.
9093 * recog.c (verify_changes): Likewise.
9094 * tree-pretty-print.c (dump_decl_name): Likewise.
9095 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9097 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9098 (SET_DECL_ASSEMBLER_NAME): Use it.
9099 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9100 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9101 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9103 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9105 * config.gcc (i386, x86_64): Add extra objects.
9106 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9107 (ix86_min_insn_size): Declare.
9108 (ix86_issue_rate): Declare.
9109 (ix86_adjust_cost): Declare.
9110 (ia32_multipass_dfa_lookahead): Declare.
9111 (ix86_macro_fusion_p): Declare.
9112 (ix86_macro_fusion_pair_p): Declare.
9113 (ix86_bd_has_dispatch): Declare.
9114 (ix86_bd_do_dispatch): Declare.
9115 (ix86_core2i7_init_hooks): Declare.
9116 (ix86_atom_sched_reorder): Declare.
9117 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9118 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9119 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9120 (rip_relative_addr_p): Rename to ...
9121 (ix86_rip_relative_addr_p): ... this one; export.
9122 (memory_address_length): Update.
9123 (ix86_issue_rate): Move to x86-tune-sched.c.
9124 (ix86_flags_dependent): Move to x86-tune-sched.c.
9125 (ix86_agi_dependent): Move to x86-tune-sched.c.
9126 (exact_dependency_1): Move to x86-tune-sched.c.
9127 (exact_store_load_dependency): Move to x86-tune-sched.c.
9128 (ix86_adjust_cost): Move to x86-tune-sched.c.
9129 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9130 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9131 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9132 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9133 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9134 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9135 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9136 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9137 (min_insn_size): Rename to ...
9138 (ix86_min_insn_size): ... this one; export.
9139 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9140 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9141 (core2i7_first_cycle_multipass_backtrack): Move to
9142 x86-tune-sched-core.c.
9143 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9144 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9145 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9146 (ix86_avoid_jump_mispredicts): Update.
9147 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9148 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9149 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9150 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9151 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9152 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9153 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9154 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9155 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9156 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9157 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9158 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9159 (BIG): Move to ix86-tune-sched-bd.c.
9160 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9161 (enum insn_path): Move to ix86-tune-sched-bd.c.
9162 (get_mem_group): Move to ix86-tune-sched-bd.c.
9163 (is_cmp): Move to ix86-tune-sched-bd.c.
9164 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9165 (is_branch): Move to ix86-tune-sched-bd.c.
9166 (is_prefetch): Move to ix86-tune-sched-bd.c.
9167 (init_window): Move to ix86-tune-sched-bd.c.
9168 (allocate_window): Move to ix86-tune-sched-bd.c.
9169 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9170 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9171 (process_end_window): Move to ix86-tune-sched-bd.c.
9172 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9173 (find_constant): Move to ix86-tune-sched-bd.c.
9174 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9175 (has_immediate): Move to ix86-tune-sched-bd.c.
9176 (get_insn_path): Move to ix86-tune-sched-bd.c.
9177 (get_insn_group): Move to ix86-tune-sched-bd.c.
9178 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9179 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9180 (add_insn_window): Move to ix86-tune-sched-bd.c.
9181 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9182 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9183 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9184 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9185 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9186 (do_dispatch): Move to ix86-tune-sched-bd.c.
9187 (has_dispatch): Move to ix86-tune-sched-bd.c.
9188 * config/i386/t-i386: Add new object files.
9189 * config/i386/x86-tune-costs.h: New file.
9190 * config/i386/x86-tune-sched-atom.c: New file.
9191 * config/i386/x86-tune-sched-bd.c: New file.
9192 * config/i386/x86-tune-sched-core.c: New file.
9193 * config/i386/x86-tune-sched.c: New file.
9195 2017-10-11 Liu Hao <lh_mouse@126.com>
9197 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9198 the generic version below instead.
9199 (should_colorize): Recognize Windows consoles as terminals
9201 * pretty-print.c [__MINGW32__] (write_all): New function.
9202 [__MINGW32__] (find_esc_head): Likewise.
9203 [__MINGW32__] (find_esc_terminator): Likewise.
9204 [__MINGW32__] (eat_esc_sequence): Likewise.
9205 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9207 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9210 2017-10-11 Richard Biener <rguenther@suse.de>
9212 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9213 Properly call analyze_scalar_evolution with the loop of the stmt.
9215 2017-10-11 Richard Biener <rguenther@suse.de>
9217 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9218 * tree-core.h (tree_base): Add chrec_var union member.
9219 * tree.h (CHREC_VAR): Remove.
9220 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9221 * tree-chrec.h (build_polynomial_chrec): Adjust.
9222 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9223 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9225 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9227 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9228 * match.pd: ... here.
9229 ((T) X == (T) Y): Relax condition.
9231 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9233 PR tree-optimization/82472
9234 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9236 (break_alias_scc_partitions): Update postorder number.
9238 2017-10-11 Martin Liska <mliska@suse.cz>
9241 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9243 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9245 2017-10-11 Martin Liska <mliska@suse.cz>
9247 * print-rtl.c (print_insn): Move declaration of idbuf
9248 to same scope as name.
9250 2017-10-11 Martin Liska <mliska@suse.cz>
9255 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9258 2017-10-11 Martin Liska <mliska@suse.cz>
9261 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9264 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9266 * config.gcc (default_gnu_indirect_function): Default to yes for
9267 arm*-*-linux* with glibc.
9269 2017-10-11 Richard Biener <rguenther@suse.de>
9271 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9272 default-defs and types we do not want to analyze.
9273 (interpret_loop_phi): Replace unreachable code with an assert.
9274 (compute_scalar_evolution_in_loop): Remove and inline ...
9275 (analyze_scalar_evolution_1): ... here, replacing condition with
9276 what makes the intent clearer. Remove handling of cases
9277 get_scalar_evolution now handles.
9279 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9281 PR rtl-optimization/81434
9282 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9283 comment for main loop. In sched_group_found if, also add checks for
9284 pass and min_cost_group.
9286 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9288 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9289 (rs6000_insn_cost): New function.
9290 * config/rs6000/rs6000.md (cost): New attribute.
9292 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9293 H.J. Lu <hongjiu.lu@intel.com>
9297 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9298 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9299 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9300 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9301 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9302 OPTION_MASK_ISA_MMX and some other bit set require both
9303 mmx and the ISAs without the mmx bit.
9304 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9305 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9306 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9307 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9308 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9309 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9310 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9311 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9312 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9313 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9314 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9315 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9316 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9317 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9318 Add OPTION_MASK_ISA_MMX.
9320 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9322 * config.gcc (armv7*-*-freebsd*): New target.
9323 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9325 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9327 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9328 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9331 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9333 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9334 a template parameter.
9335 (WIDE_INT_REF_FOR): Update accordingly.
9336 * tree.h (wi::int_traits <const_tree>): Delete.
9337 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9338 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9339 (wi::tree_to_wide_ref): New typedef.
9340 (wi::to_wide): New function.
9341 * calls.c (get_size_range): Use wi::to_wide when operating on
9343 * cgraph.c (cgraph_node::create_thunk): Likewise.
9344 * config/i386/i386.c (ix86_data_alignment): Likewise.
9345 (ix86_local_alignment): Likewise.
9346 * dbxout.c (stabstr_O): Likewise.
9347 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9348 * expr.c (const_vector_from_tree): Likewise.
9349 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9350 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9351 (fold_negate_expr_1, int_const_binop_1, const_binop)
9352 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9353 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9354 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9355 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9356 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9357 (fold_not_const, round_up_loc): Likewise.
9358 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9359 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9360 (alloca_call_type): Likewise.
9361 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9362 * godump.c (go_output_typedef): Likewise.
9363 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9364 * internal-fn.c (get_min_precision): Likewise.
9365 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9366 * ipa-polymorphic-call.c
9367 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9368 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9369 (ipa_modify_call_arguments): Likewise.
9370 * match.pd: Likewise.
9371 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9372 * print-tree.c (print_node_brief, print_node): Likewise.
9373 * stmt.c (expand_case): Likewise.
9374 * stor-layout.c (layout_type): Likewise.
9375 * tree-affine.c (tree_to_aff_combination): Likewise.
9376 * tree-cfg.c (group_case_labels_stmt): Likewise.
9377 * tree-data-ref.c (dr_analyze_indices): Likewise.
9378 (prune_runtime_alias_test_list): Likewise.
9379 * tree-dump.c (dequeue_and_dump): Likewise.
9380 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9381 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9382 * tree-pretty-print.c (dump_generic_node): Likewise.
9383 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9384 (simple_iv_with_niters): Likewise.
9385 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9386 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9387 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9388 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9389 (refine_value_range_using_guard, number_of_iterations_ne_max)
9390 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9391 (get_cst_init_from_scev, record_nonwrapping_iv)
9392 (scev_var_range_cant_overflow): Likewise.
9393 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9394 * tree-ssa-pre.c (compute_avail): Likewise.
9395 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9396 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9397 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9398 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9399 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9400 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9401 (dump_case_nodes, try_switch_expansion): Likewise.
9402 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9403 (vect_do_peeling): Likewise.
9404 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9405 * tree-vect-stmts.c (vectorizable_load): Likewise.
9406 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9407 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9408 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9409 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9410 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9411 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9412 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9413 (evrp_dom_walker::before_dom_children): Likewise.
9414 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9415 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9416 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9417 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9418 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9419 (get_range_pos_neg): Likewise.
9420 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9421 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9422 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9423 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9424 * config/avr/avr.c (avr_fold_builtin): Likewise.
9425 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9426 * config/msp430/msp430.c (msp430_attr): Likewise.
9427 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9428 * config/powerpcspe/powerpcspe-c.c
9429 (altivec_resolve_overloaded_builtin): Likewise.
9430 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9431 (rs6000_expand_ternop_builtin): Likewise.
9432 * config/rs6000/rs6000-c.c
9433 (altivec_resolve_overloaded_builtin): Likewise.
9434 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9435 (rs6000_expand_ternop_builtin): Likewise.
9436 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9438 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9440 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9441 when copying loop nest with only one inner loop.
9443 2017-10-10 Richard Biener <rguenther@suse.de>
9445 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9446 blocks if SCEV is active.
9447 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9449 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9450 (scev_initialize): Assert we are not yet initialized.
9452 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9454 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9455 inner loop's exit stmt by making it always exit the loop, otherwise
9456 we would generate an infinite empty loop.
9458 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9460 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9461 renaming variables in new preheader if it's deleted.
9463 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9465 * tree-loop-distribution.c (struct partition): Remove unused field
9466 loops of the structure.
9467 (partition_alloc, partition_free): Ditto.
9468 (build_rdg_partition_for_vertex): Ditto.
9470 2017-10-09 Jeff Law <law@redhat.com>
9472 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9473 return type to match prototype and documentation.
9475 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9477 * config/rs6000/rs6000.c (processor_costs): Move to ...
9478 * config/rs6000/rs6000.h: ... here.
9479 (rs6000_cost): Declare.
9481 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9483 * except.c (setjmp_fn): New global variable.
9484 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9485 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9486 if DONT_USE_BUILTIN_SETJMP is defined.
9488 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9490 * target.def (insn_cost): New hook.
9491 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9492 * doc/tm.texi: Regenerate.
9493 * rtlanal.c (insn_cost): Use the new hook.
9495 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9497 * combine.c (combine_validate_cost): Compute the new insn_cost,
9498 not just pattern_cost.
9499 (try_combine): Adjust comment.
9501 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9503 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9505 * combine.c (uid_insn_cost): Adjust comment.
9506 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9508 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9509 * dse.c (find_shift_sequence): Ditto.
9510 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9511 (bb_valid_for_noce_process_p): Use pattern_cost.
9512 * rtl.h (insn_rtx_cost): Delete.
9513 (pattern_cost): New prototype.
9514 (insn_cost): New prototype.
9515 * rtlanal.c (insn_rtx_cost): Rename to...
9516 (pattern_cost): ... this.
9519 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9521 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9522 (*jcc<mode>_0_r_i387): Ditto.
9523 (*jccxf_r_i387): Ditto.
9524 (*jcc<mode>_r_i387): Ditto.
9525 (*jccu<mode>_r_i387): Ditto.
9526 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9527 (*jcc): Rename from *jcc_1.
9529 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9531 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9532 deferred rescans after the lvx/stvx recombination pre-pass.
9534 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9536 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9537 memory operation instruction support.
9538 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9539 (rs6000-ibm-aix[789]*): Likewise.
9540 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9541 Document new functions.
9543 2017-10-09 Richard Biener <rguenther@suse.de>
9545 PR tree-optimization/82397
9546 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9547 equality only for semantically equal trees.
9549 2017-10-09 Richard Biener <rguenther@suse.de>
9551 PR tree-optimization/82449
9552 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9553 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9554 allow constant addresses.
9555 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9558 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9560 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9563 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9566 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9569 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9572 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9574 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9577 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9578 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9580 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9582 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9583 (WI_BINARY_PREDICATE_RESULT): Likewise.
9584 (wi::binary_traits::operator_result): New type.
9585 (wi::binary_traits::predicate_result): Likewise.
9586 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9587 (generic_wide_int::operator==, generic_wide_int::operator!=)
9588 (generic_wide_int::operator&, generic_wide_int::and_not)
9589 (generic_wide_int::operator|, generic_wide_int::or_not)
9590 (generic_wide_int::operator^, generic_wide_int::operator+
9591 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9593 (operator~, unary operator-, operator==, operator!=, operator&)
9594 (operator|, operator^, operator+, binary operator-, operator*): New
9596 * expr.c (get_inner_reference): Use wi::bit_and_not.
9597 * fold-const.c (fold_binary_loc): Likewise.
9598 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9599 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9600 (bit_value_binop): Likewise.
9601 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9602 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9603 (extract_range_from_binary_expr_1): Likewise.
9604 (masked_increment): Likewise.
9605 (simplify_bit_ops_using_ranges): Likewise.
9607 2017-10-09 Martin Jambor <mjambor@suse.cz>
9610 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9611 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9612 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9614 (hsa_fixup_mov_insn_type): New function.
9615 (hsa_op_with_type::get_in_type): Use it.
9616 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9617 immediates in an assert.
9618 (hsa_op_with_type::extend_int_to_32bit): New method.
9619 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9620 types. Convert to dest type if necessary.
9621 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9622 (reg_for_gimple_ssa): Pass false as min32int to
9623 hsa_type_for_scalar_tree_type.
9624 (gen_hsa_addr): Fixup type when creating addresable temporary.
9625 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9626 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9627 necessary. Call hsa_fixup_mov_insn_type.
9628 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9629 extend operands and convert to dest type if necessary.
9630 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9631 to dest type if necessary.
9632 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9633 if conversion nt necessary and size matches.
9634 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9635 to dest type if necessary.
9636 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9637 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9639 (gen_hsa_clrsb): Likewise.
9640 (gen_hsa_ffs): Likewise.
9641 (gen_hsa_divmod): Extend operands and convert to dest type if
9643 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9645 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9647 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9648 Remove empty default arguments. Use a brace block as output
9650 (conditional return): Ditto.
9652 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9653 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9655 (group_ending_nop): Ditto.
9656 (doloop_end): Ditto.
9657 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9658 (splitters for those): Ditto.
9660 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9662 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9663 a conditional jump (and the compare for it) so that pc_rtx is the
9665 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9666 for the deleted and renamed ctr<mode>_internal[234] patterns.
9667 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9668 Delete second conditional return pattern.
9669 (ctr<mode>_internal2): Delete this second bdnz pattern.
9670 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9671 (ctr<mode>_internal4): Delete this second bdz pattern.
9673 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9675 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9676 (always_initialized_rtx_for_ssa_name_p): New predicate.
9677 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9678 (finish_out_of_ssa): Free new field of SA.
9679 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9680 * tree-ssa-coalesce.c: Include tree-ssa.h.
9681 (get_parm_default_def_partitions): Remove extern keyword.
9682 (get_undefined_value_partitions): New function.
9683 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9684 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9686 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9688 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9690 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9692 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9694 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9695 for targets that preffer 128bit.
9697 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9699 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9701 2017-10-08 Olivier Hainque <hainque@adacore.com>
9703 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9704 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9706 (thumb_set_return_address): Likewise.
9708 2017-10-08 Olivier Hainque <hainque@adacore.com>
9710 * common/config/arm/arm-common.c (arm_except_unwind_info):
9711 Handle DWARF2_UNWIND_INFO.
9713 2017-10-07 Michael Collison <michael.collison@arm.com>
9715 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9718 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9720 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9721 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9722 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9723 defined, force the creation of a new block for a dispatch label.
9725 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9727 * invoke.texi (Wsuggest-attribute=cold): Document.
9728 * common.opt (Wsuggest-attribute=cold): New
9729 * ipa-pure-const.c (warn_function_cold): New function.
9730 * predict.c (compute_function_frequency): Use it.
9731 * predict.h (warn_function_cold): Declare.
9733 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9735 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9738 2017-10-06 Martin Liska <mliska@suse.cz>
9740 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9741 keyword for member functions.
9742 (struct sanopt_tree_couple): New struct.
9743 (struct sanopt_tree_couple_hash): New function.
9744 (struct sanopt_ctx): Add new hash_map.
9745 (has_dominating_ubsan_ptr_check): New function.
9746 (record_ubsan_ptr_check_stmt): Likewise.
9747 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9748 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9749 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9751 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9754 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9755 aarch64_simd_valid_immediate on CONST_VECTORs.
9756 (aarch64_reg_or_bic_imm): Likewise.
9758 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9760 PR rtl-optimization/82396
9761 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9763 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9765 * graphite-dependences.c (scop_get_reads): Move code to...
9766 (scop_get_must_writes): Move code to...
9767 (scop_get_may_writes): Move code to...
9768 (scop_get_reads_and_writes): ... here.
9769 (scop_get_dependences): Call scop_get_reads_and_writes.
9771 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9773 PR tree-optimization/82434
9774 * fold-const.h (can_native_encode_type_p,
9775 can_native_encode_string_p): Remove.
9776 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9777 don't encode anything, just return what would be otherwise returned.
9778 (native_encode_fixed, native_encode_complex, native_encode_vector):
9780 (native_encode_string): Likewise. Inline by hand
9781 can_native_encode_string_p.
9782 (can_native_encode_type_p): Remove.
9783 (can_native_encode_string_p): Remove.
9784 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9785 STRING_CSTs using can_native_encode_string_p, test all
9786 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9787 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9788 argument from native_encode_expr.
9789 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9790 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9793 2017-10-06 Richard Biener <rguenther@suse.de>
9795 PR tree-optimization/82397
9796 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9797 operand_equal_p but rely on data_ref_compare_tree for detecting
9799 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9800 to match up with dr_group_sort_cmp.
9802 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9805 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9807 * config/s390/s390-builtin-types.def: Regenerate.
9809 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9812 * config/s390/s390-builtin-types.def: Regenerate.
9813 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9814 Change flag from B_VXE to B_VX.
9815 (s390_vec_min_dbl): Remove B_VXE flag.
9817 2017-10-06 Richard Biener <rguenther@suse.de>
9819 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9820 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9821 translate_isl_ast_to_gimple::is_valid_rename,
9822 translate_isl_ast_to_gimple::get_rename,
9823 translate_isl_ast_to_gimple::get_def_bb_for_const,
9824 translate_isl_ast_to_gimple::get_new_name,
9825 translate_isl_ast_to_gimple::collect_all_ssa_names,
9826 translate_isl_ast_to_gimple::copy_loop_phi_args,
9827 translate_isl_ast_to_gimple::collect_all_ssa_names,
9828 translate_isl_ast_to_gimple::copy_loop_phi_args,
9829 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9830 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9831 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9832 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9833 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9834 translate_isl_ast_to_gimple::copy_cond_phi_args,
9835 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9836 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9837 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9838 translate_isl_ast_to_gimple::rename_uses,
9839 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9840 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9841 (set_rename_for_each_def): Likewise.
9842 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9843 here. Handle rewriting SCEV analyzable uses here.
9844 (copy_bb_and_scalar_dependences): Generate code for PHI
9846 (graphite_regenerate_ast_isl): Adjust.
9847 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9848 (add_write, add_read): New functions.
9849 (build_cross_bb_scalars_def): Use it and simplify.
9850 (build_cross_bb_scalars_use): Likewise.
9851 (graphite_find_cross_bb_scalar_vars): Inline into...
9852 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9853 simulating out-of-SSA. Compute liveout and add dependencies.
9854 (build_scops): Force an empty entry block.
9855 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9857 (sese_build_liveouts): Declare.
9858 (sese_trivially_empty_bb_p): Likewise.
9859 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9860 compute liveout and debug_liveout.
9861 (sese_bad_liveouts_use): Remove.
9862 (sese_reset_debug_liveouts_bb): Likewise.
9863 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9864 (sese_build_liveouts): Build liveout and debug_liveout and store
9866 (new_sese_info): Adjust.
9867 (free_sese_info): Likewise.
9868 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9869 do not build liveout here.
9870 (move_sese_in_condition): Adjust region entry.
9871 (scev_analyzable_p): Match up with chrec_apply requirements.
9872 (sese_trivially_empty_bb_p): New.
9873 * tree-into-ssa.c (get_reaching_def): Properly support generating
9874 default-defs for incremental rewrite of anonymous names.
9876 2017-10-06 Richard Biener <rguenther@suse.de>
9878 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9879 precision do not perform modulo reduction.
9881 2017-10-06 Richard Biener <rguenther@suse.de>
9883 PR tree-optimization/82436
9884 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9885 conservatively choose the vectorization factor when checking
9886 whether we can perform the required load permutation.
9887 (vect_transform_slp_perm_load): Assert when we may not fail.
9889 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9891 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9892 message for incompatible -msdata=* and -mcall-* options.
9894 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9896 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9897 rate for post-reload scheduling.
9899 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9901 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9903 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9905 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9906 to improve monte carlo in scimark.
9908 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9910 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9911 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9912 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9913 pentium4_cost, nocona_cost): Set reassociation width to 1.
9914 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9915 width to 2 for fp operations and 1 otherwise.
9916 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9918 (atom_cost): Set reassociation width to 2.
9919 (slm_cost, generic_cost): Set fp reassociation width
9920 to 2 and 1 otherwise.
9921 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
9922 (core_cost): Set fp reassociation width to 4 and vector to 2.
9923 (ix86_reassociation_width): Rewrite using cost table; special case
9924 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
9925 and TARGET_AVX128_OPTIMAL.
9926 * config/i386/i386.h (processor_costs): Add
9927 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
9928 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
9929 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
9930 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
9931 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
9932 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
9934 2017-10-05 Nathan Sidwell <nathan@acm.org>
9936 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
9938 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9940 * config/arm/arm.c (arm_test_fpu_data): New.
9941 (arm_run_selftests): Call arm_test_fpu_data.
9943 2017-10-04 Nathan Sidwell <nathan@acm.org>
9945 * toplev.c (toplev::main): Remove excess parens on pretty_printer
9947 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
9949 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9951 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
9952 check type for aarch64_simd_valid_immediate.
9953 (aarch64_output_simd_mov_immediate): Update prototype.
9954 (aarch64_simd_valid_immediate): Update prototype.
9955 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
9956 support for ORR-immediate.
9957 (and<mode>3): modified pattern to add support for BIC-immediate.
9958 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
9959 now checks for valid immediate for BIC and ORR based on new enum
9961 (aarch64_output_simd_mov_immediate): Function now used to output
9962 BIC/ORR imm as well based on new enum argument.
9963 * config/aarch64/constraints.md (Do): New vector immediate constraint.
9965 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
9966 (aarch64_reg_or_bic_imm): Likewise.
9968 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9970 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
9971 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
9973 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9977 PR rtl-optimization/82396
9978 * haifa-sched.c (autopref_multipass_init): Simplify
9980 (autopref_rank_data): Simplify sort order.
9981 * sched-int.h (autopref_multipass_data_): Remove
9982 multi_mem_insn_p, min_offset and max_offset.
9984 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9986 * doc/sourcebuild.texi: Document vect_peeling_profitable.
9988 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9990 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
9993 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9995 * doc/sourcebuild.texi: Document vect_long_mult.
9997 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
9999 PR tree-optimization/82413
10000 * fold-const.c (build_range_check): Use widest_int when comparing
10001 the maximum ETYPE value with HIGH.
10003 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10005 PR rtl-optimization/82396
10006 * haifa-sched.c (autopref_multipass_init): Simplify
10008 (autopref_rank_data): Simplify sort order.
10009 * sched-int.h (autopref_multipass_data_): Remove
10010 multi_mem_insn_p, min_offset and max_offset.
10012 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10014 PR tree-optimization/82381
10015 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10016 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10019 PR tree-optimization/82374
10020 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10021 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10022 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10023 current_function_decl to the new decl.
10025 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10027 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10028 helper macro for IEEE float128 hardware built-in functions.
10029 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10031 (TRUNCF128_ODD): Likewise.
10032 (ADDF128_ODD): Likewise.
10033 (SUBF128_ODD): Likewise.
10034 (MULF128_ODD): Likewise.
10035 (DIVF128_ODD): Likewise.
10036 (FMAF128_ODD): Likewise.
10037 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10038 UNSPEC_TRUNC_ROUND_TO_ODD.
10039 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10040 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10041 floating point round to odd instructions.
10042 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10043 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10044 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10045 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10046 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10047 (trunc<mode>sf2_hw): Change the truncate with round to odd
10048 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10049 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10050 to odd hardware instructions.
10051 (sub<mode>3_odd): Likewise.
10052 (mul<mode>3_odd): Likewise.
10053 (div<mode>3_odd): Likewise.
10054 (sqrt<mode>2_odd): Likewise.
10055 (fma<mode>4_odd): Likewise.
10056 (fms<mode>4_odd): Likewise.
10057 (nfma<mode>4_odd): Likewise.
10058 (nfms<mode>4_odd): Likewise.
10059 (trunc<mode>df2_odd): Change the truncate with round to odd
10060 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10062 * doc/extend.texi (PowerPC built-in functions): Update documentation
10063 for existing IEEE float128-bit built-in functions. Add built-in
10064 functions that generate the IEEE 128-bit floating point round to
10067 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10069 PR rtl-optimization/77729
10070 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10071 to (X&(C1&~C2))|C2 transformations.
10073 2017-10-03 Martin Jambor <mjambor@suse.cz>
10075 PR tree-optimization/82363
10076 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10077 mismatch, mark lacc written regardless of racc.
10079 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10081 PR tree-optimization/82381
10082 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10083 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10084 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10085 fallthrough into reassoc_stmt_dominates_stmt_p.
10088 * combine.c (combine_instructions): Don't combine in unreachable
10091 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10094 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10095 function to not use the have_cpu variable. Do not set cpu_index,
10096 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10097 or the default cpu.
10098 (rs6000_valid_attribute_p): Remove duplicate initializations of
10099 old_optimize and func_optimize.
10100 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10101 (rs6000_activate_target_options): Make global.
10102 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10105 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10107 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10108 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10109 Set *poffset to 0 and *psize and *pmax_size to -1 if
10110 *poffset + *psize overflows in HOST_WIDE_INT.
10112 PR tree-optimization/82387
10113 PR tree-optimization/82388
10114 PR tree-optimization/82389
10115 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10116 instead of live_bytes non-NULL.
10118 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10121 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10124 2017-10-02 Richard Biener <rguenther@suse.de>
10126 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10127 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10128 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10130 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10132 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10133 requested precision matches the type's.
10134 * calls.c (alloc_max_size): Calculate the new candidate size as
10135 a widest_int and use wi::to_widest when comparing it with the
10136 current candidate size.
10137 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10138 zero rather than integer_zero_node.
10139 * match.pd: Check for a no-op conversion before using wi::add
10140 rather than after. Use tree_to_uhwi when summing small shift
10141 counts into an unsigned int.
10143 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10144 Alan Hayward <alan.hayward@arm.com>
10145 David Sherwood <david.sherwood@arm.com>
10148 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10149 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10150 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10151 POINTER_AND_FP_REGS.
10153 2017-10-02 Richard Biener <rguenther@suse.de>
10155 PR tree-optimization/82355
10156 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10157 a mapping for the enclosing loop but avoid generating one for
10158 the loop tree root.
10159 (copy_bb_and_scalar_dependences): Remove premature codegen
10160 error on PHIs in blocks duplicated into multiple places.
10161 * graphite-scop-detection.c
10162 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10163 in the region use it as loop and nest to analyze the DR in.
10164 (try_generate_gimple_bb): Likewise.
10165 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10166 (add_loop_constraints): For blocks in a loop not in the region
10167 create a dimension with a single iteration.
10168 * sese.h (gbb_loop_at_index): Remove assert.
10170 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10172 * omp-expand.c (adjust_context_scope): New function.
10173 (expand_parallel_call): Call adjust_context_scope.
10175 2017-10-01 Jeff Law <law@redhat.com>
10177 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10178 dom_opt_dom_walker class with direct access to private members.
10179 Add comments. Call test_for_singularity.
10180 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10181 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10182 m_dummy_cond anymore.
10183 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10185 (pass_dominator:execute): Build the dummy_cond here and pass it
10186 to the dom_opt_dom_walker ctor.
10187 (test_for_singularity): New function.
10189 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10190 Maya Rashish <coypu@sdf.org>
10192 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10193 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10194 (alpha*-*-netbsd*) Use nbsd_tm_file.
10195 (arm*-*-netbsdelf*) Likewise.
10196 (i[34567]86-*-netbsdelf*) Likewise.
10197 (x86_64-*-netbsd*) Likewise.
10198 (mips*-*-netbsd*) Likewise.
10199 (powerpc-*-netbsd*) Likewise.
10200 (sh*-*-netbsd*) Likewise.
10201 (sparc-*-netbsdelf*) Likewise.
10202 (sparc64-*-netbsd*) Likewise.
10203 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10205 (vax-*-netbsdelf*) Likewise.
10206 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10207 (UINT_FAST8_TYPE) Likewise.
10208 (INT_FAST16_TYPE) Check CHAR_FAST16.
10209 (UINT_FAST16_TYPE) Likewise.
10211 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10214 * config/i386/i386.md
10215 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10216 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10217 *divmodsi4_zext_2): New define_insn_and_split.
10218 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10219 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10220 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10221 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10222 New define_insn_and_split.
10223 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10224 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10225 operands[1] having DImode when mode is SImode.
10227 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10228 always SImode for DIV and MOD in REG_EQUAL notes.
10230 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10232 PR middle-end/82319
10233 * match.pd: Fix handling of NaNs in pattern.
10235 2017-09-29 Jeff Law <law@redhat.com>
10237 * sbitmap.c (bitmap_bit_in_range_p): New function.
10238 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10239 * tree-ssa-dse.c (live_bytes_read): New function.
10240 (dse_classify_store): Ignore reads of dead bytes.
10242 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10243 typos and whitespace errors.
10244 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10245 * config/s390/s390.c (s390_emit_prologue): Likewise.
10247 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10250 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10251 with a symbol for LRA.
10253 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10255 PR rtl-optimization/82338
10256 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10258 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10260 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10261 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10262 (qsort_chk): Declare.
10263 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10264 (qsort_chk): New function.
10266 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10268 PR tree-optimization/82337
10269 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10270 phi definition if the PHI result appears in an abnormal PHI.
10271 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10272 basis appears in an abnormal PHI.
10274 2017-09-29 Richard Biener <rguenther@suse.de>
10276 * graphite-isl-ast-to-gimple.c
10277 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10278 (binary_op_to_tree): Use it.
10279 (get_rename_from_scev): Likewise.
10280 (copy_loop_phi_nodes): Likewise.
10281 (copy_bb_and_scalar_dependences): Likewise.
10282 (translate_pending_phi_nodes): Likewise.
10284 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10287 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10288 for movabsq $(i32 << shift), r64.
10290 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10292 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10293 index when encoding %esp as %rsp to avoid 0x67 prefix.
10295 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10297 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10298 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10300 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10302 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10303 Extensions with more than 16 double VFP registers.
10304 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10305 to_clear_mask and all code related to it. Replace the remaining
10306 entry by a sbitmap and adapt code accordingly.
10308 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10310 * brig-builtins.def: Change pure attributes to const.
10312 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10314 * config.gcc (default_gnu_indirect_function): Default to yes for
10315 sparc*-*-linux* with glibc.
10317 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10319 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10320 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10321 when creating .init_array and .fini_array sections with priority
10324 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10327 * config/aarch64/aarch64.c
10328 (aarch64_builtin_support_vector_misalignment): Always return false
10329 when misalignment is unknown.
10331 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10333 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10334 this function to return false if the definition used by the swap
10335 instruction is artificial, or if the memory address from which the
10336 constant value is loaded is not represented by a base address held
10337 in a register or if the base address register is a frame or stack
10338 pointer. Additionally, return false if the base address of the
10339 loaded constant is a SYMBOL_REF but is not considered to be a
10341 (replace_swapped_load_constant): New function.
10342 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10343 loaded constant vector with a load of a swapped constant vector.
10345 2017-09-27 Carl Love <cel@us.ibm.com>
10347 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10348 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10349 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10352 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10354 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10355 first, always call autopref_rank_data otherwise.
10357 2017-09-27 Richard Biener <rguenther@suse.de>
10359 * graphite-scop-detection.c (find_scop_parameters): Move
10360 loop bound handling ...
10361 (gather_bbs::before_dom_children): ... here, avoiding the need
10362 to build scop_info->loop_nest.
10363 (record_loop_in_sese): Remove.
10364 * sese.h (sese_info_t::loop_nest): Remove.
10365 * sese.c (new_sese_info): Do not allocate loop_nest.
10366 (free_sese_info): Do not free loop_nest.
10368 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10371 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10372 lhs from calls if the lhs has addressable type.
10374 2017-09-27 Richard Biener <rguenther@suse.de>
10376 * graphite.h (scop::max_alias_set): New member.
10377 * graphite-scop-detection.c: Remove references to non-existing
10378 --param in comments.
10379 (build_alias_sets): Record the maximum alias set used for drs.
10380 (build_scops): Support zero as unlimited for
10381 --param graphite-max-arrays-per-scop.
10382 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10383 and inline into ...
10384 (build_poly_sr_1): ... here. Compute alias set based on the
10385 maximum alias set used for drs rather than
10386 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10388 2017-09-27 Richard Biener <rguenther@suse.de>
10390 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10391 --param loop-block-tile-size=0 to disable tiling.
10393 2017-09-27 Richard Biener <rguenther@suse.de>
10395 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10396 (graphite-max-nb-scop-params): Document special value zero.
10397 * domwalk.h (dom_walker::STOP): New symbolical constant.
10398 (dom_walker::dom_walker): Add optional parameter for bb to
10400 (dom_walker::~dom_walker): Declare.
10401 (dom_walker::before_dom_children): Document STOP return value.
10402 (dom_walker::m_user_bb_to_rpo): New member.
10403 (dom_walker::m_bb_to_rpo): Likewise.
10404 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10405 mapping here if not provided by the user.
10406 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10407 provided by the user.
10408 (dom_walker::STOP): Define.
10409 (dom_walker::walk): Do not compute bb to RPO mapping here.
10410 Support STOP return value from before_dom_children to stop
10412 * graphite-optimize-isl.c (optimize_isl): If the schedule
10413 is the same still generate code if -fgraphite-identity
10414 or -floop-parallelize-all are given.
10415 * graphite-scop-detection.c: Include cfganal.h.
10416 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10418 (gather_bbs::before_dom_children): Return STOP for BBs
10420 (build_scops): Compute bb to RPO mapping and pass it to
10421 the domwalk. Treat --param graphite-max-nb-scop-params=0
10422 as not limiting the number of params.
10423 * graphite.c (graphite_initialize): Remove limit on the
10424 number of basic-blocks in a function.
10425 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10426 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10427 default value of 10.
10429 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10431 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10432 Adjust code to eliminate needing to do the shift right 32-bits
10433 operation after XSCVDPSPN.
10435 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10437 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10438 ((X / Y) != 0 -> X >= Y): Likewise.
10440 2017-09-26 Carl Love <cel@us.ibm.com>
10442 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10443 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10444 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10445 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10446 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10447 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10448 definitions and overloading.
10449 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10450 statement for P9V_BUILTIN_XST_LEN_R.
10451 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10452 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10453 define_expand and define_insn for the instructions and builtins.
10454 * doc/extend.texi: Update the built-in documentation file for the new
10455 built-in functions.
10456 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10457 define_insn for the instructions
10459 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10462 * gcc/config/netbsd-protos.h: New file.
10463 * gcc/config/netbsd.c: New file.
10464 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10465 * gcc/config/t-netbsd: New file.
10466 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10467 (tmake_file) Add t-netbsd.
10468 (extra_objs) Add netbsd.o.
10470 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10474 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10476 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10478 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10479 sign extension from a vector register to a GPR by doing a 32-bit
10480 direct move and then an EXTSW.
10481 (extendsi<mode>2 splitter): Likewise.
10482 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10483 right or vector extract after doing XSCVDPSPN. Use
10484 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10486 (movdi_from_sf_zero_ext): Likewise.
10487 (reload_gpr_from_vsxsf): Likewise.
10488 (p8_mfvsrd_4_disf): Delete, no longer used.
10489 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10490 then needing to move the SFmode to a GPR to use the XSCVDPSP
10491 instruction instead of FRSP and XSCVDPSPN.
10492 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10493 it is adjacent to the other XSCVSPDP insns.
10494 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10495 SFmode to be in traditional Altivec registers.
10496 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10497 (vsx_xscvspdpn): Likewise.
10498 (vsx_xscvspdpn_scalar): Likewise.
10500 2017-09-26 Martin Jambor <mjambor@suse.cz>
10502 * tree-sra.c (compare_access_positions): Put integral types first,
10503 stabilize sorting of integral types, remove conditions putting
10504 non-full-precision integers last.
10505 (sort_and_splice_var_accesses): Disable scalarization if a
10506 non-integert would be represented by a non-full-precision integer.
10508 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10510 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10511 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10512 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10513 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10514 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10515 conditionals inside the function instead of around it. Call
10516 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10517 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10519 2017-09-26 Richard Biener <rguenther@suse.de>
10521 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10523 (scop_detection::build_scop_breadth): ... this. Removed.
10524 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10525 (scop_detection::harmful_stmt_in_bb): Likewise.
10526 (scop_detection::graphite_can_represent_stmt): Likewise.
10527 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10528 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10529 (scop_detection::can_represent_loop_1): ... this. Removed.
10530 (scop_detection::harmful_loop_in_region): Simplify after inlining
10531 the above and remove more quadraticness.
10532 (build_scops): Adjust.
10533 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10536 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10539 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10540 REGNO (base) == SP_REG if base is a REG.
10542 PR middle-end/35691
10543 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10544 if it is different SSA_NAME.
10545 (optimize_range_tests_cmp_bitwise): New function.
10546 (optimize_range_tests): Call it.
10548 2017-09-26 Richard Biener <rguenther@suse.de>
10550 PR tree-optimization/82321
10551 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10552 for the def being inside the loop.
10554 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10556 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10558 * config/s390/s390-builtins.def: Fix constraint on op4.
10560 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10562 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10563 independent expanders.
10564 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10565 ("vec_ordered", "vec_unordered"): New expanders.
10567 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10569 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10572 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10574 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10575 vec_unpacks_lo_v16qi.
10576 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10578 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10580 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10581 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10582 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10584 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10586 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10588 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10590 ("*vec_slb<mode>"): New insn pattern.
10591 ("vec_shr_<mode>"): New expander.
10592 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10593 and force the shift count operand to V16QImode.
10594 ("vec_srb<mode>"): Set shift count mode to V16QI.
10596 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10598 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10599 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10600 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10602 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10605 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10607 2017-09-26 Richard Biener <rguenther@suse.de>
10609 PR tree-optimization/82320
10610 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10613 2017-09-25 Jeff Law <law@redhat.com>
10615 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10616 prototype for new argument.
10617 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10618 mostly extracted from rs6000_emit_allocate_stack.
10619 (rs6000_emit_probe_stack_range_stack_clash): New function.
10620 (rs6000_emit_allocate_stack): Call
10621 rs6000_emit_probe_stack_range_stack_clash as needed.
10622 (rs6000_emit_probe_stack_range): Add additional argument
10623 to call to gen_probe_stack_range{si,di}.
10624 (output_probe_stack_range): New.
10625 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10626 (output_probe_stack_range_stack_clash): New.
10627 (rs6000_emit_prologue): Emit notes into dump file as requested.
10628 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10629 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10630 Add additional operand and pass it to output_probe_stack_range.
10632 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10634 PR tree-optimization/82163
10635 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10636 (checking_verify_loop_closed_ssa): New parameter.
10637 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10638 (check_loop_closed_ssa_stmt): Delete.
10639 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10640 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10641 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10644 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10646 * brig-builtins.def: Treat HSAIL barrier builtins as
10647 setjmp/longjump style functions.
10649 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10651 * target.def (constant_alignment): New hook.
10652 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10653 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10654 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10655 * doc/tm.texi: Regenerate.
10656 * targhooks.h (default_constant_alignment): Declare.
10657 (constant_alignment_word_strings): Likewise.
10658 * targhooks.c (default_constant_alignment): New function.
10659 (constant_alignment_word_strings): Likewise.
10660 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10661 instead of CONSTANT_ALIGNMENT.
10662 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10663 (force_const_mem): Likewise.
10664 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10665 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10666 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10667 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10668 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10670 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10671 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10672 constant_alignment_word_strings.
10673 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10674 (CONSTANT_ALIGNMENT): Likewise.
10675 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10676 (arm_constant_alignment): New function.
10677 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10678 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10679 constant_alignment_word_strings.
10680 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10681 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10682 constant_alignment_word_strings.
10683 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10684 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10685 (cris_constant_alignment): New function.
10686 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10687 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10688 (epiphany_constant_alignment): New function.
10689 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10690 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10691 constant_alignment_word_strings.
10692 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10693 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10694 constant_alignment_word_strings.
10695 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10696 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10697 constant_alignment_word_strings.
10698 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10699 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10700 * config/i386/i386.c (ix86_constant_alignment): Make static.
10701 Use the same interface as the target hook.
10702 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10703 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10704 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10705 constant_alignment_word_strings.
10706 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10707 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10708 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10709 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10710 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10711 constant_alignment_word_strings.
10712 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10713 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10714 constant_alignment_word_strings.
10715 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10716 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10717 constant_alignment_word_strings.
10718 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10719 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10721 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10722 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10723 * config/mips/mips.c (mips_constant_alignment): New function.
10724 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10725 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10726 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10727 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10728 (mmix_constant_alignment): Make static. Use the same interface
10729 as the target hook.
10730 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10731 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10732 constant_alignment_word_strings.
10733 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10734 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10735 constant_alignment_word_strings.
10736 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10737 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10738 constant_alignment_word_strings.
10739 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10740 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10741 (rs6000_constant_alignment): New function.
10742 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10743 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10744 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10745 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10746 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10747 (rs6000_constant_alignment): New function.
10748 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10749 * config/s390/s390.c (s390_constant_alignment): New function.
10750 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10751 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10752 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10753 constant_alignment_word_strings.
10754 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10755 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10756 (sparc_constant_alignment): New function.
10757 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10758 * config/spu/spu.c (spu_constant_alignment): New function.
10759 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10760 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10761 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10762 constant_alignment_word_strings.
10763 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10764 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10765 constant_alignment_word_strings.
10766 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10767 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10768 constant_alignment_word_strings.
10769 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10770 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10771 (visium_constant_alignment): New function.
10772 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10773 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10774 (xtensa_constant_alignment): New function.
10775 * system.h (CONSTANT_ALIGNMENT): Poison.
10777 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10779 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10780 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10781 (rs6000_builtin_valid_without_lhs): New helper function.
10782 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10783 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10785 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10787 * target.h (vec_perm_indices): Use unsigned short rather than
10789 (auto_vec_perm_indices): Likewise.
10790 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10791 Use unsigned int rather than unsigned char.
10792 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10794 2017-09-25 Richard Biener <rguenther@suse.de>
10796 * cfgloop.h (sort_sibling_loops): Declare.
10797 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10798 (sort_sibling_loops): New function sorting the sibling loop list
10800 * graphite.c (graphite_transform_loops): Sort sibling loops.
10802 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10804 * target.def (vec_perm_const_ok): Change sel parameter to
10806 * optabs-query.c (can_vec_perm_p): Update accordingly.
10807 * doc/tm.texi: Regenerate.
10808 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10809 auto_vec_perm_indices and remove separate nelt field.
10810 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10811 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10812 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10813 (aarch64_expand_vec_perm_const): Update accordingly.
10814 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10815 to vec_perm_indices.
10816 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10817 auto_vec_perm_indices and remove separate nelt field.
10818 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10819 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10820 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10822 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10823 to vec_perm_indices.
10824 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10825 sel to vec_perm_indices.
10826 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10827 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10828 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10830 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10833 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10836 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10837 on the FUNCTION_DECL function context if it has a DIE that is a
10840 2017-09-25 Richard Biener <rguenther@suse.de>
10842 PR tree-optimization/82285
10843 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10846 2017-09-25 Tom de Vries <tom@codesourcery.com>
10850 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10853 2017-09-25 Richard Biener <rguenther@suse.de>
10855 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10856 ISL errors other than isl_error_quota happen. Dump if the
10857 schedule is the same.
10858 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10859 errors instead of aborting inside ISL.
10861 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10864 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10865 of libgcc_eh for m64.
10866 * config/i386/darwin64.h: Likewise.
10868 2017-09-25 Richard Biener <rguenther@suse.de>
10870 PR middle-end/82144
10871 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10872 attribute for incomplete types nor twice for complete ones.
10874 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10877 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10878 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10881 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10884 * config/i386/i386.opt (mprefer-avx256): Use
10885 ix86_target_flags variable.
10886 * config/i386/i386.c (ix86_target_string): Move
10887 -mprefer-avx256 to flag2_opts.
10889 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10891 PR middle-end/35691
10892 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10893 and x != -1 | y != -1 into (x & y) != -1.
10895 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10897 * config.gcc: Add new case statement to set
10898 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10899 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10900 s390x-*-linux* case statements. Added aarch64 to the list of
10901 supported architectures.
10903 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10905 PR tree-optimization/82289
10906 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10907 STMT_VINFO_RELEVANT_P.
10909 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10910 Alan Hayward <alan.hayward@arm.com>
10911 David Sherwood <david.sherwood@arm.com>
10913 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10914 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10915 (extract_range_from_binary_expr_1): Don't call
10916 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10918 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10919 Alan Hayward <alan.hayward@arm.com>
10920 David Sherwood <david.sherwood@arm.com>
10922 * target.def (preferred_vector_alignment): New hook.
10923 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
10925 * doc/tm.texi: Regenerate.
10926 * targhooks.h (default_preferred_vector_alignment): Declare.
10927 * targhooks.c (default_preferred_vector_alignment): New function.
10928 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
10930 (DR_TARGET_ALIGNMENT): New macro.
10931 (aligned_access_p): Update commentary.
10932 (vect_known_alignment_in_bytes): New function.
10933 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
10935 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
10936 Calculate the misalignment based on the target alignment rather than
10938 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
10939 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
10940 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
10941 the target alignment, rather than masking the element misalignment
10942 with the number of elements in a vector. Also use the target
10943 alignment when calculating the maximum number of peels.
10944 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
10945 instead of TYPE_ALIGN_UNIT.
10946 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
10947 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
10948 (vect_create_addr_base_for_vector_ref): Update call accordingly.
10949 (vect_create_data_ref_ptr): Likewise.
10950 (vect_setup_realignment): Realign by ANDing with
10951 -DR_TARGET_MISALIGNMENT.
10952 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
10953 the number of peels based on DR_TARGET_ALIGNMENT.
10954 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
10955 with the guaranteed alignment boundary when deciding whether
10957 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
10958 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
10959 (ensure_base_align): Remove stmt_info parameter. Get the
10960 target base alignment from DR_TARGET_ALIGNMENT.
10961 (vectorizable_store): Update call accordingly. Interpret
10962 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
10964 (vectorizable_load): Likewise.
10966 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10967 Alan Hayward <alan.hayward@arm.com>
10968 David Sherwood <david.sherwood@arm.com>
10970 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
10971 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
10972 (vect_enhance_data_refs_alignment): Likewise.
10974 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10976 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10977 error. Only quit immediately if parsing is complete.
10978 (BEGIN): Initialize fatal_err and parse_done.
10979 (begin fpu, end fpu): Check number of arguments.
10980 (begin arch, end arch): Likewise.
10981 (begin cpu, end cpu): Likewise.
10982 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10983 (optalias): Likewise.
10985 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10987 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10988 * config/arm/arm-isa.h: Delete. Move definitions to ...
10989 * arm-cpus.in: ... here. Use new feature and fgroup values.
10990 * config/arm/arm.c (arm_option_override): Use lower case for feature
10992 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10993 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10994 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10996 (print_isa_bits_for): New function.
10997 (gen_isa): New function.
10998 (gen_comm_data): Use print_isa_bits_for.
10999 (define feature): New keyword.
11000 (define fgroup): New keyword.
11001 * config/arm/t-arm (TM_H): Remove.
11002 (GTM_H): Add arm-isa.h.
11003 (arm-isa.h): Add rule to generate file.
11004 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11005 case for feature bit names.
11007 2017-09-22 Richard Biener <rguenther@suse.de>
11009 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11011 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11012 print of no dependency loops ...
11013 * graphite.c (graphite_transform_loops): ... here.
11014 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11016 (same_close_phi_node, remove_duplicate_close_phi,
11017 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11018 (canonicalize_loop_closed_ssa): ... here and simplify.
11019 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11020 (optimize_isl): Use dump_printf_loc to tell when we stopped
11021 optimizing because of an ISL timeout.
11023 2017-09-22 Richard Biener <rguenther@suse.de>
11025 PR tree-optimization/82291
11026 * tree-if-conv.c (predicate_mem_writes): Make sure to
11027 remove writes in blocks predicated with false.
11029 2017-09-22 Richard Biener <rguenther@suse.de>
11031 * sese.c: Include cfganal.h.
11032 (if_region_set_false_region): Remove.
11033 (create_if_region_on_edge): Likewise.
11034 (move_sese_in_condition): Re-implement without destroying
11037 2017-09-22 Richard Biener <rguenther@suse.de>
11039 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11040 Verify both BBs contain loop PHI nodes before dispatching to
11041 copy_loop_phi_args.
11042 (graphite_regenerate_ast_isl): Do not recompute dominators,
11043 do not verify three times. Restructure for clarity.
11044 * graphite-scop-detection.c (same_close_phi_node,
11045 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11046 defined_in_loop_p, canonicalize_loop_closed_ssa,
11047 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11048 checking and SSA rewrite, move to ...
11049 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11050 (graphite_initialize): Do not pass in ctx, do not reset the
11051 SCEV cache, compute only dominators.
11052 (graphite_transform_loops): Allocate ISL ctx after
11053 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11054 Maintain post-dominators only around build_scops.
11055 * sese.c (if_region_set_false_region): Make static. Free
11056 and recompute dominators.
11057 (move_sese_in_condition): Assert we don't get called with
11058 post-dominators computed.
11059 * sese.h (if_region_set_false_region): Remove.
11061 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11063 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11064 mode attribute for TARGET_AVX512VL.
11066 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11068 * config/i386/i386.opt (mprefer-avx256): New option.
11069 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11071 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11072 for TARGET_PREFER_AVX256.
11073 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11075 2017-09-21 Jeff Law <law@redhat.com>
11077 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11078 Fix dump output if the only stack space is for pushed registers.
11080 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11082 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11085 2017-09-21 Martin Sebor <msebor@redhat.com>
11088 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11089 code (in C++) or code that triggers warnings.
11091 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11093 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11095 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11097 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11098 * sched-rgn.c: Ditto.
11099 * sel-sched-ir.c: Ditto.
11101 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11103 * toplev.h (set_random_seed): Adjust return type.
11104 * toplev.c (init_local_tick): Move eager initialization of random_seed
11105 to get_random_seed. Adjust comment.
11106 (init_random_seed): Inline to get_random_seed, delete.
11107 (get_random_seed): Initialize random_seed lazily.
11108 (set_random_seed): Do not return previous value.
11109 (print_switch_value): Do not call get_random_seed.
11111 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11113 * cgraph.c (delete_function_version): New, broken out from...
11114 (cgraph_node::delete_function_version): ...here. Rename to
11115 cgraph_node::delete_function_version_by_decl. Update all uses.
11116 (cgraph_node::remove): Call delete_function_version.
11118 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11121 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11122 VAR_DECLs to which addressable non-volatile parameters are mapped
11123 and for id->retvar after the return value assignment. Clear
11124 id->retval and id->retbnd after inlining.
11126 2017-09-21 Richard Biener <rguenther@suse.de>
11128 PR tree-optimization/82276
11129 PR tree-optimization/82244
11130 * tree-vrp.c (build_assert_expr_for): Set
11131 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11133 (remove_range_assertions): Revert earlier change.
11135 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11138 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11140 2017-09-21 Richard Biener <rguenther@suse.de>
11142 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11143 Restore valid IL after code generation errors.
11144 * graphite.c (graphite_transform_loops): Diagnose code
11145 generation issues as MSG_MISSED_OPTIMIZATION and continue
11146 with processing SCOPs.
11148 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11149 Alan Hayward <alan.hayward@arm.com>
11150 David Sherwood <david.sherwood@arm.com>
11152 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11153 rather than choosing between plus_constant and gen_rtx_<CODE>.
11154 * expr.c (emit_push_insn): Likewise.
11155 (expand_expr_real_2): Likewise.
11157 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11158 Alan Hayward <alan.hayward@arm.com>
11159 David Sherwood <david.sherwood@arm.com>
11161 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11163 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11164 Alan Hayward <alan.hayward@arm.com>
11165 David Sherwood <david.sherwood@arm.com>
11167 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11168 calling tree_to_uhwi.
11170 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11171 Alan Hayward <alan.hayward@arm.com>
11172 David Sherwood <david.sherwood@arm.com>
11174 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11175 INTEGER_CST rather than a negative test for ADDR_EXPR.
11177 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11178 Alan Hayward <alan.hayward@arm.com>
11179 David Sherwood <david.sherwood@arm.com>
11181 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11182 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11184 2017-09-21 Richard Biener <rguenther@suse.de>
11186 PR tree-optimization/71351
11187 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11188 graphite_create_new_loop_guard): Remove, fold remaining parts
11190 (translate_isl_ast_node_for): ... here and simplify.
11192 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11195 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11196 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11197 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11198 alternative always use QI mode, for -Os imov (=R,R) alternative
11199 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11202 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11203 Jeff Law <law@redhat.com>
11205 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11206 (allocate_stack_space): New function, partially extracted from
11207 s390_emit_prologue.
11208 (s390_emit_prologue): Track offset to most recent stack probe.
11209 Code to allocate space moved into allocate_stack_space.
11210 Dump actions when no stack is allocated.
11211 (s390_prologue_plus_offset): New function.
11212 (s390_emit_stack_probe): Likewise.
11214 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11216 * common.opt (Wa, Wl, Wp, g, gz=): Add
11218 (gno-column-info): Remove.
11219 (gcolumn-info): Drop RejectNegative.
11220 (gno-): New prefix.
11221 (gno-record-gcc-switches): Remove.
11222 (grecord-gcc-switches): Drop RejectNegative.
11223 (gno-split-dwarf): Remove.
11224 (gsplit-dwarf): Drop RejectNegative.
11225 (gno-strict-dwarf): Remove.
11226 (gstrict-dwarf): Drop RejectNegative.
11227 * config/darwin.opt (gfull, gused): Add RejectNegative.
11228 * dwarf2out.c (gen_producer_string): Drop
11229 gno-record-gcc-switches handler.
11230 * optc-gen.awk: Add g to prefixes with negative forms.
11231 * opts-common.c (remapping_prefix_p): New.
11232 (find_opt): Check it.
11233 (generate_canonical_option): Test g prefix.
11234 (option_map): Add -gno- mapping.
11235 (add_misspelling_candidates): Check remapping_prefix_p.
11237 2017-09-20 Jeff Law <law@redhat.com>
11239 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11240 thinko in stack clash protection support.
11242 * explow.c (compute_stack_clash_protection_loop_data): Use
11243 CONST_INT_P instead of explicit test. Verify object is a
11244 CONST_INT_P before looking at INTVAL.
11245 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11246 instead of explicit test.
11248 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11251 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11252 address instead of to r1 and r11.
11254 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11256 * config.gcc: Support "knm".
11257 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11258 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11260 * config/i386/i386.c (m_KNM): Define.
11261 (processor_target_table): Add "knm".
11263 (ix86_option_override_internal): Add "knm".
11264 (ix86_issue_rate): Add PROCESSOR_KNM.
11265 (ix86_adjust_cost): Ditto.
11266 (ia32_multipass_dfa_lookahead): Ditto.
11267 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11268 (fold_builtin_cpu): Add M_INTEL_KNM.
11269 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11270 (processor_type): Add PROCESSOR_KNM.
11271 * config/i386/x86-tune.def: Add m_KNM.
11272 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11274 2017-09-20 Richard Biener <rguenther@suse.de>
11276 PR tree-optimization/80213
11277 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11278 are allowed in empty BBs as well.
11279 (canonicalize_loop_closed_ssa): Also look for other complex
11281 (scop_detection::get_sese): Include the loop-closed PHI block
11283 (scop_detection::merge_sese): Remove code adding extra blocks.
11284 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11285 (build_scops): Assert the final returned scop is invalid.
11287 2017-09-20 Richard Biener <rguenther@suse.de>
11289 PR tree-optimization/82264
11290 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11292 (vn_phi_lookup): Likewise.
11293 (vn_phi_insert): Likewise.
11295 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11297 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11298 that fits into uhwi or shwi, add DW_AT_const_value regardless
11299 of early_dwarf without going through RTL, using add_AT_unsigned
11302 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11303 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11304 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11305 (DEBUG_MACRO_SECTION): Likewise.
11306 (DEBUG_LTO_MACRO_SECTION): Likewise.
11307 (DEBUG_STR_DWO_SECTION): Likewise.
11308 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11309 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11310 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11311 (DEBUG_LTO_LINE_STR_SECTION): Define.
11312 (init_sections_and_labels): Initialize debug_line_str_section
11313 variable. Initialize debug_loc_section for -gdwarf-5 to
11314 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11316 2017-09-20 Richard Biener <rguenther@suse.de>
11318 * graphite-sese-to-poly.c (extract_affine): Properly handle
11319 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11321 2017-09-20 Richard Biener <rguenther@suse.de>
11323 PR tree-optimization/81373
11324 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11325 Force SESE live-out defs to be handled even if they are
11328 2017-09-19 Jeff Law <law@redhat.com>
11330 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11331 nothing for stack adjustments with REG_STACK_CHECK.
11332 * sched-deps.c (parse_add_or_inc): Reject insns with
11333 REG_STACK_CHECK from dependency breaking.
11334 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11335 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11336 * reg-notes.def (STACK_CHECK): New note.
11338 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11339 (ix86_expand_prologue): Dump stack clash info as needed.
11340 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11342 * function.c (dump_stack_clash_frame_info): New function.
11343 * function.h (dump_stack_clash_frame_info): Prototype.
11344 (enum stack_clash_probes): New enum.
11346 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11347 flag_stack_clash_protection.
11348 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11349 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11350 (arm_frame_pointer_required): Likewise.
11351 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11352 (ia64_expand_prologue): Likewise.
11353 * config/mips/mips.c (mips_expand_prologue): Likewise.
11354 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11355 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11356 (sparc_flat_expand_prologue): Likewise.
11357 * config/spu/spu.c (spu_expand_prologue): Likewise.
11359 * explow.c: Include "params.h".
11360 (anti_adjust_stack_and_probe_stack_clash): New function.
11361 (get_stack_check_protect): Likewise.
11362 (compute_stack_clash_protection_loop_data): Likewise.
11363 (emit_stack_clash_protection_loop_start): Likewise.
11364 (emit_stack_clash_protection_loop_end): Likewise.
11365 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11366 Use anti_adjust_stack_and_probe_stack_clash.
11367 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11368 (emit_stack_clash_protection_loop_start): Likewise.
11369 (emit_stack_clash_protection_loop_end): Likewise.
11370 * rtl.h (get_stack_check_protect): Prototype.
11371 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11372 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11373 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11375 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11377 * doc/tm.texi: Rebuilt.
11378 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11379 get_stack_check_protect.
11380 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11381 * config/arm/arm.c (arm_expand_prologue): Likewise.
11382 (arm_frame_pointer_required): Likewise.
11383 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11384 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11385 * config/mips/mips.c (mips_expand_prologue): Likewise.
11386 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11387 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11388 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11389 (sparc_flat_expand_prologue): Likewise.
11391 * common.opt (-fstack-clash-protection): New option.
11392 * flag-types.h (enum stack_check_type): Note difference between
11393 -fstack-check= and -fstack-clash-protection.
11394 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11395 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11396 * toplev.c (process_options): Issue warnings/errors for cases
11397 not handled with -fstack-clash-protection.
11398 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11399 (-fstack-check): Note additional problem with -fstack-check=generic.
11400 Note that -fstack-check is primarily for Ada and refer users
11401 to -fstack-clash-protection for stack-clash-protection.
11402 Document new params for stack clash protection.
11404 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11406 * config/i386/i386.c (ix86_split_long_move): Do not handle
11407 address used for LEA in a special way.
11409 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11411 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11413 2017-09-19 Martin Sebor <msebor@redhat.com>
11416 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11417 of incompatible types.
11419 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11421 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11422 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11423 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11424 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11426 2017-09-19 Richard Biener <rguenther@suse.de>
11428 PR tree-optimization/82244
11429 * tree-vrp.c (remove_range_assertions): Do not propagate
11430 a constant to abnormals but replace the assert with a copy.
11432 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11434 PR rtl-optimization/57878
11435 PR rtl-optimization/68988
11436 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11437 avoidance test involving non_reload_pseudos. Move frequency test
11438 below the general fragmentation avoidance test.
11440 2017-09-19 Richard Biener <rguenther@suse.de>
11442 PR tree-optimization/69728
11443 * graphite-sese-to-poly.c (schedule_error): New global.
11444 (add_loop_schedule): Handle empty domain by failing the
11446 (build_original_schedule): Handle schedule_error.
11448 2017-09-19 Richard Biener <rguenther@suse.de>
11450 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11451 Do not iterate to sibling loops but only to siblings of inner
11454 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11457 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11458 that the registers are different.
11460 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11462 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11463 to processor_model and "amdfam17h" to arch_names_table.
11464 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11466 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11469 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11471 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11472 Alan Hayward <alan.hayward@arm.com>
11473 David Sherwood <david.sherwood@arm.com>
11475 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11477 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11479 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11480 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11481 vect_schedule_slp_instance.
11482 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11483 Update call to vect_slp_analyze_node_operations. Simplify return
11485 (vect_slp_analyze_bb_1): Update call accordingly.
11486 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11487 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11488 (vect_schedule_slp): Update call accordingly.
11490 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11491 Alan Hayward <alan.hayward@arm.com>
11492 David Sherwood <david.sherwood@arm.com>
11494 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11495 with types that aren't in fact scalar.
11497 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11499 * tree-vect-slp.c (vect_record_max_nunits): New function,
11501 (vect_build_slp_tree_1): ...here.
11502 (vect_build_slp_tree_2): Call it for phis too.
11504 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11506 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11507 to vect_get_vec_def_for_operand when getting the mask operand.
11509 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11510 Alan Hayward <alan.hayward@arm.com>
11511 David Sherwood <david.sherwood@arm.com>
11513 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11516 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11517 Alan Hayward <alan.hayward@arm.com>
11518 David Sherwood <david.sherwood@arm.com>
11520 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11521 calculation for vector booleans.
11523 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11524 Alan Hayward <alan.hayward@arm.com>
11525 David Sherwood <david.sherwood@arm.com>
11527 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11529 (vect_transform_stmt): ...here.
11530 (vect_analyze_stmt): Use it instead of calling
11531 vectorizable_live_operation directly.
11533 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11535 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11536 non-SIMT targets in acc vector loops.
11538 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11540 * configure.ac: Add arc and check if assembler supports gdwarf2.
11541 * configure: Regenerate.
11543 2017-09-18 Richard Biener <rguenther@suse.de>
11545 PR tree-optimization/82220
11546 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11547 epilogue niters from the min_profitable_iters compute.
11549 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11552 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11553 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11554 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11556 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11559 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11560 switching to a new text section.
11562 2017-09-18 Richard Biener <rguenther@suse.de>
11564 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11566 (build_alias_set): Reject aliases with no access function.
11568 2017-09-18 Richard Biener <rguenther@suse.de>
11570 PR tree-optimization/79622
11571 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11573 (build_cross_bb_scalars_use): Likewise.
11575 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11577 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11579 2017-09-18 Alan Modra <amodra@gmail.com>
11582 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11583 stack_pointer_rtx for count 0. Update comments. Break up
11584 large rtl expression.
11586 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11588 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11589 Increase to 20 bytes.
11590 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11591 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11592 or avx version of the stub.
11594 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11597 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11598 compute the minimum stack alignment. Also update preferred stack
11599 boundary for leaf functions.
11601 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11603 PR tree-optimization/82228
11604 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11607 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11609 * common/config/nds32/nds32-common.c
11610 (nds32_option_optimization_table): Refine formatting.
11611 (nds32_option_optimization_table): Use -fsched-pressure and
11612 -fomit-frame-pointer for specific optimization level.
11614 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11616 * config/nds32/nds32.c: Refine formatting and comments.
11617 * config/nds32/nds32.h: Likewise.
11618 * config/nds32/nds32.md: Likewise.
11619 * config/nds32/nds32-cost.c: Likewise.
11620 * config/nds32/nds32-isr.c: Likewise.
11621 * config/nds32/nds32-md-auxiliary.c: Likewise.
11622 * config/nds32/nds32-multiple.md: Likewise.
11623 * config/nds32/nds32-predicates.c: Likewise.
11625 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11626 Jakub Jelinek <jakub@redhat.com>
11628 Add support for -std=c++2a.
11629 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11631 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11633 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11636 * doc/extend.texi (Common Function Attributes): Add
11637 references to ARM, AArch64, and S/390 specific attributes.
11638 (Function Specific Option Pragmas): Add AArch64 and S/390
11639 to list of back ends that support the target pragma.
11641 2017-09-15 Nathan Sidwell <nathan@acm.org>
11643 * doc/standards.texi: Fix C++17 description. Update URLs for
11646 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11648 * common.opt (Wcast-align=strict): New warning option.
11649 * doc/invoke.texi: Document -Wcast-align=strict.
11651 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11653 * cgraph.h (cgraph_thunk_info): Add comments.
11654 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11655 assert for VIRTUAL_* arguments stricter.
11657 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11659 PR tree-optimization/71026
11660 * match.pd: Move RDIV patterns from fold-const.c
11661 * fold-const.c (distribute_real_division): Removed.
11662 (fold_binary_loc): Remove calls to distribute_real_divison.
11664 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11666 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11667 c++1z and gnu++1z as deprecated. Change other references to
11668 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11669 Change -Wc++1z-compat to -Wc++17-compat.
11670 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11671 * dwarf2out.c (highest_c_language): Handle C++17.
11672 (gen_compile_unit_die): Likewise.
11674 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11676 PR rtl-optimization/82192
11677 * combine.c (make_extraction): Don't look through non-paradoxical
11678 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11681 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11682 Alan Hayward <alan.hayward@arm.com>
11683 David Sherwood <david.sherwood@arm.com>
11685 * target.def (function_arg_offset): New hook.
11686 * targhooks.h (default_function_arg_offset): Declare.
11687 * targhooks.c (default_function_arg_offset): New function.
11688 * function.c (locate_and_pad_parm): Use
11689 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11690 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11691 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11692 * doc/tm.texi: Regenerate.
11693 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11694 * config/spu/spu.c (spu_function_arg_offset): New function.
11695 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11696 * system.h (FUNCTION_ARG_OFFSET): Poison.
11698 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11699 Alan Hayard <alan.hayward@arm.com>
11700 David Sherwood <david.sherwood@arm.com>
11702 * target.def (truly_noop_truncation): New hook.
11703 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11704 than TRULY_NOOP_TRUNCATION.
11705 * hooks.h (hook_bool_uint_uint_true): Declare.
11706 * hooks.c (hook_bool_uint_uint_true): New function.
11707 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11708 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11709 * doc/tm.texi: Regenerate.
11710 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11711 rather than TRULY_NOOP_TRUNCATION in comments.
11712 (simplify_comparison): Likewise.
11713 (record_truncated_value): Likewise.
11714 * expmed.c (extract_bit_field_1): Likewise.
11715 (extract_split_bit_field): Likewise.
11716 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11717 instead of TRULY_NOOP_TRUNCATION.
11718 * function.c (assign_parm_setup_block): Likewise.
11719 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11720 * rtlhooks.c: Include target.h.
11721 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11722 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11723 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11724 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11725 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11726 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11727 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11728 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11729 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11730 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11731 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11732 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11733 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11734 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11735 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11736 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11737 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11738 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11739 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11740 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11741 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11742 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11743 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11744 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11745 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11746 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11747 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11748 rather than TRULY_NOOP_TRUNCATION in comments.
11749 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11750 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11751 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11752 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11753 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11754 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11755 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11756 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11757 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11758 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11759 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11760 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11761 rather than TRULY_NOOP_TRUNCATION in comments.
11762 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11763 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11764 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11765 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11766 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11767 TRULY_NOOP_TRUNCATION condition.
11768 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11769 (TRULY_NOOP_TRUNCATION): Delete.
11770 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11771 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11772 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11773 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11774 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11775 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11776 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11777 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11778 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11779 rather than TRULY_NOOP_TRUNCATION in comments.
11780 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11781 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11782 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11783 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11784 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11785 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11787 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11790 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11791 (*cmp_ior): Likewise.
11792 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11793 (*ior_scc_scc_cmp): Likewise.
11794 (*and_scc_scc): Likewise.
11795 (*and_scc_scc_cmp): Likewise.
11797 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11798 Alan Hayard <alan.hayward@arm.com>
11799 David Sherwood <david.sherwood@arm.com>
11801 * target.def (can_change_mode_class): New hook.
11802 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11803 (hard_regno_nregs): Likewise.
11804 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11805 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11806 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11807 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11808 (LOAD_EXTEND_OP): Update accordingly.
11809 * doc/tm.texi: Regenerate.
11810 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11811 CANNOT_CHANGE_MODE_CLASS.
11812 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11813 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11814 * combine.c (simplify_set): Update accordingly.
11815 * emit-rtl.c (validate_subreg): Likewise.
11816 * recog.c (general_operand): Likewise.
11817 * regcprop.c (mode_change_ok): Likewise.
11818 * reload1.c (choose_reload_regs): Likewise.
11819 (inherit_piecemeal_p): Likewise.
11820 * rtlanal.c (simplify_subreg_regno): Likewise.
11821 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11822 instead of CANNOT_CHANGE_MODE_CLASS.
11823 (reload_cse_simplify_operands): Likewise.
11824 * reload.c (push_reload): Use targetm.can_change_mode_class
11825 instead of CANNOT_CHANGE_MODE_CLASS.
11826 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11827 REG_CANNOT_CHANGE_MODE_P.
11828 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11829 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11830 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11831 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11832 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11833 (arm_can_change_mode_class): New function.
11834 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11835 than CANNOT_CHANGE_MODE_CLASS in comments.
11836 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11837 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11838 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11839 (ix86_can_change_mode_class): ...this new function, inverting the
11840 sense of the return value.
11841 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11842 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11843 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11844 (ia64_can_change_mode_class): New function.
11845 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11846 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11847 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11848 (m32c_can_change_mode_class): ...this new function, inverting the
11849 sense of the return value.
11850 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11851 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11852 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11853 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11854 (mips_can_change_mode_class): ...this new function, inverting the
11855 sense of the return value.
11856 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11857 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11858 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11859 (msp430_can_change_mode_class): New function.
11860 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11861 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11862 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11863 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11864 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11865 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11866 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11867 (pa_cannot_change_mode_class): Replace with...
11868 (pa_can_change_mode_class): ...this new function, inverting the
11869 sense of the return value.
11870 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11871 than CANNOT_CHANGE_MODE_CLASS in comments.
11872 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11873 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11874 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11875 (pdp11_cannot_change_mode_class): Replace with...
11876 (pdp11_can_change_mode_class): ...this new function, inverting the
11877 sense of the return value.
11878 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11879 * config/powerpcspe/powerpcspe-protos.h
11880 (rs6000_cannot_change_mode_class_ptr): Delete.
11881 * config/powerpcspe/powerpcspe.c
11882 (rs6000_cannot_change_mode_class_ptr): Delete.
11883 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11884 (rs6000_option_override_internal): Assign to
11885 targetm.can_change_mode_class instead of
11886 rs6000_cannot_change_mode_class_ptr.
11887 (rs6000_cannot_change_mode_class): Replace with...
11888 (rs6000_can_change_mode_class): ...this new function, inverting the
11889 sense of the return value.
11890 (rs6000_debug_cannot_change_mode_class): Replace with...
11891 (rs6000_debug_can_change_mode_class): ...this new function.
11892 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11893 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11894 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11895 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11896 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11898 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11899 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11900 (rs6000_option_override_internal): Assign to
11901 targetm.can_change_mode_class instead of
11902 rs6000_cannot_change_mode_class_ptr.
11903 (rs6000_cannot_change_mode_class): Replace with...
11904 (rs6000_can_change_mode_class): ...this new function, inverting the
11905 sense of the return value.
11906 (rs6000_debug_cannot_change_mode_class): Replace with...
11907 (rs6000_debug_can_change_mode_class): ...this new function.
11908 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11909 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11910 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11911 (s390_can_change_mode_class): ...this new function, inverting the
11912 sense of the return value.
11913 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11914 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11915 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11916 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11917 (sh_cannot_change_mode_class): Replace with...
11918 (sh_can_change_mode_class): ...this new function, inverting the
11919 sense of the return value.
11920 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11921 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11922 (sparc_can_change_mode_class): New function.
11923 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11924 * config/spu/spu.c (spu_can_change_mode_class): New function.
11925 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11926 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11927 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11928 (visium_can_change_mode_class): New function.
11929 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
11931 2017-09-15 Richard Biener <rguenther@suse.de>
11933 PR tree-optimization/82217
11934 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
11935 but not undefined case.
11937 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11940 * postreload.c (reload_cse_simplify_operands): Skip
11941 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
11943 2017-09-15 Richard Biener <rguenther@suse.de>
11945 PR tree-optimization/68823
11946 * graphite-scop-detection.c (build_alias_set): If we have a
11947 possible dependence check whether we can handle them by just
11948 looking at the DRs DR_ACCESS_FNs.
11949 (build_scops): If build_alias_set fails, fail the SCOP.
11951 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11953 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
11954 to support float128 built-in functions that require the ISA 3.0
11956 (BU_FLOAT128_3_HW): Likewise.
11957 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11958 built-in functions.
11959 (FMAF128): Likewise.
11961 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
11962 support for built-in functions that need the ISA 3.0 IEEE 128-bit
11963 floating point instructions.
11964 (rs6000_invalid_builtin): Likewise.
11965 (rs6000_builtin_mask_names): Likewise.
11966 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
11967 (RS6000_BTM_FLOAT128_HW): Likewise.
11968 (RS6000_BTM_COMMON): Likewise.
11969 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
11971 * doc/extend.texi (RS/6000 built-in functions): Document the
11972 IEEE 128-bit floating point square root and fused multiply-add
11973 built-in functions.
11975 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
11977 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
11978 reg (r2) isn't in the set of registers defined in the prologue.
11980 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11981 Alan Hayward <alan.hayward@arm.com>
11982 David Sherwood <david.sherwood@arm.com>
11984 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
11985 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
11986 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
11987 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
11988 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
11990 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
11991 max_vectorization_factor.
11992 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
11994 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11995 Alan Hayward <alan.hayward@arm.com>
11996 David Sherwood <david.sherwood@arm.com>
11998 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
11999 (vect_worthwhile_without_simd_p): Declare.
12000 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12001 (vectorizable_reduction): Use it.
12002 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12003 (vectorizable_operation): Likewise.
12005 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12006 Alan Hayward <alan.hayward@arm.com>
12007 David Sherwood <david.sherwood@arm.com>
12009 * tree-vectorizer.h (vect_get_num_copies): New function.
12010 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12011 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12012 (vectorizable_induction): Likewise.
12013 (vectorizable_live_operation): Likewise.
12014 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12015 (vectorizable_bswap): Likewise.
12016 (vectorizable_call): Likewise.
12017 (vectorizable_conversion): Likewise.
12018 (vectorizable_assignment): Likewise.
12019 (vectorizable_shift): Likewise.
12020 (vectorizable_operation): Likewise.
12021 (vectorizable_store): Likewise.
12022 (vectorizable_load): Likewise.
12023 (vectorizable_condition): Likewise.
12024 (vectorizable_comparison): Likewise.
12025 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12027 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12028 Alan Hayward <alan.hayward@arm.com>
12029 David Sherwood <david.sherwood@arm.com>
12031 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12032 of vect_init_vector.
12034 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12035 Alan Hayward <alan.hayward@arm.com>
12036 David Sherwood <david.sherwood@arm.com>
12038 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12039 an inline wrapper that provides a location.
12040 (gimple_build_vector): Likewise.
12041 * gimple-fold.c (gimple_build_vector_from_val): New function.
12042 (gimple_build_vector): Likewise.
12043 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12044 functions to build the initial value. Always return a gimple value.
12045 (get_initial_defs_for_reduction): Likewise. Only compute
12047 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12048 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12049 (vectorizable_induction): Use gimple_build_vector rather than
12052 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12053 Alan Hayward <alan.hayward@arm.com>
12054 David Sherwood <david.sherwood@arm.com>
12056 * target.h (vec_perm_indices): New typedef.
12057 (auto_vec_perm_indices): Likewise.
12058 * optabs-query.h: Include target.h
12059 (can_vec_perm_p): Take a vec_perm_indices *.
12060 * optabs-query.c (can_vec_perm_p): Likewise.
12061 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12062 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12063 * tree-vect-generic.c (lower_vec_perm): Likewise.
12064 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12065 (vect_grouped_load_supported): Likewise.
12066 (vect_shift_permute_load_chain): Likewise.
12067 (vect_permute_store_chain): Use auto_vec_perm_indices.
12068 (vect_permute_load_chain): Likewise.
12069 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12070 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12071 Update uses of can_vec_perm_p.
12072 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12073 mode with a number of elements. Take a vec_perm_indices *.
12074 (vect_create_epilog_for_reduction): Update accordingly.
12075 Use auto_vec_perm_indices.
12076 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12077 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12078 (vect_transform_slp_perm_load): Likewise.
12079 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12080 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12081 (vect_gen_perm_mask_checked): Likewise.
12082 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12083 (vect_gen_perm_mask_checked): Likewise.
12084 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12085 (vectorizable_store): Likewise.
12086 (vectorizable_load): Likewise.
12087 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12088 (vectorizable_bswap): Likewise.
12090 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12091 Alan Hayward <alan.hayward@arm.com>
12092 David Sherwood <david.sherwood@arm.com>
12094 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12095 * tree.c (build_vector): Likewise.
12096 (build_vector_from_ctor): Update accordingly.
12097 (build_vector_from_val): Likewise.
12098 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12099 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12100 * tree-vect-generic.c (add_rshift): Likewise.
12101 (expand_vector_divmod): Likewise.
12102 (optimize_vector_constructor): Likewise.
12103 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12104 (vect_transform_slp_perm_load): Likewise.
12105 (vect_schedule_slp_instance): Likewise.
12106 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12107 (vectorizable_call): Likewise.
12108 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12109 * expmed.c (make_tree): Likewise.
12110 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12111 a vector passed to build_vector.
12112 (fold_convert_const): Likewise.
12113 (exact_inverse): Likewise.
12114 (fold_ternary_loc): Likewise.
12115 (fold_relational_const): Likewise.
12116 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12117 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12118 (const_unop): Likewise. Store the reduction accumulator in a
12119 variable rather than an array.
12120 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12121 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12122 the new vector, rather than constructing it after the input arrays.
12123 (native_interpret_vector): Use auto_vec<tree> when building
12124 a vector passed to build_vector. Add elements in order.
12125 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12126 auto_vec<tree> when building a vector passed to build_vector.
12127 (vect_create_epilog_for_reduction): Likewise.
12128 (vectorizable_induction): Likewise.
12129 (get_initial_def_for_reduction): Likewise. Fix indentation of
12131 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12133 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12134 passed to build_vector.
12136 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12137 Alan Hayward <alan.hayward@arm.com>
12138 David Sherwood <david.sherwood@arm.com>
12140 * tree-core.h (tree_base::u): Add an "nelts" field.
12141 (tree_vector): Use VECTOR_CST_NELTS as the length.
12142 * tree.c (tree_size): Likewise.
12143 (make_vector): Initialize VECTOR_CST_NELTS.
12144 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12145 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12146 TYPE_VECTOR_SUBPARTS.
12147 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12148 as the number of units, setting it from VECTOR_CST_NELTS.
12149 (const_vector_from_tree): Likewise.
12150 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12151 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12152 (fold_negate_expr_1): Likewise.
12153 (fold_convert_const): Likewise.
12154 (const_binop): Likewise. Differentiate the number of output and
12156 (const_unop): Likewise.
12157 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12158 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12159 in cases that did the opposite.
12161 2017-09-14 Richard Biener <rguenther@suse.de>
12163 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12166 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12168 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12170 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12173 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12174 if and where to split a bb, except for splitting before debug insn
12175 sequences followed by non-label real insn. Delete debug insns
12176 in between basic blocks.
12178 * combine.c (make_compound_operation_int): Formatting fixes.
12180 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12181 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12182 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12183 * config/sol2.h (LINK_EH_SPEC): Likewise.
12184 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12185 * config/s390/linux.h (LINK_SPEC): Likewise.
12186 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12187 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12188 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12189 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12190 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12191 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12192 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12193 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12195 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12197 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12199 (ENDFILE_LINUX_SPEC): Likewise.
12200 (LINK_EH_SPEC): Likewise.
12201 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12202 (LINK_OS_LINUX_SPEC32): Likewise.
12203 (LINK_OS_LINUX_SPEC64): Likewise.
12204 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12205 (LINK_OS_LINUX_SPEC): Likewise.
12207 2017-09-13 Martin Liska <mliska@suse.cz>
12209 PR middle-end/82154
12210 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12211 CASE_HIGH is NULL_TREE.
12213 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12214 Alan Hayward <alan.hayward@arm.com>
12215 David Sherwood <david.sherwood@arm.com>
12217 * target.def (secondary_memory_needed): New hook.
12218 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12219 instead of SECONDARY_MEMORY_NEEDED.
12220 (secondary_memory_needed_mode): Likewise.
12221 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12222 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12223 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12224 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12225 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12226 * doc/tm.texi: Regenerate.
12227 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12228 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12229 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12230 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12231 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12232 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12233 mode argument first and change the reg_class arguments to reg_class_t.
12234 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12235 Make static. Update the call to inline_secondary_memory_needed.
12236 (ix86_register_move_cost): Update the call to
12237 inline_secondary_memory_needed.
12238 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12239 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12241 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12242 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12244 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12245 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12246 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12247 and match hook interface. Add comment from mips.h.
12248 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12249 * config/mmix/mmix.md (truncdfsf2): Refer to
12250 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12252 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12253 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12254 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12255 (pa_secondary_memory_needed): New function.
12256 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12257 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12258 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12259 (pdp11_secondary_memory_needed): Make static and match hook interface.
12260 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12261 * config/powerpcspe/powerpcspe-protos.h
12262 (rs6000_secondary_memory_needed_ptr): Delete.
12263 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12265 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12266 (rs6000_option_override_internal): Assign to
12267 targetm.secondary_memory_needed rather than
12268 rs6000_secondary_memory_needed_ptr.
12269 (rs6000_secondary_memory_needed): Match hook interface.
12270 (rs6000_debug_secondary_memory_needed): Likewise.
12271 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12272 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12273 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12274 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12275 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12276 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12278 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12279 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12280 (rs6000_option_override_internal): Assign to
12281 targetm.secondary_memory_needed rather than
12282 rs6000_secondary_memory_needed_ptr.
12283 (rs6000_secondary_memory_needed): Match hook interface.
12284 (rs6000_debug_secondary_memory_needed): Likewise.
12285 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12286 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12287 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12288 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12289 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12290 (sparc_secondary_memory_needed): New function.
12291 * lra-constraints.c (check_and_process_move): Refer to
12292 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12294 (curr_insn_transform): Likewise.
12295 (process_alt_operands): Use targetm.secondary_memory_needed
12296 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12297 (check_secondary_memory_needed_p): Likewise.
12298 (choose_split_class): Likewise.
12299 * reload.c: Unconditionally include code that was previously
12300 conditional on SECONDARY_MEMORY_NEEDED.
12301 (push_secondary_reload): Use targetm.secondary_memory_needed
12302 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12303 (push_reload): Likewise.
12304 * reload1.c: Unconditionally include code that was previously
12305 conditional on SECONDARY_MEMORY_NEEDED.
12306 (choose_reload_regs): Use targetm.secondary_memory_needed
12307 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12308 (gen_reload): Likewise.
12309 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12311 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12312 Alan Hayward <alan.hayward@arm.com>
12313 David Sherwood <david.sherwood@arm.com>
12315 * target.def (secondary_memory_needed_mode): New hook:
12316 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12317 * targhooks.h (default_secondary_memory_needed_mode): New function.
12318 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12319 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12320 * doc/tm.texi: Regenerate.
12321 * lra-constraints.c (check_and_process_move): Use
12322 targetm.secondary_memory_needed_mode instead of
12323 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12324 (curr_insn_transform): Likewise.
12325 * reload.c (get_secondary_mem): Likewise.
12326 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12327 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12329 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12330 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12331 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12332 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12333 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12335 * config/powerpcspe/powerpcspe-protos.h
12336 (rs6000_secondary_memory_needed_mode): Delete.
12337 * config/powerpcspe/powerpcspe.c
12338 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12339 (rs6000_secondary_memory_needed_mode): Make static.
12340 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12341 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12343 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12345 (rs6000_secondary_memory_needed_mode): Make static.
12346 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12347 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12348 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12349 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12350 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12352 (sparc_secondary_memory_needed_mode): New function.
12353 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12355 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12357 * config/aarch64/constraints.md (Umq): New constraint.
12358 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12360 (mov<mode>): Update condition.
12362 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12364 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12365 when bitposition is the same.
12367 2017-09-13 Richard Biener <rguenther@suse.de>
12369 * dwarf2out.c (output_die_symbol): Remove.
12370 (output_die): Do not output a DIEs symbol.
12372 2017-09-13 Richard Biener <rguenther@suse.de>
12374 PR middle-end/82128
12375 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12376 default-def to avoid breaking iterator update with the weird
12377 interaction with cgraph_update_edges_for_call_stmt_node.
12379 2017-09-13 Richard Biener <rguenther@suse.de>
12381 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12382 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12383 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12384 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12386 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12388 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12389 Disable pc relative literal load irrespective of
12390 TARGET_FIX_ERR_A53_84341 for default.
12392 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12394 * config/sparc/sparc.c (output_return): Output the source location of
12395 the insn in the delay slot, if any.
12396 (output_sibcall): Likewise.
12398 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12401 * common.opt (-static-pie): New alias.
12402 (shared): Negate static-pie.
12403 (-no-pie): Update help text.
12405 (static-pie): New option.
12406 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12407 -static-pie support.
12408 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12409 (LINK_EH_SPEC): Likewise.
12410 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12411 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12412 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12413 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12414 (init_gcc_specs): Likewise.
12415 (init_spec): Likewise.
12416 (display_help): Update help message for -pie.
12417 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12420 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12422 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12423 (movdi_aarch64): Likewise.
12424 (movti_aarch64): Likewise.
12426 2017-09-12 Simon Wright <simon@pushface.org>
12429 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12430 calculation of the minor version, always output as 0.
12432 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12435 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12436 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12437 on it early, rather than manual conversion late. For
12438 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12439 instead of performing manual conversion.
12441 2017-09-12 Carl Love <cel@us.ibm.com>
12443 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12444 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12446 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12447 VMULOSW): Add definitions.
12448 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12449 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12450 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12451 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12452 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12454 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12456 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12458 (movti_aarch64): Likewise.
12459 (movdf_aarch64): Likewise.
12460 (movtf_aarch64): Likewise.
12461 (load_pairdi): Likewise.
12462 (store_pairdi): Likewise.
12463 (load_pairdf): Likewise.
12464 (store_pairdf): Likewise.
12465 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12466 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12467 (ldr_got_small_<mode>): Likewise.
12468 (ldr_got_small_28k_<mode>): Likewise.
12469 (ldr_got_tiny): Likewise.
12470 * config/aarch64/iterators.md (ldst_sz): New.
12471 (ldpstp_sz): Likewise.
12472 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12474 (thunderx_load): Split load_8 to load_16.
12475 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12477 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12478 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12479 (xgene1_store_pair): Split store_8 to store_16.
12480 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12481 (falkor_st_0_st_sd): Split store_8 to store_16.
12483 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12485 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12486 and store1/2/3/4 to store_4/8/12/16.
12487 * config/aarch64/aarch64.md: Update for rename.
12488 * config/arm/arm.md: Likewise.: Likewise.
12489 * config/arm/arm.c: Likewise.
12490 * config/arm/thumb1.md: Likewise.
12491 * config/arm/thumb2.md: Likewise.
12492 * config/arm/vfp.md: Likewise.
12493 * config/arm/arm-generic.md: Likewise.
12494 * config/arm/arm1020e.md: Likewise.
12495 * config/arm/arm1026ejs.md: Likewise.
12496 * config/arm/arm1136jfs.md: Likewise.
12497 * config/arm/arm926ejs.md: Likewise.
12498 * config/arm/cortex-a15.md: Likewise.
12499 * config/arm/cortex-a17.md: Likewise.
12500 * config/arm/cortex-a5.md: Likewise.
12501 * config/arm/cortex-a53.md: Likewise.
12502 * config/arm/cortex-a57.md: Likewise.
12503 * config/arm/cortex-a7.md: Likewise.
12504 * config/arm/cortex-a8.md: Likewise.
12505 * config/arm/cortex-a9.md: Likewise.
12506 * config/arm/cortex-m4.md: Likewise.
12507 * config/arm/cortex-m7.md: Likewise.
12508 * config/arm/cortex-r4.md: Likewise.
12509 * config/arm/exynos-m1.md: Likewise.
12510 * config/arm/fa526.md: Likewise.
12511 * config/arm/fa606te.md: Likewise.
12512 * config/arm/fa626te.md: Likewise.
12513 * config/arm/fa726te.md: Likewise.
12514 * config/arm/fmp626.md: Likewise.
12515 * config/arm/iwmmxt.md: Likewise.
12516 * config/arm/ldmstm.md: Likewise.
12517 * config/arm/marvell-pj4.md: Likewise.
12518 * config/arm/xgene1.md: Likewise.
12519 * config/aarch64/thunderx.md: Likewise.
12520 * config/aarch64/thunderx2t99.md: Likewise.
12521 * config/aarch64/falkor.md: Likewise.
12523 2017-09-12 Martin Liska <mliska@suse.cz>
12525 * attribs.c (private_lookup_attribute): New function.
12526 * attribs.h (private_lookup_attribute): Declared here.
12527 (lookup_attribute): Called from this place.
12529 2017-09-12 Richard Biener <rguenther@suse.de>
12531 PR tree-optimization/82157
12532 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12533 stmts with side-effects.
12535 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12536 Alan Hayward <alan.hayward@arm.com>
12537 David Sherwood <david.sherwood@arm.com>
12539 * target.def (hard_regno_nregs): New hook.
12540 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12541 * targhooks.h (default_hard_regno_nregs): Declare.
12542 * targhooks.c (default_hard_regno_nregs): New function.
12543 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12544 (TARGET_HARD_REGNO_NREGS): ...this hook.
12545 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12546 (CLASS_MAX_NREGS): Likewise.
12547 * doc/tm.texi: Regenerate.
12548 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12549 instead of HARD_REGNO_NREGS.
12550 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12551 HARD_REGNO_NREGS in the comment.
12552 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12553 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12554 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12555 Return an unsigned int.
12556 (TARGET_HARD_REGNO_NREGS): Redefine.
12557 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12558 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12559 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12560 (arc_hard_regno_nregs): New function.
12561 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12562 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12563 (arm_hard_regno_nregs): New function.
12564 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12565 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12566 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12567 (TARGET_HARD_REGNO_NREGS): Redefine.
12568 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12569 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12570 (HARD_REGNO_NREGS): Delete.
12571 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12572 (cr16_hard_regno_nregs): New function.
12573 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12574 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12575 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12576 (cris_hard_regno_nregs): New function.
12577 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12578 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12579 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12580 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12581 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12582 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12583 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12584 (frv_hard_regno_nregs): Make static. Take and return an
12586 (frv_class_max_nregs): Remove outdated copy of documentation.
12587 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12588 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12589 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12590 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12591 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12592 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12593 (TARGET_HARD_REGNO_NREGS): Redefine.
12594 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12595 (CLASS_MAX_NREGS): Update comment.
12596 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12597 (ia64_hard_regno_nregs): New function.
12598 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12599 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12600 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12601 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12602 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12604 (m32c_hard_regno_nregs): Likewise. Make static.
12605 (TARGET_HARD_REGNO_NREGS): Redefine.
12606 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12607 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12608 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12609 (m68k_hard_regno_nregs): New function.
12610 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12611 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12612 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12613 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12614 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12615 Take and return an unsigned int.
12616 (TARGET_HARD_REGNO_NREGS): Redefine.
12617 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12618 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12619 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12620 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12621 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12622 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12623 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12624 (msp430_hard_regno_nregs): Make static. Take and return an
12626 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12627 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12628 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12629 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12630 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12631 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12632 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12633 (TARGET_HARD_REGNO_NREGS): Redefine.
12634 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12635 (PA_HARD_REGNO_NREGS): ...this.
12636 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12637 (PA_HARD_REGNO_NREGS): ...this.
12638 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12639 (pa_hard_regno_nregs): New function.
12640 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12641 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12642 (pdp11_hard_regno_nregs): New function.
12643 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12644 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12645 (rs6000_hard_regno_nregs_hook): New function.
12646 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12647 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12648 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12649 Take and return an unsigned int. Move earlier in file.
12650 (TARGET_HARD_REGNO_NREGS): Redefine.
12651 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12652 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12653 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12654 (rl78_hard_regno_nregs): Make static. Take and return an
12656 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12657 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12658 (rs6000_hard_regno_nregs_hook): New function.
12659 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12660 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12661 (TARGET_HARD_REGNO_NREGS): Redefine.
12662 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12663 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12664 instead of HARD_REGNO_NREGS.
12665 (s390_hard_regno_nregs): New function.
12666 (s390_hard_regno_mode_ok): Add comment from s390.h.
12667 (TARGET_HARD_REGNO_NREGS): Redefine.
12668 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12669 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12670 (sh_hard_regno_nregs): New function.
12671 (sh_pass_in_reg_p): Use it.
12672 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12673 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12674 (sparc_hard_regno_nregs): New function.
12675 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12676 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12677 (spu_function_arg_advance): Use it, supplying a valid register number.
12678 (TARGET_HARD_REGNO_NREGS): Redefine.
12679 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12680 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12681 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12682 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12683 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12684 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12685 (CLASS_MAX_NREGS): Remove copy of old documentation.
12686 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12687 (visium_hard_regno_nregs): New function.
12688 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12689 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12690 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12691 xtensa_hard_regno_nregs): New function.
12692 * system.h (HARD_REGNO_NREGS): Poison.
12694 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12696 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12697 hard_regno_nregs instead of HARD_REGNO_NREGS.
12698 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12699 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12700 (c6x_expand_epilogue): Likewise.
12701 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12702 (frv_read_iacc_argument): Likewise.
12703 * config/sh/sh.c: Include regs.h.
12704 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12705 (regs_used): Likewise.
12706 (output_stack_adjust): Likewise.
12707 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12708 * expmed.c: Include regs.h.
12709 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12710 * ree.c: Include regs.h.
12711 (combine_reaching_defs): Use hard_regno_nregs instead of
12713 (add_removable_extension): Likewise.
12715 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12717 * regs.h (hard_regno_nregs): Turn into a function.
12718 (end_hard_regno): Update accordingly.
12719 * caller-save.c (setup_save_areas): Likewise.
12720 (save_call_clobbered_regs): Likewise.
12721 (replace_reg_with_saved_mem): Likewise.
12722 (insert_restore): Likewise.
12723 (insert_save): Likewise.
12724 * combine.c (can_change_dest_mode): Likewise.
12725 (move_deaths): Likewise.
12726 (distribute_notes): Likewise.
12727 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12728 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12729 (rs6000_split_multireg_move): Likewise.
12730 (rs6000_register_move_cost): Likewise.
12731 (rs6000_memory_move_cost): Likewise.
12732 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12733 (rs6000_split_multireg_move): Likewise.
12734 (rs6000_register_move_cost): Likewise.
12735 (rs6000_memory_move_cost): Likewise.
12736 * cselib.c (cselib_reset_table): Likewise.
12737 (cselib_lookup_1): Likewise.
12738 * emit-rtl.c (set_mode_and_regno): Likewise.
12739 * function.c (aggregate_value_p): Likewise.
12740 * ira-color.c (setup_profitable_hard_regs): Likewise.
12741 (check_hard_reg_p): Likewise.
12742 (calculate_saved_nregs): Likewise.
12743 (assign_hard_reg): Likewise.
12744 (improve_allocation): Likewise.
12745 (calculate_spill_cost): Likewise.
12746 * ira-emit.c (modify_move_list): Likewise.
12747 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12748 (ira_hard_reg_in_set_p): Likewise.
12749 * ira.c (setup_reg_mode_hard_regset): Likewise.
12750 (clarify_prohibited_class_mode_regs): Likewise.
12751 (check_allocation): Likewise.
12752 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12753 (lra_setup_reg_renumber): Likewise.
12754 (setup_try_hard_regno_pseudos): Likewise.
12755 (spill_for): Likewise.
12756 (assign_hard_regno): Likewise.
12757 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12758 * lra-constraints.c (in_class_p): Likewise.
12759 (lra_constraint_offset): Likewise.
12760 (simplify_operand_subreg): Likewise.
12761 (lra_constraints): Likewise.
12762 (split_reg): Likewise.
12763 (split_if_necessary): Likewise.
12764 (invariant_p): Likewise.
12765 (inherit_in_ebb): Likewise.
12766 * lra-lives.c (process_bb_lives): Likewise.
12767 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12768 (get_hard_regs): Likewise.
12769 (do_remat): Likewise.
12770 * lra-spills.c (assign_spill_hard_regs): Likewise.
12771 * mode-switching.c (create_pre_exit): Likewise.
12772 * postreload.c (reload_combine_recognize_pattern): Likewise.
12773 * recog.c (peep2_find_free_register): Likewise.
12774 * regcprop.c (kill_value_regno): Likewise.
12775 (set_value_regno): Likewise.
12776 (copy_value): Likewise.
12777 (maybe_mode_change): Likewise.
12778 (find_oldest_value_reg): Likewise.
12779 (copyprop_hardreg_forward_1): Likewise.
12780 * regrename.c (check_new_reg_p): Likewise.
12781 (regrename_do_replace): Likewise.
12782 * reload.c (push_reload): Likewise.
12783 (combine_reloads): Likewise.
12784 (find_dummy_reload): Likewise.
12785 (operands_match_p): Likewise.
12786 (find_reloads): Likewise.
12787 (find_equiv_reg): Likewise.
12788 (reload_adjust_reg_for_mode): Likewise.
12789 * reload1.c (count_pseudo): Likewise.
12790 (count_spilled_pseudo): Likewise.
12791 (find_reg): Likewise.
12792 (clear_reload_reg_in_use): Likewise.
12793 (free_for_value_p): Likewise.
12794 (allocate_reload_reg): Likewise.
12795 (choose_reload_regs): Likewise.
12796 (reload_adjust_reg_for_temp): Likewise.
12797 (emit_reload_insns): Likewise.
12798 (delete_output_reload): Likewise.
12799 * rtlanal.c (subreg_get_info): Likewise.
12800 * sched-deps.c (sched_analyze_reg): Likewise.
12801 * sel-sched.c (init_regs_for_mode): Likewise.
12802 (mark_unavailable_hard_regs): Likewise.
12803 (choose_best_reg_1): Likewise.
12804 (verify_target_availability): Likewise.
12805 * valtrack.c (dead_debug_insert_temp): Likewise.
12806 * var-tracking.c (track_loc_p): Likewise.
12807 (emit_note_insn_var_location): Likewise.
12808 * varasm.c (make_decl_rtl): Likewise.
12809 * reginfo.c (choose_hard_reg_mode): Likewise.
12810 (init_reg_modes_target): Refer directly to
12811 this_target_regs->x_hard_regno_nregs.
12813 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12815 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12816 instead of hard_regno_nregs.
12818 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12820 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12821 end_hard_regno instead of hard_regno_nregs.
12822 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12823 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12824 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12825 * ira-color.c (improve_allocation): Likewise.
12826 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12827 * lra-lives.c (mark_regno_live): Likewise.
12828 (mark_regno_dead): Likewise.
12829 * lra-remat.c (operand_to_remat): Likewise.
12830 * lra.c (collect_non_operand_hard_regs): Likewise.
12831 * postreload.c (reload_combine_note_store): Likewise.
12832 (move2add_valid_value_p): Likewise.
12833 * reload.c (regno_clobbered_p): Likewise.
12835 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12837 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12839 * config/v850/v850.c (v850_reorg): Likewise.
12840 * reload.c (refers_to_regno_for_reload_p): Likewise.
12841 (find_equiv_reg): Likewise.
12842 * reload1.c (reload_reg_reaches_end_p): Likewise.
12844 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12846 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12848 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12849 * config/arm/arm.c (output_move_neon): Likewise.
12850 (arm_attr_length_move_neon): Likewise.
12851 (neon_split_vcombine): Likewise.
12852 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12853 (c6x_mark_reg_written): Likewise.
12854 (c6x_dwarf_register_span): Likewise.
12855 * config/i386/i386.c (ix86_save_reg): Likewise.
12856 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12857 (rws_access_reg): Likewise.
12858 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12859 * mode-switching.c (create_pre_exit): Likewise.
12860 * ree.c (combine_reaching_defs): Likewise.
12861 (add_removable_extension): Likewise.
12862 * regcprop.c (find_oldest_value_reg): Likewise.
12863 (copyprop_hardreg_forward_1): Likewise.
12864 * reload.c (reload_inner_reg_of_subreg): Likewise.
12865 (push_reload): Likewise.
12866 (combine_reloads): Likewise.
12867 (find_dummy_reload): Likewise.
12868 (reload_adjust_reg_for_mode): Likewise.
12869 * reload1.c (find_reload_regs): Likewise.
12870 (forget_old_reloads_1): Likewise.
12871 (reload_reg_free_for_value_p): Likewise.
12872 (reload_adjust_reg_for_temp): Likewise.
12873 (emit_reload_insns): Likewise.
12874 (delete_output_reload): Likewise.
12875 * sel-sched.c (choose_best_reg_1): Likewise.
12876 (choose_best_pseudo_reg): Likewise.
12878 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12879 Alan Hayward <alan.hayward@arm.com>
12880 David Sherwood <david.sherwood@arm.com>
12882 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12883 * target.def (slow_unaligned_access): New hook.
12884 * targhooks.h (default_slow_unaligned_access): Declare.
12885 * targhooks.c (default_slow_unaligned_access): New function.
12886 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12887 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12888 * doc/tm.texi: Regenerate.
12889 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12890 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12891 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12893 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12894 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12896 (rs6000_slow_unaligned_access): New function.
12897 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12898 (expand_block_compare): Likewise.
12899 (expand_strn_compare): Likewise.
12900 (rs6000_rtx_costs): Likewise.
12901 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12902 (riscv_slow_unaligned_access): Likewise.
12903 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12904 (riscv_slow_unaligned_access_p): ...this and make static.
12905 (riscv_option_override): Update accordingly.
12906 (riscv_slow_unaligned_access): New function.
12907 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12908 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12909 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12910 (rs6000_slow_unaligned_access): New function.
12911 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12912 (rs6000_rtx_costs): Likewise.
12913 * config/rs6000/rs6000-string.c (expand_block_compare)
12914 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12915 of SLOW_UNALIGNED_ACCESS.
12916 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12917 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12918 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12919 of SLOW_UNALIGNED_ACCESS.
12920 * expmed.c (simple_mem_bitfield_p): Likewise.
12921 * expr.c (alignment_for_piecewise_move): Likewise.
12922 (emit_group_load_1): Likewise.
12923 (emit_group_store): Likewise.
12924 (copy_blkmode_from_reg): Likewise.
12925 (emit_push_insn): Likewise.
12926 (expand_assignment): Likewise.
12927 (store_field): Likewise.
12928 (expand_expr_real_1): Likewise.
12929 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12930 * lra-constraints.c (simplify_operand_subreg): Likewise.
12931 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12932 * gimple-ssa-store-merging.c: Likewise in block comment at start
12934 * tree-ssa-strlen.c: Include target.h.
12935 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
12936 of SLOW_UNALIGNED_ACCESS.
12937 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
12939 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12941 PR rtl-optimization/82185
12942 * expmed.c (emit_store_flag_int): Only test tem if it has been
12945 2017-09-12 Richard Biener <rguenther@suse.de>
12947 PR middle-end/82149
12948 * match.pd ((FTYPE) N CMP CST): Fix typo.
12950 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
12952 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
12954 (mips_near_type_p): Add 'short_call' attribute as a synonym
12956 * doc/extend.texi (short_call): Document new function attribute.
12958 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12961 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
12962 assertion check that in the condition.
12963 (get_atomic_generic_size): Likewise. Before testing if parameter
12964 has pointer type, if it has array type, call for C++
12965 default_conversion to perform array-to-pointer conversion.
12967 2017-09-12 Richard Biener <rguenther@suse.de>
12969 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
12970 for operations we cannot scalarize.
12972 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
12974 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
12975 vectors heap vectors. Clean up comments.
12976 Make visited_bbs a reference.
12977 (profitable_jump_thread_path): Make GC
12978 vectors heap vectors. Clean up comments.
12980 (convert_and_register_jump_thread_path): Make GC vectors heap
12982 (check_subpath_and_update_thread_path): Same. Clean up comments.
12983 Make visited_bbs a reference.
12984 (handle_phi): Abstract common code to to
12985 register_jump_thread_path_if_profitable.
12986 Rename VAR_BB to DEF_BB.
12988 Make GC vectors heap vectors.
12989 Make visited_bbs a reference.
12990 (handle_assignment): Same.
12991 (register_jump_thread_path_if_profitable): New.
12992 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
12994 Make GC vectors heap vectors. Clean up comments.
12995 Make visited_bbs a reference.
12996 (find_jump_threads_backwards): Make visited_bbs live in the stack.
12997 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13000 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13003 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13004 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13006 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13008 Revert r251800 and r251799.
13010 2017-09-11 Martin Jambor <mjambor@suse.cz>
13013 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13014 arguments in advance.
13015 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13016 use it to find predecessor edges.
13017 (naive_outof_ssa): Collect vector of predecessors.
13019 2017-09-08 Jason Merrill <jason@redhat.com>
13021 PR c++/70029 - ICE with ref-qualifier and -flto
13022 * langhooks.h (struct lang_hooks_for_types): Add
13023 copy_lang_qualifiers.
13024 * attribs.c (build_type_attribute_qual_variant): Use it.
13025 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13027 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13028 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13030 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13033 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13034 (*mulsi3_sp64): New instruction.
13035 (mulsi3): New expander.
13037 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13039 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13041 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13043 * sancov.c: Include memmodel.h.
13045 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13048 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13051 2017-09-07 Carl Love <cel@us.ibm.com>
13053 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13054 the sldi instruction.
13056 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13058 * sancov.c: Include tm_p.h.
13060 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13063 * output.h (switch_to_other_text_partition): New declaration.
13064 * varasm.c (switch_to_other_text_partition): New function.
13065 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13066 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13067 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13068 to the other text partition before emitting LCL label and switch back
13069 after emitting the word after it.
13071 2017-09-07 Richard Biener <rguenther@suse.de>
13073 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13074 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13075 critical edges here, after loop init.
13076 (pass_data_pre): Remove PROP_no_crit_edges flags.
13077 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13078 for valueization of call args to avoid leaking VN_TOP.
13079 (visit_use): Assert we do not visit default defs.
13080 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13081 Use error_mark_node to more easily detect leaking VN_TOP.
13082 All default-defs are varying, not VN_TOP. Mark them visited.
13083 (run_scc_vn): Make code match comment.
13085 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13087 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13088 OPTION_MASK_FLOAT128_KEYWORD.
13089 (POWERPC_MASKS): Likewise.
13090 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13091 support for the -mfloat128-type option, and make -mfloat128
13092 default on PowerPC Linux systems. Define or undefine
13093 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13094 Define __float128 to be __ieee128 if IEEE 128-bit support is
13095 enabled, or undefine it.
13096 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13097 Delete defining __FLOAT128_TYPE__.
13098 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13099 -mfloat128-type option and make -mfloat128 default on PowerPC
13101 (TARGET_FLOAT128_TYPE): Likewise.
13102 (-mfloat128-type): Likewise.
13103 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13104 Delete the -mfloat128-type option and make -mfloat128 default on
13105 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13106 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13107 128-bit floating point is enabled. Change tests from using
13108 -mfloat128-type to -mfloat128.
13109 (rs6000_mangle_type): Use the correct mangling for the __float128
13110 type even if normal long double is restricted to 64-bits.
13111 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13113 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13114 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13115 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13117 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13118 documentation for -mfloat128.
13120 2017-09-06 Olivier Hainque <hainque@adacore.com>
13122 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13124 2017-09-06 Wish Wu <wishwu007@gmail.com>
13125 Jakub Jelinek <jakub@redhat.com>
13127 * asan.c (initialize_sanitizer_builtins): Add
13128 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13129 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13130 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13131 BT_FN_VOID_UINT64_PTR variables.
13132 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13133 (BT_FN_VOID_UINT16_UINT16): Likewise.
13134 (BT_FN_VOID_UINT32_UINT32): Likewise.
13135 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13136 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13137 (BT_FN_VOID_UINT64_PTR): Likewise.
13138 * common.opt (flag_sanitize_coverage): New variable.
13139 (fsanitize-coverage=trace-pc): Remove.
13140 (fsanitize-coverage=): Add.
13141 * flag-types.h (enum sanitize_coverage_code): New enum.
13142 * fold-const.c (fold_range_test): Disable non-short-circuit
13143 optimization if flag_sanitize_coverage.
13144 (fold_truth_andor): Likewise.
13145 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13146 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13147 (coverage_sanitizer_opts): New array.
13148 (get_closest_sanitizer_option): Add OPTS argument, handle also
13149 OPT_fsanitize_coverage_.
13150 (parse_sanitizer_options): Adjusted to also handle
13151 OPT_fsanitize_coverage_.
13152 (common_handle_option): Add OPT_fsanitize_coverage_.
13153 * sancov.c (instrument_comparison, instrument_switch): New function.
13154 (sancov_pass): Add trace-cmp support.
13155 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13156 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13157 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13158 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13159 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13160 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13161 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13162 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13163 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13164 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13166 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13168 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13169 error. Only quit immediately if parsing is complete.
13170 (BEGIN): Initialize fatal_err and parse_done.
13171 (begin fpu, end fpu): Check number of arguments.
13172 (begin arch, end arch): Likewise.
13173 (begin cpu, end cpu): Likewise.
13174 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13175 (optalias): Likewise.
13177 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13179 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13180 * config/arm/arm-isa.h: Delete. Move definitions to ...
13181 * arm-cpus.in: ... here. Use new feature and fgroup values.
13182 * config/arm/arm.c (arm_option_override): Use lower case for feature
13184 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13185 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13186 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13188 (print_isa_bits_for): New function.
13189 (gen_isa): New function.
13190 (gen_comm_data): Use print_isa_bits_for.
13191 (define feature): New keyword.
13192 (define fgroup): New keyword.
13193 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13194 (arm-isa.h): Add rule to generate file.
13195 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13196 case for feature bit names.
13198 2017-09-06 Richard Biener <rguenther@suse.de>
13200 * tree-ssa-pre.c (NECESSARY): Remove.
13201 (create_expression_by_pieces): Do not touch pass-local flags.
13202 (insert_into_preds_of_block): Likewise.
13203 (do_pre_regular_insertion): Likewise.
13204 (eliminate_insert): Likewise.
13205 (eliminate_dom_walker::before_dom_children): Likewise.
13206 (fini_eliminate): Do not look at inserted_exprs.
13207 (mark_operand_necessary): Remove.
13208 (remove_dead_inserted_code): Replace with simple work-list
13209 algorithm based on inserted_exprs and SSA uses.
13210 (pass_pre::execute): Re-order fini_eliminate and
13211 remove_dead_inserted_code.
13213 2017-09-06 Olivier Hainque <hainque@adacore.com>
13215 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13216 for VxWorks 7. Adjust surrounding comments.
13218 2017-09-06 Richard Biener <rguenther@suse.de>
13220 * gimple-ssa-strength-reduction.c
13221 (find_candidates_dom_walker::before_dom_children): Also allow
13224 2017-09-06 Richard Biener <rguenther@suse.de>
13226 PR tree-optimization/82108
13227 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13228 for gap in the non-permutation SLP case.
13230 2017-09-06 Martin Jambor <mjambor@suse.cz>
13232 PR tree-optimization/82078
13233 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13234 add_access_to_work_queue...
13235 (build_accesses_from_assign): ...here.
13236 (propagate_all_subaccesses): Make sure racc is the group
13237 representative, if there is one.
13239 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13241 PR middle-end/82095
13242 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13245 2017-09-06 Richard Biener <rguenther@suse.de>
13247 * gimple-ssa-strength-reduction.c
13248 (find_candidates_dom_walker::before_doom_children): Use a
13249 type and not a mode check.
13251 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13254 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13256 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13258 2017-09-05 Jeff Law <law@redhat.com>
13260 PR tree-optimization/64910
13261 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13262 cases where we have 3 or more operands.
13264 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13266 PR middle-end/81768
13267 * omp-low.c (lower_omp_for): Recompute tree invariant if
13268 gimple_omp_for_initial/final is ADDR_EXPR.
13270 PR middle-end/81768
13271 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13272 into gimple val before gimplification fo the COND_EXPR.
13274 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13276 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13277 REGION_COPY argument.
13278 (thread_through_all_blocks): Remove unused argument to
13279 duplicate_thread_path.
13281 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13282 Alan Hayward <alan.hayward@arm.com>
13283 David Sherwood <david.sherwood@arm.com>
13285 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13286 Take a scalar_mode rather than a machine_mode.
13287 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13288 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13289 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13290 (aarch64_gen_adjusted_ldpstp): Likewise.
13291 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13293 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13294 Alan Hayward <alan.hayward@arm.com>
13295 David Sherwood <david.sherwood@arm.com>
13297 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13298 Take a scalar_int_mode instead of a machine_mode.
13299 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13300 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13301 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13302 (aarch64_simd_attr_length_rglist): Delete.
13303 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13304 a scalar_int_mode instead of a machine_mode.
13305 (aarch64_add_offset): Likewise.
13306 (aarch64_internal_mov_immediate): Likewise
13307 (aarch64_add_constant_internal): Likewise.
13308 (aarch64_add_constant): Likewise.
13309 (aarch64_movw_imm): Likewise.
13310 (aarch64_rtx_arith_op_extract_p): Likewise.
13311 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13312 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13313 Remove assert that the mode isn't a vector.
13314 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13315 (aarch64_expand_mov_immediate): Update calls after above changes.
13316 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13317 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13318 (aarch64_move_imm): Likewise.
13319 (aarch64_can_const_movi_rtx_p): Likewise.
13320 (aarch64_strip_extend): Likewise.
13321 (aarch64_extr_rtx_p): Likewise.
13322 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13323 a CONST_INT when the mode parameter is VOIDmode.
13324 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13326 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13328 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13329 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13330 (bitwise_type_for_mode): Update accordingly.
13332 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13334 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13335 * stor-layout.c (mode_for_size_tree): Likewise.
13336 (mode_for_array): Update accordingly.
13337 (layout_decl): Likewise.
13338 (compute_record_mode): Likewise. Only set the mode once.
13340 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13342 * target.def (get_mask_mode): Change return type to opt_mode.
13344 * doc/tm.texi: Regenerate.
13345 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13346 * targhooks.c (default_get_mask_mode): Likewise.
13347 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13348 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13349 targetm.get_mask_mode.
13350 * tree.c (build_truth_vector_type): Likewise.
13352 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13354 * machmode.h (mode_for_vector): Return an opt_mode.
13355 * stor-layout.c (mode_for_vector): Likewise.
13356 (mode_for_int_vector): Update accordingly.
13357 (layout_type): Likewise.
13358 * config/i386/i386.c (emit_memmov): Likewise.
13359 (ix86_expand_set_or_movmem): Likewise.
13360 (ix86_expand_vector_init): Likewise.
13361 (ix86_get_mask_mode): Likewise.
13362 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13364 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13365 * expmed.c (extract_bit_field_1): Likewise.
13366 * expr.c (expand_expr_real_2): Likewise.
13367 * optabs-query.c (can_vec_perm_p): Likewise.
13368 (can_vec_mask_load_store_p): Likewise.
13369 * optabs.c (expand_vec_perm): Likewise.
13370 * targhooks.c (default_get_mask_mode): Likewise.
13371 * tree-vect-stmts.c (vectorizable_store): Likewise.
13372 (vectorizable_load): Likewise.
13373 (get_vectype_for_scalar_type_and_size): Likewise.
13375 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13377 * machmode.h (mode_for_int_vector): New function.
13378 * stor-layout.c (mode_for_int_vector): Likewise.
13379 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13380 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13381 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13382 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13383 (s390_expand_vcond): Likewise.
13385 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13387 * machmode.h (opt_machine_mode): New type.
13388 (opt_mode<T>): Allow construction from anything that can be
13390 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13391 (mode_for_size): Return an opt_machine_mode.
13392 * stor-layout.c (mode_for_size): Likewise.
13393 (mode_for_size_tree): Update call accordingly.
13394 (bitwise_mode_for_mode): Likewise.
13395 (make_fract_type): Likewise.
13396 (make_accum_type): Likewise.
13397 * caller-save.c (replace_reg_with_saved_mem): Update call
13399 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13400 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13401 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13402 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13403 * expmed.c (extract_bit_field_1): Likewise.
13404 * reload.c (get_secondary_mem): Likewise.
13405 * varasm.c (assemble_integer): Likewise.
13406 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13409 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13411 * machmode.h (decimal_float_mode_for_size): New function.
13412 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13413 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13414 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13415 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13416 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13417 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13419 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13421 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13422 (get_builtin_sync_mode): Likewise.
13423 (expand_ifn_atomic_compare_exchange): Likewise.
13424 (expand_builtin_atomic_clear): Likewise.
13425 (expand_builtin_atomic_test_and_set): Likewise.
13426 (fold_builtin_atomic_always_lock_free): Likewise.
13427 * calls.c (compute_argument_addresses): Likewise.
13428 (emit_library_call_value_1): Likewise.
13429 (store_one_arg): Likewise.
13430 * combine.c (combine_instructions): Likewise.
13431 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13432 * config/arm/arm.c (arm_function_value): Likewise.
13433 (aapcs_allocate_return_reg): Likewise.
13434 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13435 * config/i386/i386.c (construct_container): Likewise.
13436 (ix86_gimplify_va_arg): Likewise.
13437 (ix86_expand_sse_cmp): Likewise.
13438 (emit_memmov): Likewise.
13439 (emit_memset): Likewise.
13440 (expand_small_movmem_or_setmem): Likewise.
13441 (ix86_expand_pextr): Likewise.
13442 (ix86_expand_pinsr): Likewise.
13443 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13444 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13446 * config/mips/mips.c (mips_function_value_1) Likewise.
13447 (mips_block_move_straight): Likewise.
13448 (mips_expand_ins_as_unaligned_store): Likewise.
13449 * config/powerpcspe/powerpcspe.c
13450 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13451 (rs6000_darwin64_record_arg_flush): Likewise.
13452 * config/rs6000/rs6000.c
13453 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13454 (rs6000_darwin64_record_arg_flush): Likewise.
13455 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13456 (sparc_function_value_1): Likewise.
13457 * config/spu/spu.c (adjust_operand): Likewise.
13458 (spu_emit_branch_or_set): Likewise.
13459 (arith_immediate_p): Likewise.
13460 * emit-rtl.c (gen_lowpart_common): Likewise.
13461 * expr.c (expand_expr_real_1): Likewise.
13462 * function.c (assign_parm_setup_block): Likewise.
13463 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13464 * reload1.c (alter_reg): Likewise.
13465 * stor-layout.c (mode_for_vector): Likewise.
13466 (layout_type): Likewise.
13468 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13470 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13471 (spu_convert_move): Likewise.
13472 * lower-subreg.c (resolve_simple_move): Likewise.
13474 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13477 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13478 define_insn to a define_expand.
13479 (altivec_vsum2sws_direct): New define_insn.
13480 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13482 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13484 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13487 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13492 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13493 realign stack if stack alignment needed is less than incoming
13496 2017-09-05 Marek Polacek <polacek@redhat.com>
13499 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13502 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13504 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13506 2017-09-05 Richard Biener <rguenther@suse.de>
13508 PR tree-optimization/82084
13509 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13511 2017-09-05 Richard Biener <rguenther@suse.de>
13513 PR tree-optimization/82102
13514 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13516 2017-09-05 Martin Liska <mliska@suse.cz>
13518 PR tree-optimization/82032
13519 * tree-cfg.c (generate_range_test): New function.
13520 * tree-cfg.h (generate_range_test): Declared here.
13521 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13522 (cleanup_control_expr_graph): Use it.
13523 * tree-switch-conversion.c (try_switch_expansion): Remove
13525 (emit_case_nodes): Use generate_range_test.
13527 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13530 * config/i386/i386.md (*<btsc><mode>_mask): Add
13531 TARGET_USE_BT to insn constraint.
13532 (*btr<mode>_mask): Ditto.
13534 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13536 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13537 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13539 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13542 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13543 TARGET_NEON and TARGET_IWMMXT.
13544 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13545 TARGET_NEON and TARGET_IWMMXT.
13546 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13548 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13550 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13551 (ix86_rewrite_tls_address): Ditto.
13552 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13553 (ix86_rewrite_tls_address_1): Ditto.
13554 (ix86_rewrite_tls_address): Ditto.
13555 * config/i386/predicates.md (tls_address_pattern): New predicate.
13556 * config/i386/i386.md (TLS address splitter): New splitter.
13558 2017-09-04 Richard Biener <rguenther@suse.de>
13560 PR tree-optimization/82084
13561 * fold-const.h (can_native_encode_string_p): Declare.
13562 * fold-const.c (can_native_encode_string_p): Factor out from ...
13563 (native_encode_string): ... here.
13564 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13565 vectorizing stores from constants we later cannot handle.
13567 2017-09-04 Marek Polacek <polacek@redhat.com>
13570 * doc/invoke.texi: Update -Wtautological-compare documentation.
13572 2017-09-04 Jeff Law <law@redhat.com>
13574 PR tree-optimization/64910
13575 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13576 swap the first and last operand if the last is a constant.
13578 2017-09-04 Marek Polacek <polacek@redhat.com>
13581 * convert.c (do_narrow): When sanitizing signed integer overflows,
13582 bail out for signed types.
13583 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13585 2017-09-04 Richard Biener <rguenther@suse.de>
13587 PR tree-optimization/82060
13588 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13589 Move devirtualization after stmt folding and before EH/AB/noreturn
13590 cleanup to get the stmt refs canonicalized. Use a bool instead
13591 of gimple_modified_p since that doesn't work for NOPs. Schedule
13592 NOPs generated by folding for removal.
13594 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13595 Alan Hayward <alan.hayward@arm.com>
13596 David Sherwood <david.sherwood@arm.com>
13598 * coretypes.h (pad_direction): New enum.
13599 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13600 (FUNCTION_ARG_PADDING): Likewise.
13601 * target.def (function_arg_padding): New hook.
13602 * targhooks.h (default_function_arg_padding): Declare.
13603 * targhooks.c (default_function_arg_padding): New function.
13604 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13605 (TARGET_FUNCTION_ARG_PADDING): ...this.
13606 * doc/tm.texi: Regenerate.
13607 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13608 instead of direction.
13609 (compute_argument_addresses): Likewise.
13610 (load_register_parameters): Likewise.
13611 (emit_library_call_value_1): Likewise.
13612 (store_one_arg): Use targetm.calls.function_arg_padding instead
13613 of FUNCTION_ARG_PADDING.
13614 (must_pass_in_stack_var_size_or_pad): Likewise.
13615 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13616 (emit_group_store): Likewise.
13617 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13618 instead of FUNCTION_ARG_PADDING.
13619 (emit_push_insn): Likewise, and propagate enum change throughout
13621 * function.h (direction): Delete.
13622 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13624 * function.c (assign_parm_find_stack_rtl): Likewise.
13625 (assign_parm_setup_block_p): Likewise.
13626 (assign_parm_setup_block): Likewise.
13627 (gimplify_parameters): Likewise.
13628 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13629 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13631 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13632 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13633 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13634 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13635 (aarch64_function_arg_padding): ...this new function.
13636 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13637 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13638 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13639 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13640 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13641 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13642 (arm_pad_arg_upward): Replace with...
13643 (arm_function_arg_padding): ...this new function.
13644 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13646 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13647 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13648 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13649 (ia64_hpux_function_arg_padding): Replace with...
13650 (ia64_function_arg_padding): ...this new function. Use pad_direction
13651 instead of direction. Check for TARGET_HPUX.
13652 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13653 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13654 (iq2000_function_arg_padding): New function.
13655 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13656 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13657 (mips_function_arg_padding): ...this new function.
13658 (mips_pad_reg_upward): Update accordingly.
13659 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13660 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13661 targetm.calls.function_arg_padding.
13662 (FUNCTION_ARG_PADDING): Delete.
13663 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13664 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13665 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13666 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13667 (nios2_block_reg_padding): Return pad_direction instead of direction.
13668 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13669 instead of direction.
13670 (nios2_function_arg_padding): Likewise. Make static.
13671 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13672 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13673 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13674 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13675 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13676 (pa_function_arg_padding): Make static. Return pad_direction instead
13678 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13679 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13680 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13681 instead of direction. Use targetm.calls.function_arg_padding.
13682 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13683 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13684 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13685 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13686 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13688 (function_arg_padding): Rename to...
13689 (rs6000_function_arg_padding): ...this. Make static. Return
13690 pad_direction instead of direction.
13691 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13692 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13693 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13694 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13695 instead of direction. Use targetm.calls.function_arg_padding.
13696 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13697 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13698 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13699 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13700 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13701 (function_arg_padding): Rename to...
13702 (rs6000_function_arg_padding): ...this. Make static. Return
13703 pad_direction instead of direction.
13704 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13705 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13706 * config/s390/s390.c (s390_function_arg_padding): New function.
13707 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13708 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13709 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13710 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13711 (function_arg_padding): Rename to...
13712 (sparc_function_arg_padding): ...this. Make static. Return
13713 pad_direction instead of direction.
13714 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13715 * config/spu/spu.c (spu_function_arg_padding): New function.
13716 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13717 * system.h (FUNCTION_ARG_PADDING): Poison.
13719 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13720 Alan Hayward <alan.hayward@arm.com>
13721 David Sherwood <david.sherwood@arm.com>
13723 * target.def (modes_tieable_p): New hook.
13724 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13725 (TARGET_MODES_TIEABLE_P): ...this.
13726 * doc/tm.texi.in: Regenerate.
13727 * hooks.h (hook_bool_mode_mode_true): Declare.
13728 * hooks.c (hook_bool_mode_mode_true): New function.
13729 * combine.c (subst): Use targetm.modes_tieable_p instead of
13731 * dse.c (find_shift_sequence): Likewise.
13732 * expmed.c (extract_low_bits): Likewise.
13733 * lower-subreg.c: Include target.h.
13734 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13736 * rtlanal.c (rtx_cost): Likewise.
13737 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13738 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13739 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13740 (TARGET_MODES_TIEABLE_P): Redefine.
13741 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13742 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13743 (TARGET_MODES_TIEABLE_P): Redefine.
13744 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13745 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13746 (arc_modes_tieable_p): New function.
13747 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13748 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13749 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13750 (arm_modes_tieable_p): Make static.
13751 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13752 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13753 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13754 (TARGET_MODES_TIEABLE_P): Redefine.
13755 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13756 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13757 (TARGET_MODES_TIEABLE_P): Redefine.
13758 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13759 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13760 (cr16_modes_tieable_p): New function.
13761 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13762 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13763 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13764 (TRULY_NOOP_TRUNCATION): Update comment.
13765 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13766 (TRULY_NOOP_TRUNCATION): Update comment.
13767 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13768 (frv_modes_tieable_p): New function.
13769 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13770 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13771 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13772 (TARGET_MODES_TIEABLE_P): Redefine.
13773 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13774 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13775 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13776 (TARGET_MODES_TIEABLE_P): Redefine.
13777 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13778 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13779 (ia64_modes_tieable_p): New function.
13780 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13781 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13782 (iq2000_modes_tieable_p): New function.
13783 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13784 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13785 (lm32_modes_tieable_p): New function.
13786 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13787 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13788 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13789 (TARGET_MODES_TIEABLE_P): Redefine.
13790 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13791 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13792 (m32r_modes_tieable_p): New function.
13793 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13794 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13795 (m68k_modes_tieable_p): New function.
13796 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13797 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13798 (mcore_modes_tieable_p): New function.
13799 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13800 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13802 (TARGET_MODES_TIEABLE_P): Redefine.
13803 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13804 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13805 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13806 (TARGET_MODES_TIEABLE_P): Redefine.
13807 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13808 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13809 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13810 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13811 (mn10300_modes_tieable_p): ...this and make static.
13812 (TARGET_MODES_TIEABLE_P): Redefine.
13813 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13814 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13815 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13816 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13817 (msp430_modes_tieable_p): Make static.
13818 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13819 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13820 (TARGET_MODES_TIEABLE_P): Redefine.
13821 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13822 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13823 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13824 (TARGET_MODES_TIEABLE_P): Redefine.
13825 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13826 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13827 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13828 (TARGET_MODES_TIEABLE_P): Redefine.
13829 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13830 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13831 (pdp11_modes_tieable_p): New function.
13832 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13833 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13834 (rs6000_modes_tieable_p): New function.
13835 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13836 * config/powerpcspe/powerpcspe.md: Update comment.
13837 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13838 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13839 (TARGET_MODES_TIEABLE_P): Redefine.
13840 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13841 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13842 (rl78_modes_tieable_p): New function.
13843 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13844 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13845 (rs6000_modes_tieable_p): New function.
13846 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13847 * config/rs6000/rs6000.md: Update comment.
13848 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13849 * config/rx/rx.c (rx_modes_tieable_p): New function.
13850 (TARGET_MODES_TIEABLE_P): Redefine.
13851 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13852 * config/s390/s390.c (s390_modes_tieable_p): New function.
13853 (TARGET_MODES_TIEABLE_P): Redefine.
13854 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13855 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13856 (sh_modes_tieable_p): New function.
13857 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13858 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13859 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13860 (sparc_modes_tieable_p): Make static.
13861 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13862 * config/spu/spu.c (spu_modes_tieable_p): New function.
13863 (TARGET_MODES_TIEABLE_P): Redefine.
13864 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13865 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13866 (TARGET_MODES_TIEABLE_P): Redefine.
13867 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13868 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13869 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13870 * config/v850/v850.c (v850_modes_tieable_p): New function.
13871 (TARGET_MODES_TIEABLE_P): Redefine.
13872 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13873 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13874 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13875 (visium_modes_tieable_p): New function.
13876 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13877 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13878 (xtensa_modes_tieable_p): New function.
13879 * system.h (MODES_TIEABLE_P): Poison.
13881 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13882 Alan Hayward <alan.hayward@arm.com>
13883 David Sherwood <david.sherwood@arm.com>
13885 * target.def (hard_regno_mode_ok): New hook.
13886 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13887 (TARGET_HARD_REGNO_MODE_OK): ...this.
13888 * doc/tm.texi.in: Regenerate.
13889 * hooks.h (hook_bool_uint_mode_true): Declare.
13890 * hooks.c (hook_bool_uint_mode_true): New function.
13891 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13892 HARD_REGNO_MODE_OK.
13893 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13894 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13895 instead of HARD_REGNO_MODE_OK.
13896 * caller-save.c: Include target.h.
13897 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13898 HARD_REGNO_MODE_OK.
13899 * combine.c (can_combine_p): Likewise.
13900 (combinable_i3pat): Likewise.
13901 (can_change_dest_mode): Likewise.
13902 * expr.c (init_expr_target): Likewise.
13903 (convert_move): Likewise.
13904 (convert_modes): Likewise.
13905 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13906 (setup_prohibited_mode_move_regs): Likewise.
13907 * ira.h (target_ira): Likewise.
13908 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13909 * lra-constraints.c (process_alt_operands): Likewise.
13910 (split_reg): Likewise.
13911 * recog.c (peep2_find_free_register): Likewise.
13912 * ree.c (combine_reaching_defs): Likewise.
13913 * regcprop.c (maybe_mode_change): Likewise.
13914 * reginfo.c (init_reg_sets_1): Likewise.
13915 (choose_hard_reg_mode): Likewise.
13916 (simplifiable_subregs): Likewise.
13917 * regrename.c (check_new_reg_p): Likewise.
13918 * reload.c (find_valid_class): Likewise.
13919 (find_valid_class_1): Likewise.
13920 (reload_inner_reg_of_subreg): Likewise.
13921 (push_reload): Likewise.
13922 (combine_reloads): Likewise.
13923 (find_dummy_reload): Likewise.
13924 (find_reloads): Likewise.
13925 * reload1.c (find_reg): Likewise.
13926 (set_reload_reg): Likewise.
13927 (allocate_reload_reg): Likewise.
13928 (choose_reload_regs): Likewise.
13929 (reload_adjust_reg_for_temp): Likewise.
13930 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
13931 (simplify_subreg_regno): Likewise.
13932 * sel-sched.c (init_regs_for_mode): Likewise.
13933 * varasm.c (make_decl_rtl): Likewise.
13934 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
13935 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
13936 HARD_REGNO_MODE_OK.
13937 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
13938 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
13939 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13940 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
13941 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
13942 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13943 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
13944 (arc_mode_class): Delete.
13945 (HARD_REGNO_MODE_OK): Delete.
13946 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13947 (arc_hard_regno_mode_ok): Rename old array to...
13948 (arc_hard_regno_mode_ok_modes): ...this.
13949 (arc_conditional_register_usage): Update accordingly.
13950 (arc_mode_class): Make static.
13951 (arc_hard_regno_mode_ok): New function.
13952 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
13953 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
13954 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13955 (arm_hard_regno_mode_ok): Make static.
13956 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
13957 HARD_REGNO_MODE_OK.
13958 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
13959 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
13960 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
13962 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13963 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
13964 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
13965 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
13966 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
13967 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13968 * config/bfin/predicates.md (valid_reg_operand): Use
13969 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
13970 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
13971 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
13972 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13973 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
13974 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
13975 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13976 (cr16_hard_regno_mode_ok): Make static and return a bool.
13977 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
13978 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13979 (cris_hard_regno_mode_ok): New function.
13980 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
13981 (epiphany_mode_class): Delete.
13982 (HARD_REGNO_MODE_OK): Delete.
13983 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
13984 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13985 (hard_regno_mode_ok): Rename to...
13986 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
13987 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
13988 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
13989 HARD_REGNO_MODE_OK.
13990 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
13991 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
13992 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13993 (frv_hard_regno_mode_ok): Make static and return a bool.
13994 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
13995 HARD_REGNO_MODE_OK.
13996 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
13997 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
13998 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
13999 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14002 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14003 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14004 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14006 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14007 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14008 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14009 (ia64_hard_regno_mode_ok): New function.
14010 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14011 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14012 (iq2000_hard_regno_mode_ok): New function.
14013 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14014 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14015 (lm32_hard_regno_mode_ok): New function.
14016 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14017 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14018 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14019 instead of HARD_REGNO_MODE_OK.
14020 (m32c_hard_regno_ok): Rename to...
14021 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14022 (m32c_cannot_change_mode_class): Update accordingly.
14023 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14024 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14025 (m32r_mode_class): Delete.
14026 (HARD_REGNO_MODE_OK): Delete.
14027 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14028 (m32r_hard_regno_mode_ok): Rename to...
14029 (m32r_hard_regno_modes): ...this.
14030 (m32r_mode_class): Make static.
14031 (m32r_hard_regno_mode_ok): New function.
14032 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14033 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14034 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14035 (m68k_hard_regno_mode_ok): Make static.
14036 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14037 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14038 (mcore_hard_regno_mode_ok): New function.
14039 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14040 (HARD_REGNO_MODE_OK): Delete.
14041 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14043 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14044 (microblaze_hard_regno_mode_ok): New function.
14045 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14046 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14047 (mips_hard_regno_mode_ok): Delete.
14048 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14049 (mips_hard_regno_mode_ok_p): ...this and make static.
14050 (mips_hard_regno_mode_ok_p): Rename to...
14051 (mips_hard_regno_mode_ok_uncached): ...this.
14052 (mips_hard_regno_mode_ok): New function.
14053 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14054 of HARD_REGNO_MODE_OK.
14055 (mips_option_override): Update after above name changes.
14056 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14057 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14058 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14059 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14060 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14061 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14062 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14063 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14064 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14065 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14066 (msp430_hard_regno_mode_ok): Make static and return a bool.
14067 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14068 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14069 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14071 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14072 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14073 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14074 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14075 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14076 (PA_HARD_REGNO_MODE_OK): ...this
14077 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14078 (PA_HARD_REGNO_MODE_OK): ...this.
14079 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14080 (pa_hard_regno_mode_ok): New function.
14081 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14082 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14083 (pdp11_hard_regno_mode_ok): New function.
14084 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14085 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14087 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14089 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14090 (rs6000_hard_regno_mode_ok): Rename to...
14091 (rs6000_hard_regno_mode_ok_uncached): ...this.
14092 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14093 (rs6000_hard_regno_mode_ok): New function.
14094 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14095 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14096 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14097 (riscv_hard_regno_mode_ok): ...this and make static.
14098 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14099 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14100 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14101 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14102 (rl78_hard_regno_mode_ok): Make static and return bool.
14103 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14104 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14106 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14107 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14108 (rs6000_hard_regno_mode_ok): Rename to...
14109 (rs6000_hard_regno_mode_ok_uncached): ...this.
14110 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14111 (rs6000_hard_regno_mode_ok): New function.
14112 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14113 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14114 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14115 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14116 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14117 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14118 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14119 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14120 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14121 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14122 (sh_hard_regno_mode_ok): Make static.
14123 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14124 instead of HARD_REGNO_MODE_OK.
14125 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14126 (sparc_mode_class): Delete.
14127 (HARD_REGNO_MODE_OK): Delete.
14128 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14129 (hard_regno_mode_classes): Make static.
14130 (sparc_mode_class): Likewise.
14131 (sparc_hard_regno_mode_ok): New function.
14132 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14133 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14134 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14136 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14137 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14138 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14139 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14140 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14141 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14142 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14143 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14144 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14145 (visium_hard_regno_mode_ok): New function.
14146 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14147 instead of HARD_REGNO_MODE_OK.
14148 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14149 (HARD_REGNO_MODE_OK): Delete.
14150 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14151 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14152 (xtensa_option_override): Update accordingly.
14153 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14154 (xtensa_hard_regno_mode_ok): New function.
14155 * system.h (HARD_REGNO_MODE_OK): Poison.
14157 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14158 Alan Hayward <alan.hayward@arm.com>
14159 David Sherwood <david.sherwood@arm.com>
14161 * target.def (hard_regno_call_part_clobbered): New hook.
14162 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14163 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14164 * doc/tm.texi: Regenerate.
14165 * hooks.h (hook_bool_uint_mode_false): Declare.
14166 * hooks.c (hook_bool_uint_mode_false): New function.
14167 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14168 * cselib.c (cselib_process_insn): Use
14169 targetm.hard_regno_call_part_clobbered instead of
14170 HARD_REGNO_CALL_PART_CLOBBERED.
14171 * ira-conflicts.c (ira_build_conflicts): Likewise.
14172 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14173 * lra-constraints.c (need_for_call_save_p): Likewise.
14174 * lra-lives.c: Include target.h.
14175 (check_pseudos_live_through_calls): Use
14176 targetm.hard_regno_call_part_clobbered instead of
14177 HARD_REGNO_CALL_PART_CLOBBERED.
14178 * regcprop.c: Include target.h.
14179 (copyprop_hardreg_forward_1): Use
14180 targetm.hard_regno_call_part_clobbered instead of
14181 HARD_REGNO_CALL_PART_CLOBBERED.
14182 * reginfo.c (choose_hard_reg_mode): Likewise.
14183 * regrename.c (check_new_reg_p): Likewise.
14184 * reload.c (find_equiv_reg): Likewise.
14185 * reload1.c (emit_reload_insns): Likewise.
14186 * sched-deps.c (deps_analyze_insn): Likewise.
14187 * sel-sched.c (init_regs_for_mode): Likewise.
14188 (mark_unavailable_hard_regs): Likewise.
14189 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14190 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14191 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14193 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14194 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14195 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14197 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14199 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14200 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14201 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14203 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14204 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14205 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14207 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14208 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14210 * config/powerpcspe/powerpcspe.c
14211 (rs6000_hard_regno_call_part_clobbered): New function.
14212 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14213 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14214 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14216 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14217 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14218 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14220 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14221 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14222 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14224 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14225 Alan Hayward <alan.hayward@arm.com>
14226 David Sherwood <david.sherwood@arm.com>
14228 * rtl.h (subreg_memory_offset): Declare.
14229 * emit-rtl.c (subreg_memory_offset): New function.
14230 * expmed.c (store_bit_field_1): Use it.
14231 * expr.c (undefined_operand_subword_p): Likewise.
14232 * simplify-rtx.c (simplify_subreg): Likewise.
14234 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14236 PR rtl-optimization/57448
14239 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14240 using atomic_load pattern.
14241 (expand_atomic_store): Likewise.
14243 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14246 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14247 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14248 handling. Use replace_call_with_value with NULL instead of
14249 gsi_replace, unlink_stmt_vdef and release_defs.
14251 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14254 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14256 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14259 * rtl.h (emit_library_call_value_1): Declare.
14260 (emit_library_call): Replace declaration with a series of overloads.
14261 Remove the parameter count argument.
14262 (emit_library_call_value): Likewise.
14263 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14264 with an "rtx_mode_t *".
14265 (emit_library_call_value): Delete.
14266 (emit_library_call): Likewise.
14267 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14268 (asan_emit_allocas_unpoison): Likewise.
14269 * builtins.c (expand_builtin_powi): Likewise.
14270 (expand_asan_emit_allocas_unpoison): Likewise.
14271 * cfgexpand.c (expand_main_function): Likewise.
14272 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14273 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14274 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14275 * config/arm/arm.c (arm_trampoline_init): Likewise.
14276 (arm_call_tls_get_addr): Likewise.
14277 (arm_expand_divmod_libfunc): Likewise.
14278 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14279 (smulsi3_highpart): Likewise.
14280 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14281 (c6x_expand_compare): Likewise.
14282 (c6x_expand_movmem): Likewise.
14283 * config/frv/frv.c (frv_trampoline_init): Likewise.
14284 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14285 (ix86_expand_divmod_libfunc): Likewise.
14286 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14287 (ia64_expand_compare): Likewise.
14288 (ia64_profile_hook): Likewise.
14289 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14290 (nonlocal_goto): Likewise.
14291 (restore_stack_nonlocal): Likewise.
14292 * config/m32r/m32r.c (block_move_call): Likewise.
14293 (m32r_trampoline_init): Likewise.
14294 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14295 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14296 (m68k_call_m68k_read_tp): Likewise.
14297 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14298 (microblaze_expand_divide): Likewise.
14299 * config/mips/mips.h (mips_args): Likewise.
14300 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14301 (MIPS_ICACHE_SYNC): Likewise.
14302 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14303 (nios2_trampoline_init): Likewise.
14304 * config/pa/pa.c (hppa_tls_call): Likewise.
14305 (pa_trampoline_init): Likewise.
14306 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14307 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14308 (expand_strn_compare): Likewise.
14309 (rs6000_generate_compare): Likewise.
14310 (rs6000_expand_float128_convert): Likewise.
14311 (output_profile_hook): Likewise.
14312 (rs6000_trampoline_init): Likewise.
14313 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14314 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14315 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14316 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14317 (rs6000_generate_compare): Likewise.
14318 (rs6000_expand_float128_convert): Likewise.
14319 (output_profile_hook): Likewise.
14320 (rs6000_trampoline_init): Likewise.
14321 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14322 * config/sh/sh.c (sh_trampoline_init): Likewise.
14323 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14324 (sparc_emit_float_lib_cmp): Likewise.
14325 (sparc32_initialize_trampoline): Likewise.
14326 (sparc64_initialize_trampoline): Likewise.
14327 (sparc_profile_hook): Likewise.
14328 * config/spu/spu.c (ea_load_store): Likewise.
14329 * config/spu/spu.md (floatunssidf2): Likewise.
14330 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14331 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14332 * config/visium/visium.c (expand_block_move_4): Likewise.
14333 (expand_block_move_2): Likewise.
14334 (expand_block_move_1): Likewise.
14335 (expand_block_set_4): Likewise.
14336 (expand_block_set_2): Likewise.
14337 (expand_block_set_1): Likewise.
14338 (visium_trampoline_init): Likewise.
14339 (visium_profile_hook): Likewise.
14340 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14341 (xtensa_setup_frame_addresses): Likewise.
14342 (xtensa_trampoline_init): Likewise.
14343 * except.c (sjlj_emit_function_enter): Likewise.
14344 (sjlj_emit_function_exit): Likewise.
14345 * explow.c (allocate_dynamic_stack_space): Likewise.
14346 (probe_stack_range): Likewise.
14347 * expr.c (convert_mode_scalar): Likewise.
14348 * optabs.c (expand_binop): Likewise.
14349 (expand_twoval_binop_libfunc): Likewise.
14350 (expand_unop): Likewise.
14351 (prepare_cmp_insn): Likewise.
14352 (prepare_float_lib_cmp): Likewise.
14353 (expand_float): Likewise.
14354 (expand_fix): Likewise.
14355 (expand_fixed_convert): Likewise.
14356 (maybe_emit_sync_lock_test_and_set): Likewise.
14357 (expand_atomic_compare_and_swap): Likewise.
14358 (expand_mem_thread_fence): Likewise.
14359 (expand_atomic_fetch_op): Likewise.
14361 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14363 * doc/generic.texi (OpenACC): Adjust URL.
14364 * doc/invoke.texi (C Dialect Options): Ditto.
14366 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14368 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14369 predicate for operand 1. Add (m,<S>) constraint.
14370 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14371 Prevent memory operand 1 with register operand 2.
14373 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14375 PR rtl-optimization/82024
14376 * combine.c (try_combine): If the combination result is a PARALLEL,
14377 and we only need to retain the SET in there that would be placed
14378 at I2, check that we can place that at I3 instead, before doing so.
14380 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14383 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14385 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14386 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14389 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14390 Jeff Law <law@redhat.com>
14392 * varasm.c (bss_initializer_p): Do not put constants into .bss
14393 (categorize_decl_for_section): Handle bss_initializer_p returning
14394 false when DECL_INITIAL is NULL.
14396 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14399 * config/s390/s390.c (s390_can_inline_p): New function.
14401 2017-09-01 Jeff Law <law@redhat.com>
14403 PR tree-optimization/82052
14404 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14405 Always initialize the returned slot after a hash table miss
14406 when INSERT is true.
14408 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14410 * config/s390/s390.md (mem_signal_fence): Remove.
14411 * doc/md.texi (mem_signal_fence): Remove.
14412 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14414 * target-insns.def (mem_signal_fence): Remove.
14416 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14419 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14422 * asan.c (create_odr_indicator): Strip name encoding from assembler
14423 name before appending it after __odr_asan_.
14425 2017-09-01 Martin Liska <mliska@suse.cz>
14427 PR tree-optimization/82059
14428 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14429 frequency only when an edge is redirected.
14431 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14433 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14434 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14435 (arc_conditional_register_usage): Remove ARC600 lp_count
14437 (arc_file_start): Emit Tag_ARC_CPU_variation.
14438 (arc_can_use_doloop_p): New conditions to use ZOLs.
14439 (hwloop_fail): New function.
14440 (hwloop_optimize): Likewise.
14441 (hwloop_pattern_reg): Likewise.
14442 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14443 (arc_reorg_loops): New function, calls reorg_loops.
14444 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14445 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14447 (arc_loop_hazard): Remove function, functionality moved into
14449 (arc_hazard): Remove arc_loop_hazard call.
14450 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14451 into hwloop_optimize.
14452 (arc_label_align): Remove ZOL handling.
14453 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14454 * config/arc/arc.md (doloop_begin): Remove pattern.
14455 (doloop_begin_i): Likewise.
14456 (doloop_end_i): Likewise.
14457 (doloop_fallback): Likewise.
14458 (doloop_fallback_m): Likewise.
14459 (doloop_end): Reimplement expand.
14460 (arc_lp): New pattern for LP instruction.
14461 (loop_end): New pattern.
14462 (loop_fail): Likewise.
14463 (decrement_and_branch_until_zero): Likewise.
14464 * config/arc/arc.opt (mlpc-width): New option.
14465 * doc/invoke.texi (mlpc-width): Document option.
14467 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14469 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14470 (arc_ccfsm_advance): Fix checking for delay slots.
14471 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14473 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14475 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14477 (movhi_insn): Update store instruction constraint which are saving
14478 6-bit short immediates.
14479 (movsi_insn): Consider also short scaled load operations.
14480 (zero_extendhisi2_i): Use Usd constraint instead of T.
14481 (extendhisi2_i): Add q constraint.
14482 (arc_clzsi2): Add type and length attributes.
14483 (arc_ctzsi2): Likewise.
14484 * config/arc/constraints.md (Usc): Update constraint, the
14485 assembler can parse two relocations for a single instruction.
14487 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14489 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14490 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14492 2017-08-31 Olivier Hainque <hainque@adacore.com>
14494 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14495 match as powerpc-wrs-vxworks*.
14497 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14499 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14500 register constraint for by-element operand.
14501 (aarch64_mls_elt_merge<mode>): Likewise.
14503 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14505 * config/arc/arc.c (arc_can_follow_jump): Check for short
14508 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14510 * config.gcc: Use g.opt for arc.
14511 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14512 functionality moved to ...
14513 (legitimate_scaled_address_p): New function, ...here.
14514 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14515 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14516 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14518 (arc_override_options): Handle G option.
14519 (arc_output_pic_addr_const): Correct function definition.
14520 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14521 (arc_decl_anon_ns_mem_p): Delete.
14522 (arc_in_small_data_p): Overhaul this function to take into
14523 consideration the value given via G option.
14524 (arc_rewrite_small_data_1): Renamed and corrected old
14525 arc_rewrite_small_data function.
14526 (arc_rewrite_small_data): New function.
14527 (small_data_pattern): Don't use pic_offset_table_rtx.
14528 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14529 * config/arc/simdext.md (movmisalignv2hi): Use
14530 prepare_move_operands function.
14532 (movmisalign*): Likewise.
14533 * doc/invoke.texi (ARC options): Document -G option.
14535 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14537 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14539 * config/arc/arc.c (arc_print_operand): Output scalled address for
14540 sdata whenever is possible.
14541 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14542 load/stores are available.
14543 (compact_sda_memory_operand): Check for the alignment required by
14544 code density instructions.
14545 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14547 * config/arc/constraints.md (Usd): Update constraint.
14548 (Us0): New constraint.
14549 (Usc): Update constraint.
14551 2017-08-31 Richard Biener <rguenther@suse.de>
14553 PR middle-end/82054
14554 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14555 function only once.
14557 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14559 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14560 Resize type_signature.
14562 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14563 Alan Hayward <alan.hayward@arm.com>
14564 David Sherwood <david.sherwood@arm.com>
14566 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14567 subregs whose inner modes can be stored in GPRs.
14568 (aarch64_classify_index): Likewise.
14570 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14571 Alan Hayward <alan.hayward@arm.com>
14572 David Sherwood <david.sherwood@arm.com>
14574 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14575 (V_INT_EQUIV): ...this.
14576 (v_cmp_result): Rename to...
14577 (v_int_equiv): ...this.
14578 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14579 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14580 (copysign<mode>3): Likewise.
14581 (aarch64_simd_bsl<mode>_internal): Likewise.
14582 (aarch64_simd_bsl<mode>): Likewise.
14583 (vec_cmp<mode><mode>): Likewise.
14584 (vcond<mode><mode>): Likewise.
14585 (vcond<v_cmp_mixed><mode>): Likewise.
14586 (vcondu<mode><v_cmp_mixed>): Likewise.
14587 (aarch64_cm<optab><mode>): Likewise.
14588 (aarch64_cmtst<mode>): Likewise.
14589 (aarch64_fac<optab><mode>): Likewise.
14590 (vec_perm_const<mode>): Likewise.
14591 (vcond_mask_<mode><v_cmp_result>): Rename to...
14592 (vcond_mask_<mode><v_int_equiv>): ...this.
14593 (vec_cmp<mode><v_cmp_result>): Rename to...
14594 (vec_cmp<mode><v_int_equiv>): ...this.
14596 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14597 Alan Hayward <alan.hayward@arm.com>
14598 David Sherwood <david.sherwood@arm.com>
14600 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14602 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14603 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14604 (UNSPEC_LD4_DREG): New unspecs.
14605 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14606 (aarch64_ld2<mode>_dreg_be): Replace with...
14607 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14609 (aarch64_ld3<mode>_dreg_le)
14610 (aarch64_ld3<mode>_dreg_be): Replace with...
14611 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14613 (aarch64_ld4<mode>_dreg_le)
14614 (aarch64_ld4<mode>_dreg_be): Replace with...
14615 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14618 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14620 PR tree-optimization/81987
14621 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14622 insert an initializer in a location not dominated by the stride
14625 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14627 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14628 on the entire header of the finally block in the fallthru case.
14630 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14632 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14634 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14636 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14637 rs6000_emit_move_from_cr and call renamed function.
14638 (rs6000_emit_prologue): Call renamed functions.
14639 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14640 movesi_from_cr, remove volatile CRs.
14642 2017-08-30 Jon Beniston <jon@beniston.com>
14643 Richard Biener <rguenther@suse.de>
14645 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14646 of VECTOR_MODE_P check.
14647 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14648 element vector types.
14650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14652 * df.h (df_read_modify_subreg_p): Remove in favor of...
14653 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14654 const_rtx instead of an rtx.
14655 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14656 * df-problems.c (df_word_lr_mark_ref): Likewise.
14657 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14658 (mark_pseudo_reg_dead): Likewise.
14659 (mark_ref_dead): Likewise.
14660 * reginfo.c (init_subregs_of_mode): Likewise.
14661 * sched-deps.c (sched_analyze_1): Likewise.
14662 * df-scan.c (df_def_record_1): Likewise.
14663 (df_uses_record): Likewise.
14664 (df_read_modify_subreg_p): Remove in favor of...
14665 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14666 const_rtx instead of an rtx.
14668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14669 Alan Hayward <alan.hayward@arm.com>
14670 David Sherwood <david.sherwood@arm.com>
14672 * rtl.h (partial_subreg_p): New function.
14673 * caller-save.c (save_call_clobbered_regs): Use it.
14674 * calls.c (expand_call): Likewise.
14675 * combine.c (combinable_i3pat): Likewise.
14676 (simplify_set): Likewise.
14677 (make_extraction): Likewise.
14678 (make_compound_operation_int): Likewise.
14679 (gen_lowpart_or_truncate): Likewise.
14680 (force_to_mode): Likewise.
14681 (make_field_assignment): Likewise.
14682 (reg_truncated_to_mode): Likewise.
14683 (record_truncated_value): Likewise.
14684 (move_deaths): Likewise.
14685 * cse.c (record_jump_cond): Likewise.
14686 (cse_insn): Likewise.
14687 * cselib.c (cselib_lookup_1): Likewise.
14688 * expmed.c (extract_bit_field_using_extv): Likewise.
14689 * function.c (assign_parm_setup_reg): Likewise.
14690 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14691 * ira-build.c (create_insn_allocnos): Likewise.
14692 * lra-coalesce.c (merge_pseudos): Likewise.
14693 * lra-constraints.c (match_reload): Likewise.
14694 (simplify_operand_subreg): Likewise.
14695 (curr_insn_transform): Likewise.
14696 * lra-lives.c (process_bb_lives): Likewise.
14697 * lra.c (new_insn_reg): Likewise.
14698 (lra_substitute_pseudo): Likewise.
14699 * regcprop.c (mode_change_ok): Likewise.
14700 (maybe_mode_change): Likewise.
14701 (copyprop_hardreg_forward_1): Likewise.
14702 * reload.c (push_reload): Likewise.
14703 (find_reloads): Likewise.
14704 (find_reloads_subreg_address): Likewise.
14705 * reload1.c (alter_reg): Likewise.
14706 (eliminate_regs_1): Likewise.
14707 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14709 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14711 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14712 back to if statements, including unpack.
14714 2017-08-30 Martin Liska <mliska@suse.cz>
14716 PR inline-asm/82001
14717 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14719 (func_checker::compare_asm_inputs_outputs): ... this function.
14720 (func_checker::compare_gimple_asm): Use the function to compare
14721 also ASM constrains.
14722 * ipa-icf-gimple.h: Rename the function.
14724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14725 Alan Hayward <alan.hayward@arm.com>
14726 David Sherwood <david.sherwood@arm.com>
14728 * coretypes.h (complex_mode): New type.
14729 * gdbhooks.py (build_pretty_printer): Handle it.
14730 * machmode.h (complex_mode): New class.
14731 (complex_mode::includes_p): New function.
14732 (is_complex_int_mode): Likewise.
14733 (is_complex_float_mode): Likewise.
14734 * genmodes.c (get_mode_class): Handle complex mode classes.
14735 * function.c (expand_function_end): Use is_complex_int_mode.
14737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14738 Alan Hayward <alan.hayward@arm.com>
14739 David Sherwood <david.sherwood@arm.com>
14741 * coretypes.h (scalar_mode_pod): New typedef.
14742 * gdbhooks.py (build_pretty_printer): Handle it.
14743 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14744 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14745 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14746 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14747 as_a <scalar_mode>.
14749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14750 Alan Hayward <alan.hayward@arm.com>
14751 David Sherwood <david.sherwood@arm.com>
14753 * machmode.h (mode_for_vector): Take a scalar_mode instead
14755 * stor-layout.c (mode_for_vector): Likewise.
14756 * explow.c (promote_mode): Use as_a <scalar_mode>.
14757 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14760 Alan Hayward <alan.hayward@arm.com>
14761 David Sherwood <david.sherwood@arm.com>
14763 * target.def (preferred_simd_mode): Take a scalar_mode
14764 instead of a machine_mode.
14765 * targhooks.h (default_preferred_simd_mode): Likewise.
14766 * targhooks.c (default_preferred_simd_mode): Likewise.
14767 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14768 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14769 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14770 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14771 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14772 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14773 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14774 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14776 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14777 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14778 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14779 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14780 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14781 * doc/tm.texi: Regenerate.
14782 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14786 Alan Hayward <alan.hayward@arm.com>
14787 David Sherwood <david.sherwood@arm.com>
14789 * target.def (scalar_mode_supported_p): Take a scalar_mode
14790 instead of a machine_mode.
14791 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14792 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14793 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14794 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14795 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14796 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14797 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14798 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14799 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14800 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14801 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14802 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14803 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14804 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14806 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14807 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14808 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14809 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14810 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14812 * doc/tm.texi: Regenerate.
14814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14815 Alan Hayward <alan.hayward@arm.com>
14816 David Sherwood <david.sherwood@arm.com>
14818 * coretypes.h (opt_scalar_mode): New typedef.
14819 * gdbhooks.py (build_pretty_printers): Handle it.
14820 * machmode.h (mode_iterator::get_2xwider): Add overload for
14822 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14824 * expr.c (convert_mode_scalar): Likewise.
14825 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14826 * optabs.c (expand_float): Likewise.
14827 (expand_fix): Likewise.
14828 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14831 Alan Hayward <alan.hayward@arm.com>
14832 David Sherwood <david.sherwood@arm.com>
14834 * optabs.c (expand_float): Explicitly check for scalars before
14835 using a branching expansion.
14836 (expand_fix): Likewise.
14838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14839 Alan Hayward <alan.hayward@arm.com>
14840 David Sherwood <david.sherwood@arm.com>
14842 * expr.c (convert_mode): Split scalar handling out into...
14843 (convert_mode_scalar): ...this new function. Treat the modes
14846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14847 Alan Hayward <alan.hayward@arm.com>
14848 David Sherwood <david.sherwood@arm.com>
14850 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14852 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14855 Alan Hayward <alan.hayward@arm.com>
14856 David Sherwood <david.sherwood@arm.com>
14858 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14859 rather than a machine_mode.
14860 (fixed_from_string): Likewise.
14861 (fixed_convert): Likewise.
14862 (fixed_convert_from_int): Likewise.
14863 (fixed_convert_from_real): Likewise.
14864 (real_convert_from_fixed): Likewise.
14865 * fixed-value.c (fixed_from_double_int): Likewise.
14866 (fixed_from_string): Likewise.
14867 (fixed_convert): Likewise.
14868 (fixed_convert_from_int): Likewise.
14869 (fixed_convert_from_real): Likewise.
14870 (real_convert_from_fixed): Likewise.
14871 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
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 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14878 of separate mode class checks. Do not allow vector modes here.
14879 (immed_wide_int_const): Use as_a <scalar_mode>.
14880 * explow.c (trunc_int_for_mode): Likewise.
14881 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14882 (wi::shwi): Likewise.
14883 (wi::min_value): Likewise.
14884 (wi::max_value): Likewise.
14885 * dwarf2out.c (loc_descriptor): Likewise.
14886 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14887 for CONST_WIDE_INT.
14889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14890 Alan Hayward <alan.hayward@arm.com>
14891 David Sherwood <david.sherwood@arm.com>
14893 * tree.h (SCALAR_TYPE_MODE): New macro.
14894 * expr.c (expand_expr_addr_expr_1): Use it.
14895 (expand_expr_real_2): Likewise.
14896 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14897 (fold_convert_const_fixed_from_int): Likewise.
14898 (fold_convert_const_fixed_from_real): Likewise.
14899 (native_encode_fixed): Likewise
14900 (native_encode_complex): Likewise
14901 (native_encode_vector): Likewise.
14902 (native_interpret_fixed): Likewise.
14903 (native_interpret_real): Likewise.
14904 (native_interpret_complex): Likewise.
14905 (native_interpret_vector): Likewise.
14906 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14907 (simd_clone_adjust_argument_types): Likewise.
14908 (simd_clone_init_simd_arrays): Likewise.
14909 (simd_clone_adjust): Likewise.
14910 * stor-layout.c (layout_type): Likewise.
14911 * tree.c (build_minus_one_cst): Likewise.
14912 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14913 * tree-inline.c (estimate_move_cost): Likewise.
14914 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14915 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14916 (vectorizable_reduction): Likewise.
14917 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14918 (vect_recog_mixed_size_cond_pattern): Likewise.
14919 (check_bool_pattern): Likewise.
14920 (adjust_bool_pattern): Likewise.
14921 (search_type_for_mask_1): Likewise.
14922 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
14923 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14924 (vectorizable_load): Likewise.
14925 (vectorizable_store): Likewise.
14926 * ubsan.c (ubsan_encode_value): Likewise.
14927 * varasm.c (output_constant): Likewise.
14929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14930 Alan Hayward <alan.hayward@arm.com>
14931 David Sherwood <david.sherwood@arm.com>
14933 * coretypes.h (scalar_mode): New class.
14934 * machmode.h (scalar_mode): Likewise.
14935 (scalar_mode::includes_p): New function.
14936 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
14937 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
14938 * genmodes.c (get_mode_class): Handle remaining scalar modes.
14939 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
14940 * expmed.c (store_bit_field_1): Likewise.
14941 (extract_bit_field_1): Likewise.
14942 * expr.c (write_complex_part): Likewise.
14943 (read_complex_part): Likewise.
14944 (emit_move_complex_push): Likewise.
14945 (expand_expr_real_2): Likewise.
14946 * function.c (assign_parm_setup_reg): Likewise.
14947 (assign_parms_unsplit_complex): Likewise.
14948 * optabs.c (expand_binop): Likewise.
14949 * rtlanal.c (subreg_get_info): Likewise.
14950 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14951 * varasm.c (output_constant_pool_2): Likewise.
14953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14954 Alan Hayward <alan.hayward@arm.com>
14955 David Sherwood <david.sherwood@arm.com>
14957 * expmed.c (extract_high_half): Use scalar_int_mode and remove
14959 (expmed_mult_highpart_optab): Likewise.
14960 (expmed_mult_highpart): Likewise.
14962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14963 Alan Hayward <alan.hayward@arm.com>
14964 David Sherwood <david.sherwood@arm.com>
14966 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
14967 instead of a machine_mode.
14968 (builtin_memset_read_str): Likewise.
14969 * builtins.c (c_readstr): Likewise.
14970 (builtin_memcpy_read_str): Likewise.
14971 (builtin_strncpy_read_str): Likewise.
14972 (builtin_memset_read_str): Likewise.
14973 (builtin_memset_gen_str): Likewise.
14974 (expand_builtin_signbit): Use scalar_int_mode for local variables.
14975 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
14976 instead of a machine_mode.
14977 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
14979 (make_extraction): Likewise.
14980 (try_widen_shift_mode): Take and return scalar_int_modes instead
14982 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
14983 a scalar_int_mode instead of a machine_mode.
14984 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
14985 (avr_addr_space_pointer_mode): Likewise.
14986 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
14987 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
14988 (msp430_unwind_word_mode): Likewise.
14989 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
14990 (spu_addr_space_pointer_mode): Likewise.
14991 (spu_addr_space_address_mode): Likewise.
14992 (spu_libgcc_cmp_return_mode): Likewise.
14993 (spu_libgcc_shift_count_mode): Likewise.
14994 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
14995 (rl78_addr_space_pointer_mode): Likewise.
14996 (fl78_unwind_word_mode): Likewise.
14997 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
14999 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15000 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15001 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15002 (mips_valid_pointer_mode): Likewise.
15003 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15004 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15005 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15007 (ft32_addr_space_address_mode): Likewise.
15008 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15009 scalar_int_mode instead of a machine_mode.
15010 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15012 (m32c_addr_space_address_mode): Likewise.
15013 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15014 (rs6000_eh_return_filter_mode): Likewise.
15015 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15016 (rs6000_eh_return_filter_mode): Likewise.
15017 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15018 (s390_libgcc_shift_count_mode): Likewise.
15019 (s390_unwind_word_mode): Likewise.
15020 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15022 * target.def (mode_rep_extended): Likewise.
15023 (valid_pointer_mode): Likewise.
15024 (addr_space.valid_pointer_mode): Likewise.
15025 (eh_return_filter_mode): Return a scalar_int_mode rather than
15027 (libgcc_cmp_return_mode): Likewise.
15028 (libgcc_shift_count_mode): Likewise.
15029 (unwind_word_mode): Likewise.
15030 (addr_space.pointer_mode): Likewise.
15031 (addr_space.address_mode): Likewise.
15032 * doc/tm.texi: Regenerate.
15033 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15035 (do_jump): Use scalar_int_mode for local variables.
15036 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15037 rather than a machine_mode.
15038 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15039 (scompare_loc_descriptor_wide): Likewise.
15040 (scompare_loc_descriptor_narrow): Likewise.
15041 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15043 * except.c (sjlj_emit_dispatch_table): Likewise.
15044 (expand_builtin_eh_copy_values): Likewise.
15045 * explow.c (convert_memory_address_addr_space_1): Likewise.
15046 Take a scalar_int_mode rather than a machine_mode.
15047 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15048 than a machine_mode.
15049 (memory_address_addr_space): Use scalar_int_mode for local variables.
15050 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15051 rather than a machine_mode.
15052 * expmed.c (mask_rtx): Likewise.
15053 (init_expmed_one_conv): Likewise.
15054 (expand_mult_highpart_adjust): Likewise.
15055 (extract_high_half): Likewise.
15056 (expmed_mult_highpart_optab): Likewise.
15057 (expmed_mult_highpart): Likewise.
15058 (expand_smod_pow2): Likewise.
15059 (expand_sdiv_pow2): Likewise.
15060 (emit_store_flag_int): Likewise.
15061 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15063 (extract_low_bits): Likewise.
15064 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15066 * expr.c (pieces_addr::adjust): Likewise.
15067 (can_store_by_pieces): Likewise.
15068 (store_by_pieces): Likewise.
15069 (clear_by_pieces_1): Likewise.
15070 (expand_expr_addr_expr_1): Likewise.
15071 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15072 (expand_expr_real_1): Likewise.
15073 (try_casesi): Likewise.
15074 * final.c (shorten_branches): Likewise.
15075 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15076 type of "mode" to machine_mode.
15077 * internal-fn.c (expand_arith_overflow_result_store): Take a
15078 scalar_int_mode rather than a machine_mode.
15079 (expand_mul_overflow): Use scalar_int_mode for local variables.
15080 * loop-doloop.c (doloop_modify): Likewise.
15081 (doloop_optimize): Likewise.
15082 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15083 than a machine_mode.
15084 (expand_doubleword_shift_condmove): Likewise.
15085 (expand_doubleword_shift): Likewise.
15086 (expand_doubleword_clz): Likewise.
15087 (expand_doubleword_popcount): Likewise.
15088 (expand_doubleword_parity): Likewise.
15089 (expand_absneg_bit): Use scalar_int_mode for local variables.
15090 (prepare_float_lib_cmp): Likewise.
15091 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15092 rather than a machine_mode.
15093 (convert_memory_address_addr_space): Likewise.
15094 (get_mode_bounds): Likewise.
15095 (get_address_mode): Return a scalar_int_mode rather than a
15097 * rtlanal.c (get_address_mode): Likewise.
15098 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15099 than a machine_mode.
15100 * targhooks.c (default_mode_rep_extended): Likewise.
15101 (default_valid_pointer_mode): Likewise.
15102 (default_addr_space_valid_pointer_mode): Likewise.
15103 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15104 than a machine_mode.
15105 (default_libgcc_cmp_return_mode): Likewise.
15106 (default_libgcc_shift_count_mode): Likewise.
15107 (default_unwind_word_mode): Likewise.
15108 (default_addr_space_pointer_mode): Likewise.
15109 (default_addr_space_address_mode): Likewise.
15110 * targhooks.h (default_eh_return_filter_mode): Likewise.
15111 (default_libgcc_cmp_return_mode): Likewise.
15112 (default_libgcc_shift_count_mode): Likewise.
15113 (default_unwind_word_mode): Likewise.
15114 (default_addr_space_pointer_mode): Likewise.
15115 (default_addr_space_address_mode): Likewise.
15116 (default_mode_rep_extended): Take a scalar_int_mode rather than
15118 (default_valid_pointer_mode): Likewise.
15119 (default_addr_space_valid_pointer_mode): Likewise.
15120 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15122 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15123 rather than a machine_mode.
15124 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15125 for local variables.
15126 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15127 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15128 than a machine_mode.
15130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15131 Alan Hayward <alan.hayward@arm.com>
15132 David Sherwood <david.sherwood@arm.com>
15134 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15135 the mode argument to scalar_int_mode.
15136 (do_jump_by_parts_zero_rtx): Likewise.
15137 (do_jump_by_parts_equality_rtx): Likewise.
15138 (do_jump_by_parts_greater): Take a mode argument.
15139 (do_jump_by_parts_equality): Likewise.
15140 (do_jump_1): Update calls accordingly.
15142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15143 Alan Hayward <alan.hayward@arm.com>
15144 David Sherwood <david.sherwood@arm.com>
15146 * is-a.h (safe_dyn_cast): New function.
15147 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15148 (jump_table_for_label): Likewise.
15149 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15150 instead of an rtx_insn *.
15151 (shorten_branches): Use dyn_cast instead of LABEL_P and
15152 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15153 rtx_jump_table_data::get_data_mode.
15154 (final_scan_insn): Likewise.
15156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15157 Alan Hayward <alan.hayward@arm.com>
15158 David Sherwood <david.sherwood@arm.com>
15160 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15161 trying to combine a full-register integer set with a subreg
15164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15165 Alan Hayward <alan.hayward@arm.com>
15166 David Sherwood <david.sherwood@arm.com>
15168 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15169 that is always either address_mode or pointer_mode.
15171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15172 Alan Hayward <alan.hayward@arm.com>
15173 David Sherwood <david.sherwood@arm.com>
15175 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15176 when the two are known to be equal.
15178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15179 Alan Hayward <alan.hayward@arm.com>
15180 David Sherwood <david.sherwood@arm.com>
15182 * simplify-rtx.c (simplify_const_unary_operation): Use
15183 is_a <scalar_int_mode> instead of checking for a nonzero
15184 precision. Forcibly convert op_mode to a scalar_int_mode
15185 in that case. More clearly differentiate the operand and
15186 result modes and use the former when deciding what the value
15187 of a count-bits operation should be. Use is_int_mode instead
15188 of checking for a MODE_INT. Remove redundant check for whether
15189 this mode has a zero precision.
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 * optabs.c (widen_leading): Change the type of the mode argument
15196 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15197 (widen_bswap): Likewise.
15198 (expand_parity): Likewise.
15199 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15200 (expand_ffs): Likewise.
15201 (epand_unop): Check for scalar integer modes before calling the
15204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15205 Alan Hayward <alan.hayward@arm.com>
15206 David Sherwood <david.sherwood@arm.com>
15208 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15210 (expand_expr_real_1): Update call accordingly.
15212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15213 Alan Hayward <alan.hayward@arm.com>
15214 David Sherwood <david.sherwood@arm.com>
15216 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15217 value_mode arguments. Use scalar_int_mode internally.
15218 (store_bit_field_1): Rename the new integer mode from imode
15219 to op0_mode and use it instead of GET_MODE (op0). Update calls
15220 to store_split_bit_field, store_bit_field_using_insv and
15221 store_fixed_bit_field.
15222 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15223 Use scalar_int_mode internally. Use a bit count rather than a mode
15224 when calculating the largest bit size for get_best_mode.
15225 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15226 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15227 Remove assertion that OP0 has a scalar integer mode.
15228 (store_split_bit_field): Add op0_mode and value_mode arguments.
15229 Update calls to extract_fixed_bit_field.
15230 (extract_bit_field_using_extv): Add an op0_mode argument.
15231 Use scalar_int_mode internally.
15232 (extract_bit_field_1): Rename the new integer mode from imode to
15233 op0_mode and use it instead of GET_MODE (op0). Update calls to
15234 extract_split_bit_field, extract_bit_field_using_extv and
15235 extract_fixed_bit_field.
15236 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15237 to extract_split_bit_field and extract_fixed_bit_field_1.
15238 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15239 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15240 on the target mode.
15241 (extract_split_bit_field): Add an op0_mode argument. Update call
15242 to extract_fixed_bit_field.
15244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15245 Alan Hayward <alan.hayward@arm.com>
15246 David Sherwood <david.sherwood@arm.com>
15248 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15249 * explow.c (hard_function_value): Likewise.
15250 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15251 convert_to_mode call outside the loop.
15252 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15253 for the mode iterator. Require the mode specified by max_pieces
15255 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15257 (copy_blkmode_to_reg): Likewise.
15258 (set_storage_via_setmem): Likewise.
15259 * optabs.c (prepare_cmp_insn): Likewise.
15260 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15261 * stor-layout.c (finish_bitfield_representative): Likewise.
15263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15264 Alan Hayward <alan.hayward@arm.com>
15265 David Sherwood <david.sherwood@arm.com>
15267 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15268 * expr.c (convert_move): Use them.
15269 (convert_modes): Likewise.
15270 (store_expr_with_bounds): Likewise.
15272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15273 Alan Hayward <alan.hayward@arm.com>
15274 David Sherwood <david.sherwood@arm.com>
15276 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15277 parameter for the mode of "x". Remove the "known_x", "known_mode"
15278 and "known_ret" arguments. Change the type of the mode argument
15279 to scalar_int_mode.
15280 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15281 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15282 (reg_num_sign_bit_copies_for_combine): Likewise.
15283 * rtlanal.c (nonzero_bits1): Likewise.
15284 (num_sign_bit_copies1): Likewise.
15285 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15286 (reg_num_sign_bit_copies_general): Likewise.
15287 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15288 (reg_nonzero_bits_general): 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 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15295 than in subroutines. Return 1 for non-integer modes.
15296 (cached_num_sign_bit_copies): Change the type of the mode parameter
15297 to scalar_int_mode.
15298 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15299 classes. Handle CONST_INT_P first and then check whether X also
15300 has a scalar integer mode. Check the same thing for inner registers
15301 of a SUBREG and for values that are being extended or truncated.
15303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15304 Alan Hayward <alan.hayward@arm.com>
15305 David Sherwood <david.sherwood@arm.com>
15307 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15308 in subroutines. Return the mode mask for non-integer modes.
15309 (cached_nonzero_bits): Change the type of the mode parameter
15310 to scalar_int_mode.
15311 (nonzero_bits1): Likewise. Remove early exit for other mode
15312 classes. Handle CONST_INT_P first and then check whether X
15313 also has a scalar integer mode.
15315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15316 Alan Hayward <alan.hayward@arm.com>
15317 David Sherwood <david.sherwood@arm.com>
15319 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15320 Return a scalar_int_mode and assert that the size is greater than
15322 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15323 (op_by_pieces_d::op_by_pieces_d): Likewise.
15324 (op_by_pieces_d::run): Likewise.
15325 (can_store_by_pieces): Likewise.
15327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15328 Alan Hayward <alan.hayward@arm.com>
15329 David Sherwood <david.sherwood@arm.com>
15331 * combine.c (extract_left_shift): Add a mode argument and update
15333 (make_compound_operation_int): Change the type of the mode parameter
15334 to scalar_int_mode and update the call to extract_left_shift.
15336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15337 Alan Hayward <alan.hayward@arm.com>
15338 David Sherwood <david.sherwood@arm.com>
15340 * combine.c (simplify_and_const_int): Change the type of the mode
15341 parameter to scalar_int_mode.
15342 (simplify_and_const_int_1): Likewise. Update recursive call.
15344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15345 Alan Hayward <alan.hayward@arm.com>
15346 David Sherwood <david.sherwood@arm.com>
15348 * combine.c (simplify_compare_const): Check that the mode is a
15349 scalar_int_mode (rather than VOIDmode) before testing its
15351 (simplify_comparison): Move COMPARISON_P handling out of the
15352 loop and restrict the latter part of the loop to scalar_int_modes.
15353 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15354 and when considering SUBREG_REGs. Use is_int_mode instead of
15355 checking GET_MODE_CLASS against MODE_INT.
15357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15358 Alan Hayward <alan.hayward@arm.com>
15359 David Sherwood <david.sherwood@arm.com>
15361 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15362 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15363 shift_unit_mode and for modes involved in scalar shifts.
15365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15366 Alan Hayward <alan.hayward@arm.com>
15367 David Sherwood <david.sherwood@arm.com>
15369 * combine.c (force_int_to_mode): New function, split out from...
15370 (force_to_mode): ...here. Keep xmode up-to-date and use it
15371 instead of GET_MODE (x).
15373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15374 Alan Hayward <alan.hayward@arm.com>
15375 David Sherwood <david.sherwood@arm.com>
15377 * optabs-query.h (extraction_insn::struct_mode): Change type to
15378 opt_scalar_int_mode and update comment.
15379 (extraction_insn::field_mode): Change type to scalar_int_mode.
15380 (extraction_insn::pos_mode): Likewise.
15381 * combine.c (make_extraction): Update accordingly.
15382 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15383 (get_optab_extraction_insn): Likewise.
15384 * recog.c (simplify_while_replacing): Likewise.
15385 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15386 parameter to opt_scalar_int_mode.
15388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15389 Alan Hayward <alan.hayward@arm.com>
15390 David Sherwood <david.sherwood@arm.com>
15392 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15393 to a scalar_int_mode instead of a machine_mode.
15394 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15395 (get_best_mode): Return a boolean and use a pointer argument to store
15396 the selected mode. Replace the limit mode parameter with a bit limit.
15397 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15398 for the values returned by bit_field_mode_iterator::next_mode.
15399 (store_bit_field): Update call to get_best_mode.
15400 (store_fixed_bit_field): Likewise.
15401 (extract_fixed_bit_field): Likewise.
15402 * expr.c (optimize_bitfield_assignment_op): Likewise.
15403 * fold-const.c (optimize_bit_field_compare): Likewise.
15404 (fold_truth_andor_1): Likewise.
15405 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15406 Update for new type of m_mode.
15407 (get_best_mode): As above.
15409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15410 Alan Hayward <alan.hayward@arm.com>
15411 David Sherwood <david.sherwood@arm.com>
15413 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15414 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15415 (store_bit_field): Check is_a <scalar_int_mode> before calling
15416 strict_volatile_bitfield_p.
15417 (extract_bit_field): Likewise.
15419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15420 Alan Hayward <alan.hayward@arm.com>
15421 David Sherwood <david.sherwood@arm.com>
15423 * target.def (cstore_mode): Return a scalar_int_mode.
15424 * doc/tm.texi: Regenerate.
15425 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15426 * targhooks.h (default_cstore_mode): Likewise.
15427 * targhooks.c (default_cstore_mode): Likewise, using a forced
15429 * expmed.c (emit_cstore): Expect the target of the cstore to be
15432 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15433 Alan Hayward <alan.hayward@arm.com>
15434 David Sherwood <david.sherwood@arm.com>
15436 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15438 (niter_desc): Likewise mode.
15439 (iv_analyze): Add a mode parameter.
15441 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15442 and change its type to scalar_int_mode.
15443 * loop-iv.c: Update commentary at head of file.
15444 (iv_constant): Pass the mode paraeter before the rtx it describes
15445 and change its type to scalar_int_mode. Remove VOIDmode handling.
15446 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15447 (iv_extend): Likewise.
15448 (shorten_into_mode): Likewise.
15449 (iv_add): Use scalar_int_mode.
15450 (iv_mult): Likewise.
15451 (iv_shift): Likewise.
15452 (canonicalize_iv_subregs): Likewise.
15453 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15454 it describes and change its mode to scalar_int_mode. Also change
15455 the type of the returned inner_mode to scalar_int_mode.
15456 (get_biv_step): Likewise, turning outer_mode from a pointer
15457 into a direct parameter. Update call to get_biv_step_1.
15458 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15459 iv_constant and get_biv_step.
15460 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15461 and change its type to scalar_int_mode. Don't initialise iv->mode
15462 to VOIDmode and remove later checks for its still being VOIDmode.
15463 Update calls to iv_analyze_op and iv_analyze_expr. Check
15464 is_a <scalar_int_mode> when changing the mode under consideration.
15465 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15466 Update call to iv_analyze_expr.
15467 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15468 inner register is not also a scalar_int_mode. Update call to
15470 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15471 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15472 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15473 separate mode class checks. Update calls to iv_analyze. Remove
15474 fix-up of VOIDmodes after iv_analyze_biv.
15475 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15476 don't have a scalar_int_mode. Update call to biv_p.
15478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15479 Alan Hayward <alan.hayward@arm.com>
15480 David Sherwood <david.sherwood@arm.com>
15482 * cfgexpand.c (convert_debug_memory_address): Use
15483 as_a <scalar_int_mode>.
15484 * combine.c (expand_compound_operation): Likewise.
15485 (make_extraction): Likewise.
15486 (change_zero_ext): Likewise.
15487 (simplify_comparison): Likewise.
15488 * cse.c (cse_insn): Likewise.
15489 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15490 (mem_loc_descriptor): Likewise.
15491 (loc_descriptor): Likewise.
15492 * expmed.c (init_expmed_one_mode): Likewise.
15493 (synth_mult): Likewise.
15494 (emit_store_flag_1): Likewise.
15495 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15496 of a comparison with size.
15497 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15498 (reduce_to_bit_field_precision): Likewise.
15499 * function.c (expand_function_end): Likewise.
15500 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15501 * loop-doloop.c (doloop_modify): Likewise.
15502 * optabs.c (expand_binop): Likewise.
15503 (expand_unop): Likewise.
15504 (expand_copysign_absneg): Likewise.
15505 (prepare_cmp_insn): Likewise.
15506 (maybe_legitimize_operand): Likewise.
15507 * recog.c (const_scalar_int_operand): Likewise.
15508 * rtlanal.c (get_address_mode): Likewise.
15509 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15510 (simplify_cond_clz_ctz): Likewise.
15511 * tree-nested.c (get_nl_goto_field): Likewise.
15512 * tree.c (build_vector_type_for_mode): Likewise.
15513 * var-tracking.c (use_narrower_mode): Likewise.
15515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15516 Alan Hayward <alan.hayward@arm.com>
15517 David Sherwood <david.sherwood@arm.com>
15519 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15520 * builtins.c (expand_builtin_signbit): Use it.
15521 * cfgexpand.c (expand_debug_expr): Likewise.
15522 * dojump.c (do_jump): Likewise.
15523 (do_compare_and_jump): Likewise.
15524 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15525 * expmed.c (make_tree): Likewise.
15526 * expr.c (expand_expr_real_2): Likewise.
15527 (expand_expr_real_1): Likewise.
15528 (try_casesi): Likewise.
15529 * fold-const-call.c (fold_const_call_ss): Likewise.
15530 * fold-const.c (unextend): Likewise.
15531 (extract_muldiv_1): Likewise.
15532 (fold_single_bit_test): Likewise.
15533 (native_encode_int): Likewise.
15534 (native_encode_string): Likewise.
15535 (native_interpret_int): Likewise.
15536 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15537 * internal-fn.c (expand_addsub_overflow): Likewise.
15538 (expand_neg_overflow): Likewise.
15539 (expand_mul_overflow): Likewise.
15540 (expand_arith_overflow): Likewise.
15541 * match.pd: Likewise.
15542 * stor-layout.c (layout_type): Likewise.
15543 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15544 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15545 * tree-ssanames.c (get_range_info): Likewise.
15546 * tree-switch-conversion.c (array_value_type) Likewise.
15547 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15548 (vect_recog_divmod_pattern): Likewise.
15549 (vect_recog_mixed_size_cond_pattern): Likewise.
15550 * tree-vrp.c (extract_range_basic): Likewise.
15551 (simplify_float_conversion_using_ranges): Likewise.
15552 * tree.c (int_fits_type_p): Likewise.
15553 * ubsan.c (instrument_bool_enum_load): Likewise.
15554 * varasm.c (mergeable_string_section): Likewise.
15555 (narrowing_initializer_constant_valid_p): Likewise.
15556 (output_constant): Likewise.
15558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15559 Alan Hayward <alan.hayward@arm.com>
15560 David Sherwood <david.sherwood@arm.com>
15562 * machmode.h (NARROWEST_INT_MODE): New macro.
15563 * expr.c (alignment_for_piecewise_move): Use it instead of
15564 GET_CLASS_NARROWEST_MODE (MODE_INT).
15565 (push_block): Likewise.
15566 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15568 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15570 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15571 Alan Hayward <alan.hayward@arm.com>
15572 David Sherwood <david.sherwood@arm.com>
15574 * postreload.c (move2add_valid_value_p): Change the type of the
15575 mode parameter to scalar_int_mode.
15576 (move2add_use_add2_insn): Add a mode parameter and use it instead
15578 (move2add_use_add3_insn): Likewise.
15579 (reload_cse_move2add): Update accordingly.
15581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15582 Alan Hayward <alan.hayward@arm.com>
15583 David Sherwood <david.sherwood@arm.com>
15585 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15587 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15588 * optabs.c (expand_unop): Likewise.
15590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15591 Alan Hayward <alan.hayward@arm.com>
15592 David Sherwood <david.sherwood@arm.com>
15594 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15595 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15596 (popcount_loc_descriptor): Likewise.
15597 (bswap_loc_descriptor): Likewise.
15598 (rotate_loc_descriptor): Likewise.
15599 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15600 calling the functions above.
15602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15603 Alan Hayward <alan.hayward@arm.com>
15604 David Sherwood <david.sherwood@arm.com>
15606 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15608 (try_combine): Likewise.
15609 (simplify_if_then_else): Likewise.
15610 * cse.c (cse_insn): Likewise.
15611 * dwarf2out.c (mem_loc_descriptor): Likewise.
15612 * emit-rtl.c (gen_lowpart_common): Likewise.
15613 * simplify-rtx.c (simplify_truncation): Likewise.
15614 (simplify_binary_operation_1): Likewise.
15615 (simplify_const_relational_operation): Likewise.
15616 (simplify_ternary_operation): Likewise.
15617 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15620 Alan Hayward <alan.hayward@arm.com>
15621 David Sherwood <david.sherwood@arm.com>
15623 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15624 * reload.c (push_reload): Likewise.
15625 (find_reloads): Likewise.
15627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15628 Alan Hayward <alan.hayward@arm.com>
15629 David Sherwood <david.sherwood@arm.com>
15631 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15632 (make_compound_operation_int): Likewise.
15633 (change_zero_ext): Likewise.
15634 * expr.c (convert_move): Likewise.
15635 (convert_modes): Likewise.
15636 * fwprop.c (forward_propagate_subreg): Likewise.
15637 * loop-iv.c (get_biv_step_1): Likewise.
15638 * optabs.c (widen_operand): Likewise.
15639 * postreload.c (move2add_valid_value_p): Likewise.
15640 * recog.c (simplify_while_replacing): Likewise.
15641 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15642 (simplify_binary_operation_1): Likewise. Remove redundant
15643 mode equality check.
15645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15646 Alan Hayward <alan.hayward@arm.com>
15647 David Sherwood <david.sherwood@arm.com>
15649 * combine.c (combine_simplify_rtx): Add checks for
15650 is_a <scalar_int_mode>.
15651 (simplify_if_then_else): Likewise.
15652 (make_field_assignment): Likewise.
15653 (simplify_comparison): Likewise.
15654 * ifcvt.c (noce_try_bitop): Likewise.
15655 * loop-invariant.c (canonicalize_address_mult): Likewise.
15656 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15659 Alan Hayward <alan.hayward@arm.com>
15660 David Sherwood <david.sherwood@arm.com>
15662 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15663 is_a <scalar_int_mode> instead of != BLKmode.
15665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15666 Alan Hayward <alan.hayward@arm.com>
15667 David Sherwood <david.sherwood@arm.com>
15669 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15670 instead of != VOIDmode.
15671 * combine.c (if_then_else_cond): Likewise.
15672 (change_zero_ext): Likewise.
15673 * dwarf2out.c (mem_loc_descriptor): Likewise.
15674 (loc_descriptor): Likewise.
15675 * rtlanal.c (canonicalize_condition): Likewise.
15676 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15679 Alan Hayward <alan.hayward@arm.com>
15680 David Sherwood <david.sherwood@arm.com>
15682 * simplify-rtx.c (simplify_binary_operation_1): Use
15683 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
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 * wide-int.h (int_traits<unsigned char>) New class.
15690 (int_traits<unsigned short>) Likewise.
15691 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15692 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15694 * combine.c (set_nonzero_bits_and_sign_copies): Use
15695 is_a <scalar_int_mode>.
15696 (find_split_point): Likewise.
15697 (combine_simplify_rtx): Likewise.
15698 (simplify_logical): Likewise.
15699 (expand_compound_operation): Likewise.
15700 (expand_field_assignment): Likewise.
15701 (make_compound_operation): Likewise.
15702 (extended_count): Likewise.
15703 (change_zero_ext): Likewise.
15704 (simplify_comparison): Likewise.
15705 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15706 (ucompare_loc_descriptor): Likewise.
15707 (minmax_loc_descriptor): Likewise.
15708 (mem_loc_descriptor): Likewise.
15709 (loc_descriptor): Likewise.
15710 * expmed.c (init_expmed_one_mode): Likewise.
15711 * lra-constraints.c (lra_constraint_offset): Likewise.
15712 * optabs.c (prepare_libcall_arg): Likewise.
15713 * postreload.c (move2add_note_store): Likewise.
15714 * reload.c (operands_match_p): Likewise.
15715 * rtl.h (load_extend_op): Likewise.
15716 * rtlhooks.c (gen_lowpart_general): Likewise.
15717 * simplify-rtx.c (simplify_truncation): Likewise.
15718 (simplify_unary_operation_1): Likewise.
15719 (simplify_binary_operation_1): Likewise.
15720 (simplify_const_binary_operation): Likewise.
15721 (simplify_const_relational_operation): Likewise.
15722 (simplify_subreg): Likewise.
15723 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15724 * var-tracking.c (adjust_mems): Likewise.
15725 (prepare_call_arguments): Likewise.
15727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15728 Alan Hayward <alan.hayward@arm.com>
15729 David Sherwood <david.sherwood@arm.com>
15731 * machmode.h (is_int_mode): New fuction.
15732 * combine.c (find_split_point): Use it.
15733 (combine_simplify_rtx): Likewise.
15734 (simplify_if_then_else): Likewise.
15735 (simplify_set): Likewise.
15736 (simplify_shift_const_1): Likewise.
15737 (simplify_comparison): Likewise.
15738 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15739 * cse.c (notreg_cost): Likewise.
15740 (cse_insn): Likewise.
15741 * cselib.c (cselib_lookup_1): Likewise.
15742 * dojump.c (do_jump_1): Likewise.
15743 (do_compare_rtx_and_jump): Likewise.
15744 * dse.c (get_call_args): Likewise.
15745 * dwarf2out.c (rtl_for_decl_init): Likewise.
15746 (native_encode_initializer): Likewise.
15747 * expmed.c (emit_store_flag_1): Likewise.
15748 (emit_store_flag): Likewise.
15749 * expr.c (convert_modes): Likewise.
15750 (store_field): Likewise.
15751 (expand_expr_real_1): Likewise.
15752 * fold-const.c (fold_read_from_constant_string): Likewise.
15753 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15754 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15755 * optabs.c (expand_binop): Likewise.
15756 (expand_unop): Likewise.
15757 (expand_abs_nojump): Likewise.
15758 (expand_one_cmpl_abs_nojump): Likewise.
15759 * simplify-rtx.c (mode_signbit_p): Likewise.
15760 (val_signbit_p): Likewise.
15761 (val_signbit_known_set_p): Likewise.
15762 (val_signbit_known_clear_p): Likewise.
15763 (simplify_relational_operation_1): Likewise.
15764 * tree.c (vector_type_mode): Likewise.
15766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15767 Alan Hayward <alan.hayward@arm.com>
15768 David Sherwood <david.sherwood@arm.com>
15770 * machmode.h (smallest_mode_for_size): Fix formatting.
15771 (smallest_int_mode_for_size): New function.
15772 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15773 instead of smallest_mode_for_size.
15774 * combine.c (make_extraction): Likewise.
15775 * config/arc/arc.c (arc_expand_movmem): Likewise.
15776 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15777 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15778 * config/s390/s390.c (s390_expand_insv): Likewise.
15779 * config/sparc/sparc.c (assign_int_registers): Likewise.
15780 * config/spu/spu.c (spu_function_value): Likewise.
15781 (spu_function_arg): Likewise.
15782 * coverage.c (get_gcov_type): Likewise.
15783 (get_gcov_unsigned_t): Likewise.
15784 * dse.c (find_shift_sequence): Likewise.
15785 * expmed.c (store_bit_field_1): Likewise.
15786 * expr.c (convert_move): Likewise.
15787 (store_field): Likewise.
15788 * internal-fn.c (expand_arith_overflow): Likewise.
15789 * optabs-query.c (get_best_extraction_insn): Likewise.
15790 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15791 * stor-layout.c (layout_type): Likewise.
15792 (initialize_sizetypes): Likewise.
15793 * targhooks.c (default_get_mask_mode): Likewise.
15794 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15797 Alan Hayward <alan.hayward@arm.com>
15798 David Sherwood <david.sherwood@arm.com>
15800 * machmode.h (opt_mode::else_blk): New function.
15801 (int_mode_for_mode): Declare.
15802 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15803 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15805 * cfgexpand.c (expand_debug_expr): Likewise.
15806 * combine.c (gen_lowpart_or_truncate): Likewise.
15807 (gen_lowpart_for_combine): Likewise.
15808 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15809 * config/avr/avr.c (avr_to_int_mode): Likewise.
15810 (avr_out_plus_1): Likewise.
15811 (avr_out_plus): Likewise.
15812 (avr_out_round): Likewise.
15813 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15814 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15815 (s390_expand_vcond): Likewise.
15816 * config/spu/spu.c (spu_split_immediate): Likewise.
15817 (spu_expand_mov): Likewise.
15818 * dse.c (get_stored_val): Likewise.
15819 * expmed.c (store_bit_field_1): Likewise.
15820 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15822 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15823 (extract_low_bits): Likewise.
15824 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15825 handling rather than repeating the check.
15826 (emit_group_store): Likewise.
15827 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15828 * optabs.c (expand_absneg_bit): Likewise.
15829 (expand_copysign_absneg): Likewise.
15830 (expand_copysign_bit): Likewise.
15831 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15832 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15833 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15834 * var-tracking.c (prepare_call_arguments): Likewise.
15835 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15836 int_mode_for_mode instead of mode_for_size.
15837 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15840 Alan Hayward <alan.hayward@arm.com>
15841 David Sherwood <david.sherwood@arm.com>
15843 * machmode.h (int_mode_for_size): New function.
15844 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15845 instead of mode_for_size.
15846 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15848 * combine.c (expand_field_assignment): Use int_mode_for_size
15849 instead of mode_for_size.
15850 (make_extraction): Likewise.
15851 (simplify_shift_const_1): Likewise.
15852 (simplify_comparison): Likewise.
15853 * dojump.c (do_jump): Likewise.
15854 * dwarf2out.c (mem_loc_descriptor): Likewise.
15855 * emit-rtl.c (init_derived_machine_modes): Likewise.
15856 * expmed.c (flip_storage_order): Likewise.
15857 (convert_extracted_bit_field): Likewise.
15858 * expr.c (copy_blkmode_from_reg): Likewise.
15859 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15860 * internal-fn.c (expand_mul_overflow): Likewise.
15861 * lower-subreg.c (simple_move): Likewise.
15862 * optabs-libfuncs.c (init_optabs): Likewise.
15863 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15864 * tree.c (vector_type_mode): Likewise.
15865 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15866 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15867 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15868 * tree-vect-stmts.c (vectorizable_load): Likewise.
15869 (vectorizable_store): Likewise.
15871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15872 Alan Hayward <alan.hayward@arm.com>
15873 David Sherwood <david.sherwood@arm.com>
15875 * coretypes.h (pod_mode): New type.
15876 (scalar_int_mode_pod): New typedef.
15877 * machmode.h (pod_mode): New class.
15878 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15879 * genmodes.c (emit_mode_int_n): Update accordingly.
15880 * lower-subreg.h (target_lower_subreg): Change type to
15881 scalar_int_mode_pod.
15882 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15883 scalar_int_mode_pod.
15885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15886 Alan Hayward <alan.hayward@arm.com>
15887 David Sherwood <david.sherwood@arm.com>
15889 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15890 machine_mode to scalar_int_mode.
15891 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15892 (rs6000_option_override_internal): Remove cast to int.
15893 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15894 machine_mode to scalar_int_mode.
15895 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15896 (rs6000_option_override_internal): Remove cast to int.
15897 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15898 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15900 * config/s390/s390.c (s390_expand_builtin): Likewise.
15901 * coretypes.h (scalar_int_mode): New type.
15902 (opt_scalar_int_mode): New typedef.
15903 * machmode.h (scalar_int_mode): New class.
15904 (scalar_int_mode::includes_p): New function.
15905 (byte_mode): Change type to scalar_int_mode.
15906 (word_mode): Likewise.
15907 (ptr_mode): Likewise.
15908 * emit-rtl.c (byte_mode): Likewise.
15909 (word_mode): Likewise.
15910 (ptr_mode): Likewise.
15911 (init_derived_machine_modes): Update accordingly.
15912 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15913 and MODE_PARTIAL_INT.
15914 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15915 opt_scalar_int_mode.
15917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15918 Alan Hayward <alan.hayward@arm.com>
15919 David Sherwood <david.sherwood@arm.com>
15921 * target.def (libgcc_floating_mode_supported_p): Take a
15923 * doc/tm.texi: Regenerate.
15924 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
15926 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15927 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
15930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15931 Alan Hayward <alan.hayward@arm.com>
15932 David Sherwood <david.sherwood@arm.com>
15934 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
15935 * doc/tm.texi: Regenerate.
15936 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
15937 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
15938 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
15939 * targhooks.h (default_floatn_mode): Likewise.
15940 * targhooks.c (default_floatn_mode): Likewise.
15941 * tree.c (build_common_tree_nodes): Update accordingly.
15943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15944 Alan Hayward <alan.hayward@arm.com>
15945 David Sherwood <david.sherwood@arm.com>
15947 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
15948 (mode_iterator::iterate_p): Likewise.
15949 (mode_iterator::get_wider): Likewise.
15950 * expr.c (init_expr_target): Use opt_scalar_float_mode.
15952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15953 Alan Hayward <alan.hayward@arm.com>
15954 David Sherwood <david.sherwood@arm.com>
15956 * coretypes.h (opt_scalar_float_mode): New typedef.
15957 * machmode.h (float_mode_for_size): New function.
15958 * emit-rtl.c (double_mode): Delete.
15959 (init_emit_once): Use float_mode_for_size.
15960 * stor-layout.c (layout_type): Likewise.
15961 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
15963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15964 Alan Hayward <alan.hayward@arm.com>
15965 David Sherwood <david.sherwood@arm.com>
15967 * output.h (assemble_real): Take a scalar_float_mode.
15968 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
15969 * config/arm/arm.md (consttable_4): Likewise.
15970 (consttable_8): Likewise.
15971 (consttable_16): Likewise.
15972 * config/mips/mips.md (consttable_float): Likewise.
15973 * config/s390/s390.c (s390_output_pool_entry): Likewise.
15974 * varasm.c (assemble_real): Take a scalar_float_mode.
15975 (output_constant_pool_2): Update accordingly.
15976 (output_constant): Likewise.
15978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15979 Alan Hayward <alan.hayward@arm.com>
15980 David Sherwood <david.sherwood@arm.com>
15982 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
15983 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
15984 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
15985 (native_encode_real): Likewise.
15986 (native_interpret_real): Likewise.
15987 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
15988 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
15995 <scalar_float_mode>. Simplify.
15996 (gen_extend_conv_libfunc): Likewise.
15998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15999 Alan Hayward <alan.hayward@arm.com>
16000 David Sherwood <david.sherwood@arm.com>
16002 * coretypes.h (scalar_float_mode): New type.
16003 * machmode.h (mode_traits::from_int): Use machine_mode if
16004 USE_ENUM_MODES is defined.
16005 (is_a): New function.
16007 (dyn_cast): Likewise.
16008 (scalar_float_mode): New class.
16009 (scalar_float_mode::includes_p): New function.
16010 (is_float_mode): Likewise.
16011 * gdbhooks.py (MachineModePrinter): New class.
16012 (build_pretty_printer): Use it for scalar_float_mode.
16013 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16014 (format_helper::format_helper): Turn into a template.
16015 * genmodes.c (get_mode_class): New function.
16016 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16017 or machine_mode if none.
16018 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16019 as_a <scalar_float_mode>.
16020 * dwarf2out.c (mem_loc_descriptor): Likewise.
16021 (insert_float): Likewise.
16022 (add_const_value_attribute): Likewise.
16023 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16024 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16025 (expand_unop): Update accordingly.
16026 (expand_abs_nojump): Likewise.
16027 (expand_copysign_absneg): Take a scalar_float_mode.
16028 (expand_copysign_bit): Likewise.
16029 (expand_copysign): Update accordingly.
16031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16032 Alan Hayward <alan.hayward@arm.com>
16033 David Sherwood <david.sherwood@arm.com>
16035 * coretypes.h (opt_mode): New class.
16036 * machmode.h (opt_mode): Likewise.
16037 (opt_mode::else_void): New function.
16038 (opt_mode::require): Likewise.
16039 (opt_mode::exists): Likewise.
16040 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16041 (GET_MODE_2XWIDER_MODE): Likewise.
16042 (mode_iterator::get_wider): Update accordingly.
16043 (mode_iterator::get_2xwider): Likewise.
16044 (mode_iterator::get_known_wider): Likewise, turning into a template.
16045 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16046 forcing a wider mode to exist.
16047 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16048 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16049 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16050 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16051 * lower-subreg.c (init_lower_subreg): Likewise.
16052 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16053 on the final iteration.
16054 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16055 a wider mode exists before asking for a move pattern.
16056 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16057 forcing a wider mode to exist.
16058 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16059 returning false if no such mode exists.
16060 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16061 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16062 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16063 Avoid checking for a MODE_INT if we already know the mode is not a
16065 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16066 forcing a wider mode to exist.
16067 (expmed_mult_highpart_optab): Likewise.
16068 (expmed_mult_highpart): Likewise.
16069 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16071 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16072 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16073 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16074 * internal-fn.c (expand_mul_overflow): Update use of
16075 GET_MODE_2XWIDER_MODE.
16076 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16077 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16078 GET_MODE_WIDER_MODE.
16079 (convert_plusminus_to_widen): Likewise.
16080 * tree-switch-conversion.c (array_value_type): Likewise.
16081 * var-tracking.c (emit_note_insn_var_location): Likewise.
16082 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16083 Return false inside rather than outside the loop if no wider mode
16085 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16086 and GET_MODE_2XWIDER_MODE
16087 (can_compare_p): Use else_void.
16088 * gdbhooks.py (OptMachineModePrinter): New class.
16089 (build_pretty_printer): Use it for opt_mode.
16091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16092 Alan Hayward <alan.hayward@arm.com>
16093 David Sherwood <david.sherwood@arm.com>
16095 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16096 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16099 Alan Hayward <alan.hayward@arm.com>
16100 David Sherwood <david.sherwood@arm.com>
16102 * machmode.h (mode_traits): New structure.
16103 (get_narrowest_mode): New function.
16104 (mode_iterator::start): Likewise.
16105 (mode_iterator::iterate_p): Likewise.
16106 (mode_iterator::get_wider): Likewise.
16107 (mode_iterator::get_known_wider): Likewise.
16108 (mode_iterator::get_2xwider): Likewise.
16109 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16110 (FOR_EACH_MODE): Likewise.
16111 (FOR_EACH_MODE_FROM): Likewise.
16112 (FOR_EACH_MODE_UNTIL): Likewise.
16113 (FOR_EACH_WIDER_MODE): Likewise.
16114 (FOR_EACH_2XWIDER_MODE): Likewise.
16115 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16116 * combine.c (simplify_comparison): Likewise
16117 * config/i386/i386.c (type_natural_mode): Likewise.
16118 * cse.c (cse_insn): Likewise.
16119 * dse.c (find_shift_sequence): Likewise.
16120 * emit-rtl.c (init_derived_machine_modes): Likewise.
16121 (init_emit_once): Likewise.
16122 * explow.c (hard_function_value): Likewise.
16123 * expmed.c (extract_fixed_bit_field_1): Likewise.
16124 (extract_bit_field_1): Likewise.
16125 (expand_divmod): Likewise.
16126 (emit_store_flag_1): Likewise.
16127 * expr.c (init_expr_target): Likewise.
16128 (convert_move): Likewise.
16129 (alignment_for_piecewise_move): Likewise.
16130 (widest_int_mode_for_size): Likewise.
16131 (emit_block_move_via_movmem): Likewise.
16132 (copy_blkmode_to_reg): Likewise.
16133 (set_storage_via_setmem): Likewise.
16134 (compress_float_constant): Likewise.
16135 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16136 * optabs-query.c (get_best_extraction_insn): Likewise.
16137 * optabs.c (expand_binop): Likewise.
16138 (expand_twoval_unop): Likewise.
16139 (expand_twoval_binop): Likewise.
16140 (widen_leading): Likewise.
16141 (widen_bswap): Likewise.
16142 (expand_parity): Likewise.
16143 (expand_unop): Likewise.
16144 (prepare_cmp_insn): Likewise.
16145 (prepare_float_lib_cmp): Likewise.
16146 (expand_float): Likewise.
16147 (expand_fix): Likewise.
16148 (expand_sfix_optab): Likewise.
16149 * postreload.c (move2add_use_add2_insn): Likewise.
16150 * reg-stack.c (reg_to_stack): Likewise.
16151 * reginfo.c (choose_hard_reg_mode): Likewise.
16152 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16153 * stor-layout.c (mode_for_size): Likewise.
16154 (smallest_mode_for_size): Likewise.
16155 (mode_for_vector): Likewise.
16156 (finish_bitfield_representative): Likewise.
16157 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16158 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16159 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16160 * var-tracking.c (prepare_call_arguments): Likewise.
16162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16163 Alan Hayward <alan.hayward@arm.com>
16164 David Sherwood <david.sherwood@arm.com>
16166 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16167 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16168 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16169 * machmode.h (mode_size): Move earlier in file.
16170 (mode_precision): Likewise.
16171 (mode_inner): Likewise.
16172 (mode_nunits): Likewise.
16173 (mode_unit_size): Likewise.
16174 (unit_unit_precision): Likewise.
16175 (mode_wider): Likewise.
16176 (mode_2xwider): Likewise.
16177 (machine_mode): New class.
16178 (mode_to_bytes): New function.
16179 (mode_to_bits): Likewise.
16180 (mode_to_precision): Likewise.
16181 (mode_to_inner): Likewise.
16182 (mode_to_unit_size): Likewise.
16183 (mode_to_unit_precision): Likewise.
16184 (mode_to_nunits): Likewise.
16185 (GET_MODE_SIZE): Use mode_to_bytes.
16186 (GET_MODE_BITSIZE): Use mode_to_bits.
16187 (GET_MODE_PRECISION): Use mode_to_precision.
16188 (GET_MODE_INNER): Use mode_to_inner.
16189 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16190 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16191 (GET_MODE_NUNITS): Use mode_to_nunits.
16192 * system.h (ALWAYS_INLINE): New macro.
16193 * config/powerpcspe/powerpcspe-c.c
16194 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16195 int for arg1_mode and arg2_mode.
16197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16198 Alan Hayward <alan.hayward@arm.com>
16199 David Sherwood <david.sherwood@arm.com>
16201 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16202 Prefix mode names with E_ in case statements.
16203 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16204 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16205 (aarch64_split_simd_move): Likewise.
16206 (aarch64_gen_storewb_pair): Likewise.
16207 (aarch64_gen_loadwb_pair): Likewise.
16208 (aarch64_gen_store_pair): Likewise.
16209 (aarch64_gen_load_pair): Likewise.
16210 (aarch64_get_condition_code_1): Likewise.
16211 (aarch64_constant_pool_reload_icode): Likewise.
16212 (get_rsqrte_type): Likewise.
16213 (get_rsqrts_type): Likewise.
16214 (get_recpe_type): Likewise.
16215 (get_recps_type): Likewise.
16216 (aarch64_gimplify_va_arg_expr): Likewise.
16217 (aarch64_simd_container_mode): Likewise.
16218 (aarch64_emit_load_exclusive): Likewise.
16219 (aarch64_emit_store_exclusive): Likewise.
16220 (aarch64_expand_compare_and_swap): Likewise.
16221 (aarch64_gen_atomic_cas): Likewise.
16222 (aarch64_emit_bic): Likewise.
16223 (aarch64_emit_atomic_swap): Likewise.
16224 (aarch64_emit_atomic_load_op): Likewise.
16225 (aarch64_evpc_trn): Likewise.
16226 (aarch64_evpc_uzp): Likewise.
16227 (aarch64_evpc_zip): Likewise.
16228 (aarch64_evpc_ext): Likewise.
16229 (aarch64_evpc_rev): Likewise.
16230 (aarch64_evpc_dup): Likewise.
16231 (aarch64_gen_ccmp_first): Likewise.
16232 (aarch64_gen_ccmp_next): Likewise.
16233 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16234 (alpha_emit_xfloating_libcall): Likewise.
16235 (emit_insxl): Likewise.
16236 (alpha_arg_type): Likewise.
16237 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16238 (arc_preferred_simd_mode): Likewise.
16239 (arc_secondary_reload): Likewise.
16240 (get_arc_condition_code): Likewise.
16241 (arc_print_operand): Likewise.
16242 (arc_legitimate_constant_p): Likewise.
16243 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16244 * config/arc/arc.md (casesi_load): Likewise.
16245 (casesi_compact_jump): Likewise.
16246 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16247 (cc_use_register): Likewise.
16248 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16249 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16250 (arm_init_iwmmxt_builtins): Likewise.
16251 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16252 (neon_expand_vector_init): Likewise.
16253 (arm_attr_length_move_neon): Likewise.
16254 (maybe_get_arm_condition_code): Likewise.
16255 (arm_emit_vector_const): Likewise.
16256 (arm_preferred_simd_mode): Likewise.
16257 (arm_output_iwmmxt_tinsr): Likewise.
16258 (thumb1_output_casesi): Likewise.
16259 (thumb2_output_casesi): Likewise.
16260 (arm_emit_load_exclusive): Likewise.
16261 (arm_emit_store_exclusive): Likewise.
16262 (arm_expand_compare_and_swap): Likewise.
16263 (arm_evpc_neon_vuzp): Likewise.
16264 (arm_evpc_neon_vzip): Likewise.
16265 (arm_evpc_neon_vrev): Likewise.
16266 (arm_evpc_neon_vtrn): Likewise.
16267 (arm_evpc_neon_vext): Likewise.
16268 (arm_validize_comparison): Likewise.
16269 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16270 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16271 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16272 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16273 (c6x_preferred_simd_mode): Likewise.
16274 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16275 (epiphany_rtx_costs): Likewise.
16276 * config/epiphany/predicates.md (proper_comparison_operator):
16278 * config/frv/frv.c (condexec_memory_operand): Likewise.
16279 (frv_emit_move): Likewise.
16280 (output_move_single): Likewise.
16281 (output_condmove_single): Likewise.
16282 (frv_hard_regno_mode_ok): Likewise.
16283 (frv_matching_accg_mode): Likewise.
16284 * config/h8300/h8300.c (split_adds_subs): Likewise.
16285 (h8300_rtx_costs): Likewise.
16286 (h8300_print_operand): Likewise.
16287 (compute_mov_length): Likewise.
16288 (output_logical_op): Likewise.
16289 (compute_logical_op_length): Likewise.
16290 (compute_logical_op_cc): Likewise.
16291 (h8300_shift_needs_scratch_p): Likewise.
16292 (output_a_shift): Likewise.
16293 (compute_a_shift_length): Likewise.
16294 (compute_a_shift_cc): Likewise.
16295 (expand_a_rotate): Likewise.
16296 (output_a_rotate): Likewise.
16297 * config/i386/i386.c (classify_argument): Likewise.
16298 (function_arg_advance_32): Likewise.
16299 (function_arg_32): Likewise.
16300 (function_arg_64): Likewise.
16301 (function_value_64): Likewise.
16302 (ix86_gimplify_va_arg): Likewise.
16303 (ix86_legitimate_constant_p): Likewise.
16304 (put_condition_code): Likewise.
16305 (split_double_mode): Likewise.
16306 (ix86_avx256_split_vector_move_misalign): Likewise.
16307 (ix86_expand_vector_logical_operator): Likewise.
16308 (ix86_split_idivmod): Likewise.
16309 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16310 (ix86_build_const_vector): Likewise.
16311 (ix86_build_signbit_mask): Likewise.
16312 (ix86_match_ccmode): Likewise.
16313 (ix86_cc_modes_compatible): Likewise.
16314 (ix86_expand_branch): Likewise.
16315 (ix86_expand_sse_cmp): Likewise.
16316 (ix86_expand_sse_movcc): Likewise.
16317 (ix86_expand_int_sse_cmp): Likewise.
16318 (ix86_expand_vec_perm_vpermi2): Likewise.
16319 (ix86_expand_vec_perm): Likewise.
16320 (ix86_expand_sse_unpack): Likewise.
16321 (ix86_expand_int_addcc): Likewise.
16322 (ix86_split_to_parts): Likewise.
16323 (ix86_vectorize_builtin_gather): Likewise.
16324 (ix86_vectorize_builtin_scatter): Likewise.
16325 (avx_vpermilp_parallel): Likewise.
16326 (inline_memory_move_cost): Likewise.
16327 (ix86_tieable_integer_mode_p): Likewise.
16328 (x86_maybe_negate_const_int): Likewise.
16329 (ix86_expand_vector_init_duplicate): Likewise.
16330 (ix86_expand_vector_init_one_nonzero): Likewise.
16331 (ix86_expand_vector_init_one_var): Likewise.
16332 (ix86_expand_vector_init_concat): Likewise.
16333 (ix86_expand_vector_init_interleave): Likewise.
16334 (ix86_expand_vector_init_general): Likewise.
16335 (ix86_expand_vector_set): Likewise.
16336 (ix86_expand_vector_extract): Likewise.
16337 (emit_reduc_half): Likewise.
16338 (ix86_emit_i387_round): Likewise.
16339 (ix86_mangle_type): Likewise.
16340 (ix86_expand_round_sse4): Likewise.
16341 (expand_vec_perm_blend): Likewise.
16342 (canonicalize_vector_int_perm): Likewise.
16343 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16344 (expand_vec_perm_1): Likewise.
16345 (expand_vec_perm_interleave3): Likewise.
16346 (expand_vec_perm_even_odd_pack): Likewise.
16347 (expand_vec_perm_even_odd_1): Likewise.
16348 (expand_vec_perm_broadcast_1): Likewise.
16349 (ix86_vectorize_vec_perm_const_ok): Likewise.
16350 (ix86_expand_vecop_qihi): Likewise.
16351 (ix86_expand_mul_widen_hilo): Likewise.
16352 (ix86_expand_sse2_abs): Likewise.
16353 (ix86_expand_pextr): Likewise.
16354 (ix86_expand_pinsr): Likewise.
16355 (ix86_preferred_simd_mode): Likewise.
16356 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16357 * config/i386/sse.md (*andnot<mode>3): Likewise.
16358 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16359 (*<code><mode>3): Likewise.
16360 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16361 (ia64_expand_atomic_op): Likewise.
16362 (ia64_arg_type): Likewise.
16363 (ia64_mode_to_int): Likewise.
16364 (ia64_scalar_mode_supported_p): Likewise.
16365 (ia64_vector_mode_supported_p): Likewise.
16366 (expand_vec_perm_broadcast): Likewise.
16367 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16368 (iq2000_function_arg_advance): Likewise.
16369 (iq2000_function_arg): Likewise.
16370 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16371 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16372 (m68k_libcall_value): Likewise.
16373 (m68k_function_value): Likewise.
16374 (sched_attr_op_type): Likewise.
16375 * config/mcore/mcore.c (mcore_output_move): Likewise.
16376 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16378 (microblaze_function_arg): Likewise.
16379 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16380 (mips_print_operand): Likewise.
16381 (mips_mode_ok_for_mov_fmt_p): Likewise.
16382 (mips_vector_mode_supported_p): Likewise.
16383 (mips_preferred_simd_mode): Likewise.
16384 (mips_expand_vpc_loongson_even_odd): Likewise.
16385 (mips_expand_vec_unpack): Likewise.
16386 (mips_expand_vi_broadcast): Likewise.
16387 (mips_expand_vector_init): Likewise.
16388 (mips_expand_vec_reduc): Likewise.
16389 (mips_expand_msa_cmp): Likewise.
16390 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16391 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16392 (cc_flags_for_mode): Likewise.
16393 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16394 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16395 (nds32_output_casesi_pc_relative): Likewise.
16396 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16397 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16398 (nvptx_gen_unpack): Likewise.
16399 (nvptx_gen_pack): Likewise.
16400 (nvptx_gen_shuffle): Likewise.
16401 (nvptx_gen_wcast): Likewise.
16402 (nvptx_preferred_simd_mode): Likewise.
16403 * config/pa/pa.c (pa_secondary_reload): Likewise.
16404 * config/pa/predicates.md (base14_operand): Likewise.
16405 * config/powerpcspe/powerpcspe-c.c
16406 (altivec_resolve_overloaded_builtin): Likewise.
16407 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16409 (rs6000_preferred_simd_mode): Likewise.
16410 (output_vec_const_move): Likewise.
16411 (rs6000_expand_vector_extract): Likewise.
16412 (rs6000_split_vec_extract_var): Likewise.
16413 (reg_offset_addressing_ok_p): Likewise.
16414 (rs6000_legitimate_offset_address_p): Likewise.
16415 (rs6000_legitimize_address): Likewise.
16416 (rs6000_emit_set_const): Likewise.
16417 (rs6000_const_vec): Likewise.
16418 (rs6000_emit_move): Likewise.
16419 (spe_build_register_parallel): Likewise.
16420 (rs6000_darwin64_record_arg_recurse): Likewise.
16421 (swap_selector_for_mode): Likewise.
16422 (spe_init_builtins): Likewise.
16423 (paired_init_builtins): Likewise.
16424 (altivec_init_builtins): Likewise.
16425 (do_load_for_compare): Likewise.
16426 (rs6000_generate_compare): Likewise.
16427 (rs6000_expand_float128_convert): Likewise.
16428 (emit_load_locked): Likewise.
16429 (emit_store_conditional): Likewise.
16430 (rs6000_output_function_epilogue): Likewise.
16431 (rs6000_handle_altivec_attribute): Likewise.
16432 (rs6000_function_value): Likewise.
16433 (emit_fusion_gpr_load): Likewise.
16434 (emit_fusion_p9_load): Likewise.
16435 (emit_fusion_p9_store): Likewise.
16436 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16437 (fusion_gpr_mem_load): Likewise.
16438 (fusion_addis_mem_combo_load): Likewise.
16439 (fusion_addis_mem_combo_store): Likewise.
16440 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16441 (fusion_gpr_mem_load): Likewise.
16442 (fusion_addis_mem_combo_load): Likewise.
16443 (fusion_addis_mem_combo_store): Likewise.
16444 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16446 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16447 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16448 (rs6000_preferred_simd_mode): Likewise.
16449 (output_vec_const_move): Likewise.
16450 (rs6000_expand_vector_extract): Likewise.
16451 (rs6000_split_vec_extract_var): Likewise.
16452 (reg_offset_addressing_ok_p): Likewise.
16453 (rs6000_legitimate_offset_address_p): Likewise.
16454 (rs6000_legitimize_address): Likewise.
16455 (rs6000_emit_set_const): Likewise.
16456 (rs6000_const_vec): Likewise.
16457 (rs6000_emit_move): Likewise.
16458 (rs6000_darwin64_record_arg_recurse): Likewise.
16459 (swap_selector_for_mode): Likewise.
16460 (paired_init_builtins): Likewise.
16461 (altivec_init_builtins): Likewise.
16462 (rs6000_expand_float128_convert): Likewise.
16463 (emit_load_locked): Likewise.
16464 (emit_store_conditional): Likewise.
16465 (rs6000_output_function_epilogue): Likewise.
16466 (rs6000_handle_altivec_attribute): Likewise.
16467 (rs6000_function_value): Likewise.
16468 (emit_fusion_gpr_load): Likewise.
16469 (emit_fusion_p9_load): Likewise.
16470 (emit_fusion_p9_store): Likewise.
16471 * config/rx/rx.c (rx_gen_move_template): Likewise.
16472 (flags_from_mode): Likewise.
16473 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16474 (s390_slb_comparison): Likewise.
16475 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16476 (s390_vector_mode_supported_p): Likewise.
16477 (s390_cc_modes_compatible): Likewise.
16478 (s390_match_ccmode_set): Likewise.
16479 (s390_canonicalize_comparison): Likewise.
16480 (s390_emit_compare_and_swap): Likewise.
16481 (s390_branch_condition_mask): Likewise.
16482 (s390_rtx_costs): Likewise.
16483 (s390_secondary_reload): Likewise.
16484 (__SECONDARY_RELOAD_CASE): Likewise.
16485 (s390_expand_cs): Likewise.
16486 (s390_preferred_simd_mode): Likewise.
16487 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16488 * config/sh/sh.c (sh_print_operand): Likewise.
16489 (dump_table): Likewise.
16490 (sh_secondary_reload): Likewise.
16491 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16492 * config/sh/sh.md (casesi_worker_1): Likewise.
16493 (casesi_worker_2): Likewise.
16494 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16495 (fcc_comparison_operator): Likewise.
16496 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16497 (emit_soft_tfmode_cvt): Likewise.
16498 (sparc_preferred_simd_mode): Likewise.
16499 (output_cbranch): Likewise.
16500 (sparc_print_operand): Likewise.
16501 (sparc_expand_vec_perm_bmask): Likewise.
16502 (vector_init_bshuffle): Likewise.
16503 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16504 (spu_vector_mode_supported_p): Likewise.
16505 (spu_expand_insv): Likewise.
16506 (spu_emit_branch_or_set): Likewise.
16507 (spu_handle_vector_attribute): Likewise.
16508 (spu_builtin_splats): Likewise.
16509 (spu_builtin_extract): Likewise.
16510 (spu_builtin_promote): Likewise.
16511 (spu_expand_sign_extend): Likewise.
16512 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16513 (tilegx_simd_int): Likewise.
16514 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16515 (tilepro_simd_int): Likewise.
16516 * config/v850/v850.c (const_double_split): Likewise.
16517 (v850_print_operand): Likewise.
16518 (ep_memory_offset): Likewise.
16519 * config/vax/vax.c (vax_rtx_costs): Likewise.
16520 (vax_output_int_move): Likewise.
16521 (vax_output_int_add): Likewise.
16522 (vax_output_int_subtract): Likewise.
16523 * config/visium/predicates.md (visium_branch_operator): Likewise.
16524 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16525 (visium_print_operand_address): Likewise.
16526 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16527 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16528 (xtensa_expand_conditional_branch): Likewise.
16529 (xtensa_copy_incoming_a7): Likewise.
16530 (xtensa_output_literal): Likewise.
16531 * dfp.c (decimal_real_maxval): Likewise.
16532 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16534 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16535 Alan Hayward <alan.hayward@arm.com>
16536 David Sherwood <david.sherwood@arm.com>
16538 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16539 (mode_nunits_inline): Likewise.
16540 (mode_inner_inline): Likewise.
16541 (mode_unit_size_inline): Likewise.
16542 (mode_unit_precision_inline): Likewise.
16543 (emit_insn_modes_h): Likewise. Also emit a #define of the
16545 (emit_mode_wider): Add an E_ prefix to mode names.
16546 (emit_mode_complex): Likewise.
16547 (emit_mode_inner): Likewise.
16548 (emit_mode_adjustments): Likewise.
16549 (emit_mode_int_n): Likewise.
16550 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16551 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16552 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16553 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16554 (CRC32_BUILTIN, ENTRY): Likewise.
16555 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16556 (aarch64_pop_regs): Likewise.
16557 (aarch64_process_components): Likewise.
16558 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16559 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16560 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16561 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16562 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16563 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16565 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16566 (pcmpistr): Likewise.
16567 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16568 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16569 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16571 * config/rl78/rl78.c (mduc_regs): Likewise.
16572 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16573 (htm_expand_builtin): Likewise.
16574 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16575 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16576 * config/xtensa/xtensa.c (print_operand): Likewise.
16577 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16578 (NUM_MODE_VECTOR_INT): Likewise.
16579 * genoutput.c (null_operand): Likewise.
16580 (output_operand_data): Likewise.
16581 * genrecog.c (print_parameter_value): Likewise.
16582 * lra.c (debug_operand_data): Likewise.
16584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16586 * dwarf2out.c (scompare_loc_descriptor_wide)
16587 (scompare_loc_descriptor_narrow): New functions, split out from...
16588 (scompare_loc_descriptor): ...here.
16589 * expmed.c (emit_store_flag_int): New function, split out from...
16590 (emit_store_flag): ...here.
16592 2017-08-30 Richard Biener <rguenther@suse.de>
16594 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16595 (dwarf2out_early_finish): Move setting of AT_pubnames from
16596 early debug output to early finish.
16598 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16600 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16601 and -mdata-region to the assembler.
16603 2017-08-30 Richard Biener <rguenther@suse.de>
16605 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16607 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16608 (dwarf2out_early_global_decl): Only generate a DIE for the
16609 abstract origin if it doesn't already exist or is a declaration DIE.
16610 (resolve_addr): Do not add the linkage name twice when
16611 generating a stub DIE for the DW_TAG_GNU_call_site target.
16613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16615 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16616 Use machine_mode rather than int for arg1_mode.
16618 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16621 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16622 that the second argument of the built-in functions to unpack
16623 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16624 switch statement instead a lot of if statements.
16625 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16626 Allow 64-bit values to be in Altivec registers as well as
16627 traditional floating point registers.
16628 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16630 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16632 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16633 MAX_REGS_PER_ADDRESS == 1.
16635 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16637 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16638 * config/i386/i386.c (ix86_option_override_internal): Simplify
16639 setting of opts->x_flag_entry.
16641 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16642 Jakub Jelinek <jakub@redhat.com>
16643 Richard Biener <rguenther@suse.de>
16645 PR tree-optimization/81503
16646 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16647 folded constant fits in the target type; reorder tests for clarity.
16649 2017-08-29 Martin Liska <mliska@suse.cz>
16651 * passes.def: Include pass_lower_switch.
16652 * stmt.c (dump_case_nodes): Remove and move to
16653 tree-switch-conversion.
16654 (case_values_threshold): Likewise.
16655 (expand_switch_as_decision_tree_p): Likewise.
16656 (emit_case_decision_tree): Likewise.
16657 (expand_case): Likewise.
16658 (balance_case_nodes): Likewise.
16659 (node_has_low_bound): Likewise.
16660 (node_has_high_bound): Likewise.
16661 (node_is_bounded): Likewise.
16662 (emit_case_nodes): Likewise.
16663 (struct simple_case_node): New struct.
16664 (add_case_node): Remove.
16665 (emit_case_dispatch_table): Use vector instead of case_list.
16666 (reset_out_edges_aux): Remove.
16667 (compute_cases_per_edge): Likewise.
16668 (expand_case): Build list of simple_case_node.
16669 (expand_sjlj_dispatch_table): Use it.
16670 * tree-switch-conversion.c (struct case_node): Moved from
16671 stmt.c and adjusted.
16672 (emit_case_nodes): Likewise.
16673 (node_has_low_bound): Likewise.
16674 (node_has_high_bound): Likewise.
16675 (node_is_bounded): Likewise.
16676 (case_values_threshold): Likewise.
16677 (reset_out_edges_aux): Likewise.
16678 (compute_cases_per_edge): Likewise.
16679 (add_case_node): Likewise.
16680 (dump_case_nodes): Likewise.
16681 (balance_case_nodes): Likewise.
16682 (expand_switch_as_decision_tree_p): Likewise.
16683 (emit_jump): Likewise.
16684 (emit_case_decision_tree): Likewise.
16685 (try_switch_expansion): Likewise.
16686 (do_jump_if_equal): Likewise.
16687 (emit_cmp_and_jump_insns): Likewise.
16688 (fix_phi_operands_for_edge): New function.
16689 (record_phi_operand_mapping): Likewise.
16690 (class pass_lower_switch): New pass.
16691 (pass_lower_switch::execute): New function.
16692 (make_pass_lower_switch): Likewise.
16693 (conditional_probability):
16694 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16695 * tree-pass.h: Add make_pass_lower_switch.
16697 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16700 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16703 2017-08-29 Richard Biener <rguenther@suse.de>
16705 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16706 we do not add a DW_AT_inline attribute twice.
16707 (gen_subprogram_die): Remove code setting DW_AT_inline on
16708 DECL_ABSTRACT_P nodes.
16710 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16712 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16713 calls to internal functions.
16714 (gimplify_modify_expr): Likewise.
16715 * tree-call-cdce.c (use_internal_fn): Likewise.
16716 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16717 (convert_to_divmod): Set the nothrow flag.
16718 * tree-if-conv.c (predicate_mem_writes): Likewise.
16719 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16720 (vectorizable_call): Likewise.
16721 (vectorizable_store): Likewise.
16722 (vectorizable_load): Likewise.
16723 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16724 (vect_recog_mask_conversion_pattern): Likewise.
16726 2017-08-29 Martin Liska <mliska@suse.cz>
16729 * gcc.c (driver_handle_option): Add new argument.
16730 * opts-common.c (handle_option): Pass
16731 target_option_override_hook.
16732 * opts-global.c (lang_handle_option): Add new option.
16733 (set_default_handlers): Add new argument.
16734 (decode_options): Likewise.
16735 * opts.c (target_handle_option): Likewise.
16736 (common_handle_option): Call target_option_override_hook.
16737 * opts.h (struct cl_option_handler_func): Add hook for
16738 target option override.
16739 (struct cl_option_handlers): Likewise.
16740 (set_default_handlers): Add new argument.
16741 (decode_options): Likewise.
16742 (common_handle_option): Likewise.
16743 (target_handle_option): Likewise.
16744 * toplev.c (toplev::main): Pass targetm.target_option.override
16747 2017-08-29 Richard Biener <rguenther@suse.de>
16748 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16750 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16751 life to the active subtree.
16753 2017-08-28 Jeff Law <law@redhat.com>
16755 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16756 derive_equivalences.
16757 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16758 Code moved into....
16759 (edge_info::derive_equivalences): New private member function
16761 * tree-ssa-dom.c (class edge_info): Changed from a struct
16762 to a class. Add ctor/dtor, methods and data members.
16763 (edge_info::edge_info): Renamed from allocate_edge_info.
16764 Initialize additional members.
16765 (edge_info::~edge_info): New.
16766 (free_dom_edge_info): Delete the edge info.
16767 (record_edge_info): Use new class & associated member functions.
16768 Tighten forms for testing for edge equivalences.
16769 (record_temporary_equivalences): Iterate over the simple
16770 equivalences rather than assuming there's only one per edge.
16771 (cprop_into_successor_phis): Iterate over the simple
16772 equivalences rather than assuming there's only one per edge.
16773 (optimize_stmt): Use operand_equal_p rather than pointer
16774 equality for mini-DSE code.
16776 2017-08-28 Nathan Sidwell <nathan@acm.org>
16778 * gcc.c (execute): Fold SIGPIPE handling into switch
16779 statement. Adjust internal error message.
16781 2017-08-28 Richard Biener <rguenther@suse.de>
16784 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16785 Do nothing for removed DIEs.
16787 2017-08-28 Richard Biener <rguenther@suse.de>
16789 PR tree-optimization/81977
16790 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16793 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16796 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16797 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16798 using targetm.gen_mem_thread_fence.
16800 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16803 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16804 predicate to register_operand. Reorder operands.
16805 (*btr<mode>): Ditto.
16806 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16807 (*btr<mode>_mask): Ditto.
16809 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16811 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16812 * config/rs6000/xmmintrin.h: New file.
16813 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16815 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16818 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16819 parameter and_insn and return it.
16820 (recombine_lvx_pattern): Insert a copy to ensure availability of
16821 the base register of the copied masking operation at the point of
16822 the instruction replacement.
16823 (recombine_stvx_pattern): Likewise.
16825 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16827 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16828 undocumented switches.
16829 (-mpower9-dform-vector): Likewise.
16830 (-mpower9-dform): Likewise.
16831 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16832 comments to delete references to -mpower9-dform* switches.
16833 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16834 Delete reference to -mpower9-dform* switches, test for
16835 -mpower9-vector instead.
16836 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16837 (OTHER_P9_VECTOR_MASKS): Likewise.
16838 (POWERPC_MASKS): Likewise.
16839 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16840 tests against -mpower9-dform* to -mpower9-vector. Delete code
16841 that checked for -mpower9-dform* consistancy with other options.
16842 Add test for -mpower9-misc to enable other power9 switches.
16843 (rs6000_init_hard_regno_mode_ok): Likewise.
16844 (rs6000_option_override_internal): Likewise.
16845 (rs6000_emit_prologue): Likewise.
16846 (rs6000_emit_epilogue): Likewise.
16847 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16848 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16849 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16851 (emit_fusion_p9_store): Likewise.
16852 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16853 resetting these macros if the assembler does not support ISA 3.0
16855 (TARGET_P9_DFORM_VECTOR): Likewise.
16856 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16857 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16859 2017-08-25 Alan Modra <amodra@gmail.com>
16861 PR rtl-optimization/81747
16862 * cse.c (cse_extended_basic_block): Don't attempt to record
16863 equivalences for degenerate conditional branches that branch
16864 to their fall-through.
16866 2017-08-24 Martin Sebor <msebor@redhat.com>
16868 PR middle-end/81908
16869 * gimple-fold.c (size_must_be_zero_p): New function.
16870 (gimple_fold_builtin_memory_op): Call it.
16872 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16874 * config/rs6000/mm_malloc.h: New file.
16876 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16878 PR tree-optimization/81913
16879 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16880 analysis when either IVs in condition can wrap.
16882 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16884 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16885 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16887 2017-08-24 Richard Biener <rguenther@suse.de>
16890 * targhooks.c (default_target_can_inline_p): Properly
16891 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16892 is present and always compare.
16893 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16894 infer -mfpmath=sse from TARGET_SSE_P.
16895 (ix86_can_inline_p): Properly use target_option_default_node when
16896 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16898 2017-08-24 Richard Biener <rguenther@suse.de>
16901 * dwarf2out.c (output_die): Handle flag_generate_offload like
16903 (output_comp_unit): Likewise.
16904 (gen_array_type_die): Likewise.
16905 (dwarf2out_early_finish): Likewise.
16906 (note_variable_value_in_expr): Likewise.
16907 (dwarf2out_finish): Likewise. Adjust assert.
16908 * cgraphunit.c (symbol_table::compile): Move setting of
16909 flag_generate_offload earlier ...
16910 (symbol_table::finalize_compilation_unit): ... here, before
16911 early debug finalization.
16913 2017-08-24 Richard Biener <rguenther@suse.de>
16915 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16916 and ipa-fnsummary.h.
16917 (ix86_can_inline_p): When ix86_fpmath flags do not match
16918 check whether the callee uses FP math at all.
16920 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
16922 PR middle-end/81931
16923 * tree-ssanames.c (get_nonzero_bits): Use element_precision
16924 instead of TYPE_PRECISION.
16926 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
16927 Alan Hayward <alan.hayward@arm.com>
16928 David Sherwood <david.sherwood@arm.com>
16930 * combine.c (make_extraction): Use subreg_offset_from_lsb.
16932 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
16934 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
16936 (ix86_frame::stack_realign_allocate): New field.
16937 (struct machine_frame_state): Modify comments.
16938 (machine_frame_state::sp_realigned_fp_end): New field.
16939 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
16940 layout calculation.
16941 (sp_valid_at): Add assertion to assure no attempt to access invalid
16942 offset of a realigned stack.
16943 (fp_valid_at): Likewise.
16944 (choose_baseaddr): Modify comments.
16945 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
16946 ix86_expand_prologue.
16947 (ix86_expand_prologue): Modify stack realignment and allocation.
16948 (ix86_expand_epilogue): Modify comments.
16949 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
16950 avx2_runtime, avx512f, and avx512f_runtime.
16952 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
16954 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
16955 (mstackrealign): Do not init to -1.
16956 * config/i386/i386.c (ix86_option_override_internal):
16957 Check opts_set, not opts when setting default value of
16958 opts->x_ix86_force_align_arg_pointer.
16960 2017-08-23 Richard Biener <rguenther@suse.de>
16962 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
16963 lang_hooks.decl_printable_name.
16964 * print-rtl-function.c (print_rtx_function): Likewise.
16965 * tree-pretty-print.c (dump_function_header): Likewise.
16967 2017-08-23 Richard Biener <rguenther@suse.de>
16970 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
16971 -g0 at compile-time.
16973 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16975 PR middle-end/19706
16976 * doc/sourcebuild.texi (Other hardware attributes):
16979 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16981 PR middle-end/19706
16982 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
16983 Fix single-use check.
16985 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
16987 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
16989 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16991 * doc/install.texi: Modify to add more details on running selected
16994 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16996 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
16997 is combined with -mabi=ms.
16998 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17001 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17003 PR tree-optimization/81488
17004 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17005 and cached_basis fields.
17006 (MAX_SPREAD): New constant.
17007 (alloc_cand_and_find_basis): Initialize new fields.
17008 (clear_visited): New function.
17009 (create_phi_basis_1): Rename from create_phi_basis, set visited
17010 and cached_basis fields.
17011 (create_phi_basis): New wrapper function.
17012 (phi_add_costs_1): Rename from phi_add_costs, add spread
17013 parameter, set visited field, short-circuit when limits reached.
17014 (phi_add_costs): New wrapper function.
17015 (record_phi_increments_1): Rename from record_phi_increments, set
17017 (record_phi_increments): New wrapper function.
17018 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17019 (phi_incr_cost): New wrapper function.
17020 (all_phi_incrs_profitable_1): Rename from
17021 all_phi_incrs_profitable, set visited field.
17022 (all_phi_incrs_profitable): New wrapper function.
17024 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17025 Alan Hayward <alan.hayward@arm.com>
17026 David Sherwood <david.sherwood@arm.com>
17028 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17029 that takes the outer and inner modes.
17030 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17031 comparison as the canonical test for a paradoxical subreg.
17032 * combine.c (simplify_set): Use paradoxical_subreg_p.
17033 (make_extraction): Likewise.
17034 (force_to_mode): Likewise.
17035 (rtx_equal_for_field_assignment_p): Likewise.
17036 (gen_lowpart_for_combine): Likewise.
17037 (simplify_comparison): Likewise.
17038 * cse.c (equiv_constant): Likewise.
17039 * expmed.c (store_bit_field_1): Likewise.
17040 * final.c (alter_subreg): Likewise.
17041 * fwprop.c (propagate_rtx): Likewise.
17042 (forward_propagate_subreg): Likewise.
17043 * ira-conflicts.c (ira_build_conflicts): Likewise.
17044 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17045 * lra-constraints.c (curr_insn_transform): Likewise.
17046 (split_reg): Likewise.
17047 * lra-eliminations.c (move_plus_up): Likewise.
17048 (lra_eliminate_regs_1): Likewise.
17049 * recog.c (general_operand): Likewise.
17050 * ree.c (combine_reaching_defs): Likewise.
17051 * reload.c (push_reload): Likewise.
17052 (find_reloads): Likewise.
17053 * reload1.c (elimination_effects): Likewise.
17054 (compute_reload_subreg_offset): Likewise.
17055 (choose_reload_regs): Likewise.
17056 * rtlanal.c (subreg_lsb_1): Likewise.
17057 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17058 (simplify_subreg): Likewise.
17059 * var-tracking.c (track_loc_p): Likewise.
17060 * emit-rtl.c (byte_lowpart_offset): Likewise.
17061 (paradoxical_subreg_p): Delete out-of-line definition.
17063 2017-08-22 Jeff Law <law@redhat.com>
17065 PR tree-optimization/81741
17066 PR tree-optimization/71947
17067 * tree-ssa-dom.c: Include tree-inline.h.
17068 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17069 equivalences if one is more expensive to compute than the other.
17070 * tree-ssa-scopedtables.h (class const_or_copies): Make
17071 record_const_or_copy_raw method private.
17072 (class avail_exprs_stack): New method simplify_binary_operation.
17073 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17074 avail_exprs_stack::simplify_binary_operation as needed.
17075 (avail_exprs_stack::simplify_binary_operation): New function.
17077 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17079 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17080 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17081 (DOT_SYMBOLS): Likewise.
17082 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17083 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17084 (RS6000_ABI_NAME): Likewise.
17085 (TARGET_CMODEL): Likewise.
17086 (TOC_SECTION_ASM_OP): Likewise.
17087 (SET_CMODEL): New macro.
17088 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17090 2017-08-22 Richard Biener <rguenther@suse.de>
17092 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17093 to allow for free-lang-data replacements similar to verify_type_variant.
17095 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17096 Alan Hayward <alan.hayward@arm.com>
17097 David Sherwood <david.sherwood@arm.com>
17099 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17100 VOIDmode for the LABEL_REF.
17102 2017-08-22 Richard Biener <rguenther@suse.de>
17104 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17106 2017-08-22 Richard Biener <rguenther@suse.de>
17108 * common.opt (feliminate-dwarf2-dups): Ignore.
17109 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17110 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17111 same_die_p_wrap, compute_section_prefix,
17112 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17113 (comdat_symbol_id, comdat_symbol_number): Likewise.
17114 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17116 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17117 (output_die): Mark unreachable path unreachable.
17118 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17119 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17120 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17121 (dwarf2out_early_finish): Likewise.
17123 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17125 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17127 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17130 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17131 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17132 (avr_attribute_table) <io, io_low, address>: Initialize
17133 .decl_required with true.
17135 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17137 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17138 undocumented debugging options.
17139 (-mvsx-scalar-double): Likewise.
17140 (-mallow-df-permute): Likewise.
17141 (-mvectorize-builtins): Likewise.
17142 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17143 (rs6000_builtin_vectorized_function): Likewise.
17144 (rs6000_builtin_md_vectorized_function): Likewise.
17145 (rs6000_opt_vars): Likewise.
17147 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17150 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17151 (*btrq_imm): Rename from *btrq.
17152 (*btcq_imm): Rename from *btcq.
17153 (btsc): New code attribute.
17154 (*<btsc><mode>): New insn pattern.
17155 (*btr<mode>): Ditto.
17156 (*<btsc><mode>_mask): New insn_and_split pattern.
17157 (*btr<mode>_mask): Ditto.
17159 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17160 Alan Hayward <alan.hayward@arm.com>
17161 David Sherwood <david.sherwood@arm.com>
17163 * function.c (pad_below): Simplify padding calculation.
17165 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17166 Alan Hayward <alan.hayward@arm.com>
17167 David Sherwood <david.sherwood@arm.com>
17169 * target.def (function_prologue): Remove frame size argument.
17170 (function_epilogue): Likewise.
17171 * doc/tm.texi: Regenerate.
17172 * final.c (final_start_function): Update call to function_prologue.
17173 (final_end_function): Update call to function_epilogue.
17174 (default_function_pro_epilogue): Remove frame size argument.
17175 * output.h (default_function_pro_epilogue): Likewise.
17176 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17177 (arm_output_function_prologue): Likewise.
17178 * config/frv/frv.c (frv_function_prologue): Likewise.
17179 (frv_function_epilogue): Likewise.
17180 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17181 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17182 (ia64_output_function_epilogue): Likewise.
17183 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17184 (m32r_output_function_epilogue): Likewise.
17185 * config/microblaze/microblaze.c (microblaze_function_prologue)
17186 (microblaze_function_epilogue): Likewise.
17187 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17188 (mips_output_function_epilogue): Likewise.
17189 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17190 (mmix_target_asm_function_epilogue): Likewise.
17191 * config/msp430/msp430.c (msp430_start_function): Likewise.
17192 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17193 (nds32_asm_function_epilogue): Likewise.
17194 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17195 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17196 (pa_output_function_epilogue): Likewise.
17197 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17198 (rs6000_output_function_epilogue): Likewise.
17199 * config/rl78/rl78.c (rl78_start_function): Likewise.
17200 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17201 (rs6000_output_function_epilogue): Likewise.
17202 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17203 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17204 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17205 (sparc_asm_function_epilogue): Likewise.
17207 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17209 * tree.h (type_has_mode_precision_p): New function.
17210 * convert.c (convert_to_integer_1): Use it.
17211 * expr.c (expand_expr_real_2): Likewise.
17212 (expand_expr_real_1): Likewise.
17213 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17214 * match.pd: Likewise.
17215 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17216 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17217 * tree-tailcall.c (process_assignment): Likewise.
17218 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17219 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17220 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17221 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17222 (vectorizable_assignment): Likewise.
17223 (vectorizable_shift): Likewise.
17224 (vectorizable_operation): Likewise.
17225 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17227 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17229 * match.pd: Add pow (C, x) simplification.
17231 2017-08-21 Richard Biener <rguenther@suse.de>
17233 PR tree-optimization/81900
17234 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17235 for blocks with abnormal predecessors.
17236 (compute_antic): Do not set visited flag prematurely.
17238 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17241 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17243 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17245 * stor-layout.h (vector_type_mode): Move to...
17246 * tree.h (vector_type_mode): ...here.
17247 * stor-layout.c (vector_type_mode): Move to...
17248 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17250 2017-08-21 Richard Biener <rguenther@suse.de>
17252 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17253 register_external_die hooks.
17254 (debug_false_tree_charstarstar_uhwistar): Declare.
17255 (debug_nothing_tree_charstar_uhwi): Likewise.
17256 * debug.c (do_nothing_debug_hooks): Adjust.
17257 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17258 (debug_nothing_tree_charstar_uhwi): Likewise.
17259 * dbxout.c (dbx_debug_hooks): Adjust.
17260 (xcoff_debug_hooks): Likewise.
17261 * sdbout.c (sdb_debug_hooks): Likewise.
17262 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17263 * dwarf2out.c (macinfo_label_base): New global.
17264 (dwarf2out_register_external_die): New function for the
17265 register_external_die hook.
17266 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17267 (dwarf2_debug_hooks): Use them.
17268 (dwarf2_lineno_debug_hooks): Adjust.
17269 (struct die_struct): Add with_offset flag.
17270 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17271 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17272 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17273 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17274 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17275 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17276 defining section names for the early LTO debug variants.
17277 (reset_indirect_string): New helper.
17278 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17279 (print_dw_val): Add support for offsetted symbol references.
17280 (get_ultimate_context): Split out from is_cxx.
17281 (is_cxx): Use get_ultimate_context.
17282 (is_fortran): Add decl overload.
17283 (compute_comp_unit_symbol): Split out worker from
17284 compute_section_prefix.
17285 (compute_section_prefix): Call compute_comp_unit_symbol and
17286 set comdat_type_p here.
17287 (output_die): Skip DIE symbol output for the LTO added one.
17288 Handle DIE symbol references with offset.
17289 (output_comp_unit): Guard section name mangling properly.
17290 For LTO debug sections emit a symbol at the section beginning
17291 which we use to refer to its DIEs.
17292 (add_abstract_origin_attribute): For DIEs registered via
17293 dwarf2out_register_external_die directly refer to the early
17294 DIE rather than indirectly through the shadow one we created.
17295 Remove obsolete call to dwarf2out_abstract_function for
17296 non-function/block origins.
17297 (gen_array_type_die): When generating early LTO debug do
17298 not emit DW_AT_string_length.
17299 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17300 late when in LTO. As suggested place a gcc_unreachable for
17301 the DECL_ABSTRACT_P case.
17302 (gen_subprogram_die): Avoid another specification DIE
17303 for early built declarations/definitions for the late LTO case.
17304 (gen_variable_die): Add type references for late duplicated VLA dies
17306 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17307 we have the abstract instance already.
17308 (process_scope_var): Adjust decl DIE contexts in LTO which
17309 first puts them in limbo.
17310 (gen_decl_die): Do not generate type DIEs late apart from
17311 types for VLAs or for decls we do not yet have a DIE. Do not
17312 call dwarf2out_abstract_function late.
17313 (dwarf2out_early_global_decl): Make sure to create DIEs
17314 for abstract instances of a decl first.
17315 (dwarf2out_late_global_decl): Adjust comment.
17316 (output_macinfo_op): With multiple macro sections use
17317 macinfo_label_base to distinguish labels.
17318 (output_macinfo): Likewise. Update macinfo_label_base.
17319 Pass in the line info label.
17320 (note_variable_value_in_expr): When generating LTO resolve
17321 all variable values here by generating DIEs as needed.
17322 (init_sections_and_labels): Add early LTO debug flag parameter
17323 and generate different sections and names if set. Add generation
17324 counter for the labels so we can have multiple of them.
17325 (reset_dies): Helper to allow DIEs to be output multiple times.
17326 (dwarf2out_finish): When outputting DIEs to the fat part of an
17327 LTO object first reset DIEs.
17328 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17329 (modified_type_die): Check for decl_ultimate_origin being self
17331 (gen_type_die_with_usage): Likewise.
17332 (gen_typedef_die): Allow decl_ultimate_origin being self.
17333 (set_decl_abstract_flags): Remove.
17334 (set_block_abstract_flags): Likewise.
17335 (dwarf2out_abstract_function): Treat the early generated DIEs
17336 as the abstract copy and only add DW_AT_inline and
17337 DW_AT_artificial here and call set_decl_origin_self.
17338 If the DIE has an abstract origin don't do anything.
17339 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17340 if we have none yet (Go fails to build one, PR78628).
17341 (variably_modified_type_p): Prevent endless recursion for Ada
17342 cyclic pointer types.
17343 * lto-streamer-in.c: Include debug.h.
17344 (dref_queue): New global.
17345 (lto_read_tree_1): Stream in DIE references.
17346 (lto_input_tree): Register DIE references.
17347 (input_function): Stream DECL_DEBUG_ARGS.
17348 * lto-streamer-out.c: Include debug.h.
17349 (lto_write_tree_1): Output DIE references.
17350 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17351 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17352 (output_function): Stream DECL_DEBUG_ARGS.
17353 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17354 Stream DECL_ABSTRACT_ORIGIN.
17355 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17356 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17357 DECL_CONTEXT for file-scope decls.
17358 * lto-streamer.h (struct dref_entry): Declare.
17359 (dref_queue): Likewise.
17360 * cfgexpand.c (pass_expand::execute): Do not call the
17361 outlining_inline_function hook here.
17362 * lto-wrapper.c (debug_obj): New global.
17363 (tool_cleanup): Unlink it if required.
17364 (debug_objcopy): New function.
17365 (run_gcc): Handle early debug sections in the IL files by
17366 extracting them to separate files, partially linkin them and
17367 feeding the result back as result to the linker.
17368 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17369 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17370 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17371 sections into a separate segment.
17372 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17374 (darwin_asm_dwarf_section): Likewise.
17375 (darwin_asm_output_dwarf_offset): Likewise.
17376 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17378 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17379 Alan Hayward <alan.hayward@arm.com>
17380 David Sherwood <david.sherwood@arm.com>
17382 * read-md.h (md_reader::record_potential_iterator_use): Replace
17383 pointer argument with an rtx and an index.
17384 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17385 (apply_mode_iterator): Likewise.
17386 (apply_code_iterator): Likewise.
17387 (apply_int_iterator): Likewise.
17388 (apply_subst_iterator): Likewise.
17389 (record_iterator_use): Likewise.
17390 (record_attribute_use): Likewise.
17391 (md_reader::record_potential_iterator_use): Likewise. Update calls
17392 to record_iterator_use and apply_iterator.
17393 (iterator_use): Replace ptr with x and index.
17394 (attribute_use): Likewise.
17395 (apply_attribute_uses): Update calls to apply_iterator.
17396 (apply_iterators): Likewise. Update initialization of iterator_use.
17397 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17398 and record_potential_iterator_use.
17399 (rtx_reader::read_rtx_operand): Likewise.
17401 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17402 Alan Hayward <alan.hayward@arm.com>
17403 David Sherwood <david.sherwood@arm.com>
17405 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17408 2017-08-21 Richard Biener <rguenther@suse.de>
17410 PR middle-end/81884
17411 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17412 at struct end conservatively when comparing common bases.
17414 2017-08-21 Richard Biener <rguenther@suse.de>
17416 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17417 (mem_ref_in_stmt): Remove.
17418 (determine_max_movement): Use ref index to get at the reference.
17419 (invariantness_dom_walker::before_dom_children): Deal with
17420 lim data already initialized.
17421 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17423 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17425 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17426 (TARGET_ISA_ROUND): Ditto.
17427 (TARGET_ROUND): Ditto.
17428 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17429 * config/i386/i386.md: Ditto.
17430 * config/i386/sse.md: Ditto.
17431 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17432 with OPTION_MASK_ISA_SSE4_1.
17434 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17437 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17438 __builtin_ia32_lzcnt_u16.
17440 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17443 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17444 (rs6000_set_current_function): Rewrite function to use it.
17446 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17449 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17450 and TYPE_WARN_IF_NOT_ALIGN.
17451 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17452 (handle_warn_if_not_align): New.
17453 (place_union_field): Call handle_warn_if_not_align.
17454 (place_field): Call handle_warn_if_not_align.
17455 Copy TYPE_WARN_IF_NOT_ALIGN.
17456 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17457 (layout_type): Likewise.
17458 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17460 (tree_decl_common): Add warn_if_not_align.
17461 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17462 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17463 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17464 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17465 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17466 * doc/extend.texi: Document warn_if_not_aligned attribute.
17467 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17469 2017-08-17 Martin Liska <mliska@suse.cz>
17472 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17473 (get_data_dependence): Use it as pointer type.
17474 (distribute_loop): Likewise.
17476 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17478 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17479 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17480 (p8_vmrgow_<mode>_direct): New define_insn.
17481 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17482 handle endianness for vmrgew and vmrgow permute patterns.
17484 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17486 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17487 * config/rs6000/rs6000-cpus.def: Remove comment.
17488 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17489 (POWERPC_MASKS): Likewise.
17490 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17491 use of TARGET_VSX_TIMODE.
17492 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17493 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17494 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17495 (rs6000_option_override_internal): Remove dead code.
17496 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17497 (rs6000_legitimize_reload_address): Likewise.
17498 (rs6000_legitimate_address_p): Likewise.
17499 (rs6000_opt_masks): Delete "vsx-timode".
17500 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17501 from function comment.
17502 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17503 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17504 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17506 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17507 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17508 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17509 (VSX_TI): Likewise.
17511 (define_peephole2): Likewise.
17513 2017-08-17 Martin Sebor <msebor@redhat.com>
17516 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17517 past the end of an array.
17518 (test_pp_format): Add test cases.
17520 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17522 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17523 missing ECF_NOTHROW flags.
17525 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17528 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17529 operands residing in integer registers.
17530 (*vsx_le_perm_load_<mode>): Likewise.
17531 (*vsx_le_perm_store_<mode>): Likewise.
17532 (define_peephole2): Add peepholes to optimize the above.
17534 2017-08-17 Marek Polacek <polacek@redhat.com>
17536 PR middle-end/81814
17537 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17538 to mimic what shorten_compare did. Change the return type to bool.
17539 (fold_cond_expr_with_comparison): Update call to
17540 operand_equal_for_comparison_p.
17541 (fold_ternary_loc): Likewise.
17543 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17545 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17547 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17549 2017-08-17 Richard Biener <rguenther@suse.de>
17551 * tree-ssa-structalias.c (solve_graph): When propagating
17552 to successors update the graphs succ edges and avoid duplicate work.
17554 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17557 * config/i386/i386.c (ix86_option_override_internal): Save target
17558 specific options after ix86_stack_protector_guard_reg was changed.
17560 2017-08-17 Richard Biener <rguenther@suse.de>
17562 PR tree-optimization/81827
17563 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17564 (new_var_info): Initialize it conservatively.
17565 (get_call_vi): Mark register vars.
17566 (new_scalar_tmp_constraint_exp): Likewise.
17567 (handle_rhs_call): Likewise.
17568 (handle_const_call): Likewise.
17569 (create_function_info_for): Likewise.
17570 (solve_constraints): Sort varinfos to separate register from
17571 non-register vars to pack points-to solution bitmaps during
17574 2017-08-17 Marek Polacek <polacek@redhat.com>
17576 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17578 2017-08-17 Richard Biener <rguenther@suse.de>
17580 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17581 to true when overflow is undefined and we saturated the result.
17583 2017-08-17 Alan Modra <amodra@gmail.com>
17586 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17587 Don't use store multiple if only one reg needs saving.
17588 (interesting_frame_related_regno): New function.
17589 (rs6000_frame_related): Don't emit frame info for regs that
17591 (rs6000_emit_epilogue): Likewise.
17593 2017-08-16 Nathan Sidwell <nathan@acm.org>
17595 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17596 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17597 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17598 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17599 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17600 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17601 (verify_type): Adjust for TYPE_BINFO move.
17602 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17603 process TYPE_BINFO directly.
17604 (hash_tree): Likewise.
17605 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17607 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17610 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17612 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17614 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17617 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17618 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17619 Add ix86_binary_operator_ok to insn constraint.
17620 (*iordi_1_bts): Ditto.
17621 (*xordi_1_btc): Ditto.
17622 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17623 Update corresponding peephole2 pattern.
17627 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17629 PR tree-optimization/81832
17630 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17631 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17633 2017-08-16 Marek Polacek <polacek@redhat.com>
17636 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17637 regarding size_zero_node.
17639 2017-08-16 Martin Liska <mliska@suse.cz>
17642 * config.gcc: Respect previously set extra_objs in case
17645 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17647 PR tree-optimization/81835
17648 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17649 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17650 not depend on the phi.
17652 2017-08-16 Alan Modra <amodra@gmail.com>
17654 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17657 2017-08-16 Alan Modra <amodra@gmail.com>
17659 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17660 (save_reg_p): ..into this. Update all callers.
17661 (first_reg_to_save): Simplify.
17663 2017-08-16 Alan Modra <amodra@gmail.com>
17665 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17668 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17672 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17673 constant count if that count is less than 32.
17675 2017-08-15 Nathan Sidwell <nathan@acm.org>
17677 * gcc.c (execute): Emit friendlier message if inferior is killed
17678 by an external cause.
17680 2017-08-15 Richard Biener <rguenther@suse.de>
17682 PR tree-optimization/81790
17683 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17684 CONSTRUCTORs from simplifying and VN.
17686 2017-08-14 Martin Sebor <msebor@redhat.com>
17688 * builtin-attrs.def: Add comments.
17690 2017-08-14 Martin Sebor <msebor@redhat.com>
17693 * doc/extend.texi (attribute nonstring): Document new attribute.
17695 2017-08-14 Martin Sebor <msebor@redhat.com>
17698 * tree-diagnostic.c (default_tree_printer): Handle %G.
17699 * gimple-pretty-print.h (percent_G_format): Declare new function.
17700 * gimple-pretty-print.c (percent_G_format): Define.
17701 * tree-pretty-print.c (percent_K_format): Add argument.
17703 2017-08-14 Martin Sebor <msebor@redhat.com>
17705 PR translation/79998
17706 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17707 Remove a stray space.
17709 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17712 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17713 (*iordi_1_bts): Ditto.
17714 (*xordi_1_btc): Ditto.
17716 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17719 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17720 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17722 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17723 quoted strings, and make more translator-friendly.
17724 (darwin_rs6000_override_options): Likewise.
17725 (rs6000_option_override_internal): Likewise.
17726 (rs6000_return_in_memory): Fix overlong line.
17727 (init_cmulative_args): Use quoted strings, and make more
17728 translator-friendly.
17729 (rs6000_pass_by_reference): Fix overlong line.
17730 (def_builtin): Use quoted strings.
17731 (altivec_expand_predicate_builtin): Use quoted strings, and make
17732 more translator-friendly.
17733 (htm_expand_builtin): Use quoted strings.
17734 (cpu_expand_builtin): Use quoted strings, and make more
17735 translator-friendly.
17736 (altivec_expand_builtin): Likewise.
17737 (paired_expand_predicate_builtin): Likewise.
17738 (rs6000_invalid_builtin): Likewise.
17739 (builtin_function_type): Use quoted strings.
17740 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17741 more translator-friendly.
17742 (rs6000_trampoline_init): Likewise.
17743 (rs6000_handle_altivec_attribute): Likewise.
17744 (rs6000_inner_target_options): Use quoted strings.
17745 (rs6000_disable_incompatible_switches): Likewise.
17746 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17747 strings, and make more translator-friendly.
17748 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17750 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17752 PR tree-optimization/81799
17753 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17754 cond_expr to simple gimple operand.
17756 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17758 PR middle-end/46932
17759 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17761 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17765 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17766 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17767 TARGET_GASISR_PROLOGUES.
17768 * config/avr/avr.c (avr_option_override): Same.
17769 (avr_pass_pre_proep::execute): Same.
17771 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17774 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17775 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17777 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17779 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17780 define_insn_and_split. Split to a memory load from 0 in
17781 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17782 using PTR mode iterator.
17783 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17784 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17785 (*add_tp_<mode>): Redefine as define_insn_and_split.
17786 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17787 address space. Merge with *add_tp_x32 using PTR mode iterator.
17788 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17789 Split to an add with a memory load from 0 in
17790 DEFAULT_TLS_SEG_REG address space.
17792 2017-08-12 Andrew Pinski <apinski@cavium.com>
17794 * config/aarch64/aarch64-option-extensions.def (rdma):
17795 Fix feature string to what Linux prints out in /proc/cpuinfo.
17797 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17800 * dwarf2out.c (modified_type_die): For C typedef types that have
17801 an ultimate origin, process the ultimate origin instead of the
17803 (gen_typedef_die): Assert that input DECLs have no ultimate
17805 (gen_type_die_with_usage): For typedef variants that have an
17806 ultimate origin, just call gen_decl_die on the original DECL.
17807 (process_scope_var): Avoid creating DIEs for local typedefs and
17808 concrete static variables.
17810 2017-08-12 Alan Modra <amodra@gmail.com>
17814 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17815 match gnu-user.h startfile.
17816 (ENDFILE_LINUX_SPEC): Similarly.
17818 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17821 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17823 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17825 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17826 * config/aarch64/aarch64.md (mov<mode>): Change.
17827 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17828 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17829 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17831 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17833 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17834 for storage order barriers.
17836 2017-08-11 Martin Liska <mliska@suse.cz>
17839 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17840 variables of void type.
17842 2017-08-11 Martin Liska <mliska@suse.cz>
17844 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17845 TARGET_SUPPORTS_ALIASES.
17846 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17847 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17848 (optimize_weakref): Likewise.
17849 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17850 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17851 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17853 2017-08-11 Martin Liska <mliska@suse.cz>
17856 * config/i386/i386.c (make_resolver_func): Do complete
17857 refactoring of the function.
17859 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17862 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17863 * config/i386/i386.c (ix86_stack_protect_guard): Use
17864 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17865 * doc/invoke.texi (x86 Options): Document
17866 -mstack-protector-guard-symbol= option.
17868 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17870 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17871 * config/i386/i386.c (ix86_split_stack_guard): New function.
17872 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17873 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17874 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17875 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17876 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17877 (split_stack_space_check): Call ix86_split_stack_guard.
17879 2017-08-10 Martin Sebor <msebor@redhat.com>
17881 * print-tree.c (print_node): Print location using the established
17883 Replace spaces with colons.
17884 (debug_raw, debug): Ditto.
17886 2017-08-10 Martin Sebor <msebor@redhat.com>
17889 * pretty-print.c (pp_format): Correct the handling of %s precision.
17891 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17894 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17896 (ix86_finalize_stack_frame_flags): This. Also clear
17897 frame_pointer_needed if -fno-omit-frame-pointer is used without
17899 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17900 with ix86_finalize_stack_frame_flags.
17901 (ix86_expand_epilogue): Likewise.
17902 (ix86_expand_split_stack_prologue): Likewise.
17903 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17905 2017-08-10 Martin Liska <mliska@suse.cz>
17908 * c-attribs.c (handle_target_attribute):
17909 Report warning for an empty string argument of target attribute.
17911 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17914 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17916 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17917 or DECL_NONLOCAL labels.
17918 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17919 or DECL_NONLOCAL labels here.
17921 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
17923 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
17924 to indicate when early gimple folding has been disabled.
17925 (rs6000_gimple_fold_builtin): Add debug content.
17926 (rs6000_invalid_builtin): Fix whitespace.
17927 (rs6000_expand_builtin): Fix whitespace.
17928 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
17930 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
17933 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
17934 SAVE_MULTIPLE if not all the registers that saves, should be saved.
17936 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
17938 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
17939 (qdf24xx): Likewise.
17940 * config/aarch64/aarch64.md: Include falkor.md.
17941 * config/aarch64/falkor.md: New.
17943 2017-08-09 Marek Polacek <polacek@redhat.com>
17946 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
17947 * diagnostic.c (emit_diagnostic): Add a comment.
17948 (emit_diagnostic_valist): New function.
17950 2017-08-09 Marek Polacek <polacek@redhat.com>
17953 * input.c (make_location): New overload.
17954 * input.h (make_location): Declare.
17956 2017-08-08 Alan Modra <amodra@gmail.com>
17957 H.J. Lu <hongjiu.lu@intel.com>
17960 * gcc.c (NO_PIE_SPEC): Delete.
17961 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
17963 (LINK_PIE_SPEC): ..to here.
17964 (LINK_COMMAND_SPEC): Support -no-pie.
17965 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
17966 chain of crtbegin*.o selection, update for PIE_SPEC changes and
17968 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
17969 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
17970 (ENDFILE_CRTEND_SPEC): Similarly.
17972 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
17975 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
17976 (mstack-protector-guard-offset=): Ditto.
17977 * config/i386/i386.c (ix86_option_override): Handle
17978 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
17980 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
17981 ix86_stack_protect_guard_offset variables.
17982 (TARGET_STACK_PROTECT_GUARD): Always define.
17983 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
17984 and -mstack-protector-guard-offset= options.
17986 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17988 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
17989 boundary case for the last candidate.
17991 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17993 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
17994 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
17996 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17998 PR middle-end/19706
17999 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18000 * config/aarch64/aarch64-builtins.c
18001 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18002 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18003 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18005 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18006 Andrew Pinski <pinskia@gmail.com>
18008 PR middle-end/19706
18009 * internal-fn.def (XORSIGN): New.
18010 * optabs.def (xorsign_optab): New.
18011 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18012 (convert_expand_mult_copysign): New.
18013 (pass_optimize_widening_mul::execute): Call
18014 convert_expand_mult_copysign.
18016 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18018 PR tree-optimization/81354
18019 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18020 Insert on edges rather than explicitly creating landing pads.
18021 (analyze_candidates_and_replace): Commit edge inserts.
18023 2017-08-08 Richard Biener <rguenther@suse.de>
18025 PR middle-end/81719
18026 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18027 (expand_simple_operations): Also look through ADDR_EXPRs with
18028 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18030 2017-08-08 Richard Biener <rguenther@suse.de>
18032 PR tree-optimization/81723
18033 * tree-vect-slp.c (struct bst_traits): New hash traits.
18034 (bst_fail): New global.
18035 (vect_build_slp_tree_2): New worker, split out from ...
18036 (vect_build_slp_tree): ... this now wrapping it with using
18037 bst_fail set to cache SLP tree build fails. Properly handle
18039 (vect_analyze_slp_instance): Allocate and free bst_fail.
18041 2017-08-08 Martin Liska <mliska@suse.cz>
18044 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18045 LABEL_DECLs that can be from a different function.
18047 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18049 PR tree-optimization/81744
18050 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18051 loop's number of iterations.
18053 2017-08-08 Martin Liska <mliska@suse.cz>
18055 * asan.c: Include header files.
18056 * attribs.c (build_decl_attribute_variant): New function moved
18058 (build_type_attribute_qual_variant): Likewise.
18059 (cmp_attrib_identifiers): Likewise.
18060 (simple_cst_list_equal): Likewise.
18061 (omp_declare_simd_clauses_equal): Likewise.
18062 (attribute_value_equal): Likewise.
18063 (comp_type_attributes): Likewise.
18064 (build_type_attribute_variant): Likewise.
18065 (lookup_ident_attribute): Likewise.
18066 (remove_attribute): Likewise.
18067 (merge_attributes): Likewise.
18068 (merge_type_attributes): Likewise.
18069 (merge_decl_attributes): Likewise.
18070 (merge_dllimport_decl_attributes): Likewise.
18071 (handle_dll_attribute): Likewise.
18072 (attribute_list_equal): Likewise.
18073 (attribute_list_contained): Likewise.
18074 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18075 (lookup_attribute_by_prefix): Likewise.
18076 * bb-reorder.c: Include header files.
18077 * builtins.c: Likewise.
18078 * calls.c: Likewise.
18079 * cfgexpand.c: Likewise.
18080 * cgraph.c: Likewise.
18081 * cgraphunit.c: Likewise.
18082 * convert.c: Likewise.
18083 * dwarf2out.c: Likewise.
18084 * final.c: Likewise.
18085 * fold-const.c: Likewise.
18086 * function.c: Likewise.
18087 * gimple-expr.c: Likewise.
18088 * gimple-fold.c: Likewise.
18089 * gimple-pretty-print.c: Likewise.
18090 * gimple.c: Likewise.
18091 * gimplify.c: Likewise.
18092 * hsa-common.c: Likewise.
18093 * hsa-gen.c: Likewise.
18094 * internal-fn.c: Likewise.
18095 * ipa-chkp.c: Likewise.
18096 * ipa-cp.c: Likewise.
18097 * ipa-devirt.c: Likewise.
18098 * ipa-fnsummary.c: Likewise.
18099 * ipa-inline.c: Likewise.
18100 * ipa-visibility.c: Likewise.
18102 * lto-cgraph.c: Likewise.
18103 * omp-expand.c: Likewise.
18104 * omp-general.c: Likewise.
18105 * omp-low.c: Likewise.
18106 * omp-offload.c: Likewise.
18107 * omp-simd-clone.c: Likewise.
18108 * opts-global.c: Likewise.
18109 * passes.c: Likewise.
18110 * predict.c: Likewise.
18111 * sancov.c: Likewise.
18112 * sanopt.c: Likewise.
18113 * symtab.c: Likewise.
18114 * toplev.c: Likewise.
18115 * trans-mem.c: Likewise.
18116 * tree-chkp.c: Likewise.
18117 * tree-eh.c: Likewise.
18118 * tree-into-ssa.c: Likewise.
18119 * tree-object-size.c: Likewise.
18120 * tree-parloops.c: Likewise.
18121 * tree-profile.c: Likewise.
18122 * tree-ssa-ccp.c: Likewise.
18123 * tree-ssa-live.c: Likewise.
18124 * tree-ssa-loop.c: Likewise.
18125 * tree-ssa-sccvn.c: Likewise.
18126 * tree-ssa-structalias.c: Likewise.
18127 * tree-ssa.c: Likewise.
18128 * tree-streamer-in.c: Likewise.
18129 * tree-vectorizer.c: Likewise.
18130 * tree-vrp.c: Likewise.
18131 * tsan.c: Likewise.
18132 * ubsan.c: Likewise.
18133 * varasm.c: Likewise.
18134 * varpool.c: Likewise.
18135 * tree.c: Remove functions moved to attribs.[ch].
18136 * tree.h: Likewise.
18137 * config/aarch64/aarch64.c: Add attrs.h header file.
18138 * config/alpha/alpha.c: Likewise.
18139 * config/arc/arc.c: Likewise.
18140 * config/arm/arm.c: Likewise.
18141 * config/avr/avr.c: Likewise.
18142 * config/bfin/bfin.c: Likewise.
18143 * config/c6x/c6x.c: Likewise.
18144 * config/cr16/cr16.c: Likewise.
18145 * config/cris/cris.c: Likewise.
18146 * config/darwin.c: Likewise.
18147 * config/epiphany/epiphany.c: Likewise.
18148 * config/fr30/fr30.c: Likewise.
18149 * config/frv/frv.c: Likewise.
18150 * config/ft32/ft32.c: Likewise.
18151 * config/h8300/h8300.c: Likewise.
18152 * config/i386/winnt.c: Likewise.
18153 * config/ia64/ia64.c: Likewise.
18154 * config/iq2000/iq2000.c: Likewise.
18155 * config/lm32/lm32.c: Likewise.
18156 * config/m32c/m32c.c: Likewise.
18157 * config/m32r/m32r.c: Likewise.
18158 * config/m68k/m68k.c: Likewise.
18159 * config/mcore/mcore.c: Likewise.
18160 * config/microblaze/microblaze.c: Likewise.
18161 * config/mips/mips.c: Likewise.
18162 * config/mmix/mmix.c: Likewise.
18163 * config/mn10300/mn10300.c: Likewise.
18164 * config/moxie/moxie.c: Likewise.
18165 * config/msp430/msp430.c: Likewise.
18166 * config/nds32/nds32-isr.c: Likewise.
18167 * config/nds32/nds32.c: Likewise.
18168 * config/nios2/nios2.c: Likewise.
18169 * config/nvptx/nvptx.c: Likewise.
18170 * config/pa/pa.c: Likewise.
18171 * config/pdp11/pdp11.c: Likewise.
18172 * config/powerpcspe/powerpcspe.c: Likewise.
18173 * config/riscv/riscv.c: Likewise.
18174 * config/rl78/rl78.c: Likewise.
18175 * config/rx/rx.c: Likewise.
18176 * config/s390/s390.c: Likewise.
18177 * config/sh/sh.c: Likewise.
18178 * config/sol2.c: Likewise.
18179 * config/sparc/sparc.c: Likewise.
18180 * config/spu/spu.c: Likewise.
18181 * config/stormy16/stormy16.c: Likewise.
18182 * config/tilegx/tilegx.c: Likewise.
18183 * config/tilepro/tilepro.c: Likewise.
18184 * config/v850/v850.c: Likewise.
18185 * config/vax/vax.c: Likewise.
18186 * config/visium/visium.c: Likewise.
18187 * config/xtensa/xtensa.c: Likewise.
18189 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18192 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18193 constraints since the -mupper-regs-* switches have been
18195 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18196 into a vector from a double word element that was extracted from
18197 another vector, and eliminate extra XXPERMDI instructions.
18198 (vsx_concat_<mode>_2): Likewise.
18199 (vsx_concat_<mode>_3): Likewise.
18200 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18201 concat to allow optimizing inserts from previous extracts.
18203 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18205 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18206 memory reference to a SSP offset in TLS address space.
18207 (ix86_print_operand) <case '@'>: Remove.
18208 (ix86_print_operand_punct_valid_p): Remove '@' code.
18209 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18210 UNSPEC_SP_TLS_TEST.
18211 (stack_tls_protect_set_<mode>): Remove.
18212 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18213 (stack_tls_protect_test_<mode>): Remove.
18214 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18216 2017-08-07 Olivier Hainque <hainque@adacore.com>
18219 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18221 2017-08-07 Douglas Rupp <rupp@adacore.com>
18223 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18224 variable was referenced as multidir in command.
18226 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18229 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18232 2017-08-07 Martin Liska <mliska@suse.cz>
18234 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18235 * config/rl78/rl78.c: Add include of attribs.h.
18236 * config/sh/sh.c: Likewise.
18237 * config/v850/v850.c: Likewise.
18239 2017-08-07 Tom de Vries <tom@codesourcery.com>
18241 PR middle-end/78266
18242 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18244 2017-08-07 Martin Liska <mliska@suse.cz>
18246 * config/mips/mips.c: Include attribs.h.
18248 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18251 * doc/invoke.texi: Document change in behvaior for -Ofast for
18254 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18256 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18258 (aarch64_pop_regs): Likewise.
18259 (aarch64_gen_load_pair): Likewise.
18260 (aarch64_save_callee_saves): Likewise.
18261 (aarch64_restore_callee_saves): Likewise.
18263 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18265 * config/i386/i386.c: Revert the last change.
18267 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18270 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18272 (ix86_finalize_stack_frame_flags): This. Also clear
18273 frame_pointer_needed if -fno-omit-frame-pointer is used without
18275 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18276 with ix86_finalize_stack_frame_flags.
18277 (ix86_expand_epilogue): Likewise.
18278 (ix86_expand_split_stack_prologue): Likewise.
18280 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18283 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18284 to P_AES for Westmere.
18286 2017-08-07 Jonathan Yong <10walls@gmail.com>
18288 * config/i386/mingw.opt (fset-stack-executable): Removed.
18289 * config/i386/cygming.opt (fset-stack-executable): Moved
18291 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18293 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18295 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18297 2017-08-07 Marek Polacek <polacek@redhat.com>
18299 PR middle-end/81737
18300 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18302 2017-08-07 Martin Liska <mliska@suse.cz>
18304 * attribs.h (canonicalize_attr_name): New function.
18305 (cmp_attribs): Move from c-format.c and adjusted.
18306 (is_attribute_p): Moved from tree.h.
18307 * tree-inline.c: Add new includes.
18308 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18309 (private_is_attribute_p): Remove.
18310 (private_lookup_attribute): Likewise.
18311 (private_lookup_attribute_by_prefix): Simplify.
18312 (remove_attribute): Use is_attribute_p.
18313 * tree.h: Remove removed declarations.
18315 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18317 PR middle-end/81698
18318 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18319 instead of computing it in the function. Formatting fix.
18320 (expand_case): Don't rely on default_edge being the first edge,
18321 clear it if removing it, pass default_edge to
18322 emit_case_dispatch_table.
18323 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18326 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18328 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18329 insn in the function, emit NOP after the insn.
18331 2017-08-06 Tom de Vries <tom@codesourcery.com>
18333 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18336 2017-08-06 Tom de Vries <tom@codesourcery.com>
18338 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18341 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18343 PR tree-optimization/57371
18344 * match.pd: New pattern.
18346 2017-08-04 Marek Polacek <polacek@redhat.com>
18348 PR middle-end/81695
18349 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18350 perform the computation in offset_int.
18352 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18354 PR tree-optimization/81136
18355 * tree-vectorizer.h: Include tree-hash-traits.h.
18356 (vec_base_alignments): New typedef.
18357 (vec_info): Add a base_alignments field.
18358 (vect_record_base_alignments): Declare.
18359 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18361 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18362 (create_data_ref): Add an is_conditional_in_stmt argument.
18363 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18364 the is_conditional_in_stmt field.
18365 (data_ref_loc): Add an is_conditional_in_stmt field.
18366 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18367 (find_data_references_in_stmt): Update call to create_data_ref.
18368 (graphite_find_data_references_in_stmt): Likewise.
18369 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18370 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18371 (vect_record_base_alignment): New function.
18372 (vect_record_base_alignments): Likewise.
18373 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18374 for nested statements even if we fail to compute a misalignment.
18375 Use pooled base alignments for unconditional references.
18376 (vect_find_same_alignment_drs): Compare base addresses instead
18378 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18379 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18381 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18383 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18384 Add an explicit name for the enum. Use auto_vec for slp_instances
18385 and grouped_stores.
18386 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18388 (_bb_vec_info): Add a constructor and destructor.
18389 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18390 (destroy_loop_vec_info): Delete.
18391 (vect_destroy_datarefs): Likewise.
18392 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18393 (vec_info::vec_info): New function.
18394 (vec_info::~vec_info): Likewise.
18395 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18396 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18397 destroy_loop_vec_info.
18398 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18399 (_loop_vec_info::_loop_vec_info): ...this.
18400 (destroy_loop_vec_info): Replace with...
18401 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18402 the stmt_vec_infos. Leave handling of vec_info information to its
18403 destructor. Remove explicit vector releases.
18404 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18405 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18406 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18407 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18408 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18409 (destroy_bb_vec_info): Replace with...
18410 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18411 information to its destructor.
18412 (vect_slp_analyze_bb_1): Use new and delete instead of
18413 new_bb_vec_info and destroy_bb_vec_info.
18414 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18417 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18419 * tree-data-ref.h (subscript): Add access_fn field.
18420 (data_dependence_relation): Add could_be_independent_p.
18421 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18422 (same_access_functions): Move to tree-data-ref.c.
18423 * tree-data-ref.c (ref_contains_union_access_p): New function.
18424 (access_fn_component_p): Likewise.
18425 (access_fn_components_comparable_p): Likewise.
18426 (dr_analyze_indices): Add a reference to access_fn_component_p.
18427 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18429 (constant_access_functions): Likewise.
18430 (add_other_self_distances): Likewise.
18431 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18432 (initialize_data_dependence_relation): Use XCNEW and remove
18433 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18434 of access functions that have the same type. Allow the
18435 subsequence to end with different bases in some circumstances.
18436 Record the chosen access functions in SUB_ACCESS_FN.
18437 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18438 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18439 (subscript_dependence_tester_1): Likewise dra and drb.
18440 (build_classic_dist_vector): Update calls accordingly.
18441 (subscript_dependence_tester): Likewise.
18442 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18443 DDR_COULD_BE_INDEPENDENT_P.
18444 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18445 comp_alias_ddrs instead of may_alias_ddrs.
18446 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18448 (vect_analyze_data_ref_dependence): Use it if
18449 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18450 distance vectors if that fails.
18451 (dependence_distance_ge_vf): New function.
18452 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18453 LOOP_VINFO_MAY_ALIAS_DDRS.
18455 2017-08-04 Richard Biener <rguenther@suse.de>
18457 PR middle-end/81705
18458 * fold-const.c (fold_binary_loc): Properly restrict
18459 minus_var0 && minus_var1 case when associating undefined overflow
18462 2017-08-04 Tom de Vries <tom@codesourcery.com>
18464 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18466 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18468 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18469 Don't start diagnostic messages with a capital letter.
18470 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18472 (rs6000_invalid_builtin): Likewise.
18473 (rs6000_trampoline_init): Likewise.
18475 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18478 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18479 after setting changeable df flags.
18481 2017-08-03 Richard Biener <rguenther@suse.de>
18483 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18484 up if the use is in USE - X.
18486 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18488 * toplev.c (dumpfile.h): New include.
18489 (internal_error_reentered): New static function. Use it...
18490 (internal_error_function): ...here to handle reentered internal_error.
18492 2017-08-03 Richard Biener <rguenther@suse.de>
18494 PR middle-end/81148
18495 * fold-const.c (split_tree): Add minus_var and minus_con
18496 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18497 here but always use minus_*.
18498 (associate_trees): Assert we never associate with MINUS_EXPR
18499 and NULL first operand. Do not recurse for PLUS_EXPR operands
18500 when associating as MINUS_EXPR either.
18501 (fold_binary_loc): Track minus_var and minus_con.
18503 2017-08-03 Tom de Vries <tom@codesourcery.com>
18506 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18507 ACCEL_COMPILER, apply finish_options on
18508 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18510 2017-08-03 Tom de Vries <tom@codesourcery.com>
18513 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18514 function_entry_patch_area_size > 0.
18516 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18519 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18520 instead of 10??LU, perform unit multiplication in wide_int,
18521 don't change alloc_object_size_limit if the limit is larger
18524 PR tree-optimization/81655
18525 PR tree-optimization/81588
18526 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18527 the case when ranges[i].low and high are 1 for unsigned type with
18530 PR middle-end/81052
18531 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18532 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18534 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18536 * tree-vrp.h: Add include guard.
18538 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18541 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18542 (ud2): New insn pattern.
18543 * config/i386/i386.c (ix86_expand_epilogue):
18544 For naked functions, generate ud2 instead of trap insn.
18546 2017-08-02 Marek Polacek <polacek@redhat.com>
18549 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18551 2017-08-02 Tom de Vries <tom@codesourcery.com>
18552 Cesar Philippidis <cesar@codesourcery.com>
18554 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18555 Add missing edge probabilities.
18557 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18559 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18560 Correct endianness.
18562 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18564 PR middle-end/79499
18565 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18566 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18567 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18568 prologue_seq sequences - if any.
18570 2017-08-02 Richard Biener <rguenther@suse.de>
18572 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18573 via vectors if supported, integer extracts via punning if supported
18574 or otherwise vector extracts.
18576 2017-08-02 Richard Biener <rguenther@suse.de>
18578 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18580 (bitmap_insert_into_set): ... this.
18582 2017-08-02 Richard Biener <rguenther@suse.de>
18584 PR tree-optimization/81633
18586 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18588 PR tree-optimization/71752
18589 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18591 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18593 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18594 (machine_function::call_ms2sysv_pad_out): Remove field.
18595 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18596 (ix86_compute_frame_layout): Likewise.
18598 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18601 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18602 attribute with interrupt attribute.
18604 2017-08-01 Andrew Pinski <apinski@cavium.com>
18606 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18607 BIT_INSERT_EXPR's operand 1
18608 to see if the types precision matches.
18610 2017-08-01 Martin Liska <mliska@suse.cz>
18612 PR middle-end/70140
18613 * builtins.c (expand_builtin_memcpy_args): Remove.
18614 (expand_builtin_memcpy): Call newly added function
18615 expand_builtin_memory_copy_args.
18616 (expand_builtin_memcpy_with_bounds): Likewise.
18617 (expand_builtin_mempcpy): Remove last argument.
18618 (expand_builtin_mempcpy_with_bounds): Likewise.
18619 (expand_builtin_memory_copy_args): New function created from
18620 expand_builtin_mempcpy_args with small modifications.
18621 (expand_builtin_mempcpy_args): Remove.
18622 (expand_builtin_stpcpy): Remove unused argument.
18623 (expand_builtin): Likewise.
18624 (expand_builtin_with_bounds): Likewise.
18626 2017-08-01 Martin Liska <mliska@suse.cz>
18629 Make mempcpy more optimal (PR middle-end/70140).
18631 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18634 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18635 __builtin_vec_cmpne verify both arguments are compatible vectors
18636 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18637 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18638 move computation of aligned to after checking the argument types.
18642 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18643 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18646 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18647 Doug Rupp <rupp@adacore.com>
18648 Olivier Hainque <hainque@adacore.com>
18650 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18651 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18653 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18654 for TARGET_OS_CPP_BUILTIN.
18655 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18656 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18658 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18659 passing required abi options to the assembler for EABI configurations.
18660 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18661 of .text.hot and .text.unlikely sections for kernel modules when
18662 using ARM style exceptions.
18663 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18664 options. Add EXTRA_CC1_SPEC.
18665 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18667 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18669 (ARM_TARGET2_DWARF_FORMAT): Define.
18670 * config/arm/t-vxworks: Adjust multilib control to removal of the
18671 Diab command line options.
18673 2017-08-01 Martin Liska <mliska@suse.cz>
18675 PR gcov-profile/81561
18676 * gcov.c (unblock): Make unblocking safe as we need to preserve
18677 index correspondence of blocks and block_lists.
18679 2017-08-01 Richard Biener <rguenther@suse.de>
18681 PR tree-optimization/81181
18682 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18683 (compute_antic): ... end of iteration here.
18685 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18687 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18688 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18689 (ftree-slp-vectorize): Likewise.
18690 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18691 can no longer be set independent of flag_tree_loop_vectorize.
18692 * omp-general.c (emp_max_vf): Likewise.
18693 * opts.c (enable_fdo_optimizations): Remove references to
18694 flag_tree_vectorize, these are now implicit.
18695 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18696 and leave it for the options machinery.
18698 2017-08-01 Martin Liska <mliska@suse.cz>
18700 PR middle-end/70140
18701 * builtins.c (expand_builtin_memcpy_args): Remove.
18702 (expand_builtin_memcpy): Call newly added function
18703 expand_builtin_memory_copy_args.
18704 (expand_builtin_memcpy_with_bounds): Likewise.
18705 (expand_builtin_mempcpy): Remove last argument.
18706 (expand_builtin_mempcpy_with_bounds): Likewise.
18707 (expand_builtin_memory_copy_args): New function created from
18708 expand_builtin_mempcpy_args with small modifications.
18709 (expand_builtin_mempcpy_args): Remove.
18710 (expand_builtin_stpcpy): Remove unused argument.
18711 (expand_builtin): Likewise.
18712 (expand_builtin_with_bounds): Likewise.
18714 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18717 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18718 print "ds:" only for immediates in generic address space.
18720 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18723 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18724 (ix86_function_ok_for_sibcall): Return false for naked functions.
18726 2017-08-01 Richard Biener <rguenther@suse.de>
18728 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18729 (compute_antic): Seed worklist with exit block predecessors.
18730 * cfganal.c (dfs_find_deadend): For a cycle return the source
18731 of the edge closing it.
18733 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18735 * config/aarch64/aarch64.c
18736 (aarch64_can_const_movi_rtx_p): Move 0 check.
18738 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18740 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18741 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18743 * match.pd: Ditto in address comparison pattern.
18745 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18747 PR tree-optimization/81627
18748 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18749 closed ssa form for store-store chain.
18751 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18753 PR tree-optimization/81620
18754 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18755 for store-store chain.
18757 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18759 PR tree-optimization/81588
18760 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18761 ranges[i].in_p, invert comparison code ccode. For >/>=,
18762 swap rhs1 and rhs2 and comparison code unconditionally,
18763 for </<= don't do that. Don't swap rhs1/rhs2 again if
18764 ranges[i].in_p, instead invert comparison code ccode if
18765 opcode or oe->rank is BIT_IOR_EXPR.
18768 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18769 a direct optab to conversion optab.
18770 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18771 with GET_MODE_INNER as last argument instead of optab_handler.
18772 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18773 vector extraction if possible and optab is available.
18774 * expr.c (store_constructor): Use convert_optab_handler instead
18775 of optab_handler. Use vector initialization from smaller
18776 vectors if possible and optab is available.
18777 * tree-vect-stmts.c (vectorizable_load): Likewise.
18778 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18779 now have two modes.
18780 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18781 of vec_init from half-sized vectors with the same element mode.
18782 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18783 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18784 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18785 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18786 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18787 after mode in gen_vec_extract* calls.
18788 (vec_extract<mode>): Renamed to ...
18789 (vec_extract<mode><ssescalarmodelower>): ... this.
18790 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18791 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18792 element mode after mode in gen_vec_init* calls.
18793 (VEC_INIT_HALF_MODE): New mode iterator.
18794 (vec_init<mode>): Renamed to ...
18795 (vec_init<mode><ssescalarmodelower>): ... this.
18796 (vec_init<mode><ssehalfvecmodelower>): New expander.
18797 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18798 (vec_extractv2sfsf): ... this.
18799 (vec_initv2sf): Renamed to ...
18800 (vec_initv2sfsf): ... this.
18801 (vec_extractv2si): Renamed to ...
18802 (vec_extractv2sisi): ... this.
18803 (vec_initv2si): Renamed to ...
18804 (vec_initv2sisi): ... this.
18805 (vec_extractv4hi): Renamed to ...
18806 (vec_extractv4hihi): ... this.
18807 (vec_initv4hi): Renamed to ...
18808 (vec_initv4hihi): ... this.
18809 (vec_extractv8qi): Renamed to ...
18810 (vec_extractv8qiqi): ... this.
18811 (vec_initv8qi): Renamed to ...
18812 (vec_initv8qiqi): ... this.
18813 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18814 (vec_init<mode>): Renamed to ...
18815 (vec_init<mode><VEC_base_l>): ... this.
18816 (vec_extract<mode>): Renamed to ...
18817 (vec_extract<mode><VEC_base_l>): ... this.
18818 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18819 (vec_initv2sfsf): ... this.
18820 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18821 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18822 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18823 element mode after mode in gen_vec_init* calls.
18824 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18825 (vec_init<mode><Vel>): ... this.
18826 (vec_extract<mode>): Renamed to ...
18827 (vec_extract<mode><Vel>): ... this.
18828 * config/aarch64/iterators.md (Vel): New mode attribute.
18829 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18830 Add element mode after mode in gen_vec_extract* calls.
18831 * config/s390/vector.md (non_vec_l): New mode attribute.
18832 (vec_extract<mode>): Renamed to ...
18833 (vec_extract<mode><non_vec_l>): ... this.
18834 (vec_init<mode>): Renamed to ...
18835 (vec_init<mode><non_vec_l>): ... this.
18836 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18837 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18839 * config/arm/iterators.md (V_elem_l): New mode attribute.
18840 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18841 (vec_extract<mode><V_elem_l>): ... this.
18842 (vec_extractv2di): Renamed to ...
18843 (vec_extractv2didi): ... this.
18844 (vec_init<mode>): Renamed to ...
18845 (vec_init<mode><V_elem_l>): ... this.
18846 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18847 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18848 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18849 Add element mode after gen_vec_extract* calls.
18850 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18851 (vec_init<mode><unitmode>): ... this.
18852 (vec_extract<mode>): Renamed to ...
18853 (vec_extract<mode><unitmode>): ... this.
18854 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18855 (vec_init<mode><unitmode>): ... this.
18856 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18857 (vec_initv2sfsf): ... this.
18858 (vec_extractv2sf): Renamed to ...
18859 (vec_extractv2sfsf): ... this.
18860 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18861 Add element mode after gen_vec_extract* calls.
18862 * config/mips/mips.md (unitmode): New mode iterator.
18863 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18864 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18865 * config/spu/spu.md (inner_l): New mode attribute.
18866 (vec_init<mode>): Renamed to ...
18867 (vec_init<mode><inner_l>): ... this.
18868 (vec_extract<mode>): Renamed to ...
18869 (vec_extract<mode><inner_l>): ... this.
18870 * config/sparc/sparc.md (veltmode): New mode iterator.
18871 (vec_init<VMALL:mode>): Renamed to ...
18872 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18873 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18874 (vec_initv2sisi): ... this.
18875 (vec_initv2sf): Renamed to ...
18876 (vec_initv2sfsf): ... this.
18877 (vec_extractv2sf): Renamed to ...
18878 (vec_extractv2sfsf): ... this.
18879 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18880 (vec_init<mode>): Renamed to ...
18881 (vec_init<mode><VEC_base_l>): ... this.
18882 (vec_extract<mode>): Renamed to ...
18883 (vec_extract<mode><VEC_base_l>): ... this.
18884 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18885 (vec_initv2sfsf): ... this.
18886 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18887 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18888 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18889 gen_vec_init* calls.
18891 2017-08-01 Richard Biener <rguenther@suse.de>
18893 PR tree-optimization/81297
18894 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18895 TREE_OVERFLOW from INTEGER_CSTs.
18897 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18899 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18901 2017-07-31 Carl Love <cel@us.ibm.com>
18903 * config/rs6000/rs6000-c: Add support for built-in functions
18904 vector signed char vec_xl_be (signed long long, signed char *);
18905 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18906 vector signed int vec_xl_be (signed long long, signed int *);
18907 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18908 vector signed long long vec_xl_be (signed long long, signed long long *);
18909 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18910 vector signed short vec_xl_be (signed long long, signed short *);
18911 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18912 vector double vec_xl_be (signed long long, double *);
18913 vector float vec_xl_be (signed long long, float *);
18914 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18915 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18916 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18918 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18919 (altivec_expand_builtin): Add switch statement to call
18920 altivec_expand_xl_be for each builtin.
18921 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
18922 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
18923 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
18924 __builtin_vsx_le_be_v16qi.
18925 * doc/extend.texi: Update the built-in documentation file for the
18926 new built-in functions.
18928 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18931 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
18933 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18935 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18937 * config.gcc: Add z14.
18938 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
18939 CPU model numbers for z13s and z14.
18940 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
18942 * config/s390/s390-opts.h (enum processor_type): Rename
18943 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18944 * config/s390/s390.c (processor_table): Add field for CPU name to
18945 be passed to Binutils.
18946 (s390_asm_output_machine_for_arch): Use the new field in
18947 processor_table for Binutils.
18948 (s390_expand_builtin): Replace arch12 with z14.
18949 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18950 (s390_get_sched_attrmask): Likewise.
18951 (s390_get_unit_mask): Likewise.
18952 * config/s390/s390.opt: Add z14 to processor_type enum.
18954 2017-07-31 Martin Jambor <mjambor@suse.cz>
18957 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
18958 regardless of optimization level.
18960 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
18961 Martin Liska <mliska@suse.cz>
18963 * predict.def: Remove old comment and adjust probability.
18964 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
18965 PREDICT statements.
18967 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18970 * config/i386/i386.c (ix86_function_naked): New function.
18971 (ix86_can_use_return_insn_p): Return false for naked functions.
18972 (ix86_expand_prologue): Skip prologue for naked functions.
18973 (ix86_expand_epilogue): Skip epilogue for naked functions
18974 and emit trap instruction.
18975 (ix86_warn_func_return): New function.
18976 (ix86_attribute_table): Add "naked" attribute specification.
18977 (TARGET_WARN_FUNC_RETURN): Define.
18978 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
18980 2017-07-31 Martin Liska <mliska@suse.cz>
18982 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
18983 (dump_gimple_bb_header): Always dump BB info.
18984 (pp_cfg_jump): Do not append info about BB when dumping a jump.
18986 2017-07-31 Martin Liska <mliska@suse.cz>
18989 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
18990 also with current_function_decl non-null equality.
18992 2017-07-31 Jakub Jelinek <jakub@redhat.com>
18995 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
18996 change type to the element type, instead add eltype variable and
18997 use it where we are interested in the element type.
18999 PR tree-optimization/81603
19000 * ipa-polymorphic-call.c
19001 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19002 offset arithmetic in offset_int, bail out if the resulting bit offset
19003 doesn't fit into shwi.
19005 2017-07-31 Martin Liska <mliska@suse.cz>
19007 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19008 (gimplify_save_expr): Fix comment.
19010 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19013 * config/i386/i386.c (ix86_function_arg): Update arguments for
19015 (ix86_compute_frame_layout): Set the initial stack offset to
19016 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19017 INCOMING_FRAME_SP_OFFSET.
19018 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19019 stack before exception handler returns.
19020 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19021 the 'ERROR_CODE' for exception handler.
19023 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19025 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19026 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19027 (ASM_OUTPUT_REG_POP): Ditto.
19028 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19029 instead of asm_fprintf to output pure string.
19031 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19033 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19034 to imported_module_or_decl hook.
19035 (debug_nothing_tree_tree_tree_bool): Remove.
19036 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19037 * debug.c (do_nothing_debug_hooks): Use
19038 debug_nothing_tree_tree_tree_bool_bool instead of
19039 debug_nothing_tree_tree_tree_bool.
19040 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19041 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19042 * sdbout.c (sdb_debug_hooks): Likewise.
19043 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19044 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19046 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19047 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19048 attribute, don't add anything.
19050 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19052 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19053 (fold_build2_stat_loc): Likewise.
19054 (fold_build3_stat_loc): Likewise.
19055 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19056 (fold_build1_loc): Remove macro.
19057 (fold_build2_loc): Likewise.
19058 (fold_build3_loc): Likewise.
19060 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19062 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19063 (gimple_build_debug_bind_source_stat): Likewise.
19064 * gimple.h (gimple_build_debug_bind): Remove macro.
19065 (gimple_build_debug_bind_source): Likewise.
19067 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19069 * bitmap.c (bitmap_alloc): Adjust.
19070 (bitmap_gc_alloc): Likewise.
19071 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19073 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19075 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19076 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19077 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19078 (bitmap_gc_alloc_stat): Likewise.
19079 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19081 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19083 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19084 * rtl.h (shallow_copy_rtx): Remove macro.
19086 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19088 * emit-rtl.c (gen_raw_REG): Adjust.
19089 * gengenrtl.c (gendef): Likewise.
19090 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19091 * rtl.h (rtx_alloc): Remove macro.
19093 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19095 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19096 (build_tree_list_stat): Likewise.
19097 * tree.h (build_tree_list): Remove macro.
19098 (build_tree_list_vec): Likewise.
19100 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19102 * tree.c (make_vector_stat): Remove _stat from name.
19103 (build_vector_stat): Likewise.
19104 * tree.h (make_vector_stat): Remove macro.
19105 (build_vector_stat): Likewise.
19107 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19109 * tree.h (build_var_debug_value): Remove prototype.
19111 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19113 * tree.c (tree_cons_stat): Remove _stat from name.
19114 * tree.h (tree_cons): Remove macro.
19116 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19118 * tree.c (build_vl_exp_stat): Remove _stat from name.
19119 * tree.h (build_vl_exp): Remove macro.
19121 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19123 * tree.c (build_decl_stat): Remove _stat from name.
19124 * tree.h (build_decl): Remove macro.
19126 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19128 * gimple.c (gimple_build_with_ops_stat): Adjust.
19129 (gimple_alloc_stat): Remove _stat from name.
19130 * gimple.h (gimple_alloc): Remove macro.
19132 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19134 * tree.c (make_tree_vec_stat): Remove _stat from name.
19135 (grow_tree_vec_stat): Likewise.
19136 * tree.h (make_tree_vec_stat): Adjust prototype.
19137 (grow_tree_vec_stat): Likewise.
19138 (make_tree_vec): Remove macro.
19139 (grow_tree_vec): Likewise.
19141 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19143 * fold-const.c (fold_build1_stat_loc): Adjust.
19144 (fold_build2_stat_loc): Likewise.
19145 (fold_build3_stat_loc): Likewise.
19146 * tree.c (build0_stat): Remove _stat from name.
19147 (build1_stat): Likewise.
19148 (build2_stat): Likewise.
19149 (build3_stat): Likewise.
19150 (build4_stat): Likewise.
19151 (build5_stat): Likewise.
19152 * tree.h (build1_loc): Remove macro, and rename _stat function
19154 (build2_loc): Likewise.
19155 (build3_loc): Likewise.
19156 (build4_loc): Likewise.
19157 (build5_loc): Likewise.
19159 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19161 * tree.c (make_int_cst_stat): Remove _stat from name.
19162 * tree.h (make_int_cst_stat): Adjust prototype.
19163 (make_int_cst): Remove macro.
19165 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19167 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19168 * tree.h (make_tree_binfo_stat): Adjust prototype.
19169 (make_tree_binfo): Remove.
19171 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19173 * tree.c (copy_node_stat): Rename to copy_node.
19174 (build_distinct_type_copy): Adjust.
19175 * tree.h (copy_node_stat): Adjust prototype.
19176 (copy_node): Remove macro.
19178 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19180 * tree.c (make_node_stat): rename to make_node.
19181 (build_tree_list_stat): Adjust.
19182 (build0_stat): Likewise.
19183 (build2_stat): Likewise.
19184 (build3_stat): Likewise.
19185 (build4_stat): Likewise.
19186 (build5_stat): Likewise.
19187 (build_decl_stat): Likewise.
19188 * tree.h (make_node_stat): Adjust prototype.
19189 (make_node): remove macro.
19191 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19193 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19194 (PPC_FEATURE2_SCV): Likewise.
19195 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19197 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19199 * config/aarch64/aarch64.c
19200 (aarch64_internal_mov_immediate): Add new special pattern.
19201 * config/aarch64/aarch64.md (*movdi_aarch64):
19202 Add reg/32bit const mov case.
19204 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19205 Richard Sandiford <richard.sandiford@linaro.org>
19207 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19208 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19209 Add integer and movi cases.
19210 (movi-split-hf-df-sf split, fp16): New.
19211 (enabled): Added TARGET_FP_F16INST.
19212 * config/aarch64/iterators.md (GPF_HF): New.
19213 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19215 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19217 * config/aarch64/aarch64.c
19218 (aarch64_simd_container_mode): Add prototype.
19219 (aarch64_expand_mov_immediate): Add HI support.
19220 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19221 (aarch64_can_const_movi_rtx_p): New.
19222 (aarch64_preferred_reload_class):
19223 Remove restrictions of using FP registers for certain SIMD operations.
19224 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19225 (aarch64_valid_floating_const): Add integer move validation.
19226 (aarch64_simd_imm_scalar_p): Remove.
19227 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19228 (aarch64_legitimate_constant_p): Expand list of supported cases.
19229 * config/aarch64/aarch64-protos.h
19230 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19231 (aarch64_reinterpret_float_as_int): New.
19232 (aarch64_simd_imm_scalar_p): Remove.
19233 * config/aarch64/constraints.md (Uvi): New.
19234 (Dd): Split into Ds and new Dd.
19235 * config/aarch64/aarch64.md (*movsi_aarch64):
19237 (*movdi_aarch64): Add SIMD mov case.
19239 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19241 * tree-predcom.c: (struct chain): Handle store-store chain in which
19242 stores for elimination only store loop invariant values.
19243 (execute_pred_commoning_chain): Ditto.
19244 (prepare_initializers_chain_store_elim): Ditto.
19245 (prepare_finalizers): Ditto.
19246 (is_inv_store_elimination_chain): New function.
19247 (initialize_root_vars_store_elim_1): New function.
19249 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19251 * tree-predcom.c: Revise general description of the pass.
19252 (enum chain_type): New enum type for store elimination.
19253 (struct chain): New field supporting store elimination.
19254 (struct component): Ditto.
19255 (dump_chain): Dump store-stores chain.
19256 (release_chain): Release resources.
19257 (split_data_refs_to_components): Compute and create component
19258 contains only stores for elimination.
19259 (get_chain_last_ref_at): New function.
19260 (make_invariant_chain): Initialization.
19261 (make_rooted_chain): Specify chain type in parameter and record it.
19262 (add_looparound_copies): Skip for store-stores chain.
19263 (determine_roots_comp): Compute type of chain and pass it to
19265 (initialize_root_vars_store_elim_2): New function.
19266 (finalize_eliminated_stores): New function.
19267 (remove_stmt): Handle store for elimination.
19268 (execute_pred_commoning_chain): Execute predictive commoning on
19269 store-store chains.
19270 (determine_unroll_factor): Skip unroll for store-stores chain.
19271 (prepare_initializers_chain_store_elim): New function.
19272 (prepare_initializers_chain): Hanlde store-store chain.
19273 (prepare_finalizers_chain, prepare_finalizers): New function.
19274 (tree_predictive_commoning_loop): Return integer value indicating
19275 if loop is unrolled or lcssa form is corrupted.
19276 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19278 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19280 * tree-predcom.c (initialize_root): Delete.
19281 (execute_pred_commoning_chain): Initialize root vars and replace
19282 reference of non-combined chain directly, rather than call above
19285 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19287 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19288 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19290 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19292 * tree-predcom.c (struct chain): New field init_seq.
19293 (release_chain): Release init_seq.
19294 (prepare_initializers_chain): Record intialization stmts in above
19296 (insert_init_seqs): New function.
19297 (tree_predictive_commoning_loop): Call insert_init_seqs.
19299 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19301 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19303 2017-07-28 Richard Biener <rguenther@suse.de>
19305 * match.pd: Remove superfluous :c.
19306 * genmatch.c (simplify::id): Add member.
19307 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19309 (current_id): New global.
19310 (dt_node::parent): Move from ...
19311 (dt_operand::parent): ... here. Add for_id member.
19312 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19313 (decision_tree::find_node): Relax order requirement when
19314 merging DT_TRUE nodes to ones inbetween the current simplify
19315 and the one we try to merge with. Add diagnostic whenever
19316 we need to enforce pattern order by not merging.
19317 (decision_tree::insert): Set current_id.
19318 (decision_tree::print_node): Dump parent node and for_id.
19319 (parser::last_id): Add member.
19320 (parser::push_simplify): Assign unique id.
19321 (parser::parser): Initialize last_id.
19323 2017-07-28 Martin Liska <mliska@suse.cz>
19326 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19327 gimple_build_debug_bind.
19329 2017-07-28 Richard Biener <rguenther@suse.de>
19331 PR tree-optimization/81502
19332 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19334 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19336 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19337 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19338 for BIT_FIELD_REF args.
19339 * fold-const.c (make_bit_field_ref): Likewise.
19340 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19342 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19345 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19346 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19347 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19348 Or it into SANITIZER_UNDEFINED.
19349 * ubsan.c: Include gimple-fold.h and varasm.h.
19350 (ubsan_expand_ptr_ifn): New function.
19351 (instrument_pointer_overflow): New function.
19352 (maybe_instrument_pointer_overflow): New function.
19353 (instrument_object_size): Formatting fix.
19354 (pass_ubsan::execute): Call instrument_pointer_overflow
19355 and maybe_instrument_pointer_overflow.
19356 * internal-fn.c (expand_UBSAN_PTR): New function.
19357 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19358 * sanitizer.def (__ubsan_handle_pointer_overflow,
19359 __ubsan_handle_pointer_overflow_abort): New builtins.
19360 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19361 * internal-fn.def (UBSAN_PTR): New internal function.
19362 * opts.c (sanitizer_opts): Add pointer-overflow.
19363 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19364 * fold-const.c (build_range_check): Compute pointer range check in
19365 integral type if pointer arithmetics would be needed. Formatting
19368 2017-07-28 Martin Liska <mliska@suse.cz>
19371 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19372 parameters that are of a variable-length.
19374 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19376 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19378 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19379 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19380 (CRT_CALL_STATIC_FUNCTION): Likewise.
19381 (ASM_DEFAULT_SPEC): New define.
19382 (ASM_SPEC32): Likewise.
19383 (ASM_SPEC64): Likewise.
19384 (ASM_SPEC_COMMON): Likewise.
19385 (ASM_SPEC): Likewise.
19386 (INVALID_64BIT): Likewise.
19387 (LINK_OS_DEFAULT_SPEC): Likewise.
19388 (LINK_OS_SPEC32): Likewise.
19389 (LINK_OS_SPEC64): Likewise.
19390 (POWERPC_LINUX): Likewise.
19391 (PTRDIFF_TYPE): Likewise.
19392 (RESTORE_FP_PREFIX): Likewise.
19393 (RESTORE_FP_SUFFIX): Likewise.
19394 (SAVE_FP_PREFIX): Likewise.
19395 (SAVE_FP_SUFFIX): Likewise.
19396 (SIZE_TYPE): Likewise.
19397 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19398 (TARGET_64BIT): Likewise.
19399 (TARGET_64BIT): Likewise.
19400 (TARGET_AIX): Likewise.
19401 (WCHAR_TYPE_SIZE): Likewise.
19402 (WCHAR_TYPE): Undefine.
19403 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19404 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19405 (CPP_OS_RTEMS_SPEC): Delete.
19406 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19407 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19409 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19411 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19413 PR tree-optimization/81578
19414 * tree-parloops.c (build_new_reduction): Bail out if
19415 reduction_code isn't one of the standard OpenMP reductions.
19416 Move the details printing after that decision.
19418 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19420 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19421 related to reload_in_progress.
19422 (splat_input_operand): Likewise.
19423 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19425 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19427 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19428 (TARGET_INSTANTIATE_DECLS): Likewise.
19429 (legitimate_indexed_address_p): Delete reload_in_progress code.
19430 (rs6000_debug_legitimate_address_p): Likewise.
19431 (rs6000_eliminate_indexed_memrefs): Likewise.
19432 (rs6000_emit_le_vsx_store): Likewise.
19433 (rs6000_emit_move_si_sf_subreg): Likewise.
19434 (rs6000_emit_move): Likewise.
19435 (register_to_reg_type): Likewise.
19436 (rs6000_pre_atomic_barrier): Likewise.
19437 (rs6000_machopic_legitimize_pic_address): Likewise.
19438 (rs6000_allocate_stack_temp): Likewise.
19439 (rs6000_address_for_fpconvert): Likewise.
19440 (rs6000_address_for_altivec): Likewise.
19441 (rs6000_secondary_memory_needed_rtx): Delete function.
19442 (rs6000_check_sdmode): Likewise.
19443 (rs6000_alloc_sdmode_stack_slot): Likewise.
19444 (rs6000_instantiate_decls): Likewise.
19445 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19446 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19447 Delete reload_in_progress.
19448 (*vec_reload_and_plus_<mptrsize>): Likewise.
19449 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19450 (vsx_div_v2di): Likewise.
19451 (vsx_udiv_v2di): Likewise.
19453 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19455 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19456 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19457 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19458 (rs6000_debug_reg_global): Delete print of LRA status.
19459 (rs6000_option_override_internal): Delete dead LRA related code.
19460 (rs6000_lra_p): Delete function.
19461 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19463 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19465 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19466 * config/riscv/rtems.h: New file.
19468 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19469 Sudakshina Das <sudi.das@arm.com>
19471 * config/aarch64/aarch64.md
19472 (define_split for and<mode>3nr_compare): Move
19473 non aarch64_logical_operand to a register.
19474 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19475 register immediate operand to a register.
19476 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19478 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19480 PR middle-end/81564
19481 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19483 2017-07-27 Richard Biener <rguenther@suse.de>
19485 PR tree-optimization/81573
19486 PR tree-optimization/81494
19487 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19488 multi defuse cycle case.
19490 2017-07-27 Richard Biener <rguenther@suse.de>
19492 PR tree-optimization/81571
19493 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19496 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19498 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19499 earlier and only if MASK_FPU is set. Adjust formatting.
19501 2017-07-27 Martin Liska <mliska@suse.cz>
19503 * opt-functions.awk: Add validation of value of Init.
19504 * optc-gen.awk: Pass new argument.
19506 2017-07-27 Martin Liska <mliska@suse.cz>
19508 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19509 Fix wrong condition.
19511 2017-07-27 Martin Liska <mliska@suse.cz>
19513 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19514 BBs and edges seen by autoFDO.
19516 2017-07-27 Richard Biener <rguenther@suse.de>
19518 PR tree-optimization/81502
19519 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19520 with incompatible but same sized type.
19521 (execute_update_addresses_taken): Likewise.
19523 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19525 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19526 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19528 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19531 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19532 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19533 Change s_operand to memory_operand.
19535 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19537 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19538 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19539 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19540 Emit instructions rather than returning an expression. Handle TFmode
19541 and KFmode by casting to TImode.
19542 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19543 (rs6000_emit_le_vsx_store): Likewise.
19544 * config/rs6000/vsx.md (VSX_TI): New iterator.
19545 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19546 (*vsx_le_undo_permute_<mode>): Likewise.
19547 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19548 emit the split sequence.
19549 (*vsx_le_perm_store_<mode>): Likewise.
19551 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19553 PR tree-optimization/81555
19554 PR tree-optimization/81556
19555 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19556 if true, force CHANGED for the recursive invocation.
19557 (reassociate_bb): Remember original length of ops array, pass
19558 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19560 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19561 attributes for noipa attribute. For naked attribute use
19562 lookup_attribute first before lookup_attribute_spec.
19563 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19565 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19566 for functions with noipa attribute.
19567 (cgraph_externally_visible_p): Return true for functions with noipa
19569 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19570 for functions with noipa attribute.
19571 * doc/extend.texi: Document noipa function attribute.
19572 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19573 also for functions with noipa attribute.
19574 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19576 2017-07-26 Andrew Pinski <apinski@cavium.com>
19578 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19579 vec_unalign_load_cost and vec_unalign_store_cost.
19581 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19583 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19584 -mvsx-small-integer option.
19585 (ISA_3_0_MASKS_IEEE): Likewise.
19586 (OTHER_VSX_VECTOR_MASKS): Likewise.
19587 (POWERPC_MASKS): Likewise.
19588 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19589 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19590 code, only testing for DImode being allowed in non-VSX floating
19592 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19593 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19595 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19596 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19597 TARGET_P8_VECTOR test.
19598 (rs6000_secondary_reload_simple_move): Likewise.
19599 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19600 since TARGET_P9_VECTOR was already tested.
19601 (rs6000_opt_masks): Remove -mvsx-small-integer.
19602 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19603 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19605 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19606 test for TARGET_VEXTRACTUB was used, and that uses
19608 (p9 extract splitter): Likewise.
19609 (vsx_extract_<mode>_di_p9): Likewise.
19610 (vsx_extract_<mode>_store_p9): Likewise.
19611 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19612 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19613 the elimination of TARGET_VSX_SMALL_INTEGER.
19614 (vsx_extract_<mode>_p8): Likewise.
19615 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19616 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19617 (vsx_set_<mode>_p9): Likewise.
19618 (vsx_set_v4sf_p9): Likewise.
19619 (vsx_set_v4sf_p9_zero): Likewise.
19620 (vsx_insert_extract_v4sf_p9): Likewise.
19621 (vsx_insert_extract_v4sf_p9_2): Likewise.
19622 * config/rs6000/rs6000.md (sign extend splitter): Change
19623 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19624 (floatsi<mode>2_lfiwax_mem): Likewise.
19625 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19626 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19627 since a test for TARGET_P9_VECTOR was used.
19628 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19629 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19630 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19631 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19632 TARGET_P8_VECTOR test.
19633 (fix_trunc<mode>si2_stfiwx): Likewise.
19634 (fix_trunc<mode>si2_internal): Likewise.
19635 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19636 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19638 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19639 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19640 TARGET_P8_VECTOR test.
19641 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19642 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19643 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19645 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19646 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19647 since a test for TARGET_P9_VECTOR was used.
19648 (splitter for loading small constants): Likewise.
19650 2017-07-26 Andrew Pinski <apinski@cavium.com>
19652 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19655 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19658 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19659 (fp_valid_at): Likewise.
19661 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19663 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19664 (qdf24xx_addrcost_table): Likewise.
19665 (cortexa57_tunings): Update to use generic_branch_cost.
19666 (cortexa72_tunings): Likewise.
19667 (cortexa73_tunings): Likewise.
19668 (qdf24xx_tunings): Likewise.
19670 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19672 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19673 (thunderx2t99_branch_cost): Likewise.
19674 (cortexa35_tunings): Update to use generic_branch_cost.
19675 (cortexa53_tunings): Likewise.
19676 (cortexa57_tunings): Likewise.
19677 (cortexa72_tunings): Likewise.
19678 (cortexa73_tunings): Likewise.
19679 (thunderx2t99_tunings): Likewise.
19681 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19683 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19684 (sparc_option_override): Honour MASK_FSMULD.
19685 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19686 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19687 * config/sparc/sparc.opt (mfsmuld): New option.
19688 * doc/invoke.texi (mfsmuld): Document option.
19690 2017-07-26 Marek Polacek <polacek@redhat.com>
19692 PR middle-end/70992
19693 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19695 2017-07-26 Richard Biener <rguenther@suse.de>
19697 * gimple-match-head.c (do_valueize): Return OP if valueize
19699 (get_def): New helper to get at the def stmt of a SSA name
19700 if valueize allows.
19701 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19702 do_valueize to get at the def stmt.
19703 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19705 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19707 PR middle-end/46932
19708 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19710 2017-07-26 Martin Liska <mliska@suse.cz>
19713 * function.c (expand_function_start): Make expansion of
19714 nonlocal_goto_save_area after parm_birth_insn.
19716 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19718 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19719 from all CPU target flags enable members.
19721 2017-07-26 Richard Biener <rguenther@suse.de>
19723 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19724 (decision_tree::gen): Make 'type' const.
19725 (write_predicate): Likewise.
19727 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19729 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19730 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19731 (rs6000_option_override_internal): Likewise.
19732 (rs6000_expand_vector_set): Likewise.
19733 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19734 (TARGET_UPPER_REGS_SF): Likewise.
19735 (TARGET_UPPER_REGS_DI): Likewise.
19736 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19737 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19738 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19739 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19740 (Splitters for DI constants in Altivec registers): Likewise.
19741 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19742 (vsx_set_v4sf_p9): Likewise.
19743 (vsx_set_v4sf_p9_zero): Likewise.
19744 (vsx_insert_extract_v4sf_p9): Likewise.
19745 (vsx_insert_extract_v4sf_p9_2): Likewise.
19747 2017-07-25 Carl Love <cel@us.ibm.com>
19749 * doc/extend.texi: Update the built-in documentation file for the
19750 existing built-in functions
19751 vector signed char vec_cnttz (vector signed char);
19752 vector unsigned char vec_cnttz (vector unsigned char);
19753 vector signed short vec_cnttz (vector signed short);
19754 vector unsigned short vec_cnttz (vector unsigned short);
19755 vector signed int vec_cnttz (vector signed int);
19756 vector unsigned int vec_cnttz (vector unsigned int);
19757 vector signed long long vec_cnttz (vector signed long long);
19758 vector unsigned long long vec_cnttz (vector unsigned long long);
19760 2017-07-25 Andrew Pinski <apinski@cavium.com>
19762 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19763 accesses where the use is for the first operand of a BIT_INSERT.
19765 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19768 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19769 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19771 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19773 * config/i386/gstabs.h: Delete.
19774 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19776 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19778 * config/i386/i386.c (ix86_decompose_address): Do not check for
19779 register RTX when looking at index_reg or base_reg.
19780 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19782 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19784 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19785 to update EH info here.
19787 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19789 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19791 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19793 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19795 2017-07-25 Torsten Duwe <duwe@suse.de>
19797 * common.opt: Introduce -fpatchable-function-entry
19798 command line option, and its variables function_entry_patch_area_size
19799 and function_entry_patch_area_start.
19800 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19801 including a two-value parser.
19802 * target.def (print_patchable_function_entry): New target hook.
19803 * targhooks.h (default_print_patchable_function_entry): New function.
19804 * targhooks.c (default_print_patchable_function_entry): Likewise.
19805 * toplev.c (process_options): Switch off IPA-RA if
19806 patchable function entries are being generated.
19807 * varasm.c (assemble_start_function): Look at the
19808 patchable-function-entry command line switch and current
19809 function attributes and maybe generate NOP instructions by
19810 calling the print_patchable_function_entry hook.
19811 * doc/extend.texi: Document patchable_function_entry attribute.
19812 * doc/invoke.texi: Document -fpatchable_function_entry
19813 command line option.
19814 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19816 * doc/tm.texi: Re-generate.
19818 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19821 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19822 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19824 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19826 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19828 2017-07-25 Richard Biener <rguenther@suse.de>
19830 PR tree-optimization/81455
19831 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19832 not walk in cycles when looking for guards.
19834 2017-07-25 Richard Biener <rguenther@suse.de>
19836 PR tree-optimization/81529
19837 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19838 when optimizing backedge uses.
19840 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19842 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19844 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19845 to dl_section_ref. On AIX, append an expression to subtract
19846 the size of the section length to dl_section_ref.
19848 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19850 * configure.ac: If any of the config.* scripts fail, exit 1.
19851 * configure: Regenerate.
19853 2017-07-25 Richard Biener <rguenther@suse.de>
19855 PR middle-end/81546
19856 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19857 of immediate uses, be more verbose on errors.
19859 2017-07-25 Richard Biener <rguenther@suse.de>
19861 PR tree-optimization/81510
19862 * tree-vect-loop.c (vect_is_simple_reduction): When the
19863 reduction stmt is not inside the loop bail out.
19865 2017-07-25 Richard Biener <rguenther@suse.de>
19867 PR tree-optimization/81303
19868 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19869 profitability check against LOOP_VINFO_NITERSM1.
19871 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19873 * domwalk.c (cmp_bb_postorder): Simplify.
19874 (sort_bbs_postorder): New function. Use it...
19875 (dom_walker::walk): ...here to optimize common cases.
19877 2017-07-25 Martin Liska <mliska@suse.cz>
19880 * ipa-visibility.c (function_and_variable_visibility): Make the
19881 redirection just on target that supports aliasing.
19882 Fix GNU coding style.
19884 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19887 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19889 * config/arm/rtems.h: Likewise.
19890 * config/bfin/rtems.h: Likewise.
19891 * config/i386/rtemself.h: Likewise.
19892 * config/lm32/rtems.h: Likewise.
19893 * config/m32c/rtems.h: Likewise.
19894 * config/m68k/rtemself.h: Likewise.
19895 * config/microblaze/rtems.h: Likewise.
19896 * config/mips/rtems.h: Likewise.
19897 * config/moxie/rtems.h: Likewise.
19898 * config/nios2/rtems.h: Likewise.
19899 * config/powerpcspe/rtems.h: Likewise.
19900 * config/rs6000/rtems.h: Likewise.
19901 * config/rtems.h: Likewise.
19902 * config/sh/rtems.h: Likewise.
19903 * config/sh/rtemself.h: Likewise.
19904 * config/sparc/rtemself.h: Likewise.
19906 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19909 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19910 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19911 * tree-ssa-structalias.c (alias_get_name): Same.
19913 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19916 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19917 instructions if no du chain is found.
19919 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19921 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
19923 2017-07-25 Richard Biener <rguenther@suse.de>
19925 PR middle-end/81505
19926 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
19929 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19931 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
19932 upper-regs options.
19933 (ISA_2_7_MASKS_SERVER): Likewise.
19934 (ISA_3_0_MASKS_IEEE): Likewise.
19935 (OTHER_P8_VECTOR_MASKS): Likewise.
19936 (OTHER_VSX_VECTOR_MASKS): Likewise.
19937 (POWERPC_MASKS): Likewise.
19938 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
19939 duplicate list of options.
19940 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
19941 explicit -mupper-regs options.
19942 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
19943 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
19944 alias for -mupper-regs-df.
19945 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19946 (rs6000_init_hard_regno_mode_ok): Likewise.
19947 (rs6000_option_override_internal): Likewise.
19948 (rs6000_opt_masks): Likewise.
19949 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
19950 options in terms of whether -mvsx or -mpower8-vector was used.
19951 (TARGET_UPPER_REGS_DI): Likewise.
19952 (TARGET_UPPER_REGS_SF): Likewise.
19953 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
19954 -mupper-regs-* options.
19956 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19958 * passes.c (emergency_dump_function): Print some empty lines and a
19959 header before the RTL dump.
19961 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19963 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
19965 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
19968 * config/aarch64/aarch64.c (aarch64_classify_symbol):
19969 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
19971 2017-07-24 Carl Love <cel@us.ibm.com>
19973 * config/rs6000/rs6000-c.c: Add support for built-in functions
19974 vector float vec_extract_fp32_from_shorth (vector unsigned short);
19975 vector float vec_extract_fp32_from_shortl (vector unsigned short);
19976 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
19977 vec_extract_fp_from_shortl): Add defines for the two builtins.
19978 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
19979 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
19981 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
19982 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
19983 * doc/extend.texi: Update the built-in documentation file for the
19984 new built-in function.
19986 2017-07-24 Jakub Jelinek <jakub@redhat.com>
19989 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
19991 * doc/generic.texi: Likewise.
19992 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19993 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19995 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
19997 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
19998 (aarch64_mls_elt_merge<mode>): Likewise.
20000 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20002 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20003 having __cxa_atexit.
20005 2017-07-23 Michael Collison <michael.collison@arm.com>
20007 * config/arm/arm.c (arm_option_override): Deprecate
20008 use of -mstructure-size-boundary.
20009 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20010 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20012 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20015 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20016 Reduce cost estimate for direct moves.
20018 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20021 * config/i386/i386.c (ix86_option_override_internal): Disable
20022 BMI, BMI2 and TBM instructions for -m16.
20024 2017-07-21 Carl Love <cel@us.ibm.com>
20026 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20027 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20028 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20029 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20030 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20031 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20032 VMULOSW): New enum "unspec" values.
20033 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20034 altivec_vmulosw): New patterns.
20035 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20036 VMULOSW): Add definitions.
20038 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20040 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20041 (qdf24xx): Likewise.
20042 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20043 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20044 (AARCH64_FL_V8_1): Renumber.
20045 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20046 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20047 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20048 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20049 rdma to feature modifiers list.
20051 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20053 PR middle-end/56727
20054 * ipa-visibility (function_and_variable_visibility): Convert
20055 recursive PLT call to direct call if appropriate.
20057 2017-07-21 Andrew Pinski <apinski@cavium.com>
20059 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20060 operand 1 to see if the types precision matches.
20061 * fold-const.c (operand_equal_p): Likewise.
20063 2017-07-21 Richard Biener <rguenther@suse.de>
20065 PR tree-optimization/81303
20066 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20067 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20068 (vect_peeling_hash_get_lowest_cost): Adjust.
20069 (vect_enhance_data_refs_alignment): Likewise. Use
20070 vect_get_peeling_costs_all_drs to compute the penalty for no
20071 peeling to match up costs.
20073 2017-07-21 Richard Biener <rguenther@suse.de>
20075 PR tree-optimization/81500
20076 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20077 we didn't identify a reduction path.
20079 2017-07-21 Tom de Vries <tom@codesourcery.com>
20080 Cesar Philippidis <cesar@codesourcery.com>
20082 PR gcov-profile/81442
20083 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20086 2017-07-21 Tom de Vries <tom@codesourcery.com>
20089 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20091 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20092 nvptx_override_options_after_change.
20094 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20096 * dwarf2out.c (output_file_names): Avoid double testing for
20097 dwarf_version >= 5.
20099 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20101 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20103 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20105 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20107 (try_crossjump_to_edge): Do not punt on partitioned functions.
20109 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20111 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20112 Put all BBs reachable only via paths crossing cold region to cold
20114 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20116 2016-07-21 Richard Biener <rguenther@suse.de>
20118 PR tree-optimization/81303
20119 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20120 into account prologue and epilogue iterations when raising
20121 min_profitable_iters to sth at least covering one vector iteration.
20123 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20125 * config/arm/arm.c (arm_test_cpu_arch_dat):
20128 2017-07-20 Nathan Sidwell <nathan@acm.org>
20130 Remove TYPE_METHODS.
20131 * tree.h (TYPE_METHODS): Delete.
20132 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20133 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20134 (dbxout_type_methods): Scan TYPE_FIELDS.
20135 (dbxout_type): Don't check TYPE_METHODS here.
20136 * function.c (use_register_for_decl): Always ignore register for
20137 class types when not optimizing.
20138 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20139 * tree.c (free_lang_data_in_type): Stitch out member functions and
20140 templates from TYPE_FIELDS.
20141 (build_distinct_type_copy, verify_type_variant,
20142 verify_type): Member fns are on TYPE_FIELDS.
20143 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20144 * tree-pretty-print.c (dump_generic_node): Likewise.
20146 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20149 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20150 V2TImode and V4TImode.
20151 (ix86_expand_vector_extract): Likewise.
20152 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20153 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20154 (ssescalarmode): Handle V4TImode and V2TImode.
20155 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20156 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20157 (VEXTRACTI128_MODE): New mode iterator.
20158 (splitter for *vec_extractv?ti first element): New.
20159 (VEC_INIT_MODE): New mode iterator.
20160 (vec_init<mode>): Consolidate 3 expanders into one using
20161 VEC_INIT_MODE mode iterator.
20163 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20165 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20166 non_spilled_static_chain_regno_p.
20168 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20170 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20172 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20174 * bb-reorder.c (connect_traces): Allow copying of blocks within
20177 2017-07-20 Richard Biener <rguenther@suse.de>
20179 * gimple.h (gimple_phi_result): Add gphi * overload.
20180 (gimple_phi_result_ptr): Likewise.
20181 (gimple_phi_arg): Likewise. Adjust index assert to only
20182 allow actual argument accesses rather than all slots available
20184 (gimple_phi_arg_def): Add gphi * overload.
20185 * tree-phinodes.c (make_phi_node): Initialize only actual
20187 (resize_phi_node): Clear memory not covered by old node,
20188 do not initialize excess argument slots.
20189 (reserve_phi_args_for_new_edge): Initialize new argument slot
20192 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20194 PR tree-optimization/81388
20196 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20198 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20199 by removing computation of may_be_zero.
20201 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20202 Tom de Vries <tom@codesourcery.com>
20204 PR middle-end/81030
20205 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20206 when gimple level profile disagrees with what RTL expander did.
20208 2017-07-20 Richard Biener <rguenther@suse.de>
20210 PR tree-optimization/61171
20211 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20212 (vect_analyze_stmt): Add slp instance parameter.
20213 (vectorizable_reduction): Likewise.
20214 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20215 (vect_is_simple_reduction): Deal with chains not detected
20216 as SLP reduction chain, specifically not properly associated
20217 chains containing a mix of plus/minus.
20218 (get_reduction_op): Remove.
20219 (get_initial_defs_for_reduction): Simplify, pass in whether
20220 this is a reduction chain, pass in the SLP node for the PHIs.
20221 (vect_create_epilog_for_reduction): Get the SLP instance as
20223 (vectorizable_reduction): Get the SLP instance as arg.
20224 During analysis remember the SLP node with the PHIs in the
20225 instance. Simplify getting at the vectorized reduction PHIs.
20226 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20227 through SLP instance.
20228 (vect_slp_analyze_operations): Likewise.
20229 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20230 (vect_transform_stmt): Likewise.
20232 2017-07-20 Tom de Vries <tom@codesourcery.com>
20234 PR tree-optimization/81489
20235 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20236 read of phi arg location to before loop that modifies phi.
20238 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20240 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20243 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20245 PR middle-end/81331
20246 * except.c (execute): Fix ordering issue.
20248 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20250 PR rtl-optimization/81423
20251 * combine.c (make_compound_operation_int): Don't try to optimize
20252 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20254 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20256 PR rtl-optimization/81423
20257 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20258 with a constant that is -1 in the truncated to mode.
20260 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20262 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20263 (determine_unlikely_bbs): ... here.
20264 * predict.h (propagate_unlikely_bbs_forward): Declare.
20265 * cfgexpand.c (pass_expand::execute): Use it.
20266 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20268 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20269 propagate_unlikely_bbs_forward.
20271 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20273 PR middle-end/81331
20274 * except.c (maybe_add_nop_after_section_switch): New function.
20277 2017-07-19 Tom de Vries <tom@codesourcery.com>
20279 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20281 2017-07-19 Tom de Vries <tom@codesourcery.com>
20283 * gimple.h (gimple_phi_arg): Make assert more strict.
20285 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20287 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20288 * config/rs6000/mmintrin.h: New file.
20289 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20291 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20293 PR tree-optimization/81346
20294 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20296 2017-07-19 Tom de Vries <tom@codesourcery.com>
20298 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20300 2017-07-19 Tom de Vries <tom@codesourcery.com>
20302 * config/nvptx/nvptx-modes.def: Add V2DImode.
20303 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20304 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20305 (nvptx_output_mov_insn): Handle lack of mov.b128.
20306 (nvptx_print_operand): Handle 'H' and 'L' codes.
20307 (nvptx_vector_mode_supported): Allow V2DImode.
20308 (nvptx_preferred_simd_mode): New function.
20309 (nvptx_data_alignment): New function.
20310 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20311 nvptx_preferred_simd_mode.
20312 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20314 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20316 2017-07-19 Tom de Vries <tom@codesourcery.com>
20318 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20319 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20320 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20321 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20322 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20323 (mov<VECIM>_insn): New define_insn.
20324 (define_expand "mov<VECIM>): New define_expand.
20326 2017-07-19 Tom de Vries <tom@codesourcery.com>
20328 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20330 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20332 PR tree-optimization/81346
20333 * fold-const.h (fold_div_compare, range_check_type): Declare.
20334 * fold-const.c (range_check_type): New function.
20335 (build_range_check): Use range_check_type.
20336 (fold_div_compare): No longer static, rewritten into
20337 a match.pd helper function.
20338 (fold_comparison): Don't call fold_div_compare here.
20339 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20340 as helper function.
20342 2017-07-19 Nathan Sidwell <nathan@acm.org>
20344 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20345 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20346 * tree.c (find_decls_types_r, verify_type): Use
20347 TYPE_{MIN,MAX}_VALUE_RAW.
20348 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20349 (hash_tree): Likewise.
20350 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20352 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20355 2017-07-18 Tom de Vries <tom@codesourcery.com>
20357 PR middle-end/81464
20358 * omp-expand.c (expand_omp_for_static_chunk): Handle
20359 equal-argument loop exit phi.
20361 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20364 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20365 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20366 operand 2 predicate.
20367 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20368 operand 2 predicate.
20369 (ror,rol -> rorx splitters): Use const_int_operand as
20370 operand 2 predicate.
20372 2017-06-18 Richard Biener <rguenther@suse.de>
20374 PR tree-optimization/81410
20375 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20376 the gap in the ! slp_perm SLP case after each group.
20378 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20380 PR middle-end/81463
20381 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20384 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20386 PR middle-end/81462
20387 * predict.c (set_even_probabilities): Cleanup; do not affect
20388 probabilities that are already known.
20389 (combine_predictions_for_bb): Call even when count is set.
20391 2017-07-18 Nathan Sidwell <nathan@acm.org>
20393 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20396 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20399 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20400 optimization for loop niter analysis.
20402 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20405 * config/avr/avr.c (avr_optimize_casesi): Don't use
20406 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20408 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20410 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20411 body_cost_vec from _vect_peel_extended_info.
20412 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20413 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20416 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20418 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20420 2017-07-18 Richard Biener <rguenther@suse.de>
20422 PR tree-optimization/80620
20423 PR tree-optimization/81403
20424 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20425 info when re-using a VN table entry.
20427 2017-07-18 Richard Biener <rguenther@suse.de>
20429 PR tree-optimization/81418
20430 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20431 vectype_in. Verify that with lane-reducing reduction operations
20432 we have a single def-use cycle.
20434 2017-07-17 Carl Love <cel@us.ibm.com>
20436 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20438 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20439 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20440 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20441 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20442 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20443 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20444 VMULOSW): New enum "unspec" values.
20445 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20446 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20447 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20448 altivec_vmulosw): New patterns.
20449 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20450 VMULOSW): Add definitions.
20452 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20454 * config/alpha/alpha.c: Include predict.h.
20456 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20458 * tree-vrp.c (compare_assert_loc): Fix comparison function
20459 to return predictable results.
20461 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20463 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20465 (subdi3): Likewise.
20466 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20467 * doc/invoke.texi (mexpand-adddi): Update text.
20469 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20471 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20472 that also clobbers the CC register. The old expand code is moved
20474 (*arc_clzsi2): ... here.
20475 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20476 the CC register. The old expand code is moved to ...
20477 (arc_ctzsi2): ... here.
20479 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20481 * config/arc/arc.opt (mindexed-loads): Use initial value
20482 TARGET_INDEXED_LOADS_DEFAULT.
20483 (mauto-modify-reg): Use initial value
20484 TARGET_AUTO_MODIFY_REG_DEFAULT.
20485 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20486 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20487 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20488 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20490 2017-07-17 Martin Liska <mliska@suse.cz>
20493 * opts.c (finish_options): Do not allow -fgnu-tm
20494 w/ -fsanitize={kernel-,}address. Say sorry.
20496 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20499 * tree-loop-distribution.c (classify_partition): Only assert on
20500 numer of iterations.
20501 (merge_dep_scc_partitions): Delete prameter. Update function call.
20502 (distribute_loop): Remove code handling loop with unknown niters.
20503 (pass_loop_distribution::execute): Skip loop with unknown niters.
20505 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20508 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20509 function sort_partitions_by_post_order.
20511 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20513 PR tree-optimization/81374
20514 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20515 the max index of basic blocks, rather than number of basic blocks.
20517 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20519 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20521 (arc_legitimate_pic_operand_p): Likewise.
20522 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20524 (arc_needs_pcl_p): Likewise.
20525 (arc_legitimate_pc_offset_p): Likewise.
20526 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20527 function is also used in constrains.md.
20528 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20529 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20530 PLUS. Only return true/false in known cases, otherwise assert.
20531 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20532 is already called in arc_legitimate_constant_p.
20533 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20535 (LEGITIMATE_PIC_OPERAND_P): Use
20536 arc_raw_symbolic_reference_mentioned_p function.
20537 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20542 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20543 Andrew Burgess <andrew.burgess@embecosm.com>
20545 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20546 (arc_return_address_register): New function.
20547 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20548 (arc_handle_fndecl_attribute): Add naked attribute.
20549 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20550 (TARGET_WARN_FUNC_RETURN): Likewise.
20551 (arc_allocate_stack_slots_for_args): New function.
20552 (arc_warn_func_return): Likewise.
20553 (machine_function): Change type fn_type.
20554 (arc_compute_function_type): Consider new naked function type,
20555 change function return type.
20556 (arc_must_save_register): Adapt to handle new
20557 arc_compute_function_type's return type.
20558 (arc_expand_prologue): Likewise.
20559 (arc_expand_epilogue): Likewise.
20560 (arc_return_address_regs): Delete.
20561 (arc_return_address_register): New function.
20562 (arc_epilogue_uses): Use above function.
20563 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20564 (arc_function_type): Change encoding, add naked type.
20565 (ARC_INTERRUPT_P): Change to handle the new encoding.
20566 (ARC_FAST_INTERRUPT_P): Likewise.
20567 (ARC_NORMAL_P): Define.
20568 (ARC_NAKED_P): Likewise.
20569 (arc_compute_function_type): Delete prototype.
20570 * config/arc/arc.md (in_ret_delay_slot): Use
20571 arc_return_address_register function.
20572 (simple_return): Likewise.
20573 (p_return_i): Likewise.
20575 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20577 PR tree-optimization/81428
20578 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20579 can't be built for those types.
20581 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20583 Remove stuff dead since r239246.
20585 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20586 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20587 (avr_inform_devices): Remove dead stuff.
20589 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20591 * config/arm/arm_neon.h: Fix softp typo.
20593 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20595 PR tree-optimization/81365
20596 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20597 aggregate moves onto bb predecessor edges, make sure there are no
20598 loads that could alias the lhs in between the start of bb and the
20601 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20604 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20605 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20606 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20608 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20610 PR tree-optimization/81396
20611 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20612 (init_symbolic_number): Initialize it to 1.
20613 (perform_symbolic_merge): Add n_ops from both operands into the new
20615 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20616 without base_addr as useless if they need more than one operation.
20617 (bswap_replace): Handle !bswap case for NULL base_addr.
20619 2017-07-17 Tom de Vries <tom@codesourcery.com>
20622 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20625 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20627 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20628 conditional builtin define __FIX_LEON3FT_B2BST.
20630 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20632 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20633 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20636 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20638 PR rtl-optimization/81424
20639 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20640 to remove potential trapping from operands if -fnon-call-exceptions.
20642 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20644 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20645 profile_proability for scalling.
20646 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20648 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20650 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20652 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20654 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20655 fixpoint arithmetics.
20657 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20659 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20660 fixpoint arithmetics.
20662 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20664 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20665 fixpoint arithmetics.
20667 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20669 * profile-count.h (profile_probability::from_reg_br_prob_note,
20670 profile_probability::to_reg_br_prob_note): New functions.
20671 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20672 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20673 * predict.c (probability_reliable_p): Update.
20674 (edge_probability_reliable_p): Update.
20675 (br_prob_note_reliable_p): Update.
20676 (invert_br_probabilities): Update.
20677 (add_reg_br_prob_note): New function.
20678 (combine_predictions_for_insn): Update.
20679 * asan.c (asan_clear_shadow): Update.
20680 * cfgbuild.c (compute_outgoing_frequencies): Update.
20681 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20682 (update_br_prob_note): Update.
20683 (rtl_verify_edges): Update.
20684 (purge_dead_edges): Update.
20685 (fixup_reorder_chain): Update.
20686 * emit-rtl.c (try_split): Update.
20687 * ifcvt.c (cond_exec_process_insns): Update.
20688 (cond_exec_process_if_block): Update.
20689 (dead_or_predicable): Update.
20690 * internal-fn.c (expand_addsub_overflow): Update.
20691 (expand_neg_overflow): Update.
20692 (expand_mul_overflow): Update.
20693 * loop-doloop.c (doloop_modify): Update.
20694 * loop-unroll.c (compare_and_jump_seq): Update.
20695 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20696 * predict.h: Update.
20697 * reorg.c (mostly_true_jump): Update.
20699 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20700 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20701 * config/arc/arc.c: (emit_unlikely_jump): Update.
20702 * config/arm/arm.c: (emit_unlikely_jump): Update.
20703 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20704 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20705 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20706 (ix86_print_operand): Update.
20707 (ix86_split_fp_branch): Update.
20708 (predict_jump): Update.
20709 * config/ia64/ia64.c (ia64_print_operand): Update.
20710 * config/mmix/mmix.c (mmix_print_operand): Update.
20711 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20712 (rs6000_expand_split_stack_prologue): Update.
20713 * config/rs6000/rs6000.c: Update.
20714 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20715 (s390_expand_vec_movstr): Update.
20716 (s390_expand_cs_tdsi): Update.
20717 (s390_expand_split_stack_prologue): Update.
20718 * config/sh/sh.c (sh_print_operand): Update.
20719 (expand_cbranchsi4): Update.
20720 (expand_cbranchdi4): Update.
20721 * config/sparc/sparc.c (output_v9branch): Update.
20722 * config/spu/spu.c (get_branch_target): Update.
20723 (ea_load_store_inline): Update.
20724 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20725 * config/tilepro/tilepro.c: Update.
20727 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20729 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20730 (gimplify_save_expr): Likewise.
20732 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20734 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20736 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20738 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20740 * timevar.def (TV_IPA_FNSUMMARY): Define.
20742 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20744 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20745 to back store errata sensitive sequence from being generated.
20746 (sqrtdf2_fix): Likewise.
20748 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20750 * tree-ssa-threadupdate.c (compute_path_counts,
20751 update_joiner_offpath_counts): Use profile_probability.
20753 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20756 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20758 * config/arm/arm-c.c (arm_cpu_builtins): Define
20759 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20761 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20763 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20764 array entries to represent __ieee128 versions of the
20765 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20766 scalar_extract_sig, and scalar_insert_exp built-in functions.
20767 (altivec_resolve_overloaded_builtin): Add special case handling
20768 for the __builtin_scalar_insert_exp function, as represented by
20769 the P9V_BUILTIN_VEC_VSIEDP constant.
20770 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20771 exponent support for __ieee128 argument.
20772 (VSESQP): Add scalar extract signature support for __ieee128
20774 (VSTDCNQP): Add scalar test negative support for __ieee128
20776 (VSIEQP): Add scalar insert exponent support for __int128 argument
20777 with __ieee128 result.
20778 (VSIEQPF): Add scalar insert exponent support for __ieee128
20779 argument with __ieee128 result.
20780 (VSTDCQP): Add scalar test data class support for __ieee128
20782 (VSTDCNQP): Add overload support for scalar test negative with
20783 __ieee128 argument.
20784 (VSTDCQP): Add overload support for scalar test data class
20785 __ieee128 argument.
20786 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20787 UNSPEC_VSX_SXSIGDP.
20788 (UNSPEC_VSX_SIEXPQP): New constant.
20789 (xsxexpqp): New insn for VSX scalar extract exponent quad
20791 (xsxsigqp): New insn for VSX scalar extract significand quad
20793 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20794 precision with floating point argument.
20795 (xststdcqp): New expand for VSX scalar test data class quad
20797 (xststdcnegqp): New expand for VSX scalar test negative quad
20799 (xststdcqp): New insn to match expansions for VSX scalar test data
20800 class quad precision and VSX scalar test negative quad precision.
20801 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20802 special case operand checking to enforce that second operand of
20803 VSX scalar test data class with quad precision argument is a 7-bit
20805 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20806 prototypes and descriptions of __ieee128 versions of
20807 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20808 scalar_test_data_class, and scalar_test_neg built-in functions.
20810 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20812 PR tree-optimization/81162
20813 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20814 replace a negate with an add.
20816 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20818 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20820 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20822 * config/arm/arm-c.c (arm_cpu_builtins): Define
20823 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20825 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20827 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20828 (armv8-r): Set ARM Cortex-R52 as default CPU.
20829 * config/arm/arm-tables.opt: Regenerate.
20830 * config/arm/arm-tune.md: Regenerate.
20831 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20833 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20834 extension for -mcpu=cortex-r52.
20836 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20838 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20839 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20840 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20841 (fp-armv8): Define it as FP_ARMv8 only.
20842 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20843 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20845 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20846 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20847 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20848 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20849 than TARGET_FPU_ARMV8.
20850 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20851 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20852 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20854 * config/arm/neon.md (neon_vrint): Likewise.
20855 (neon_vcvt): Likewise.
20856 (neon_<fmaxmin_op><mode>): Likewise.
20857 (<fmaxmin><mode>3): Likewise.
20858 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20859 * config/arm/predicates.md (arm_cond_move_operator): Check against
20860 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20862 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20864 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20865 to top of function.
20867 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20869 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20870 loop in comment with memset.
20872 2017-07-14 Martin Liska <mliska@suse.cz>
20874 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20875 * dwarf2out.c (is_java): Remove the function.
20876 (output_pubname): Remove usage of the function.
20877 (lower_bound_default): Remove usage of DW_LANG_Java.
20878 (gen_compile_unit_die): Likewise.
20879 * gcc.c: Remove compiler defaults for .java and .zip files.
20880 * gimple-expr.c (remove_suffix): Change as there's no longer
20881 extension than 4-letter one.
20882 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20883 (gimplify_save_expr): Likewise.
20884 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20885 as it's possible even for other languages than Java.
20886 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20887 * lto-opts.c (lto_write_options): Remove reference to Java.
20888 * opts.c (strip_off_ending): Update file extension handling.
20889 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20890 * tree-eh.c (lower_resx): Likewise.
20891 * tree.c (free_lang_data_in_type): Remove dead code.
20892 (find_decls_types_r): Likewise.
20893 (build_common_builtin_nodes): Remove Java from a comment.
20894 (verify_type): Remove dead code.
20895 * varasm.c (assemble_external): Remove Java from a comment.
20897 2017-07-14 Martin Liska <mliska@suse.cz>
20899 * opts.c (finish_options): Add quotes.
20900 (common_handle_option): Likewise.
20902 2017-07-14 Martin Liska <mliska@suse.cz>
20904 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20905 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20906 Remove N_SO_PASCAL.
20907 * dwarf2out.c (lower_bound_default): Do not handle
20909 (gen_compile_unit_die): Likewise.
20910 * gcc.c: Remove default extension binding for GNU Pascal.
20911 * stmt.c: Remove Pascal language from a comment.
20912 * xcoffout.c: Likewise.
20914 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20917 * diagnostic-show-locus.c (fixit_cmp): New function.
20918 (layout::layout): Sort m_fixit_hints.
20919 (column_range::column_range): Assert that the values are valid.
20920 (struct char_span): New struct.
20921 (correction::overwrite): New method.
20922 (struct source_line): New struct.
20923 (line_corrections::add_hint): Add assertions. Reimplement memcpy
20924 calls in terms of classes source_line and char_span, and
20925 correction::overwrite.
20926 (selftest::test_overlapped_fixit_printing_2): New function.
20927 (selftest::diagnostic_show_locus_c_tests): Call it.
20929 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
20931 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
20932 early if there is no lhs.
20934 2017-07-13 Martin Liska <mliska@suse.cz>
20936 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
20937 (gen_reference_type_die): Likewise.
20938 * stor-layout.c: Remove Pascal-related comment.
20940 2017-07-13 Martin Liska <mliska@suse.cz>
20942 * opts.c (finish_options): Add quotes to error messages.
20943 (parse_sanitizer_options): Likewise.
20945 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
20947 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
20949 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
20951 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
20953 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
20955 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
20957 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
20959 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
20962 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
20963 provides the hardware capability bits, define the macro
20964 __BUILTIN_CPU_SUPPORTS__.
20965 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
20966 if GLIBC does not provide the hardware capability bits. Add a
20967 gcc_unreachable call if the built-in cpu function is neither
20968 __builtin_cpu_is nor __builtin_cpu_supports.
20969 (rs6000_get_function_versions_dispatcher): Change the warning
20970 that an old GLIBC is used which does not export the capability
20971 bits to be an error.
20972 * doc/extend.texi (target_clones attribute): Document the
20973 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
20974 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
20975 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
20976 the macros defined by GCC if the newer GLIBC is available.
20978 2017-07-12 Jeff Law <law@redhat.com>
20980 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
20981 remaining includes slightly.
20982 * config/riscv/riscv-builtins.c: Include profile-count.h.
20984 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20987 * config/avr/avr.c (avr_set_current_function): In diagnostic
20988 messages: Quote keywords and (parts of) identifiers.
20989 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
20992 2017-07-12 Carl Love <cel@us.ibm.com>
20994 * config/rs6000/rs6000-c.c: Add support for built-in functions
20995 vector bool char vec_revb (vector bool char);
20996 vector bool short vec_revb (vector short char);
20997 vector bool int vec_revb (vector bool int);
20998 vector bool long long vec_revb (vector bool long long);
20999 * doc/extend.texi: Update the built-in documentation file for the
21000 new built-in functions.
21002 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21004 * config/s390/s390.md: Remove movcc splitter.
21006 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21008 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21009 load/store on condition.
21011 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21014 * config/avr/avr.c (avr_encode_section_info)
21015 [progmem && !TREE_READONLY]: Error if progmem object needs
21018 2017-07-11 Michael Collison <michael.collison@arm.com>
21020 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21023 2017-07-11 Carl Love <cel@us.ibm.com>
21025 * config/rs6000/rs6000-c.c: Add support for builtins
21026 vector unsigned int vec_parity_lsbb (vector signed int);
21027 vector unsigned int vec_parity_lsbb (vector unsigned int);
21028 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21029 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21030 vector unsigned long long vec_parity_lsbb (vector signed long long);
21031 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21032 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21033 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21034 * doc/extend.texi: Update the built-in documentation file for the
21035 new built-in functions.
21037 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21039 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21040 (layout::m_primary_loc): New field.
21041 (layout::layout): Initialize new field. Move location filtering
21042 logic from here to...
21043 (layout::maybe_add_location_range): ...this new method. Add
21044 support for filtering to just the lines already specified by other
21046 (layout::will_show_line_p): New method.
21047 (gcc_rich_location::add_location_if_nearby): New method.
21048 (selftest::test_add_location_if_nearby): New test function.
21049 (selftest::diagnostic_show_locus_c_tests): Call it.
21050 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21053 2017-07-11 Tom de Vries <tom@codesourcery.com>
21055 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21056 (bb_first_real_insn): New function.
21057 (nvptx_single): Add extra initialization of broadcasted condition
21060 2017-07-11 Nathan Sidwell <nathan@acm.org>
21062 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21064 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21066 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21067 Binutils doc as TEXI will mess them up.
21068 * doc/invoke.texi (AVR Options): Same here.
21070 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21072 * config/sparc/sparc.opt (mfix-ut700): New option.
21073 (mfix-gr712rc): Likewise.
21074 (sparc_fix_b2bst): New variable.
21075 * doc/invoke.texi (SPARC options): Document them.
21076 (ARM options): Fix warnings.
21077 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21078 instructions to prevent sequences that can trigger the store-store
21079 errata for certain LEON3FT processors.
21080 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21081 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21082 * config/sparc/sparc.md (fix_b2bst): New attribute.
21083 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21085 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21088 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21090 (*rsqrtsf2_sse): Ditto.
21092 (div<mode>3): Macroize insn from divdf3 and divsf3
21093 using MODEF mode iterator.
21095 2017-07-10 Martin Sebor <msebor@redhat.com>
21097 PR tree-optimization/80397
21098 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21099 instead of testing for equality to INTEGER_TYPE.
21101 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21103 * config.gcc: Remove uclibc from arc target spec.
21105 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21107 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21109 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21112 * lto-wrapper.c (remove_option): New function.
21113 (merge_and_complain): Merge PIC/PIE options more realistically.
21115 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21117 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21121 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21122 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21123 * config.in: Regenerate.
21124 * configure: Regenerate.
21125 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21126 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21127 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21128 (TARGET_GASISR_PROLOGUES): ...target mask.
21129 * common/config/avr/avr-common.c
21130 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21131 Set -mgas-isr-prologues.
21132 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21133 INSERT_PASS_BEFORE for it.
21134 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21135 * config/avr/avr.c (avr_option_override)
21136 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21137 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21138 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21139 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21140 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21142 (make_avr_pass_pre_proep): New function.
21143 (emit_push_sfr) <treg>: Add argument to function and use it
21144 instead of TMP_REG.
21145 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21146 and set machine->gasisr.yes.
21147 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21148 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21149 __gcc_isr.n_pushed to .L__stack_usage.
21150 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21151 (avr_asm_final_postscan_insn): ...this new static function.
21152 * config/avr/avr.h (machine_function)
21153 <is_no_gccisr, use_L__stack_usage>: New fields.
21154 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21155 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21156 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21157 (gasisr, *gasisr): New expander and insn.
21158 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21159 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21160 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21162 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21164 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21167 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21169 Move jump-tables out of .text again.
21172 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21173 (ASM_OUTPUT_ADDR_VEC): New function.
21174 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21175 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21176 INSN_ADDRESSes as asm comment.
21177 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21178 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21179 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21180 * config/avr/avr.md (*tablejump): Adjust comment.
21181 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21182 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21184 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21185 (avr_output_addr_vec): New proto.
21186 (avr_log_t) <insn_addresses>: New field.
21188 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21191 * config/i386/i386.c (ix86_function_arg_advance): Set
21192 outgoing_args_on_stack to true if there are outgoing arguments
21194 (ix86_function_arg): Likewise.
21195 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21196 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21197 * config/i386/i386.h (machine_function): Add
21198 outgoing_args_on_stack.
21200 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21202 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21203 supporting pthreds.
21204 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21206 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21208 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21209 (REAL_H): Remove $(MACHMODE_H).
21210 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21212 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21213 $(MACHMODE_H) and double-int.h.
21214 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21216 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21219 2017-07-07 Andrew Pinski <apinski@cavium.com>
21221 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21222 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21224 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21226 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21227 Add warning if GCC was not configured to link against a GLIBC that
21228 exports the hardware capability bits.
21229 (make_resolver_func): Make resolver function private and not a
21230 COMDAT function. Create the name with clone_function_name instead
21231 of make_unique_name.
21234 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21235 correct operand in doing the split.
21237 2017-07-07 Carl Love <cel@us.ibm.com>
21239 * config/rs6000/rs6000-c: Add support for built-in function
21240 vector unsigned short vec_pack_to_short_fp32 (vector float,
21242 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21243 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21244 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21245 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21246 (convert_4f32_8i16): Add define_expand.
21247 * doc/extend.texi: Update the built-in documentation file for the
21248 new built-in function.
21250 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21252 * config/sparc/m8.md: New file.
21253 * config/sparc/sparc.md: Include m8.md.
21255 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21257 * config/sparc/sparc.opt: New option -mvis4b.
21258 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21259 (sparc_option_override): Handle VIS4B.
21260 (enum sparc_builtins): Define
21261 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21262 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21263 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21264 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21265 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21266 (check_constant_argument): New function.
21267 (sparc_vis_init_builtins): Define builtins
21268 __builtin_vis_dictunpack{8,16,32},
21269 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21270 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21271 __builtin_vis_fpcmpde{8,16,32}shl and
21272 __builtin_vis_fpcmpur{8,16,32}shl.
21273 (sparc_expand_builtin): Check that the constant operands to
21274 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21275 constant and in range.
21276 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21278 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21279 (SPARC_IMM5_P): Likewise.
21280 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21281 (enabled): Handle vis4b.
21282 (UNSPEC_DICTUNPACK): New unspec.
21283 (UNSPEC_FPCMPSHL): Likewise.
21284 (UNSPEC_FPUCMPSHL): Likewise.
21285 (UNSPEC_FPCMPDESHL): Likewise.
21286 (UNSPEC_FPCMPURSHL): Likewise.
21287 (cpu_feature): New CPU feature `vis4b'.
21288 (dictunpack{8,16,32}): New insns.
21289 (FPCSMODE): New mode iterator.
21290 (fpcscond): New code iterator.
21291 (fpcsucond): Likewise.
21292 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21293 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21294 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21295 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21296 * config/sparc/constraints.md: Define constraints `q' for unsigned
21297 2-bit integer constants and `t' for unsigned 5-bit integer
21299 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21301 (imm5_operand_dictunpack16): Likewise.
21302 (imm5_operand_dictunpack32): Likewise.
21303 (imm2_operand): Likewise.
21304 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21305 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21306 ditunpack* and fpcmp*shl builtins.
21308 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21310 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21311 * config.in: Add HAVE_AS_SPARC6 define.
21312 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21314 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21316 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21317 (CPP_CPU_SPEC): Handle m8.
21318 (ASM_CPU_SPEC): Likewise.
21319 * config/sparc/sparc-opts.h (enum processor_type): Add
21321 * config/sparc/sparc.c (m8_costs): New struct.
21322 (sparc_option_override): Handle TARGET_CPU_m8.
21323 (sparc32_initialize_trampoline): Likewise.
21324 (sparc64_initialize_trampoline): Likewise.
21325 (sparc_issue_rate): Likewise.
21326 (sparc_register_move_cost): Likewise.
21327 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21328 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21329 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21330 (CPP_CPU_SPEC): Handle M8.
21331 (ASM_CPU_SPEC): Likewise.
21332 (AS_M8_FLAG): Define.
21333 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21334 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21335 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21337 * configure: Regenerate.
21338 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21341 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21343 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21345 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21346 ("*movdi_insn_sp32"): Do not set v3pipe.
21347 ("*movsi_insn"): Likewise.
21348 ("*movdi_insn_sp64"): Likewise.
21349 ("*movsf_insn"): Likewise.
21350 ("*movdf_insn_sp32"): Likewise.
21351 ("*movdf_insn_sp64"): Likewise.
21352 ("*zero_extendsidi2_insn_sp64"): Likewise.
21353 ("*sign_extendsidi2_insn"): Likewise.
21354 ("*mov<VM32:mode>_insn"): Likewise.
21355 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21356 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21357 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21358 ("<vlop:code><VL:mode>3"): Likewise.
21359 ("*not_<vlop:code><VL:mode>3"): Likewise.
21360 ("*nand<VL:mode>_vis"): Likewise.
21361 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21362 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21363 ("one_cmpl<VL:mode>2"): Likewise.
21364 ("faligndata<VM64:mode>_vis"): Likewise.
21365 ("alignaddrsi_vis"): Likewise.
21366 ("alignaddrdi_vis"): Likweise.
21367 ("alignaddrlsi_vis"): Likewise.
21368 ("alignaddrldi_vis"): Likewise.
21369 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21370 ("bmaskdi_vis"): Likewise.
21371 ("bmasksi_vis"): Likewise.
21372 ("bshuffle<VM64:mode>_vis"): Likewise.
21373 ("cmask8<P:mode>_vis"): Likewise.
21374 ("cmask16<P:mode>_vis"): Likewise.
21375 ("cmask32<P:mode>_vis"): Likewise.
21376 ("pdistn<P:mode>_vis"): Likewise.
21377 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21379 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21381 * config/sparc/sparc.md ("subtype"): New insn attribute.
21382 ("*wrgsr_sp64"): Set insn subtype.
21383 ("*rdgsr_sp64"): Likewise.
21384 ("alignaddrsi_vis"): Likewise.
21385 ("alignaddrdi_vis"): Likewise.
21386 ("alignaddrlsi_vis"): Likewise.
21387 ("alignaddrldi_vis"): Likewise.
21388 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21389 ("fexpand_vis"): Likewise.
21390 ("fpmerge_vis"): Likewise.
21391 ("faligndata<VM64:mode>_vis"): Likewise.
21392 ("bshuffle<VM64:mode>_vis"): Likewise.
21393 ("cmask8<P:mode>_vis"): Likewise.
21394 ("cmask16<P:mode>_vis"): Likewise.
21395 ("cmask32<P:mode>_vis"): Likewise.
21396 ("fchksm16_vis"): Likewise.
21397 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21398 ("fmean16_vis"): Likewise.
21399 ("fp<plusminus_insn>64_vis"): Likewise.
21400 ("<plusminus_insn>v8qi3"): Likewise.
21401 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21402 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21403 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21404 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21405 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21406 ("*movqi_insn"): Likewise.
21407 ("*movhi_insn"): Likewise.
21408 ("*movsi_insn"): Likewise.
21409 ("movsi_pic_gotdata_op"): Likewise.
21410 ("*movdi_insn_sp32"): Likewise.
21411 ("*movdi_insn_sp64"): Likewise.
21412 ("movdi_pic_gotdata_op"): Likewise.
21413 ("*movsf_insn"): Likewise.
21414 ("*movdf_insn_sp32"): Likewise.
21415 ("*movdf_insn_sp64"): Likewise.
21416 ("*zero_extendhisi2_insn"): Likewise.
21417 ("*zero_extendqihi2_insn"): Likewise.
21418 ("*zero_extendqisi2_insn"): Likewise.
21419 ("*zero_extendqidi2_insn"): Likewise.
21420 ("*zero_extendhidi2_insn"): Likewise.
21421 ("*zero_extendsidi2_insn_sp64"): Likewise.
21422 ("ldfsr"): Likewise.
21423 ("prefetch_64"): Likewise.
21424 ("prefetch_32"): Likewise.
21425 ("tie_ld32"): Likewise.
21426 ("tie_ld64"): Likewise.
21427 ("*tldo_ldub_sp32"): Likewise.
21428 ("*tldo_ldub1_sp32"): Likewise.
21429 ("*tldo_ldub2_sp32"): Likewise.
21430 ("*tldo_ldub_sp64"): Likewise.
21431 ("*tldo_ldub1_sp64"): Likewise.
21432 ("*tldo_ldub2_sp64"): Likewise.
21433 ("*tldo_ldub3_sp64"): Likewise.
21434 ("*tldo_lduh_sp32"): Likewise.
21435 ("*tldo_lduh1_sp32"): Likewise.
21436 ("*tldo_lduh_sp64"): Likewise.
21437 ("*tldo_lduh1_sp64"): Likewise.
21438 ("*tldo_lduh2_sp64"): Likewise.
21439 ("*tldo_lduw_sp32"): Likewise.
21440 ("*tldo_lduw_sp64"): Likewise.
21441 ("*tldo_lduw1_sp64"): Likewise.
21442 ("*tldo_ldx_sp64"): Likewise.
21443 ("*mov<VM32:mode>_insn"): Likewise.
21444 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21445 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21447 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21449 * config/sparc/sparc.md ("type"): New insn type viscmp.
21450 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21452 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21453 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21454 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21455 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21457 ("n7_vis_logical_11cycle"): Likewise.
21458 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21459 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21460 * config/sparc/niagara.md ("niag_vis"): Likewise.
21461 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21462 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21464 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21466 * config/sparc/sparc.md: New instruction type `bmask'.
21467 (bmaskdi_vis): Use the `bmask' type.
21468 (bmasksi_vis): Likewise.
21469 * config/sparc/ultra3.md (us3_array): Likewise.
21470 * config/sparc/niagara7.md (n7_array): Likewise.
21471 * config/sparc/niagara4.md (n4_array): Likewise.
21472 * config/sparc/niagara2.md (niag2_vis): Likewise.
21473 (niag3_vis): Likewise.
21474 * config/sparc/niagara.md (niag_vis): Likewise.
21476 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21478 * ipa-comdats.c: Remove optimize check from gate.
21479 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21480 for functions not optimized.
21481 (ipa_fn_summary_read): Skip optimize check.
21482 (ipa_fn_summary_write): Likewise.
21483 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21485 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21487 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21488 (check_callers): Give up on caller which is not optimized.
21489 (inline_small_functions): Likewise.
21490 (ipa_inline): Do not give up when not optimizing.
21491 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21492 away unoptimizes cdtors.
21493 (whole_program_function_and_variable_visibility): Do
21494 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21495 * ipa.c (process_references): Do not check optimize.
21496 (symbol_table::remove_unreachable_nodes): Update optimize check.
21497 (set_writeonly_bit): Update optimize check.
21498 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21499 (pass_ipa_single_use::gate): Remove.
21501 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21503 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21504 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21505 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21506 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21507 permute_load, permute_store, adjust_extract, adjust_splat,
21508 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21509 replace_swap_with_copy, dump_swap_insn_table,
21510 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21511 recombine_lvx_pattern, recombine_stvx_pattern,
21512 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21513 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21514 to file rs6000-p8swap.c.
21515 * config/rs6000/rs6000-p8swap.c: New file.
21516 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21517 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21518 and rs6000*-*-* targets.
21520 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21522 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21524 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21526 * lto-wrapper.c (merge_and_complain): Do not merge
21527 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21528 fsigned_zeros, ftrapping_math, fwrapv.
21529 (append_compiler_options): Do not track these options.
21530 (append_linker_options): Likewie
21532 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21534 * cgraphunit.c (cgraph_node::finalize_function): When
21535 !flag_toplevel_reorde set no_reorder flag.
21536 (varpool_node::finalize_decl): Likewise.
21537 (symbol_table::compile): Drop no toplevel reorder path.
21539 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21541 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21542 edges; zero probability is not better than uninitialized.
21544 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21546 * asan.h (asan_sanitize_allocas_p): Declare.
21547 * asan.c (asan_sanitize_allocas_p): New function.
21548 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21549 (handle_builtin_alloca): Likewise.
21550 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21551 if !asan_sanitize_allocas_p.
21552 * params.def (asan-instrument-allocas): Add new option.
21553 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21554 * opts.c (common_handle_option): Disable allocas sanitization for
21557 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21559 * asan.c: Include gimple-fold.h.
21560 (get_last_alloca_addr): New function.
21561 (handle_builtin_stackrestore): Likewise.
21562 (handle_builtin_alloca): Likewise.
21563 (asan_emit_allocas_unpoison): Likewise.
21564 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21565 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21566 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21567 (instrument_builtin_call): Pass gimple iterator to
21568 get_mem_refs_of_builtin_call.
21569 (last_alloca_addr): New global.
21570 * asan.h (asan_emit_allocas_unpoison): Declare.
21571 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21572 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21573 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21574 if function calls alloca.
21575 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21576 * gimple-fold.h (replace_call_with_value): Declare.
21577 * internal-fn.c: Include asan.h.
21578 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21579 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21581 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21583 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21584 (C_SELFTEST_FLAGS): New.
21585 (CPP_SELFTEST_FLAGS): New.
21586 (SELFTEST_DEPS): New, from deps of s-selftest.
21587 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21588 (CPP_SELFTEST_DEPS): New.
21589 (selftest): Add dependency on s-selftest-c++.
21590 (s-selftest): Rename to...
21591 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21592 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21593 than SELFTEST_FLAGS.
21594 (selftest-gdb): Rename to...
21595 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21597 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21598 (selftest-valgrind): Rename to...
21599 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21601 (selftest-valgrind): Reintroduce as an alias for
21602 selftest-c-valgrind.
21603 (s-selftest-c++): New.
21604 (selftest-c++-gdb): New.
21605 (selftest-c++-valgrind): New.
21607 2017-07-06 Olivier Hainque <hainque@adacore.com>
21609 * gcc.c (process_command): When deciding if undefined variables
21610 should be ignored when processing specs, accept "gcc -v" as well.
21612 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21614 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21615 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21617 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21619 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21620 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21621 * config/arm/arm-tables.opt: Regenerate.
21622 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21624 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21626 2017-07-06 Carl Love <cel@us.ibm.com>
21628 * ChangeLog: Clean up from mid air collision
21630 2017-07-06 Carl Love <cel@us.ibm.com>
21632 * config/rs6000/rs6000-c.c: Add support for built-in functions
21633 vector signed int vec_subc (vector signed int, vector signed int);
21634 vector signed __int128 vec_subc (vector signed __int128,
21635 vector signed __int128);
21636 vector unsigned __int128 vec_subc (vector unsigned __int128,
21637 vector unsigned __int128);
21638 vector signed int vec_sube (vector signed int, vector signed int,
21639 vector signed int);
21640 vector unsigned int vec_sube (vector unsigned int,
21641 vector unsigned int,
21642 vector unsigned int);
21643 vector signed __int128 vec_sube (vector signed __int128,
21644 vector signed __int128,
21645 vector signed__int128);
21646 vector unsigned __int128 vec_sube (vector unsigned __int128,
21647 vector unsigned __int128,
21648 vector unsigned __int128);
21649 vector signed int vec_subec (vector signed int, vector signed int,
21650 vector signed int);
21651 vector unsigned int vec_subec (vector unsigned int,
21652 vector unsigned int,
21653 vector unsigned int);
21654 vector signed __int128 vec_subec (vector signed __int128,
21655 vector signed __int128,
21656 vector signed__int128);
21657 vector unsigned __int128 vec_subec (vector unsigned __int128,
21658 vector unsigned __int128,
21659 vector unsigned __int128);
21660 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21661 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21662 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21663 BU_ALTIVEC_OVERLOAD_X definitions.
21664 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21665 * doc/extend.texi: Update the built-in documentation file for the new
21666 built-in functions.
21668 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21671 * diagnostic-show-locus.c (layout::layout): Use start and finish
21672 spelling location for the start and finish of each range.
21673 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21674 Add unused aspect param.
21675 * input.c (expand_location_1): Add "aspect" param, and use it
21676 to access the correct part of the location.
21677 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21679 (expand_location_to_spelling_point): Likewise.
21680 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21681 param, and pass it to expand_location_1.
21683 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21685 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21686 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21687 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21688 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21689 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21690 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21691 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21692 _mm_maskz_getmant_ss): New intrinsics.
21693 (__builtin_ia32_getexpss128_mask): Changed to ...
21694 __builtin_ia32_getexpss128_round ... this.
21695 (__builtin_ia32_getexpsd128_mask): Changed to ...
21696 __builtin_ia32_getexpsd128_round ... this.
21697 * config/i386/i386-builtin-types.def
21698 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21699 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21700 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21701 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21702 __builtin_ia32_getmantss_mask_round): New builtins.
21703 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21704 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21705 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21706 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21707 * config/i386/sse.md
21708 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21709 avx512f_sgetexp<mode><mask_scalar_name>
21710 <round_saeonly_scalar_name> ... this.
21711 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21712 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21713 vgetexp<ssescalarmodesuffix>
21714 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21715 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21716 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21717 avx512f_vgetmant<mode><mask_scalar_name>
21718 <round_saeonly_scalar_name> ... this.
21719 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21720 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21721 vgetmant<ssescalarmodesuffix>
21722 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21723 %0<mask_scalar_operand4>, %1, %2
21724 <round_saeonly_scalar_mask_op4>, %3} ... this.
21725 * config/i386/subst.md (mask_scalar_operand4,
21726 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21727 round_saeonly_scalar_nimm_predicate): New subst attributes.
21729 2017-07-06 Julia Koval <julia.koval@intel.com>
21731 * config/i386/i386.c (ix86_erase_embedded_rounding):
21732 Remove code for old rounding pattern.
21734 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21736 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21738 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21740 * doc/sourcebuild.texi (Test Directives, Variants of
21741 dg-require-support): Add documentation for dg-require-stack-check.
21743 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21745 * config/i386/subst.md (mask_scalar, round_scalar,
21746 round_saeonly_scalar): New meta-templates.
21747 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21748 round_scalar_mask_operand3, round_scalar_mask_op3,
21749 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21750 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21751 round_saeonly_scalar_constraint,
21752 round_saeonly_scalar_prefix): New subst attribute.
21753 * config/i386/sse.md
21754 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21755 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21756 <round_scalar_name> ... this.
21757 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21758 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21759 <round_scalar_name> ... this.
21760 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21761 <sse>_vm<code><mode>3<mask_scalar_name>
21762 <round_saeonly_scalar_name> ... this.
21763 (v<plusminus_mnemonic><ssescalarmodesuffix>
21764 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21765 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21766 v<plusminus_mnemonic><ssescalarmodesuffix>
21767 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21768 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21769 (v<multdiv_mnemonic><ssescalarmodesuffix>
21770 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21771 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21772 v<multdiv_mnemonic><ssescalarmodesuffix>
21773 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21774 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21775 (v<maxmin_float><ssescalarmodesuffix>
21776 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21777 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21778 v<maxmin_float><ssescalarmodesuffix>
21779 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21780 %0<mask_scalar_operand3>, %1, %<iptr>2
21781 <round_saeonly_scalar_mask_op3>} ... this.
21783 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21785 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21786 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21788 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21789 Alan Hayward <alan.hayward@arm.com>
21790 David Sherwood <david.sherwood@arm.com>
21792 * combine.c (simplify_if_then_else): Remove "enum" before
21794 * compare-elim.c (can_eliminate_compare): Likewise.
21795 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21797 (aarch64_lookup_simd_builtin_type): Likewise.
21798 (aarch64_simd_builtin_type): Likewise.
21799 (aarch64_init_simd_builtin_types): Likewise.
21800 (aarch64_simd_expand_args): Likewise.
21801 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21803 (aarch64_reverse_mask): Likewise.
21804 (aarch64_simd_emit_reg_reg_move): Likewise.
21805 (aarch64_gen_adjusted_ldpstp): Likewise.
21806 (aarch64_ccmp_mode_to_code): Likewise.
21807 (aarch64_operands_ok_for_ldpstp): Likewise.
21808 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21809 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21811 (aarch64_min_divisions_for_recip_mul): Likewise.
21812 (aarch64_reassociation_width): Likewise.
21813 (aarch64_get_condition_code_1): Likewise.
21814 (aarch64_simd_emit_reg_reg_move): Likewise.
21815 (aarch64_simd_attr_length_rglist): Likewise.
21816 (aarch64_reverse_mask): Likewise.
21817 (aarch64_operands_ok_for_ldpstp): Likewise.
21818 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21819 (aarch64_gen_adjusted_ldpstp): Likewise.
21820 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21822 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21823 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21824 (arm_lookup_simd_builtin_type): Likewise.
21825 (arm_simd_builtin_type): Likewise.
21826 (arm_init_simd_builtin_types): Likewise.
21827 (arm_expand_builtin_args): Likewise.
21828 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21829 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21830 (ft32_setup_incoming_varargs): Likewise.
21831 (ft32_function_arg): Likewise.
21832 (ft32_function_arg_advance): Likewise.
21833 (ft32_pass_by_reference): Likewise.
21834 (ft32_arg_partial_bytes): Likewise.
21835 (ft32_valid_pointer_mode): Likewise.
21836 (ft32_addr_space_pointer_mode): Likewise.
21837 (ft32_addr_space_legitimate_address_p): Likewise.
21838 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21840 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21841 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21842 (iamcu_alignment): Likewise.
21843 (canonicalize_vector_int_perm): Likewise.
21844 (ix86_noce_conversion_profitable_p): Likewise.
21845 (ix86_mpx_bound_mode): Likewise.
21846 (ix86_operands_ok_for_move_multiple): Likewise.
21847 * config/microblaze/microblaze-protos.h
21848 (microblaze_expand_conditional_branch_reg): Likewise.
21849 * config/microblaze/microblaze.c
21850 (microblaze_expand_conditional_branch_reg): Likewise.
21851 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21853 (rs6000_reassociation_width): Likewise.
21854 (rs6000_invalid_binary_op): Likewise.
21855 (fusion_p9_p): Likewise.
21856 (emit_fusion_p9_load): Likewise.
21857 (emit_fusion_p9_store): Likewise.
21858 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21860 (riscv_hard_regno_mode_ok_p): Likewise.
21861 (riscv_address_insns): Likewise.
21862 (riscv_split_symbol): Likewise.
21863 (riscv_legitimize_move): Likewise.
21864 (riscv_function_value): Likewise.
21865 (riscv_hard_regno_nregs): Likewise.
21866 (riscv_expand_builtin): Likewise.
21867 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21868 (riscv_build_integer): Likewise.
21869 (riscv_split_integer): Likewise.
21870 (riscv_legitimate_constant_p): Likewise.
21871 (riscv_cannot_force_const_mem): Likewise.
21872 (riscv_regno_mode_ok_for_base_p): Likewise.
21873 (riscv_valid_base_register_p): Likewise.
21874 (riscv_valid_offset_p): Likewise.
21875 (riscv_valid_lo_sum_p): Likewise.
21876 (riscv_classify_address): Likewise.
21877 (riscv_legitimate_address_p): Likewise.
21878 (riscv_address_insns): Likewise.
21879 (riscv_load_store_insns): Likewise.
21880 (riscv_force_binary): Likewise.
21881 (riscv_split_symbol): Likewise.
21882 (riscv_force_address): Likewise.
21883 (riscv_legitimize_address): Likewise.
21884 (riscv_move_integer): Likewise.
21885 (riscv_legitimize_const_move): Likewise.
21886 (riscv_legitimize_move): Likewise.
21887 (riscv_address_cost): Likewise.
21888 (riscv_subword): Likewise.
21889 (riscv_output_move): Likewise.
21890 (riscv_canonicalize_int_order_test): Likewise.
21891 (riscv_emit_int_order_test): Likewise.
21892 (riscv_function_arg_boundary): Likewise.
21893 (riscv_pass_mode_in_fpr_p): Likewise.
21894 (riscv_pass_fpr_single): Likewise.
21895 (riscv_pass_fpr_pair): Likewise.
21896 (riscv_get_arg_info): Likewise.
21897 (riscv_function_arg): Likewise.
21898 (riscv_function_arg_advance): Likewise.
21899 (riscv_arg_partial_bytes): Likewise.
21900 (riscv_function_value): Likewise.
21901 (riscv_pass_by_reference): Likewise.
21902 (riscv_setup_incoming_varargs): Likewise.
21903 (riscv_print_operand): Likewise.
21904 (riscv_elf_select_rtx_section): Likewise.
21905 (riscv_save_restore_reg): Likewise.
21906 (riscv_for_each_saved_reg): Likewise.
21907 (riscv_register_move_cost): Likewise.
21908 (riscv_hard_regno_mode_ok_p): Likewise.
21909 (riscv_hard_regno_nregs): Likewise.
21910 (riscv_class_max_nregs): Likewise.
21911 (riscv_memory_move_cost): Likewise.
21912 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21913 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21914 (rl78_addr_space_address_mode): Likewise.
21915 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21917 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21918 (rs6000_reassociation_width): Likewise.
21919 (rs6000_invalid_binary_op): Likewise.
21920 (fusion_p9_p): Likewise.
21921 (emit_fusion_p9_load): Likewise.
21922 (emit_fusion_p9_store): Likewise.
21923 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
21924 (ok_for_simple_move_operands): Likewise.
21925 (ok_for_simple_move_strict_operands): Likewise.
21926 (ok_for_simple_arith_logic_operands): Likewise.
21927 (visium_legitimize_reload_address): Likewise.
21928 (visium_select_cc_mode): Likewise.
21929 (output_cbranch): Likewise.
21930 (visium_split_double_move): Likewise.
21931 (visium_expand_copysign): Likewise.
21932 (visium_expand_int_cstore): Likewise.
21933 (visium_expand_fp_cstore): Likewise.
21934 * config/visium/visium.c (visium_pass_by_reference): Likewise.
21935 (visium_function_arg): Likewise.
21936 (visium_function_arg_advance): Likewise.
21937 (visium_libcall_value): Likewise.
21938 (visium_setup_incoming_varargs): Likewise.
21939 (visium_legitimate_constant_p): Likewise.
21940 (visium_legitimate_address_p): Likewise.
21941 (visium_legitimize_address): Likewise.
21942 (visium_secondary_reload): Likewise.
21943 (visium_register_move_cost): Likewise.
21944 (visium_memory_move_cost): Likewise.
21945 (prepare_move_operands): Likewise.
21946 (ok_for_simple_move_operands): Likewise.
21947 (ok_for_simple_move_strict_operands): Likewise.
21948 (ok_for_simple_arith_logic_operands): Likewise.
21949 (visium_function_value_1): Likewise.
21950 (rtx_ok_for_offset_p): Likewise.
21951 (visium_legitimize_reload_address): Likewise.
21952 (visium_split_double_move): Likewise.
21953 (visium_expand_copysign): Likewise.
21954 (visium_expand_int_cstore): Likewise.
21955 (visium_expand_fp_cstore): Likewise.
21956 (visium_split_cstore): Likewise.
21957 (visium_select_cc_mode): Likewise.
21958 (visium_split_cbranch): Likewise.
21959 (output_cbranch): Likewise.
21960 (visium_print_operand_address): Likewise.
21961 * expmed.c (flip_storage_order): Likewise.
21962 * expmed.h (emit_cstore): Likewise.
21963 (flip_storage_order): Likewise.
21964 * genrecog.c (validate_pattern): Likewise.
21965 * hsa-gen.c (gen_hsa_addr): Likewise.
21966 * internal-fn.c (expand_arith_overflow): Likewise.
21967 * ira-color.c (allocno_copy_cost_saving): Likewise.
21968 * lra-assigns.c (find_hard_regno_for_1): Likewise.
21969 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
21970 (process_invariant_for_inheritance): Likewise.
21971 * lra-eliminations.c (move_plus_up): Likewise.
21972 * omp-low.c (lower_oacc_reductions): Likewise.
21973 * simplify-rtx.c (simplify_subreg): Likewise.
21974 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
21975 (TARGET_CHKP_BOUND_MODE): Likewise..
21976 * targhooks.c (default_chkp_bound_mode): Likewise.
21977 (default_setup_incoming_vararg_bounds): Likewise.
21978 * targhooks.h (default_chkp_bound_mode): Likewise.
21979 (default_setup_incoming_vararg_bounds): Likewise.
21980 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
21981 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
21982 (have_whole_vector_shift): Likewise.
21983 * tree-vect-stmts.c (vectorizable_load): Likewise.
21984 * doc/tm.texi: Regenerate.
21986 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21988 Graceful degrade if Binutils PR21472 is not available.
21991 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
21992 .rodata in flash test fails.
21993 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
21994 * confgure: Regenerate.
21995 * config.in: Regenerate.
21996 * config/avr/avr.c (avr_asm_named_section)
21997 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
21998 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
21999 (avr_asm_init_sections): Same.
22001 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22003 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22004 (fma<VH:mode>4_intrinsic): Likewise.
22005 (*fmsub<VCVTF:mode>4): Likewise.
22006 (*fmsub<VH:mode>4_intrinsic): Likewise.
22008 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22011 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22012 Don't depend on "optimize > 0".
22013 (out_movhi_r_mr, out_movqi_mr_r): Same.
22014 (out_movhi_mr_r, out_movqi_r_mr): Same.
22015 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22016 io_address_operand on "optimize > 0".
22018 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22020 * tree-loop-distribution.c: Add general explanantion on the pass.
22021 (generate_loops_for_partition): Mark distributed loop.
22022 (pg_add_dependence_edges): New parameter. Handle alias data
22023 dependence specially and record it in the parameter if asked.
22024 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22025 (init_partition_graph_vertices, add_partition_graph_edge): New.
22026 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22027 (free_partition_graph_vdata, build_partition_graph): New.
22028 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22029 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22030 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22031 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22032 (version_for_distribution_p, finalize_partitions): New.
22033 (distribute_loop): Handle alias data dependence specially. Factor
22034 out loop fusion code as functions and call these functions.
22036 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22038 * tree-loop-distribution.c (classify_partition): New parameter and
22039 better handle reduction statement.
22040 (rdg_build_partitions): Revise comment.
22041 (distribute_loop): Compute statements in all partitions and pass it
22042 to classify_partition.
22044 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22046 * tree-loop-distribution.c (enum partition_type): New.
22047 (struct partition): New field type.
22048 (partition_merge_into): Add parameter. Update partition type.
22049 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22050 (build_rdg_partition_for_vertex): Compute partition type.
22051 (rdg_build_partitions): Dump partition type.
22052 (distribute_loop): Update calls to partition_merge_into.
22054 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22056 * tree-loop-distribution.c (struct ddr_hasher): New.
22057 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22059 (classify_partition): Call get_data_dependence.
22060 (pg_add_dependence_edges): Ditto.
22061 (distribute_loop): Release data dependence hash table.
22063 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22065 * tree-loop-distribution.c (ref_base_address): Delete.
22066 (similar_memory_accesses): Rename ...
22067 (share_memory_accesses): ... to this. Check if partitions access
22068 the same memory reference.
22069 (distribute_loop): Call share_memory_accesses.
22071 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22073 * tree-loop-distribution.c (struct partition): New field recording
22074 its data reference.
22075 (partition_alloc, partition_free): Init and release data refs.
22076 (partition_merge_into): Merge data refs.
22077 (build_rdg_partition_for_vertex): Collect data refs for partition.
22078 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22080 (distribute_loop): Remve data refs from vertice data of partition
22083 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22085 * tree-loop-distribution.c (params.h): Include header file.
22086 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22087 (datarefs_vec): New global var.
22088 (create_rdg_vertices): Use datarefs_vec directly.
22089 (free_rdg): Don't free data references.
22090 (build_rdg): Update use. Don't free data references.
22091 (distribute_loop): Compute global variable for data references.
22092 Bail out if there are too many data references.
22094 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22096 * tree-loop-distribution.c (loop_nest): New global var.
22097 (build_rdg): Use loop directly, rather than loop nest.
22098 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22100 (distribute_loop): Compute global variable loop nest. Update use.
22102 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22104 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22105 (partition_merge_into): New parameter. Dump reason for fusion.
22106 (distribute_loop): Update use of partition_merge_into.
22108 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22110 * tree-loop-distribution.c (bb_top_order_index): New.
22111 (bb_top_order_index_size, bb_top_order_cmp): New.
22112 (stmts_from_loop): Use topological order.
22113 (pass_loop_distribution::execute): Compute and release topological
22114 order for basic blocks.
22116 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22118 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22121 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22123 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22124 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22125 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22126 * internal-fn.def (LOOP_DIST_ALIAS): New.
22127 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22128 (fold_loop_internal_call): ... this.
22129 (vect_loop_dist_alias_call): New function.
22130 (set_uid_loop_bbs): Call fold_loop_internal_call.
22131 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22134 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22137 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22138 Require dead FLAGS_REG at the beginning of a peephole.
22140 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22143 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22144 arguments in the call to __builtin_ia32_sbb_u32.
22145 (_subborrow_u64): Swap _X and _Y arguments in the call to
22146 __builtin_ia32_sbb_u64.
22148 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22151 * tree-vrp.c (compare_assert_loc): Turn into a function template
22152 with stable template parameter. Only test if a->e is NULL,
22153 !a->e == !b->e has been verified already. Use e == NULL or
22154 e != NULL instead of e or ! e tests. If stable is true, don't use
22155 iterative_hash_expr, on the other side allow a or b or both NULL
22156 and sort the NULLs last.
22157 (process_assert_insertions): Sort using compare_assert_loc<false>
22158 instead of compare_assert_loc, later sort using
22159 compare_assert_loc<true> before calling process_assert_insertions_for
22160 in a loop. Use break instead of continue once seen NULL pointer.
22162 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22164 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22165 Cortex-R7 and Cortex-R8 processors.
22167 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22169 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22170 uninitialized while src is not.
22172 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22174 * common/config/arm/arm-common.c: Adjust include path for
22176 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22177 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22178 (arm-cpu-data.h): Likewise.
22179 (arm-cpu-cdata.h): Likewise.
22180 * config/arm/arm-cpu.h: Delete.
22181 * config/arm/arm-cpu-cdata.h: Delete.
22182 * config/arm/arm-cpu-data.h: Delete.
22184 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22186 * config/arm/arm-cpus.in (cortex-a55): New.
22187 (cortex-a75): Likewise.
22188 (cortex-a75.cortex-a55): Likewise.
22189 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22191 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22192 * config/arm/arm-cpu-cdata.h: Regenerate.
22193 * config/arm/arm-cpu-data.h: Regenerate.
22194 * config/arm/arm-cpu.h: Regenerate.
22195 * config/arm/arm-tables.opt: Regenerate.
22196 * config/arm/arm-tune.md: Regenerate.
22198 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22200 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22202 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22204 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22207 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22209 PR tree-optimization/81292
22210 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22211 full_string_p, also call adjust_related_strinfos if the adjustment
22212 is simple, otherwise invalidate related strinfos.
22214 2017-07-04 Martin Liska <mliska@suse.cz>
22217 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22218 newly created variable as DECL_IGNORED_P.
22220 2017-07-04 Martin Liska <mliska@suse.cz>
22223 * ipa-inline.c (inline_small_functions):
22224 Use xstrdup_for_dump.
22226 2017-07-04 Tom de Vries <tom@codesourcery.com>
22228 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22230 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22233 * config/darwin.c (darwin_function_switched_text_sections):
22236 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22238 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22240 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22242 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22244 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22246 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22247 min_profitable_iters, and th as inclusive lower bounds.
22248 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22249 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22250 for min_profitable_iters and min_profitable_estimate.
22251 (vect_transform_loop): Treat th as an inclusive lower bound.
22252 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22254 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22257 * config/darwin.c (darwin_function_switched_text_sections):
22258 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22259 in two pieces, and suppress the use of buf.
22261 2017-07-03 Nathan Sidwell <nathan@acm.org>
22263 * hash-table.h (hash_table_mod1): Fix indentation.
22265 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22267 PR middle-end/81290
22268 * predict.c (force_edge_cold): Be more careful about propagation
22270 * profile-count.h (profile_probability::guessed,
22271 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22273 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22275 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22277 * doc/invoke.texi (rcpc architecture extension): Document it.
22279 2017-07-03 Richard Biener <rguenther@suse.de>
22281 PR tree-optimization/60510
22282 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22283 the scalar reduction PHI and use it.
22284 (vectorizable_reduction): Properly guard the single_defuse_cycle
22285 path for non-SLP reduction chains where we cannot use it.
22286 Rework reduc_def/index and vector type deduction. Rework
22287 vector operand gathering during reduction op code-gen.
22288 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22289 chains dissolve the chain and leave it to non-SLP reduction
22292 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22294 * tree-data-ref.h (dr_alignment): Declare.
22295 * tree-data-ref.c (dr_alignment): New function.
22296 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22297 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22299 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22301 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22303 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22304 and base_misalignment fields.
22305 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22306 * tree-data-ref.c: Include builtins.h.
22307 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22308 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22309 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22310 * tree-vect-data-refs.c: Include tree-cfg.h.
22311 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22312 fields instead of calculating an alignment here.
22313 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22314 innermost_loop_behavior fields.
22316 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22318 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22320 (DR_STEP_ALIGNMENT): New macro.
22321 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22322 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22323 (create_data_ref): Print it.
22324 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22325 to tell whether the step preserves vector (mis)alignment.
22326 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22327 Move the check for an integer step and generalise to all INTEGER_CST.
22328 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22329 Print the outer step alignment.
22331 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22333 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22334 with offset_alignment.
22335 (DR_ALIGNED_TO): Delete.
22336 (DR_OFFSET_ALIGNMENT): New macro.
22337 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22338 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22339 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22340 (create_data_ref): Likewise.
22341 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22342 (vect_analyze_data_refs): Likewise.
22343 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22344 creating dummy innermost behavior.
22346 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22348 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22349 with a "innermost_loop_behavior *" and refeence tree.
22350 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22351 (create_data_ref): Update call accordingly.
22352 * tree-predcom.c (find_looparound_phi): Likewise.
22354 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22356 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22357 fields with dr_wrt_vec_loop.
22358 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22359 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22360 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22361 (vect_dr_behavior): New function.
22362 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22363 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22364 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22365 track whether the step preserves the misalignment.
22366 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22367 Use vect_dr_behavior.
22368 (vect_setup_realignment): Update call accordingly.
22369 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22370 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22371 call to vect_create_addr_base_for_vector_ref.
22372 (vect_create_cond_for_align_checks): Likewise.
22373 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22374 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22375 (vect_recog_mask_conversion_pattern): Likewise.
22376 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22377 (new_stmt_vec_info): Remove redundant zeroing.
22379 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22381 * common/config/arm/arm-common.c (arm_be8_option): New function.
22382 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22383 (ISA_ARMv6): Add isa_bit_be8.
22384 * config/arm/arm.h (arm_be8_option): Add prototype.
22385 (BE8_SPEC_FUNCTION): New define.
22386 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22387 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22388 (mlittle-endian): Similarly.
22389 (mbe8, mbe32): New options.
22390 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22391 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22393 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22395 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22397 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22399 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22400 (cleanup_tree_cfg_bb): Use it.
22401 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22403 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22405 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22408 * loop-doloop.c (add_test): Update profile.
22410 2017-07-03 Martin Liska <mliska@suse.cz>
22413 * sanopt.c (rewrite_usage_of_param): New function.
22414 (sanitize_rewrite_addressable_params): Likewise.
22415 (pass_sanopt::execute): Call rewrite_usage_of_param.
22417 2017-07-03 Richard Biener <rguenther@suse.de>
22419 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22420 back to using VIEW_CONVERT_EXPR.
22422 2017-07-03 Martin Liska <mliska@suse.cz>
22425 * doc/extend.texi: Document when a resolver function is
22426 generated for target_clones.
22428 2017-07-03 Martin Liska <mliska@suse.cz>
22430 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22431 and shadow memory of auto variables which are subject of
22432 use-after-scope sanitization.
22433 (asan_expand_mark_ifn): Add do set only when is_poison.
22435 2016-07-03 Richard Biener <rguenther@suse.de>
22437 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22439 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22440 (vectorizable_reduction): Perform reduction PHI creation when
22441 visiting a reduction PHI and adjust and simplify code generation
22442 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22443 (vect_transform_loop): Visit reduction PHIs.
22444 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22445 defs into the SLP tree.
22446 (vect_build_slp_tree): Reduction defs terminate the recursion.
22447 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22449 (vect_get_vec_defs_for_stmt_copy): Export.
22450 (vect_get_vec_defs): Likewise.
22451 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22453 (vect_get_vec_defs_for_stmt_copy): Declare.
22454 (vect_get_vec_defs): Likewise.
22456 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22458 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22459 parameter with a "loop" parameter and use it instead of the
22460 loop containing DR_STMT. Don't check simple_iv when doing
22461 BB analysis. Describe the two analysis modes in the comment.
22463 2017-07-03 Tom de Vries <tom@codesourcery.com>
22465 PR tree-optimization/69468
22466 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22467 (find_same_succ_bb): Handle ignore_edge_flags.
22469 2017-07-03 Tom de Vries <tom@codesourcery.com>
22471 PR tree-optimization/81192
22472 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22474 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22476 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22478 2017-07-03 Tom de Vries <tom@codesourcery.com>
22480 PR tree-optimization/81192
22481 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22482 BB_SAME_SUCC (bb) == NULL.
22484 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22486 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22489 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22491 * dumpfile.c: Include profile-count.h
22492 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22494 (insert_cond_bb): Update profile.
22495 * tree-cfg.h (insert_cond_bb): Update prototype.
22496 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22497 * tree-dump.c: Do not include tree-cfg.
22499 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22501 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22503 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22505 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22508 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22510 * tree-complex.c (expand_complex_div_wide): update profile.
22512 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22513 Alan Hayward <alan.hayward@arm.com>
22514 David Sherwood <david.sherwood@arm.com>
22516 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22517 (CORETYPES_H): New define.
22518 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22519 (insn-modes-inline.h, s-modes-inline-h): New rules.
22520 (generated_files): Add insn-modes-inline.h.
22521 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22522 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22523 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22524 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22525 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22526 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22527 (build/gencodes.o, build/genconditions.o): Likewise.
22528 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22529 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22530 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22531 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22532 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22533 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22534 * coretypes.h: Include everything up to real.h for generators.
22535 Include insn-modes.h first. Include wide-int-print.h after
22536 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22537 * machmode.h: Don't include insn-modes.h here.
22538 * function-tests.c: Remove includes of signop.h, machmode.h,
22539 double-int.h and wide-int.h.
22541 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22543 * optc-save-gen.awk: Likewise.
22544 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22545 * godump.c: Remove include of wide-int-print.h.
22546 * pretty-print.h: Likewise.
22547 * wide-int-print.cc: Likewise.
22548 * wide-int.cc: Likewise.
22549 * hash-map-tests.c: Remove include of signop.h.
22550 * hash-set-tests.c: Likewise.
22551 * rtl-tests.c: Likewise.
22552 * mkconfig.sh: Remove include of machmode.h.
22553 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22555 (emit_insn_modes_inline_h): ...this new function. Emit the code
22556 into an insn-modes-inline.h header file, adding appropriate
22557 include guards and end comments.
22558 (emit_insn_modes_c_header): Remove include of machmode.h.
22559 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22561 (main): Handle -i flag and call emit_insn_modes_inline_h when
22564 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22566 * tree-ssa-strlen.c (strinfo): Rename the length field to
22567 nonzero_chars. Add a full_string_p field.
22568 (compare_nonzero_chars, zero_length_string_p): New functions.
22569 (get_addr_stridx): Add an offset_out parameter.
22570 Use compare_nonzero_chars.
22571 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22572 (new_strinfo): Update after above changes to strinfo.
22573 (set_endptr_and_length): Set full_string_p.
22574 (get_string_length): Update after above changes to strinfo.
22575 (unshare_strinfo): Update call to new_strinfo.
22576 (maybe_invalidate): Likewise.
22577 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22578 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22579 as a uhwi instead of an shwi. Update after above changes to
22580 strinfo and new_strinfo.
22581 (zero_length_string): Assert that chainsi contains full strings.
22582 Use zero_length_string_p. Update call to new_strinfo.
22583 (adjust_related_strinfos): Update after above changes to strinfo.
22584 Copy full_string_p from origsi.
22585 (adjust_last_stmt): Use zero_length_string_p.
22586 (handle_builtin_strlen): Update after above changes to strinfo and
22587 new_strinfo. Install the lhs as the string length if the previous
22588 entry didn't describe a full string.
22589 (handle_builtin_strchr): Update after above changes to strinfo
22591 (handle_builtin_strcpy): Likewise.
22592 (handle_builtin_strcat): Likewise.
22593 (handle_builtin_malloc): Likewise.
22594 (handle_pointer_plus): Likewise.
22595 (handle_builtin_memcpy): Likewise. Track nonzero characters
22596 that aren't necessarily followed by a nul terminator.
22597 (handle_char_store): Likewise.
22599 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22601 PR tree-optimization/80769
22602 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22603 for malloc and calloc. Document the new invariant that all related
22604 strinfos have delayed lengths or none do.
22605 (verify_related_strinfos): Move earlier in file.
22606 (set_endptr_and_length): New function, split out from...
22607 (get_string_length): ...here. Also set the lengths of related
22609 (zero_length_string): Assert that chainsi has known (rather than
22611 (adjust_related_strinfos): Likewise.
22613 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22615 PR tree-optimization/81136
22616 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22617 assert that two references with the same misalignment have the same
22618 compile-time misalignment if those compile-time misalignments
22621 2017-07-01 Andi Kleen <ak@linux.intel.com>
22623 * print-tree.c (print_node): Print all attributes.
22625 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22627 * cfg.c (scale_bbs_frequencies): New function.
22628 * cfg.h (scale_bbs_frequencies): Declare it.
22629 * cfgloopanal.c (single_likely_exit): Cleanup.
22630 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22632 (scale_loop_profile): Likewise.
22633 (loop_version): Likewise.
22634 (create_empty_loop_on_edge): Update.
22635 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22636 scale_loop_frequencies, scale_loop_profile, loopify,
22637 loop_version): Update prototypes.
22638 * modulo-sched.c (sms_schedule): Update.
22639 * predict.c (unlikely_executed_edge_p): Also check probability.
22640 (probably_never_executed_edge_p): Fix typo.
22641 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22642 * tree-parloops.c (gen_parallel_loop): Update.
22643 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22644 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22645 * tree-ssa-loop-split.c (split_loop): Update.
22646 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22647 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22648 (vect_loop_versioning): Update.
22649 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22651 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22653 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22655 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22657 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22658 to keep profile consistent.
22660 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22662 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22663 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22664 * profile-count.h (max_safe_multiplier): Make unsigned.
22665 (profile_count::guessed_zero): New.
22667 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22669 * bb-reorder.c (fix_up_crossing_landing_pad,
22670 fix_crossing_conditional_branches): Use make_single_succ_edge
22671 to keep profile consistent.
22673 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22675 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22678 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22681 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22682 the right scopes, make sure cond_jump isn't preserved between multiple
22683 iterations. Search for fallthru edge whenever there are 3+ edges and
22684 use find_fallthru_edge for it.
22686 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22688 Patch by Alexander Monakov <amonakov@ispras.ru>
22689 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22690 probabilities consistently.
22692 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22694 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22695 emit_cmp_and_jump_insns.
22697 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22700 * tree-inline.c (expand_call_inline): Combine profile statuses.
22702 2017-06-30 Andrew Pinski <apinski@cavium.com>
22704 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22705 fold_stmt returned true.
22707 2017-06-30 Nathan Sidwell <nathan@acm.org>
22709 * ggc.h (empty_string): Delete.
22710 * cfgexpand.c (expand_asm_stmt): Use plain "".
22711 * optabs.c (expand_asm_memory_barrier): Likewise.
22712 * stringpool.c (empty_string): Delete.
22713 (digit_vector, digit_string): Delete.
22714 (ggc_alloc_string): Use plain "", don't optimize single digit
22715 strings. Use ggc_alloc_atomic.
22717 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22719 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22720 comparison set and one other set, use the cost of the non-comparison
22723 2017-06-30 Nathan Sidwell <nathan@acm.org>
22725 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22728 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22730 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22731 loops. Remove now unneeded calls to gimple_switch_set_label() that
22732 just set removed labels to NULL_TREE.
22734 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22736 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22737 (set_range_info): ...here. Only call set_range_info_raw if domain
22739 (set_nonzero_bits): Call set_range_info_raw.
22740 * tree-ssanames.h (set_range_info_raw): New.
22742 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22745 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22746 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22747 of nonimmediate_operand and <store_mask_constraint> instead of m
22748 for the input operand. For V8FI iterator, always split if input
22749 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22750 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22751 <store_mask_predicate> instead of register_operand and
22752 <store_mask_constraint> instead of v for the input operand. Make
22753 sure both operands aren't MEMs for if not <mask_applied>.
22755 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22757 * lto-wrapper.c (copy_file) Close both file descriptors before
22760 2017-06-30 Martin Liska <mliska@suse.cz>
22763 * multiple_target.c (create_dispatcher_calls): Make ifunc
22764 also for function that don't have calls or are not referenced.
22766 2017-06-30 Richard Biener <rguenther@suse.de>
22768 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22769 analyze the first scalar stmt. Move vector type computation
22770 for the BB case here from ...
22771 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22772 live operation processing in the SLP case properly.
22774 2017-06-30 Richard Biener <rguenther@suse.de>
22776 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22778 2017-06-30 Martin Liska <mliska@suse.cz>
22781 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22782 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22784 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22786 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22789 2017-06-30 Richard Biener <rguenther@suse.de>
22791 PR tree-optimization/81249
22792 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22793 condition reduction result to original scalar type.
22795 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22797 * profile-count.h (enum profile_quality): Fix typos and whitespace
22800 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22802 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22803 type for branch probabilities.
22805 2017-06-29 Julian Brown <julian@codesourcery.com>
22806 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22808 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22809 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22810 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22811 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22813 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22815 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22816 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22817 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22818 CC usage from generic code to here.
22819 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22820 CC usage into the target macros.
22822 2017-06-29 Maya Rashish <coypu@sdf.org>
22824 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22827 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22829 * arm/arm-builtins.c: Include profile-count.h
22830 * except.c (sjlj_emit_function_enter): Use
22831 profile_probability::unlikely.
22833 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22835 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22836 and tocrel_offset be pointer args rather than implicitly using
22838 (legitimate_constant_pool_address_p, rs6000_emit_move,
22839 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22840 tocrel_offset and use in toc_relative_expr_p call.
22841 (print_operand, print_operand_address): Use static tocrel_base_oac
22842 and tocrel_offset_oac.
22843 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22846 2017-06-29 Maya Rashish <coypu@sdf.org>
22848 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22850 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22852 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22853 objects, take into account only the alignment of 'op0' and 'mode1' if
22856 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22858 * ccmp.c (ccmp_tree_comparison_p): New function.
22859 (ccmp_candidate_p): Update to use above function.
22860 (get_compare_parts): New function.
22861 (expand_ccmp_next): Update to use new functions.
22862 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22864 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22865 take mode as argument.
22866 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22867 * expr.c (expand_expr_real_1): Pass mode as argument.
22869 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22871 * combine.c (combine_instructions): Print insns to dump_file, together
22874 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22876 * asan.c (asan_emit_stack_protection): Update.
22877 (create_cond_insert_point): Update.
22878 * auto-profile.c (afdo_propagate_circuit): Update.
22879 * basic-block.h (struct edge_def): Turn probability to
22880 profile_probability.
22881 (EDGE_FREQUENCY): Update.
22882 * bb-reorder.c (find_traces_1_round): Update.
22883 (better_edge_p): Update.
22884 (sanitize_hot_paths): Update.
22885 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22886 (make_single_succ_edge): Update.
22887 (check_bb_profile): Update.
22888 (dump_edge_info): Update.
22889 (update_bb_profile_for_threading): Update.
22890 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22891 probabilitycount to 0.
22892 * cfgbuild.c (compute_outgoing_frequencies): Update.
22893 * cfgcleanup.c (try_forward_edges): Update.
22894 (outgoing_edges_match): Update.
22895 (try_crossjump_to_edge): Update.
22896 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22897 (expand_gimple_tailcall): Update.
22898 (construct_init_block): Use make_single_succ_edge.
22899 (construct_exit_block): Use make_single_succ_edge.
22900 * cfghooks.c (verify_flow_info): Update.
22901 (redirect_edge_succ_nodup): Update.
22902 (split_edge): Update.
22903 (account_profile_record): Update.
22904 * cfgloopanal.c (single_likely_exit): Update.
22905 * cfgloopmanip.c (scale_loop_profile): Update.
22906 (set_zero_probability): Remove.
22907 (duplicate_loop_to_header_edge): Update.
22908 * cfgloopmanip.h (loop_version): Update prototype.
22909 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22910 (force_nonfallthru_and_redirect): Update.
22911 (update_br_prob_note): Update.
22912 (rtl_verify_edges): Update.
22913 (purge_dead_edges): Update.
22914 (rtl_lv_add_condition_to_bb): Update.
22915 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22916 * cgraphunit.c (init_lowered_empty_function): Update.
22917 (cgraph_node::expand_thunk): Update.
22918 * cilk-common.c: Include profile-count.h
22919 * dojump.c (inv): Remove.
22920 (jumpifnot): Update.
22921 (jumpifnot_1): Update.
22922 (do_jump_1): Update.
22924 (do_jump_by_parts_greater_rtx): Update.
22925 (do_compare_rtx_and_jump): Update.
22926 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
22927 do_jump_1. do_compare_rtx_and_jump): Update prototype.
22928 * dwarf2cfi.c: Include profile-count.h
22929 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
22930 (sjlj_emit_dispatch_table): Likewise.
22931 * explow.c: Include profile-count.h
22932 * expmed.c (emit_store_flag_force): Update.
22933 (do_cmp_and_jump): Update.
22934 * expr.c (compare_by_pieces_d::generate): Update.
22935 (compare_by_pieces_d::finish_mode): Update.
22936 (emit_block_move_via_loop): Update.
22937 (store_expr_with_bounds): Update.
22938 (store_constructor): Update.
22939 (expand_expr_real_2): Update.
22940 (expand_expr_real_1): Update.
22941 * expr.h (try_casesi, try_tablejump): Update prototypes.
22942 * gimple-pretty-print.c (dump_probability): Update.
22943 (dump_profile): New.
22944 (dump_gimple_label): Update.
22945 (dump_gimple_bb_header): Update.
22946 * graph.c (draw_cfg_node_succ_edges): Update.
22947 * hsa-gen.c (convert_switch_statements): Update.
22948 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
22949 (find_if_case_1): Update.
22950 (find_if_case_2): Update.
22951 * internal-fn.c (expand_arith_overflow_result_store): Update.
22952 (expand_addsub_overflow): Update.
22953 (expand_neg_overflow): Update.
22954 (expand_mul_overflow): Update.
22955 (expand_vector_ubsan_overflow): Update.
22956 * ipa-cp.c (good_cloning_opportunity_p): Update.
22957 * ipa-split.c (split_function): Use make_single_succ_edge.
22958 * ipa-utils.c (ipa_merge_profiles): Update.
22959 * loop-doloop.c (add_test): Update.
22960 (doloop_modify): Update.
22961 * loop-unroll.c (compare_and_jump_seq): Update.
22962 (unroll_loop_runtime_iterations): Update.
22963 * lra-constraints.c (lra_inheritance): Update.
22964 * lto-streamer-in.c (input_cfg): Update.
22965 * lto-streamer-out.c (output_cfg): Update.
22966 * mcf.c (adjust_cfg_counts): Update.
22967 * modulo-sched.c (sms_schedule): Update.
22968 * omp-expand.c (expand_omp_for_init_counts): Update.
22969 (extract_omp_for_update_vars): Update.
22970 (expand_omp_ordered_sink): Update.
22971 (expand_omp_for_ordered_loops): Update.
22972 (expand_omp_for_generic): Update.
22973 (expand_omp_for_static_nochunk): Update.
22974 (expand_omp_for_static_chunk): Update.
22975 (expand_cilk_for): Update.
22976 (expand_omp_simd): Update.
22977 (expand_omp_taskloop_for_outer): Update.
22978 (expand_omp_taskloop_for_inner): Update.
22979 * omp-simd-clone.c (simd_clone_adjust): Update.
22980 * optabs.c (expand_doubleword_shift): Update.
22981 (expand_abs): Update.
22982 (emit_cmp_and_jump_insn_1): Update.
22983 (expand_compare_and_swap_loop): Update.
22984 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
22985 * predict.c (predictable_edge_p): Update.
22986 (edge_probability_reliable_p): Update.
22987 (set_even_probabilities): Update.
22988 (combine_predictions_for_insn): Update.
22989 (combine_predictions_for_bb): Update.
22990 (propagate_freq): Update.
22991 (estimate_bb_frequencies): Update.
22992 (force_edge_cold): Update.
22993 * profile-count.c (profile_count::dump): Add missing space into dump.
22994 (profile_count::debug): Add newline.
22995 (profile_count::differs_from_p): Explicitly convert to unsigned.
22996 (profile_count::stream_in): Update.
22997 (profile_probability::dump): New member function.
22998 (profile_probability::debug): New member function.
22999 (profile_probability::differs_from_p): New member function.
23000 (profile_probability::differs_lot_from_p): New member function.
23001 (profile_probability::stream_in): New member function.
23002 (profile_probability::stream_out): New member function.
23003 * profile-count.h (profile_count_quality): Rename to ...
23004 (profile_quality): ... this one.
23005 (profile_probability): New.
23006 (profile_count): Update.
23007 * profile.c (compute_branch_probabilities): Update.
23008 * recog.c (peep2_attempt): Update.
23009 * sched-ebb.c (schedule_ebbs): Update.
23010 * sched-rgn.c (find_single_block_region): Update.
23011 (compute_dom_prob_ps): Update.
23012 (schedule_region): Update.
23013 * sel-sched-ir.c (compute_succs_info): Update.
23014 * stmt.c (struct case_node): Update.
23015 (do_jump_if_equal): Update.
23016 (get_outgoing_edge_probs): Update.
23017 (conditional_probability): Update.
23018 (emit_case_dispatch_table): Update.
23019 (expand_case): Update.
23020 (expand_sjlj_dispatch_table): Update.
23021 (emit_case_nodes): Update.
23022 * targhooks.c: Update.
23023 * tracer.c (better_p): Update.
23024 (find_best_successor): Update.
23025 * trans-mem.c (expand_transaction): Update.
23026 * tree-call-cdce.c: Update.
23027 * tree-cfg.c (gimple_split_edge): Upate.
23028 (move_sese_region_to_fn): Upate.
23029 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23030 * tree-eh.c (lower_resx): Upate.
23031 (cleanup_empty_eh_move_lp): Upate.
23032 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23033 * tree-inline.c (copy_edges_for_bb): Update.
23034 (copy_cfg_body): Update.
23035 * tree-parloops.c (gen_parallel_loop): Update.
23036 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23037 (gimple_gen_time_profiler): Update.
23038 * tree-ssa-dce.c (remove_dead_stmt): Update.
23039 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23040 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23041 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23042 (unloop_loops): Update.
23043 (try_peel_loop): Update.
23044 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23045 * tree-ssa-loop-split.c (connect_loops): Update.
23046 (split_loop): Update.
23047 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23048 (hoist_guard): Update.
23049 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23050 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23051 (value_replacement): Update.
23052 * tree-ssa-reassoc.c (branch_fixup): Update.
23053 * tree-ssa-tail-merge.c (replace_block_by): Update.
23054 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23055 (create_edge_and_update_destination_phis): Update.
23056 (compute_path_counts): Update.
23057 (recompute_probabilities): Update.
23058 (update_joiner_offpath_counts): Update.
23059 (freqs_to_counts_path): Update.
23060 (duplicate_thread_path): Update.
23061 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23062 (struct switch_conv_info): Update.
23063 (gen_inbound_check): Update.
23064 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23065 (vect_do_peeling): Update.
23066 (vect_loop_versioning): Update.
23067 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23068 (optimize_mask_stores): Update.
23069 * ubsan.c (ubsan_expand_null_ifn): Update.
23070 * value-prof.c (gimple_divmod_fixed_value): Update.
23071 (gimple_divmod_fixed_value_transform): Update.
23072 (gimple_mod_pow2): Update.
23073 (gimple_mod_pow2_value_transform): Update.
23074 (gimple_mod_subtract): Update.
23075 (gimple_mod_subtract_transform): Update.
23076 (gimple_ic): Update.
23077 (gimple_stringop_fixed_value): Update.
23078 (gimple_stringops_transform): Update.
23079 * value-prof.h: Update.
23081 2017-06-29 Carl Love <cel@us.ibm.com>
23083 * config/rs6000/rs6000-c.c: Add support for built-in functions
23084 vector signed int vec_signed (vector float);
23085 vector signed long long vec_signed (vector double);
23086 vector signed int vec_signed2 (vector double, vector double);
23087 vector signed int vec_signede (vector double);
23088 vector signed int vec_signedo (vector double);
23089 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23090 instruction generator.
23091 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23092 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23093 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23095 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23096 vunsignede_v2df): Add define_expands.
23097 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23098 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23099 VEC_UNSIGNEDO): Add definitions.
23100 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23101 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23102 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23103 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23104 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23105 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23106 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23107 vec_unsignede, vec_unsignedo): Add builtin defines.
23108 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23110 * doc/extend.texi: Update the built-in documentation file for the
23111 new built-in functions.
23113 2017-06-29 Richard Biener <rguenther@suse.de>
23115 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23116 reduction chains to LOOP_VINFO_REDUCTIONS.
23117 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23118 SLP reductions after processing reduction chains.
23120 2017-06-29 Nathan Sidwell <nathan@acm.org>
23122 * builtins.c (fold_builtin_FUNCTION): Use
23123 lang_hooks.decl_printable_name.
23125 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23127 PR middle-end/81194
23128 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23129 with only one label.
23130 * stmt.c (expand_case): Assert NCASES is greater than one.
23132 2017-06-29 Richard Biener <rguenther@suse.de>
23134 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23136 (group_case_labels): Likewise.
23137 (find_taken_edge): Push sanity checking on val to workers...
23138 (find_taken_edge_cond_expr): ... here
23139 (find_taken_edge_switch_expr): ... and here, handle cases
23140 with just a default label.
23141 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23142 (group_case_labels): Likewise.
23143 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23144 group_case_labels does anything cleanup the CFG again.
23146 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23148 PR tree-optimization/81196
23149 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23150 exit condition comparing two IVs.
23152 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23154 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23155 profile to the dummy entry at the end of the list of architectures.
23156 * config/arm/arm-cpu-cdata.h: Regenerated.
23158 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23159 Michael Collison <michael.collison@arm.com>
23162 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23164 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23165 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23166 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23167 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23168 of shift when the shift amount is masked with constant equal to
23169 the size of the mode.
23170 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23173 2017-06-29 Martin Liska <mliska@suse.cz>
23175 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23177 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23180 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23183 (edge_growth_cache_entry::edge_growth_cache_entry): New
23185 (reset_edge_growth_cache): Update to use constructor.
23187 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23189 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23190 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23191 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23193 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23195 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23196 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23198 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23200 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23201 (*-linux-uclibc*): Add t-uclibc tmake_file.
23202 * config/t-musl: New.
23203 * config/t-uclibc: New.
23205 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23207 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23209 (gen_comm_data): Emit architectural setting of arch_prof.
23210 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23212 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23213 (armv8-m.base, armv8-m.main): Likewise.
23214 * arm-protos.h (arm_build_target): Add profile field.
23215 (arch_option): Likewise.
23216 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23218 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23219 arm_active_target.profile.
23221 2017-06-28 Richard Biener <rguenther@suse.de>
23223 PR middle-end/81227
23224 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23225 TYPE_OVERFLOW_WRAPS.
23226 * match.pd (negate_expr_p): Likewise.
23227 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23228 fold_build2, not fold_binary.
23230 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23232 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23233 Convert memory address to Pmode.
23234 (aarch64_print_operand): Assert MEM operands are always Pmode.
23236 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23239 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23240 Remove redundant if.
23241 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23242 * config/arm/aarch-common-protos.h
23243 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23244 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23246 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23249 * multiple_target.c (create_dispatcher_calls): Set the default
23250 clone to be static, not public.
23252 2017-06-28 Richard Biener <rguenther@suse.de>
23254 * tree-vect-loop.c (vectorizable_reduction): Move special
23255 cond reduction IV var creation ...
23256 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23257 parameter. Use STMT_VINFO_VECTYPE.
23258 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23261 2017-06-28 Martin Liska <mliska@suse.cz>
23264 * ipa-visibility.c (non_local_p): Handle visibility.
23266 2017-06-28 Martin Liska <mliska@suse.cz>
23269 * common.opt: Add IntegerRange to various options.
23270 * opt-functions.awk (integer_range_info): New function.
23271 * optc-gen.awk: Add integer_range_info to cl_options struct.
23272 * opts-common.c (decode_cmdline_option): Handle
23273 CL_ERR_INT_RANGE_ARG.
23274 (cmdline_handle_error): Likewise.
23275 * opts.c (print_filtered_help): Show valid interval in
23276 when --help is provided.
23277 * opts.h (struct cl_option): Add range_min and range_max fields.
23278 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23280 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23282 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23283 (x * C EQ/NE y * C): New transformation.
23285 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23287 * genmultilib (combination_space): Accept '+' in option names.
23289 2017-06-28 Martin Liska <mliska@suse.cz>
23292 * asan.c (instrument_derefs): Bail out inner references
23293 that are hard register variables.
23295 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23298 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23299 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23301 2017-06-28 Richard Biener <rguenther@suse.de>
23303 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23304 (vect_get_slp_defs): Adjust.
23305 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23307 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23309 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23310 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23311 (vectorizable_reduction): Adjust.
23312 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23314 (vect_get_slp_defs): Likewise.
23315 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23316 (vectorizable_bswap): Adjust.
23317 (vectorizable_call): Likewise.
23318 (vectorizable_conversion): Likewise.
23319 (vectorizable_assignment): Likewise.
23320 (vectorizable_shift): Likewise.
23321 (vectorizable_operation): Likewise.
23322 (vectorizable_store): Likewise.
23323 (vectorizable_condition): Likewise.
23324 (vectorizable_comparison): Likewise.
23326 2017-06-28 Michael Collison <michael.collison@arm.com>
23329 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23331 (arm_gen_movmemqi): Removed unused variable 'i'.
23332 Convert 'for' loop into 'while' loop.
23333 (arm_expand_prologue): Remove last unnecessary set of insn.
23334 (thumb_pop): Remove unused variable 'pushed_words'.
23335 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23337 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23339 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23340 * config/s390/s390-protos.h: Add prototype of
23341 s390_rel_address_ok_p.
23342 * config/s390/s390.c (s390_got_symbol): New function.
23343 (s390_rel_address_ok_p): New function.
23344 (legitimize_pic_address): Use s390_rel_address_ok_p.
23345 (s390_load_got): Use s390_got_symbol.
23346 (s390_option_override): Issue error if
23347 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23348 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23350 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23352 2017-06-27 Andrew Pinski <apinski@cavium.com>
23354 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23355 (X * copysign (1.0, X)): New pattern.
23356 (X * copysign (1.0, -X)): New pattern.
23357 (copysign (-1.0, CST)): New pattern.
23359 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23361 * genmultilib (combination_space): Remove variable.
23362 Validate reuse rules against regular expression for any sequence
23363 of multilib options in any order.
23365 2017-06-27 Michael Collison <michael.collison@arm.com>
23367 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23368 call aarch64_split_simd_combine.
23369 * (aarch64_combine_internal<mode>): Delete pattern.
23370 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23371 Allow register and subreg operands.
23373 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23375 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23376 specific need, just fallback on defaults.
23377 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23379 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23380 Olivier Hainque <hainque@adacore.com>
23382 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23384 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23385 targets. Pick a default if no particular attempt applied.
23386 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23389 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23391 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23392 (x86_64-wrs-vxworks7): Likewise.
23394 2017-06-27 Marek Polacek <polacek@redhat.com>
23397 * ubsan.c (instrument_null): Check get_base_address's result for null.
23399 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23401 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23403 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23405 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23406 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23407 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23408 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23409 New function types.
23410 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23411 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23412 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23413 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23414 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23415 BUILT_IN_FEUPDATEENV): New builtins.
23416 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23417 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23418 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23419 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23421 (builtin_structptr_types): Adjust size.
23422 * tree.c (builtin_structptr_types): Add four entries.
23424 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23425 Olivier Hainque <hainque@adacore.com>
23427 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23428 (TLS_SYM): New local macro, forcing reference to __tls__ on
23429 link command lines for VxWorks 7 RTPs, triggering initialization
23431 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23432 OS features TLS support, true for RTPs on VxWorks 7.
23433 * config/vxworks.c (vxworks_override_options): Setup emutls
23436 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23438 * predict.c (test_prediction_value_range): Use -1U instead of -1
23439 to avoid narrowing conversion warning.
23440 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23441 to avoid narrowing conversion warning.
23442 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23444 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23446 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23448 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23449 64bit configurations.
23450 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23451 (SIZE_TYPE): Likewise.
23452 * config/vxworks.c (vxworks_emutls_var_fields): Use
23453 long_unsigned_type_node instead of unsigned_type_node as the offset
23454 field type, which is "pointer" mode in emutls.c.
23456 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23459 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23461 PR middle-end/81207
23462 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23463 gimple_vuse copying separately from gimple_vdef copying.
23465 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23467 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23470 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23471 Olivier Hainque <hainque@adacore.com>
23473 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23474 for all vxworks7 targets.
23475 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23476 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23477 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23478 variations for VX6/VX7 and 32/64bits later on in ...
23479 (VXWORKS_LIB_SPEC): Leverage new macros.
23480 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23481 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23483 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23485 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23486 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23488 2017-06-26 Carl Love <cel@us.ibm.com>
23490 * config/rs6000/rs6000-c.c: Add support for built-in functions
23491 vector bool char vec_reve (vector bool char);
23492 vector signed char vec_reve (vector signed char);
23493 vector unsigned char vec_reve (vector unsigned char);
23494 vector bool int vec_reve (vector bool int);
23495 vector signed int vec_reve (vector signed int);
23496 vector unsigned int vec_reve (vector unsigned int);
23497 vector bool long long vec_reve (vector bool long long);
23498 vector signed long long vec_reve (vector signed long long);
23499 vector unsigned long long vec_reve (vector unsigned long long);
23500 vector bool short vec_reve (vector bool short);
23501 vector signed short vec_reve (vector signed short);
23502 vector double vec_reve (vector double);
23503 vector float vec_reve (vector float);
23504 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23505 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23506 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23507 (altivec_vreve): New pattern.
23508 * config/rs6000/altivec.h (vec_reve): New define.
23509 * doc/extend.texi (vec_rev): Update the built-in documentation file
23510 for the new built-in functions.
23512 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23514 PR tree-optimization/71815
23515 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23517 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23519 (slsr_process_phi): Likewise.
23520 (replace_uncond_cands_and_profitable_phis): Don't replace a
23521 multiply candidate with a stride of 1 (copy or cast).
23522 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23524 (lowest_cost_path): Likewise.
23525 (total_savings): Likewise.
23527 2017-06-26 Richard Biener <rguenther@suse.de>
23530 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23531 Use def_builtin_pure for all gather builtins.
23533 2017-06-26 Richard Biener <rguenther@suse.de>
23535 PR tree-optimization/81203
23536 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23537 non-dominating BBs.
23539 2017-06-26 Marek Polacek <polacek@redhat.com>
23542 * doc/invoke.texi: Document -Wmultistatement-macros.
23544 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23546 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23547 arm_neon_ok_no_float_abi effective target.
23549 2017-06-26 Richard Biener <rguenther@suse.de>
23551 PR tree-optimization/80928
23552 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23553 (copy_bbs): Set BB_DUPLICATED flag early.
23554 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23556 (execute_on_shrinking_pred): Likewise.
23557 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23558 BB_DUPLICATED blocks.
23559 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23560 iterate over all PHIs considering removal of *gsi.
23562 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23564 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23567 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23569 * config/rs6000/rs6000-string.c: (expand_block_clear,
23570 do_load_for_compare, select_block_compare_mode,
23571 compute_current_alignment, expand_block_compare,
23572 expand_strncmp_align_check, expand_strn_compare,
23573 expand_block_move, rs6000_output_load_multiple)
23574 Move functions related to string/block move/compare
23575 to a separate file.
23576 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23577 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23578 for this function which is now used in two files.
23579 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23580 * config.gcc: Add rs6000-string.o to extra_objs for
23581 targets powerpc*-*-* and rs6000*-*-*.
23583 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23586 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23587 32-bit, since indexed is not valid for DImode.
23588 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23589 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23590 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23591 (define_peephole2 for Altivec d-form store): Likewise.
23594 * multiple_target.c (create_dispatcher_calls): Only create the
23595 dispatcher call if the function is the default clone of a
23596 versioned function.
23598 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23600 PR middle-end/80902
23601 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23602 a call, force the call to not be a tail call.
23604 2017-06-23 Jeff Law <law@redhat.com>
23606 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23609 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23611 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23612 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23613 handling for early expansion of vector shifts (sl,sr,sra,rl).
23614 (builtin_function_type): Add vector shift right instructions
23615 to the unsigned argument list.
23617 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23619 rtl-optimizatoin/79286
23620 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23621 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23622 trap. PIC register plus a const unspec without offset can never trap.
23624 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23626 * tree.h (builtin_structptr_type): New type.
23627 (builtin_structptr_types): Declare new array.
23628 * tree.c (builtin_structptr_types): New array.
23629 (free_lang_data, build_common_tree_nodes): Use it.
23631 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23634 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23637 2017-06-22 Matt Turner <mattst88@gmail.com>
23639 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23640 Lake models to skylake case. Assume skylake for unknown
23641 models with clflushopt.
23643 2017-06-22 Jeff Law <law@redhat.com>
23645 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23646 frame sizes that do not satisfy aarch64_uimm12_shift.
23648 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23650 * profile-count.h (apply_probability,
23651 apply_scale, probability_in): Fix checks for zero.
23653 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23655 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23656 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23658 2016-06-22 Richard Biener <rguenther@suse.de>
23660 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23661 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23662 REDUC_MAX_EXPR support.
23663 (vectorizable_reduction): Likewise.
23664 (vect_create_epilog_for_reduction): Likewise.
23666 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23668 * match.pd (A / (1 << B) -> A >> B): New.
23669 * generic-match-head.c: Include optabs-tree.h.
23670 * gimple-match-head.c: Likewise.
23671 * optabs-tree.h (target_supports_op_p): New.
23672 * optabs-tree.c (target_supports_op_p): New.
23674 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23676 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23677 $gcc_cv_ld --help output.
23678 (gcc_cv_ld_demangle): Likewise.
23679 (gcc_cv_ld_eh_frame_hdr): Likewise.
23680 (gcc_cv_ld_pie): Likewise.
23681 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23682 (gcc_cv_ld_buildid): Likewise.
23683 (gcc_cv_ld_sysroot): Likewise.
23684 (ld_bndplt_support): Likewise.
23685 (ld_pushpopstate_support): Likewise.
23686 * configure: Regenerate.
23687 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23689 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23692 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23693 operands indexes to avoid gap between operands and match_dups.
23695 2017-06-21 Andrew Pinski <apinski@cavium.com>
23697 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23698 Increment Arith_shift and Arith_shift_reg by 1.
23699 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23701 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23702 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23703 (aarch64_strip_extend): Add new argument and test for it.
23704 (aarch64_cheap_mult_shift_p): New function.
23705 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23706 add a cost if it is true.
23707 Update calls to aarch64_strip_extend.
23708 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23710 2017-06-21 Andrew Pinski <apinski@cavium.com>
23712 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23714 (thunderxt88): Likewise.
23715 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23716 (thunderx_prefetch_tune): New variable.
23717 (thunderx2t99_prefetch_tune): Update for the correct values.
23718 (thunderxt88_tunings): New variable.
23719 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23720 generic_prefetch_tune.
23721 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23723 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23725 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23726 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23727 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23728 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23729 (aarch64_atomic_cas<mode>, GPI): Likewise.
23731 2017-06-21 Martin Liska <mliska@suse.cz>
23733 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23734 statements on cold and hot labels.
23735 * predict.c (tree_estimate_probability_bb): Remove the
23736 prediction from this place.
23738 2017-06-21 Martin Liska <mliska@suse.cz>
23740 PR tree-optimization/79489
23741 * gimplify.c (maybe_add_early_return_predict_stmt): New
23743 (gimplify_return_expr): Call the function.
23744 * predict.c (tree_estimate_probability_bb): Remove handling
23746 * predict.def: Update comment about early return predictor.
23747 * gimple-predict.h (is_gimple_predict): New function.
23748 * predict.def: Change default value of early return to 66.
23749 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23751 * passes.def: Put pass_strip_predict_hints to the beginning of
23754 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23756 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23757 FUNCTION_DECL declarations.
23758 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23760 (dwaf2out_decl): Likewise.
23761 * godump.c (go_early_global_decl): Skip call to the real debug hook
23762 for FUNCTION_DECL declarations.
23763 * passes.c (rest_of_decl_compilation): Skip call to the
23764 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23765 -fdump-go-spec is passed.
23767 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23769 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23770 Reorder for compactness.
23771 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23772 (def_builtin_pure, def_builtin_pure2): New functions.
23773 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23775 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23777 * match.pd (nop_convert): New predicate.
23778 ((A +- CST1) +- CST2): Allow some NOP conversions.
23780 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23783 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23784 with ctors/dtors if GOVD_SHARED is set.
23786 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23788 * config/aarch64/aarch64.md (movti_aarch64):
23789 Emit mov rather than orr.
23790 (movtf_aarch64): Likewise.
23791 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23792 Emit mov rather than orr.
23794 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23796 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23797 Swap alternatives, make integer dup more expensive.
23799 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23801 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23802 Return true for non-tls symbols.
23804 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23806 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23807 (cortex-a75): Likewise.
23808 (cortex-a75.cortex-a55): Likewise.
23809 * config/aarch64/aarch64-tune.md: Regenerate.
23810 * doc/invoke.texi (-mtune): Document new values for -mtune.
23812 2017-06-21 Tom de Vries <tom@codesourcery.com>
23814 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23815 stack_size feature.
23816 (Effective-Target Keywords, Other attributes): Suggest using
23817 dg-add-options stack_size feature to get stack limit in stack_size
23818 effective target documentation.
23820 2017-06-21 Julian Brown <julian@codesourcery.com>
23821 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23823 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23824 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23825 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23827 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23828 attribute type list for neon_multiply.
23829 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23830 attribute type list for neon_multiply.
23831 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23832 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23833 attribute type list for neon_multiply.
23834 * config/arm/types.md (crypto_pmull): Add.
23835 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23836 attribute type list.
23838 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23840 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23843 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23845 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23846 to make sure not to dereference a NULL cost_classes_ptr pointer.
23848 2017-06-20 Carl Love <cel@us.ibm.com>
23850 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23851 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23852 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23853 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23854 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23855 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23856 VMULOSW): New enum "unspec" values.
23857 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23858 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23859 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23860 altivec_vmulosw): New patterns.
23861 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23862 VMULOSW): Add definitions.
23864 2017-06-20 Julia Koval <julia.koval@intel.com>
23866 * config/i386/i386.c: Fix rounding expand for new pattern.
23867 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23869 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23871 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23872 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23874 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23876 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23879 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23881 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23882 architecture, then by implementer ID.
23883 * config/aarch64/aarch64-tune.md: Regenerate.
23885 2017-06-20 Richard Biener <rguenther@suse.de>
23887 PR middle-end/81097
23888 * fold-const.c (split_tree): Fold to type before negating.
23890 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23892 * diagnostic-show-locus.c
23893 (selftest::test_fixit_deletion_affecting_newline): New function.
23894 (selftest::diagnostic_show_locus_c_tests): Call it.
23896 2017-06-20 Andreas Schwab <schwab@suse.de>
23899 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23902 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23904 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23905 __ARM_FEATURE_COPROC according to support.
23907 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23909 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23910 Rewritten to avoid overflow for > 32-bit pointers.
23913 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23914 by removing enum keyword.
23915 (ubsan_type_descriptor): Likewise. Formatting fix.
23918 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23919 splitter): Require TARGET_SSE2 in the condition.
23921 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
23924 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
23925 for doing vector set of SFmode on ISA 3.0.
23926 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
23927 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
23929 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
23930 SFmode value into a V4SF variable that was extracted from another
23931 V4SF variable without converting the element to double precision
23932 and back to single precision vector format.
23933 (vsx_insert_extract_v4sf_p9_2): Likewise.
23935 2017-06-19 Jakub Jelinek <jakub@redhat.com>
23937 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
23938 in UWHI to avoid undefined overflow.
23941 * ubsan.h (enum ubsan_encode_value_phase): New.
23942 (ubsan_encode_value): Change second argument to
23943 enum ubsan_encode_value_phase with default value of
23944 UBSAN_ENCODE_VALUE_GENERIC.
23945 * ubsan.c (ubsan_encode_value): Change second argument to
23946 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
23947 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
23948 create_tmp_var_raw instead of create_tmp_var and use a
23950 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
23951 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
23952 ubsan_encode_value callers.
23955 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
23956 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
23957 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
23959 2017-06-19 Richard Biener <rguenther@suse.de>
23961 PR middle-end/81118
23962 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
23963 estimates if we changed anything.
23965 2017-06-19 Richard Biener <rguenther@suse.de>
23967 PR tree-optimization/80887
23968 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
23969 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
23970 simplified lookups, then reset mprts_hook.
23971 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
23973 (try_to_simplify): Likewise.
23975 2017-06-19 Martin Liska <mliska@suse.cz>
23978 * gimplify.c (gimplify_switch_expr):
23979 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
23981 2017-06-19 Martin Liska <mliska@suse.cz>
23983 * doc/install.texi: Document that PGO runs in 4 stages.
23985 2017-06-19 Martin Liska <mliska@suse.cz>
23988 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
23989 to dispatcher function name.
23990 * multiple_target.c (replace_function_decl): New function.
23991 (create_dispatcher_calls): Redirect both edges and references.
23993 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
23995 * profile-count.c (profile_count::dump): Dump quality.
23996 (profile_count::differs_from_p): Update for unsigned val.
23997 * profile-count.h (profile_count_quality): New enum.
23998 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24000 2017-06-19 Richard Biener <rguenther@suse.de>
24002 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24003 struct function as arg.
24004 (estimate_numbers_of_iterations): Export overload with loop arg.
24005 (free_numbers_of_iterations_estimates_loop): Use an overload of
24006 free_numbers_of_iterations_estimates instead.
24007 * tree-cfg.c (remove_bb): Adjust.
24008 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24009 * tree-parloops.c (gen_parallel_loop): Likewise.
24010 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24012 (tree_unroll_loops_completely): Likewise.
24013 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24014 Use an overload instead and export.
24015 (estimated_loop_iterations): Adjust.
24016 (max_loop_iterations): Likewise.
24017 (likely_max_loop_iterations): Likewise.
24018 (estimate_numbers_of_iterations): Take struct function as arg
24020 (loop_exits_before_overflow): Adjust.
24021 (free_numbers_of_iterations_estimates_loop): Use an overload.
24022 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24023 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24025 2017-06-19 Richard Biener <rguenther@suse.de>
24028 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24029 RANGE_EXPR conservatively.
24031 2017-06-16 Carl Love <cel@us.ibm.com>
24033 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24034 definitions for vec_float, vec_float2, vec_floato,
24035 vec_floate built-ins.
24036 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24037 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24039 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24040 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24041 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24042 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24043 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24044 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24045 vec_floato): Add builtin defines.
24046 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24047 Update the built-in documentation file for the new built-in
24050 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24052 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24053 (mthumb): Mark as the negative of -marm.
24055 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24057 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24059 (ARM Options, -mtune): Document that this accepts the same
24060 extension options as -mcpu.
24061 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24063 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24065 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24066 permitted extensions.
24068 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24070 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24071 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24072 (armv8-m.main): Add option +nodsp.
24073 * config/arm/arm-cpu-cdata.h: Regenerated.
24075 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24077 * config/arm/t-fuchsia: New file.
24078 * config.gcc (arm*-*-fuchsia*): Use it.
24080 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24082 * config/arm/t-symbian: Rewrite for new option infrastructure.
24084 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24086 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24087 (MULTILIB_REQUIRED): Likewise.
24089 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24091 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24092 (MULTILIB_RESUE): Likewise.
24093 (MULTILIB_MATCHES): Likewise.
24094 (MULTLIB_REQUIRED): Likewise.
24096 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24098 * config/arm/t-rtems: Rewrite for new option framework.
24100 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24102 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24103 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24104 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24105 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24106 * config/arm/t-multilib: ... here.
24107 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24108 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24109 armv7-a and armv8*-a when A-profile libraries have not been built.
24110 * config/arm/t-rmprofile: Rewrite.
24112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24114 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24115 with a backslash. Remove the backslash after substituting unescaped
24117 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24119 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24121 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24122 the driver to pass through the default setting of -mfloat-abi.
24123 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24125 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24126 (all_feat_combs): New rule.
24127 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24129 * config/arm/t-aprofile: Rewrite.
24131 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24133 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24134 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24135 fpu is not specified by the user/command-line.
24136 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24137 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24138 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24139 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24140 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24141 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24143 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24145 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24146 * config/arm/t-arm-elf: Rewritten.
24148 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24150 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24151 have some floating-point instructions.
24152 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24153 (TARGET_MAYBE_HARD_FLOAT): New macro.
24154 * config/arm/arm-builtins.c (arm_init_builtins): Use
24155 TARGET_MAYBE_HARD_FLOAT.
24156 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24158 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24160 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24161 (configargs.h): Include it.
24162 (arm_print_hint_for_fpu_option): New function.
24163 (arm_parse_fpu_option): New function.
24164 (candidate_extension): New class.
24165 (arm_canon_for_multilib): New function.
24166 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24167 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24168 (ARCH_CANONICAL_SPECS): New macro.
24169 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24171 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24173 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24174 are set after handling multilib fragments. Set target_cpu_default2
24177 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24179 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24181 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24182 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24184 (check_arch): Likewise.
24185 * config/arm/arm.c (arm_configure_build_target): Handle
24186 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24187 options in the default.
24189 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24191 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24192 when an option is an alias of another.
24193 * config/arm/parsecpu.awk (optalias): New parser token.
24194 (gen_comm_data): Mark non-alias options as such. Emit entries
24195 for extension aliases.
24196 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24197 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24198 (armv6kz, armv6zk, armv6t2): Likewise.
24199 (armv7): Make vfpv3-d16 an alias.
24200 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24202 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24203 Sort in canonical order.
24204 (armv8-a): Sort in canonical order.
24205 (armv8.1-a, armv8.2-a): Likewise.
24206 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24208 (cortex-a9): Sort in canonical order.
24209 * config/arm/arm.c (selftests.h): Include it.
24210 (arm_test_cpu_arch_data): New function.
24211 (arm_run_self_tests): New function.
24212 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24213 (targetm): Move declaration to the end of the file.
24214 * arm-cpu-cdata.h: Regenerated.
24216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24218 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24219 call to target_mode_check describing the type of option passed.
24220 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24221 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24222 arm_parse_cpu_option_name to match parameters against list of
24224 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24225 arm_arch_core_flags data structure.
24226 * config/arm/arm-cpu_cdata.h: Regenerated.
24228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24230 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24232 (arm_print_hint_for_cpu_option): Likewise.
24233 (arm_print_hint_for_arch_option): Likewise.
24234 (arm_parse_cpu_option_name): Likewise.
24235 (arm_parse_arch_option_name): Likewise.
24236 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24237 of entries in the all_fpus list.
24238 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24239 (arm_parse_cpu_option_name): Declare.
24240 (arm_parse_arch_option_name): Declare.
24241 (arm_parse_option_features): Declare.
24242 (arm_intialize_isa): Declare.
24243 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24245 (gen_comm_data): ... here. Make definitions non-static.
24246 * config/arm/arm-cpu-data.h: Regenerated.
24247 * config/arm/arm-cpu-cdata.h: Regenerated.
24249 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24251 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24252 (cpu_arch_extension): New structure.
24253 (cpu_arch_option, arch_option, cpu_option): New structures.
24254 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24255 architecture types.
24256 (gen_data): Generate new format data tables.
24257 * config/arm/arm.c (cpu_tune): New structure.
24258 (cpu_option, processors): Delete.
24259 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24260 (arm_print_hint_for_cpu_option): ... this and ...
24261 (arm_print_hint_for_arch_option): ... this.
24262 (arm_parse_arch_cpu_name): Delete. Replace with ...
24263 (arm_parse_cpu_option_name): ... this and ...
24264 (arm_parse_arch_option_name): ... this.
24265 (arm_unrecognized_feature): Change type of target parameter to
24267 (arm_parse_arch_cpu_features): Delete. Replace with ...
24268 (arm_parse_option_features): ... this.
24269 (arm_configure_build_target): Rework to use new configuration data
24271 (arm_print_tune_info): Rework for new configuration data tables.
24272 * config/arm/arm-cpu-data.h: Regenerated.
24273 * config/arm/arm-cpu.h: Regenerated.
24275 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24277 * Makefile.in (OBJS): Move sbitmap.o from here ...
24278 (OBJS-libcommon): ... to here.
24280 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24282 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24283 (ISA_ALL_CRYPTO): New macro.
24284 (ISA_ALL_SIMD): New macro
24285 (ISA_ALL_FP): New macro.
24286 * config/arm/arm.c (fpu_bitlist): Update initializer.
24287 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24289 (arm9e): Add fpu. Add option for nofp
24290 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24291 (arm926ej-s, arm1026ej-s): Likewise.
24292 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24293 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24294 neon-fp16, neon-vfpv4, nofp and nosimd.
24295 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24296 (cortex-a8): Add fpu. Add option for nofp.
24297 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24298 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24299 (cortex-r4f): Add fpu.
24300 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24301 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24303 (cortex-r8): Likewise.
24304 (cortex-m4): Add fpu. Add option for nofp.
24305 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24306 (cortex-a17.cortex-a7): Likewise.
24307 (cortex-a32): Add fpu. Add options for crypto and nofp.
24308 (cortex-a35, cortex-a53): Likewise.
24309 (cortex-a57): Add fpu. Add option for crypto.
24310 (cortex-a72, cortex-a73): Likewise.
24311 (exynos-m1): Likewise.
24312 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24313 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24314 (cortex-m33): Add fpu. Add option for nofp.
24315 * config/arm/arm-cpu-cdata.h: Regenerated
24316 * config/arm/arm-cpu-data.h: Regenerated.
24318 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24320 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24321 (armv5te, armv5tej): Likewise.
24322 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24323 (armv7): Add options fp and vfpv3-d16.
24324 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24325 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24327 (armv7ve): Likewise.
24328 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24329 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24330 (armv8-a): Add nocrypto option.
24331 (armv8.1-a, armv8.2-a): Likewise.
24332 (armv8-m.main): add options fp, fp.dp and nofp.
24334 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24336 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24338 (armv8-a+crc): Delete.
24339 (armv8.1-a): Add options simd, crypto and nofp.
24340 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24341 (armv8.2-a+fp16): Delete.
24342 (armv8-m.main): Add option dsp.
24343 (armv8-m.main+dsp): Delete.
24344 (cortex-a8): Add fpu. Add nofp option.
24345 (cortex-a9): Add fpu. Add nofp and nosimd options.
24346 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24347 link to main cpu and architecture data structures.
24348 (gen_comm_data): Only put isa attributes from the main architecture
24350 (option): New statement for architecture and CPU entries.
24351 * arm.c (struct cpu_option): New structure.
24352 (struct processors): Add entry for options.
24353 (arm_unrecognized_feature): New function.
24354 (arm_parse_arch_cpu_name): Ignore any characters after the first
24356 (arm_parse_arch_cpu_feature): New function.
24357 (arm_configure_build_target): Separate out any CPU and architecture
24358 features and parse separately. Don't error out if -mfpu=auto is
24359 used with only an architecture string.
24360 (arm_print_asm_arch_directives): New function.
24361 (arm_file_start): Call it.
24362 * config/arm/arm-cpu-cdata.h: Regenerated.
24363 * config/arm/arm-cpu-data.h: Likewise.
24364 * config/arm/arm-tables.opt: Likewise.
24366 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24368 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24369 assembler when it is not -mfpu=auto.
24371 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24373 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24374 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24375 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24376 (ASM_CPU_SPEC): Rewrite.
24377 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24378 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24379 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24380 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24381 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24382 copied string is NUL-terminated. Also strip any characters prefixed
24384 (arm_rewrite_selected_arch): New function.
24385 (arm_rewrite_march): New function.
24387 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24389 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24390 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24391 (march, mcpu, mtune): Convert to string-based options.
24392 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24393 (arm_parse_arch_cpu_name): New function.
24394 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24395 identify selected architecture or CPU.
24396 (arm_option_save): New function.
24397 (TARGET_OPTION_SAVE): Redefine.
24398 (arm_option_restore): Restore string options.
24399 (arm_option_print): Print string options.
24401 2017-06-16 Martin Sebor <msebor@redhat.com>
24403 PR tree-optimization/80933
24404 PR tree-optimization/80934
24405 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24406 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24407 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24408 (gimple_fold_builtin): Call them.
24410 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24412 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24413 as unlikely; update profile.
24415 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24417 * predict.c (force_edge_cold): Handle declaring edges impossible
24420 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24422 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24424 (try_unroll_loop_completely): Fix reporting.
24426 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24428 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24430 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24433 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24434 if given a non-constant argument for an intrinsic which requires a
24437 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24439 * profile.c (compare_freqs): New function.
24440 (branch_prob): Sort edge list.
24441 (find_spanning_tree): Assume that the list is priority sorted.
24443 2017-06-16 Richard Biener <rguenther@suse.de>
24445 PR tree-optimization/81090
24446 * passes.def (pass_record_bounds): Remove.
24447 * tree-pass.h (make_pass_record_bounds): Likewise.
24448 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24449 make_pass_record_bounds): Likewise.
24450 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24451 not free niter estimates at the beginning but at the end.
24452 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24454 2017-06-16 Richard Biener <rguenther@suse.de>
24456 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24457 initializer to workaround ICE in host GCC 4.8.
24459 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24461 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24463 (clone_inlined_nodes): Update.
24465 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24467 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24468 prefetch settings, and enable prefetching by default at -O3.
24470 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24472 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24473 Set flag_prefetch_loop_arrays according to tuning data.
24475 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24477 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24478 New tune structure.
24479 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24480 [Unrelated to main purpose of the patch] Place the pointer field last
24481 to enable type checking errors when tune structure are wrongly merged.
24482 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24483 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24484 (thunderx2t99_prefetch_tune): New tune constants.
24485 (tune_params *_tunings): Update all tunings (no functional change).
24486 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24487 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24488 from tunings structures.
24490 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24493 * ubsan.c (instrument_null): Add T argument, use it instead
24494 of computing it based on IS_LHS.
24495 (instrument_object_size): Likewise.
24496 (pass_ubsan::execute): Adjust instrument_null and
24497 instrument_object_size callers to pass gimple_get_lhs or
24498 gimple_assign_rhs1 result to it. Use instrument_null instead of
24499 calling get_base_address and instrument_mem_ref. Handle
24500 aggregate call arguments for object-size sanitization.
24502 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24504 PR tree-optimization/81089
24505 * tree-vrp.c (is_masked_range_test): Validate operands of
24508 2017-06-15 Martin Sebor <msebor@redhat.com>
24511 * dumpfile.c (dump_register): Avoid calling memset to initialize
24512 a class with a default ctor.
24513 * gcc.c (struct compiler): Remove const qualification.
24514 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24515 * hash-table.h: Ditto.
24516 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24518 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24519 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24521 * params.h (struct param_info): Make struct members non-const.
24522 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24523 with default initialization.
24524 * vec.h (vec_copy_construct, vec_default_construct): New helper
24526 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24527 with vec_copy_construct.
24528 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24529 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24530 * doc/invoke.texi (-Wclass-memaccess): Document.
24532 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24534 * emit-rtl.h (is_leaf): Update comment about local
24535 register allocator.
24537 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24540 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24541 for a variable to have a section before checking if the section has a
24543 Set section to.persistent if persistent attribute is set.
24544 Warn if .persistent attribute is used on an automatic variable.
24546 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24548 PR rtl-optimization/80474
24549 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24551 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24553 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24556 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24558 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24559 (casesi): Emit bounds checking as RTL.
24560 (casesi_internal_mips16_<mode>): Remove bounds checking.
24562 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24564 * config/xtensa/xtensa.c (xtensa_option_override): Append
24565 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24566 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24567 xtensa_doloop_hooks): Define unconditionally.
24568 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24570 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24571 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24572 for it in xtensa_option_override.
24573 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24574 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24576 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24578 * doc/cppopts.texi: Document '-' special value to -MF.
24580 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24582 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24583 (cortex_a53_fconst): Likewise.
24584 (cortex_a53_fpmul): Likewise.
24585 (cortex_a53_f_load_64): Likewise.
24586 (cortex_a53_f_load_many): Likewise.
24587 (cortex_a53_advsimd_alu): Likewise.
24588 (cortex_a53_advsimd_alu_q): Likewise.
24589 (cortex_a53_advsimd_mul): Likewise.
24590 (cortex_a53_advsimd_mul_q): Likewise.
24591 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24592 Add missing fmul, r2f_cvt and fconst cases.
24594 2017-06-14 Richard Biener <rguenther@suse.de>
24596 PR middle-end/81088
24597 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24599 (fold_binary_loc): When associating do not treat pre-existing
24600 TREE_OVERFLOW on literal constants as a reason to allow
24601 TREE_OVERFLOW on associated literal constants.
24603 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24605 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24606 (MASK_FEATURES): New macro.
24607 * config/sparc/sparc.c (sparc_option_override): Remove the special
24608 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24610 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24612 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24613 a division of 0 if non-call exceptions are enabled.
24615 2017-06-14 Andrew Pinski <apinski@cavium.com>
24616 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24619 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24620 Improve vector initialization code gen for only variable case.
24622 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24624 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24626 2017-06-14 Richard Biener <rguenther@suse.de>
24628 PR tree-optimization/81083
24629 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24632 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24634 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24635 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24636 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24637 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24638 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24639 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24641 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24643 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24644 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24646 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24648 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24650 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24652 * config/rs6000/t-rtems: Don't handle SPE.
24654 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24656 * config/rs6000/t-linux: Don't handle SPE.
24658 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24660 * config/rs6000/eabispe.h: Delete file.
24662 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24664 * config/rs6000/t-spe: Delete file.
24666 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24668 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24669 (rs6000_legitimate_offset_address_p): Return false for anything in
24670 V2SImode or V2SFmode.
24672 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24674 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24675 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24676 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24678 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24679 (rs6000_legitimate_offset_address_p): Ditto.
24680 (rs6000_emit_move): Ditto.
24681 (rs6000_init_builtins): Remove V4HI_type_node.
24683 2017-06-13 Martin Liska <mliska@suse.cz>
24686 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24687 (gate_asan): Likewise.
24688 * asan.h (asan_no_sanitize_address_p): Remove the function.
24689 (sanitize_flags_p): New function.
24690 * builtins.def: Fix coding style.
24691 * common.opt: Use renamed enum value.
24692 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24693 * doc/extend.texi: Document no_sanitize attribute.
24694 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24695 to SANITIZE_UNDEFINED_NONDEFAULT.
24696 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24697 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24698 Use sanitize_flags_p.
24699 * gimplify.c (gimplify_function_tree): Likewise.
24700 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24701 * opts.c (parse_no_sanitize_attribute): New function.
24702 (common_handle_option): Use renamed enum value.
24703 * opts.h (parse_no_sanitize_attribute): Declare.
24704 * tree.c (sanitize_flags_p): New function.
24705 * tree.h: Declared here.
24706 * tsan.c: Use sanitize_flags_p.
24707 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24708 (instrument_mem_ref): Likewise.
24709 (instrument_bool_enum_load): Likewise.
24710 (do_ubsan_in_current_function): Remove the function.
24711 (pass_ubsan::execute): Use sanitize_flags_p.
24712 * ubsan.h: Remove do_ubsan_in_current_function
24713 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24714 (dump_function_to_file): Use it here.
24716 2017-06-13 Martin Jambor <mjambor@suse.cz>
24718 PR tree-optimization/80803
24719 PR tree-optimization/81063
24720 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24721 (propagate_subaccesses_across_link): Enqueue subtree whenever
24722 necessary instead of relying on the caller.
24724 2017-06-13 Martin Jambor <mjambor@suse.cz>
24726 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24727 that have a first_link.
24728 (sort_and_splice_var_accesses): Do not check first_link before
24730 (subtree_mark_written_and_enqueue): Likewise.
24731 (propagate_all_subaccesses): Likewise and do not stop at first
24732 parent with a first_link.
24734 2017-06-13 Martin Jambor <mjambor@suse.cz>
24736 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24739 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24741 * match.pd: New pattern.
24743 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24745 * tree-vrp.c (is_masked_range_test): New function.
24746 (register_edge_assert_for): Determine ranges for
24749 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24751 PR tree-optimization/67328
24752 * fold-const.c (maskable_range_p): New function.
24753 (build_range_check): Generate bittests if possible.
24755 2017-06-13 Martin Liska <mliska@suse.cz>
24757 * gimple-pretty-print.c (dump_probability): Add new argument.
24758 (dump_edge_probability): Dump both probability and count.
24759 (dump_gimple_label): Likewise.
24760 (dump_gimple_bb_header): Likewise.
24762 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24765 * config/avr/avr-devices.c: Fix indentation.
24766 * config/avr/gen-avr-mmcu-specs.c: Dito.
24768 2017-06-13 Richard Biener <rguenther@suse.de>
24770 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24771 instead get vector type from stmt_info.
24772 (vectorizable_reduction): Adjust. Remove dead code.
24774 2017-06-13 Richard Biener <rguenther@suse.de>
24776 PR middle-end/81065
24777 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24778 case of C * (x * C2 + C3).
24779 (fold_addr_of_array_ref_difference): Properly fold index difference.
24781 2017-06-12 David S. Miller <davem@davemloft.net>
24784 * config/sparc/sparc.md (return expander): Emit frame blockage if
24785 function uses alloca.
24787 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24789 * combine.c (make_field_assignment): Check len rather than the mode
24790 precision when calling force_to_mode.
24792 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24794 Support multilibs and devices that see flash in RAM address range.
24797 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24798 (avr_mcu_t) <flash_pm_offset>: New field.
24799 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24800 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24801 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24802 (AVR_TINY_PM_OFFSET): Remove macro.
24803 * config/avr/avr.opt (-mshort-calls): New option.
24804 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24805 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24806 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24807 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24808 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24809 instead of avr_arch->have_jmp_call.
24810 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24811 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24812 avr_arch->flash_pm_offset to define.
24813 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24814 new field flash_pm_offset. Add entry for avrxmega3.
24815 (avr_texinfo): Add entry for avrxmega3.
24816 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24817 attiny212, attiny214,
24818 attiny412, attiny414, attiny416, attiny417,
24819 attiny814, attiny816, attiny817,
24820 attiny1614, attiny1616, attiny1617,
24821 attiny3214, attiny3216, attiny3217.
24822 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24823 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24824 (avr_print_operand_address) [AVR_TINY]: Same.
24825 (avr_asm_init_sections) <readonly_data_section>: Only patch
24826 callback if avr_arch->flash_pm_offset = 0.
24827 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24828 for rodata if avr_arch->flash_pm_offset != 0.
24829 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24830 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24831 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24832 (m_options): Append opt_rcall.
24833 (m_dirnames): Append dir_rcall.
24834 * config/avr/t-multilib: Regenerate.
24836 * configure.ac [target=avr]: Check whether avrxmega3 default
24837 linker description file works as needed.
24838 * configure: Regenerate.
24839 * doc/avr-mmcu.texi: Regenerate.
24840 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24841 <__AVR_ARCH__>: Document avrxmega3 and 103.
24842 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24843 <__AVR_SHORT_CALLS__>: Document it.
24844 <__AVR_PM_BASE_ADDRESS__>: Document it.
24845 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24846 (AVR Variable Attributes) <progmem>: Document this is
24847 not needed for avrxmega3.
24848 (AVR Named Address Spaces) <__flash>: Dito.
24850 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24852 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24854 2017-06-12 Doug Rupp <rupp@adacore.com>
24856 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24857 Append vxworks-stdint.h to the tm_file list.
24858 * config/vxworks-stdint.h: New file.
24860 2017-06-12 Martin Liska <mliska@suse.cz>
24862 PR tree-optimization/81041
24863 * tree-profile.c (gimple_gen_ic_func_profiler):
24864 Create an extra BB in profile-generate
24865 (gimple_gen_time_profiler): Likewise.
24867 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24869 PR tree-optimization/81003
24870 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24871 (update_range_test): Use it instead of force_gimple_operand_gsi.
24873 2017-06-12 Richard Biener <rguenther@suse.de>
24875 PR tree-optimization/81053
24876 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24877 with backedge value not defined in loop. Simplify def stmt
24880 2017-06-11 Tom de Vries <tom@codesourcery.com>
24883 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24885 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24886 nvptx_cannot_force_const_mem.
24888 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24890 * opts.c (finish_options): Move test for flag_split_stack after
24891 it has been initialized.
24893 2017-06-11 Jason Merrill <jason@redhat.com>
24895 * tree.h (id_equal): New.
24896 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24897 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24898 instead of strcmp of IDENTIFIER_POINTER.
24900 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24902 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24903 (mark_all_inlined_calls_cdtor): Fix formating.
24904 (inline_transform): Rescale profile before inlining.
24906 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24908 * cgraph.h (cgraph_edge::clone): Update prototype.
24909 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24910 (cgraph_node::create_clone): Update.
24911 (cgraph_node::create_version_clone): Update.
24912 * tree-inline.c (copy_bb): Update.
24913 (expand_call_inline): Update.
24915 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24917 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24918 factored out from ...
24919 (rs6000_emit_prologue): ... here.
24921 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24923 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
24924 factored out from ...
24925 (rs6000_emit_prologue): ... here.
24927 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24929 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
24931 (handle_missing_profiles): Fix computation of tp_first_run.
24932 (counts_to_freqs): Do not touch freqs when count is 0.
24934 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24936 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
24939 2017-06-10 Tom de Vries <tom@codesourcery.com>
24941 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24942 attributes): Document signal effective target.
24944 2017-06-10 Tom de Vries <tom@codesourcery.com>
24946 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24947 Document effective target stack_size.
24949 2017-06-09 David Malcolm <dmalcolm@redhat.com>
24951 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
24952 to the edit_context if they can be auto-applied.
24954 2017-06-9 Ian Lance Taylor <iant@golang.org>
24956 * opts.c (finish_options): If -fsplit-stack, disable implicit
24957 -forder-blocks-and-partition.
24958 * doc/invoke.texi (Optimize Options): Document that when using
24959 -fsplit-stack -forder-blocks-and-partition is not implicitly
24962 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24964 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
24965 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
24966 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
24967 * builtins.def (abort, trap, unreachable): Declare cold.
24968 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
24969 * tree-core.h (ECF_COLD): New.
24970 * tree.c (set_call_expr_flags): Handle ECF_COLD.
24971 (build_common_builtin_nodes): Mark unreachable and abort as cold.
24973 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24975 * predict.c (unlikely_executed_stmt_p): Cleanup.
24977 2017-06-09 Richard Biener <rguenther@suse.de>
24979 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
24980 model if the ref is always written to.
24982 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24984 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
24986 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24988 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
24991 2017-06-09 Tom de Vries <tom@codesourcery.com>
24994 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
24995 "target cannot support label values" when encountering LABEL_REF.
24997 2017-06-09 Martin Liska <mliska@suse.cz>
24999 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25000 (gimple_gen_ic_func_profiler): Emit direct comparison
25001 of __gcov_indirect_call_callee with NULL.
25002 (gimple_gen_time_profiler): Change probability from
25003 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25005 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25007 * profile.c (edge_gcov_counts): Turn to pointer.
25008 (compute_branch_probabilities, compute_branch_probabilities): Update.
25009 (branch_prob): Do not clear edge_gcov_count.
25010 * profile.h (edge_gcov_counts): Turn to pointer.
25011 (edge_gcov_count): Update.
25013 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25015 * gimple.h (gimple_check_failed): Mark cold.
25017 2017-06-09 Richard Biener <rguenther@suse.de>
25019 PR tree-optimization/66623
25020 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25021 refactor check_reduction into two parts, properly computing
25022 whether we have to check reduction validity for outer loop
25025 2017-06-09 Richard Biener <rguenther@suse.de>
25027 PR tree-optimization/79483
25028 * graphite-scop-detection.c (order): New global.
25029 (get_order): Compute bb to order mapping that satisfies code
25030 generation constraints.
25031 (cmp_pbbs): New helper.
25032 (build_scops): Start domwalk at entry block, sort generated
25035 2017-06-09 Richard Biener <rguenther@suse.de>
25037 PR middle-end/81007
25038 * ipa-polymorphic-call.c
25039 (ipa_polymorphic_call_context::restrict_to_inner_class):
25040 Skip FIELD_DECLs with error_mark_node type.
25041 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25044 2017-06-09 Martin Liska <mliska@suse.cz>
25046 * predict.c (struct branch_predictor): New struct.
25047 (test_prediction_value_range): New test.
25048 (predict_c_tests): New function.
25049 * selftest-run-tests.c (selftest::run_tests): Run the function.
25050 * selftest.h: Declare new tests.
25052 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25055 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25056 gen_add3_insn did not fail.
25057 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25058 r0, construct that number in a temporary reg and add that reg to r0.
25059 If asked to put the result in r0 as well, fail.
25061 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25063 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25064 for early expansion of vec_eqv.
25066 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25068 PR middle-end/81005
25069 * ubsan.c (instrument_null): Avoid pointless code temporary.
25070 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25072 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25076 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25079 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25082 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25084 * predict.c (unlikely_executed_edge_p): Move ahead.
25085 (probably_never_executed_edge_p): Use it.
25087 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25089 PR middle-end/79988
25090 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25091 gimple_call_builtin_p call.
25093 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25095 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25096 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25097 rtl_check_failed_type2, rtl_check_failed_code1,
25098 rtl_check_failed_code2, rtl_check_failed_code_mode,
25099 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25100 rtvec_check_failed_bounds, rtl_check_failed_flag,
25101 _fatal_insn_not_found, _fatal_insn): Likewise.
25102 * tree.h (tree_contains_struct_check_failed,
25103 tree_check_failed, tree_not_check_failed,
25104 tree_class_check_failed, tree_range_check_failed,
25105 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25106 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25107 tree_operand_check_failed, omp_clause_check_failed,
25108 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25111 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25113 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25114 flag_branch_probabilities.
25115 * ipa-inline.c (edge_badness): Likewise.
25116 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25117 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25118 * predict.c (maybe_hot_frequency_p): Likewise.
25119 (probably_never_executed): Likewise.
25120 * sched-ebb.c (schedule_ebbs): Likewise.
25121 * sched-rgn.c (find_single_block_region): Likewise.
25122 * tracer.c (tail_duplicate): Likewise.
25124 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25126 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25127 longer requires x_flag_profile_use.
25129 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25131 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25132 instead of flag_reorder_blocks_and_partition.
25133 * dbxout.c (dbxout_function_end): Likewise.
25134 * dwarf2out.c (gen_subprogram_die): Likewise.
25135 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25136 * hw-doloop.c (reorg_loops): Likewise.
25137 * varasm.c (assemble_start_function,
25138 assemble_end_function): Likewise.
25139 (decide_function_section): Do not check for
25140 flag_reorder_blocks_and_partition.
25142 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25144 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25146 (chkp_get_hard_register_fake_addr_expr): Ditto.
25147 (chkp_build_addr_expr): Add check for hard reg case.
25148 (chkp_parse_array_and_component_ref): Ditto.
25149 (chkp_find_bounds_1): Ditto.
25150 (chkp_process_stmt): Don't generate bounds store for
25153 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25155 * predict.c (maybe_hot_bb_p): Do not check profile status.
25156 (maybe_hot_edge_p): Likewise.
25157 (probably_never_executed): Check for zero counts even if profile
25159 (unlikely_executed_edge_p): New function.
25160 (unlikely_executed_stmt_p): New function.
25161 (unlikely_executed_bb_p): New function.
25162 (set_even_probabilities): Use unlikely predicates.
25163 (combine_predictions_for_bb): Likewise.
25164 (predict_paths_for_bb): Likewise.
25165 (predict_paths_leading_to_edge): Likewise.
25166 (determine_unlikely_bbs): New function.
25167 (estimate_bb_frequencies): Use it.
25168 (compute_function_frequency): Use zero counts even if profile is
25170 * profile-count.h: Fix typo.
25172 2017-08-08 Julia Koval <julia.koval@intel.com>
25174 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25175 _mm512_mask_cvtsepi16_storeu_epi8,
25176 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25177 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25178 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25179 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25180 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25181 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25182 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25183 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25184 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25185 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25186 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25187 __builtin_ia32_pmovuswb256mem_mask,
25188 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25189 __builtin_ia32_pmovwb512mem_mask): New builtins.
25191 2017-08-08 Julia Koval <julia.koval@intel.com>
25193 PR target/73350,80862
25194 * config/i386/subst.md (round): Fix round pattern.
25195 * config/i386/i386.c (ix86_erase_embedded_rounding):
25196 Fix erasing rounding for the fixed pattern.
25198 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25200 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25202 2017-06-08 Martin Liska <mliska@suse.cz>
25204 PR gcov-profile/80911
25205 * gcov.c (block_info::block_info): New constructor.
25207 2017-06-07 Carl Love <cel@us.ibm.com>
25209 * config/rs6000/rs6000-c: The return type of the following
25210 built-in functions was implemented as int not long long. Fix sign
25211 of return value for the unsigned version of vec_mulo and vec_mule.
25212 vector unsigned long long vec_bperm (vector unsigned long long,
25213 vector unsigned char)
25214 vector signed long long vec_mule (vector signed int,
25216 vector unsigned long long vec_mule (vector unsigned int,
25217 vector unsigned int)
25218 vector signed long long vec_mulo (vector signed int,
25220 vector unsigned long long vec_mulo (vector unsigned int,
25221 vector unsigned int)
25222 * doc/extend.texi: Fix the documentation for the built-in
25225 2017-06-07 Carl Love <cel@us.ibm.com>
25228 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25231 2017-06-07 Carl Love <cel@us.ibm.com>
25233 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25234 support, Generate doublehv for signed int/float for BE case only.
25236 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25238 * doc/invoke.texi (mcx16): Rewrite.
25240 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25242 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25243 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25244 *mov<mode>_softfloat, and an anonymous splitter): Use
25245 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25247 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25249 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25251 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25252 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25253 (rs6000_debug_reg_global): Adjust.
25254 (rs6000_init_hard_regno_mode_ok): Adjust.
25255 (rs6000_dbx_register_number): Adjust.
25256 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25257 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25258 Remove SPE_ACC and SPEFSCR.
25259 (REG_ALLOC_ORDER): Ditto.
25260 (FRAME_POINTER_REGNUM): Change to 111.
25261 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25262 (REG_CLASS_NAMES): Ditto.
25263 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25264 (REGISTER_NAMES): Ditto.
25265 (ADDITIONAL_REG_NAMES): Ditto.
25266 (rs6000_reg_names): Ditto.
25267 * config/rs6000/rs6000.md: Renumber some register number
25270 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25272 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25274 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25275 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25277 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25278 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25279 Delete the SPE high registers.
25280 (REG_ALLOC_ORDER): Ditto.
25281 (enum reg_class): Remove SPE_HIGH_REGS.
25282 (REG_CLASS_NAMES): Ditto.
25283 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25284 (REGISTER_NAMES): Ditto.
25285 (rs6000_reg_names): Ditto.
25286 * doc/tm.texi.in: Remove SPE as example.
25287 * doc/tm.texi: Regenerate.
25289 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25291 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25292 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25293 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25294 * config/rs6000/rs6000.md (type): Remove "brinc".
25296 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25298 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25299 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25300 * config/rs6000/linuxspe.h: Delete file.
25301 * config/rs6000/rs6000.md: Don't include spe.md.
25302 * config/rs6000/spe.h: Delete file.
25303 * config/rs6000/spe.md: Delete file.
25304 * config/rs6000/t-rs6000: Remove spe.md.
25306 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25308 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25309 (reg_or_none500mem_operand): Delete.
25310 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25311 instead of reg_or_none500mem_operand.
25313 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25315 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25316 handling of SPE flags.
25317 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25319 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25321 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25323 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25324 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25325 paired_divv2sf3): Similar.
25326 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25327 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25328 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25330 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25331 Rename the paired_* instruction patterns.
25332 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25334 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25335 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25336 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25337 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25338 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25339 PAIRED_VECTOR_MODE.
25340 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25341 (spe_func_has_64bit_regs_p): Delete.
25342 (spe_expand_predicate_builtin): Delete.
25343 (spe_expand_evsel_builtin): Delete.
25344 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25345 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25346 (invalid_e500_subreg): Delete.
25347 (rs6000_legitimize_address): Always force_reg op2 as well, for
25348 paired single memory accesses.
25349 (rs6000_member_type_forces_blk): Delete.
25350 (rs6000_spe_function_arg): Delete.
25351 (rs6000_expand_unop_builtin): Delete SPE handling.
25352 (rs6000_expand_binop_builtin): Ditto.
25353 (spe_expand_stv_builtin): Delete.
25354 (bdesc_2arg_spe): Delete.
25355 (spe_expand_builtin): Delete.
25356 (spe_expand_predicate_builtin): Delete.
25357 (spe_expand_evsel_builtin): Delete.
25358 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25359 (spe_init_builtins): Delete.
25360 (spe_func_has_64bit_regs_p): Delete.
25361 (savres_routine_name): Delete "info" parameter. Adjust callers.
25362 (rs6000_emit_stack_reset): Ditto.
25363 (rs6000_dwarf_register_span): Delete.
25364 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25365 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25366 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25368 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25370 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25371 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25372 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25373 mulv2sf3, divv2sf3): Delete expanders.
25375 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25377 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25379 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25381 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25382 * config/rs6000/rs6000.c: Ditto.
25384 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25386 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25387 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25388 comparison_operator.
25390 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25392 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25393 * config/rs6000/rs6000.opt: Ditto.
25394 * config/rs6000/t-rtems: Ditto.
25396 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25398 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25399 TARGET_E500_SINGLE by 0, simplify.
25400 * config/rs6000/rs6000.c: Ditto.
25401 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25402 (spe_build_register_parallel): Delete.
25403 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25404 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25405 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25406 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25407 (E500_CONVERT): Delete.
25408 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25410 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25412 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25413 * config/rs6000/dfp.md: Ditto.
25414 (negdd2, *negdd2_fpr): Merge.
25415 (absdd2, *absdd2_fpr): Merge.
25416 (negtd2, *negtd2_fpr): Merge.
25417 (abstd2, *abstd2_fpr): Merge.
25418 * config/rs6000/e500.h: Delete file.
25419 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25420 TARGET_FPRS by 1 and simplify.
25421 * config/rs6000/rs6000-c.c: Ditto.
25422 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25423 TARGET_DF_SPE by 0.
25424 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25426 * config/rs6000/rs6000.md: Ditto.
25427 (floatdidf2, *floatdidf2_fpr): Merge.
25428 (move_from_CR_gt_bit): Delete.
25429 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25430 (E500_CR_IOR_COMPARE): Delete.
25431 (All patterns that require !TARGET_FPRS): Delete.
25432 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25434 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25436 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25438 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25440 * graphds.c (add_edge): Intitialize edge's attached data.
25441 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25442 pointer parameter. Call pointed function on each edge during
25443 graph traversing. Skip traversing the edge when the function
25445 (graphds_dfs, graphds_scc): Ditto.
25446 (for_each_edge): New parameter. Pass the new parameter to callback
25448 * graphds.h (skip_edge_callback): New function pointer type.
25449 (graphds_dfs, graphds_scc): New function pointer parameter.
25450 (graphds_edge_callback, for_each_edge): New parameter.
25452 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25454 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25455 out code checking if runtime alias check is possible to below ...
25456 Call the new function.
25457 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25458 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25460 2017-06-07 Marek Polacek <polacek@redhat.com>
25463 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25464 TYPE_OVERFLOW_WRAPS check.
25466 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25468 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25469 if versioning is required.
25470 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25471 peeling with the check for versioning.
25473 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25475 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25476 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25477 Set true to new parameter if new ssa variable is defined.
25478 (vect_gen_vector_loop_niters): Refactor. Set range information
25479 for the new vector loop bound variable.
25480 (vect_do_peeling): Ditto.
25482 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25484 * tree-affine.c (ssa.h): Include header file.
25485 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25486 has wrapping overflow behavior.
25488 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25490 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25492 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25494 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25495 (tree_to_aff_combination): ... here.
25497 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25499 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25500 reg_pressure model function.
25501 (ivopts_global_cost_for_size): Delete.
25502 (determine_set_costs, iv_ca_recount_cost): Call new model function
25503 ivopts_estimate_reg_pressure.
25505 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25507 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25508 expensive than udiv. Remove floating point cases from mod.
25510 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25512 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25513 Increase idiv cost.
25515 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25517 * config/aarch64/aarch64.md
25518 (copysignsf3): Fix mask generation.
25520 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25522 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25524 (class dump_manager): Add register_dumps method.
25525 * dumpfile.c: Include langhooks.h.
25526 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25527 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25528 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25529 (dump_manager::dump_register): Start with 512 entries instead of 32.
25530 (dump_manager::register_dumps): New method.
25531 * toplev.c (general_init): Instead of invoking register_dumps
25532 langhook, invoke register_dumps method on the dump manager.
25533 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25536 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25538 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25539 Say that # requires an associated define_split to exist, and that
25540 the define_split must be suitable for use after register allocation.
25542 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25544 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25545 (compute_outgoing_frequencies): Also initialize zero counts.
25546 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25547 around loops; preserve more of profile when nothing changes.
25549 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25551 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25553 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25554 * config/arm/arm-cpu-cdata.h: Regenerate.
25555 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25556 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25557 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25558 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25559 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25561 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25563 * config/arm/t-rmprofile: Likewise.
25564 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25566 2017-06-06 David S. Miller <davem@davemloft.net>
25569 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25570 blockage if function uses alloca.
25572 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25574 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25575 New "uid" fields to hold pretty-print IDs of group and ref.
25576 Memory references are now identified as <group_id>:<ref_id>
25577 instead of using [random] addresses.
25578 (dump_mem_details): Simplify, no functional change.
25579 (dump_mem_ref): Simplify and make output more concise.
25580 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25581 (find_or_create_group): Initialize group uid.
25582 (record_ref): Initialize ref uid. Improve debug output.
25583 (prune_group_by_reuse, should_issue_prefetch_p,)
25584 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25585 (mark_nontemporal_store, determine_loop_nest_reuse):
25586 Improve debug output.
25588 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25590 * dbgcnt.def (prefetch): New debug counter.
25591 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25592 (schedule_prefetches): Stop issueing prefetches if debug counter
25595 2017-06-06 Tom de Vries <tom@codesourcery.com>
25597 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25598 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25600 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25602 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25603 Use aarch64_reg_or_zero predicate for operand 4.
25604 (aarch64_compare_and_swap<mode> define_insn_and_split):
25605 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25606 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25608 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25610 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25611 (arm_compute_save_core_reg_mask): This.
25612 (thumb1_compute_save_reg_mask): Rename into ...
25613 (thumb1_compute_save_core_reg_mask): This.
25614 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25615 (arm_compute_frame_layout): Likewise.
25617 2017-06-06 Richard Biener <rguenther@suse.de>
25619 PR tree-optimization/80974
25620 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25621 keep or clear leaders SSA info.
25623 2017-06-06 Tom de Vries <tom@codesourcery.com>
25625 * config/nvptx/nvptx.c (split_mode_p): New function.
25626 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25628 2017-06-06 Tom de Vries <tom@codesourcery.com>
25630 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25632 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25635 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25638 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25640 * shrink-wrap.c (handle_simple_exit): Update profile.
25641 (try_shrink_wrapping): Upate profile.
25643 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25645 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25646 (tree_guess_outgoing_edge_probabilities): New.
25647 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25648 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25650 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25652 * ipa-split.c (split_function): Initialize return bb profile.
25654 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25656 * profile.c (compute_branch_probabilities): Also initialize
25657 EXIT_BLOCK profile.
25659 2017-06-06 Richard Biener <rguenther@suse.de>
25661 PR tree-optimization/80928
25662 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25663 (vect_analyze_loop_operations): Properly guard analysis for
25665 (vect_transform_loop): Likewise.
25666 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25667 (vect_model_induction_cost): Do not cost for pure SLP.
25668 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25669 of induction in inner loop vectorization.
25670 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25671 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25672 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25674 (vect_analyze_slp_cost_1): Cost induction.
25675 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25676 (vect_get_slp_vect_defs): Likewise.
25677 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25678 (vect_transform_stmt): Handle SLP reductions.
25679 * tree-vectorizer.h (vectorizable_induction): Adjust.
25681 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25683 * config/rs6000/rs6000.c (make_resolver_func): Update
25684 init_lowered_empty_function call.
25686 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25688 * doc/invoke.texi: Document the -fprofile-abs-path option.
25689 * common.opt (fprofile-abs-path): New option.
25690 * gcov-io.h (gcov_write_filename): Declare.
25691 * gcov-io.c (gcov_write_filename): New function.
25692 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25693 * profile.c (output_location): Likewise.
25695 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25697 * shring-wrap.c: Revert accidental commit.
25699 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25701 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25703 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25705 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25707 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25708 profile in callgraph edge.
25709 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25710 (apply_scale): Likewise.
25711 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25712 Also scale profile when inlining function with zero profile.
25713 (initialize_cfun): Update exit block profile even when it is zero.
25714 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25715 when profile is read.
25717 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25719 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25720 (CLONE_*): New constants to define the processors we can generate
25721 code for with the target_clone attribute.
25722 (rs6000_clone_map): New array to identify which clone processors
25723 the current program is running on.
25724 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25725 target_clone attribute.
25726 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25727 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25728 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25729 (cpu_expand_builtin): Add support for target_clone attribute.
25730 (rs6000_valid_attribute_p): Allow "default" attribute.
25731 (get_decl_name): New debug function to simplify printing the
25732 current function name in debugging statements.
25733 (rs6000_clone_priority): New functions to support the target_clone
25734 attribute, and be able to generate code to switch between ISA 2.05
25735 through ISA 3.0 (power6 through power9).
25736 (rs6000_compare_version_priority): Likewise.
25737 (rs6000_get_function_versions_dispatcher): Likewise.
25738 (make_resolver_func): Likewise.
25739 (add_condition_to_bb): Likewise.
25740 (dispatch_function_versions): Likewise.
25741 (rs6000_generate_version_dispatcher_body): Likewise.
25742 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25743 (fusion_gpr_load_p): Fix a spacing issue.
25744 * doc/extend.texi (Common Function Attributes): Document that the
25745 PowerPC supports the target_clone attribute.
25747 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25749 * config/arm/arm.h: explain F symbol found in description of ARM
25750 register allocation in its legend.
25752 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25754 * config/mips/frame-header-opt.c: Include profile-count.h.
25755 * config/riscv/riscv.c: Include profile-count.h
25757 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25759 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25761 (sm_set_flag_if_changed): Add bbs field.
25762 (execute_sm_if_changed_flag_set): Pass BBS.
25763 (execute_sm): Update.
25765 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25767 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25770 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25772 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25773 (peephole2): New peephole2 to emit the above.
25774 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25776 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25778 * config/aarch64/aarch64.c (define_peephole2 above
25779 *sub_<shift>_<mode>): New peephole.
25781 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25783 * config/i386/i386.c (make_resolver_func): Update.
25784 * Makefile.in: Add profile-count.h and profile-count.o
25785 * auto-profile.c (afdo_indirect_call): Update to new API.
25786 (afdo_set_bb_count): Update.
25787 (afdo_propagate_edge): Update.
25788 (afdo_propagate_circuit): Update.
25789 (afdo_calculate_branch_prob): Update.
25790 (afdo_annotate_cfg): Update.
25791 * basic-block.h: Include profile-count.h
25792 (struct edge_def): Turn count to profile_count.
25793 (struct basic_block_def): Likewie.
25794 (REG_BR_PROB_BASE): Move to profile-count.h
25795 (RDIV): Move to profile-count.h
25796 * bb-reorder.c (max_entry_count): Turn to profile_count.
25797 (find_traces): Update.
25798 (rotate_loop):Update.
25799 (connect_traces):Update.
25800 (sanitize_hot_paths):Update.
25801 * bt-load.c (migrate_btr_defs): Update.
25802 * cfg.c (RDIV): Remove.
25803 (init_flow): Use alloc_block.
25804 (alloc_block): Uninitialize count.
25805 (unchecked_make_edge): Uninitialize count.
25806 (check_bb_profile): Update.
25807 (dump_edge_info): Update.
25808 (dump_bb_info): Update.
25809 (update_bb_profile_for_threading): Update.
25810 (scale_bbs_frequencies_int): Update.
25811 (scale_bbs_frequencies_gcov_type): Update.
25812 (scale_bbs_frequencies_profile_count): New.
25813 * cfg.h (update_bb_profile_for_threading): Update.
25814 (scale_bbs_frequencies_profile_count): Declare.
25815 * cfgbuild.c (compute_outgoing_frequencies): Update.
25816 (find_many_sub_basic_blocks): Update.
25817 * cfgcleanup.c (try_forward_edges): Update.
25818 (try_crossjump_to_edge): Update.
25819 * cfgexpand.c (expand_gimple_tailcall): Update.
25820 (construct_exit_block): Update.
25821 * cfghooks.c (verify_flow_info): Update.
25822 (dump_bb_for_graph): Update.
25823 (split_edge): Update.
25824 (make_forwarder_block): Update.
25825 (duplicate_block): Update.
25826 (account_profile_record): Update.
25827 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25828 (get_estimated_loop_iterations): Update.
25829 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25830 (single_likely_exit): Update.
25831 * cfgloopmanip.c (scale_loop_profile): Update.
25833 (set_zero_probability): Update.
25834 (lv_adjust_loop_entry_edge): Update.
25835 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25836 (purge_dead_edges): Update.
25837 (rtl_account_profile_record): Update.
25838 * cgraph.c (cgraph_node::create): Uninitialize count.
25839 (symbol_table::create_edge): Uninitialize count.
25840 (cgraph_update_edges_for_call_stmt_node): Update.
25841 (cgraph_edge::dump_edge_flags): Update.
25842 (cgraph_node::dump): Update.
25843 (cgraph_edge::maybe_hot_p): Update.
25844 * cgraph.h: Include profile-count.h
25845 (create_clone), create_edge, create_indirect_edge): Update.
25846 (cgraph_node): Turn count to profile_count.
25847 (cgraph_edge0: Likewise.
25848 (make_speculative, clone): Update.
25849 (create_edge): Update.
25850 (init_lowered_empty_function): Update.
25851 * cgraphclones.c (cgraph_edge::clone): Update.
25852 (duplicate_thunk_for_node): Update.
25853 (cgraph_node::create_clone): Update.
25854 * cgraphunit.c (cgraph_node::analyze): Update.
25855 (cgraph_node::expand_thunk): Update.
25856 * final.c (dump_basic_block_info): Update.
25857 * gimple-streamer-in.c (input_bb): Update.
25858 * gimple-streamer-out.c (output_bb): Update.
25859 * graphite.c (print_global_statistics): Update.
25860 (print_graphite_scop_statistics): Update.
25861 * hsa-brig.c: Include basic-block.h.
25862 * hsa-dump.c: Include basic-block.h.
25863 * hsa-gen.c (T sum_slice): Update.
25864 (convert_switch_statements):Update.
25865 * hsa-regalloc.c: Include basic-block.h.
25866 * ipa-chkp.c (chkp_produce_thunks): Update.
25867 * ipa-cp.c (struct caller_statistics): Update.
25868 (init_caller_stats): Update.
25869 (gather_caller_stats): Update.
25870 (ipcp_cloning_candidate_p): Update.
25871 (good_cloning_opportunity_p): Update.
25872 (get_info_about_necessary_edges): Update.
25873 (dump_profile_updates): Update.
25874 (update_profiling_info): Update.
25875 (update_specialized_profile): Update.
25876 (perhaps_add_new_callers): Update.
25877 (decide_about_value): Update.
25878 (ipa_cp_c_finalize): Update.
25879 * ipa-devirt.c (struct odr_type_warn_count): Update.
25880 (struct decl_warn_count): Update.
25881 (struct final_warning_record): Update.
25882 (possible_polymorphic_call_targets): Update.
25883 (ipa_devirt): Update.
25884 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25885 * ipa-icf.c (sem_function::merge): Update.
25886 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25887 * ipa-inline.c (compute_uninlined_call_time): Update.
25888 (compute_inlined_call_time): Update.
25889 (want_inline_small_function_p): Update.
25890 (want_inline_self_recursive_call_p): Update.
25891 (edge_badness): Update.
25892 (lookup_recursive_calls): Update.
25893 (recursive_inlining): Update.
25894 (inline_small_functions): Update.
25895 (dump_overall_stats): Update.
25896 (dump_inline_stats): Update.
25897 * ipa-profile.c (ipa_profile_generate_summary): Update.
25898 (ipa_propagate_frequency): Update.
25899 (ipa_profile): Update.
25900 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25901 * ipa-utils.c (ipa_merge_profiles): Update.
25902 * loop-doloop.c (doloop_modify): Update.
25903 * loop-unroll.c (report_unroll): Update.
25904 (unroll_loop_runtime_iterations): Update.
25905 * lto-cgraph.c (lto_output_edge): Update.
25906 (lto_output_node): Update.
25907 (input_node): Update.
25908 (input_edge): Update.
25909 (merge_profile_summaries): Update.
25910 * lto-streamer-in.c (input_cfg): Update.
25911 * lto-streamer-out.c (output_cfg): Update.
25912 * mcf.c (create_fixup_graph): Update.
25913 (adjust_cfg_counts): Update.
25914 (sum_edge_counts): Update.
25915 * modulo-sched.c (sms_schedule): Update.
25916 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25917 * predict.c (maybe_hot_count_p): Update.
25918 (probably_never_executed): Update.
25919 (dump_prediction): Update.
25920 (combine_predictions_for_bb): Update.
25921 (propagate_freq): Update.
25922 (handle_missing_profiles): Update.
25923 (counts_to_freqs): Update.
25924 (rebuild_frequencies): Update.
25925 (force_edge_cold): Update.
25926 * predict.h: Include profile-count.h
25927 (maybe_hot_count_p, counts_to_freqs): UPdate.
25928 * print-rtl-function.c: Do not include cfg.h
25929 * print-rtl.c: Include basic-block.h
25930 * profile-count.c: New file.
25931 * profile-count.h: New file.
25932 * profile.c (is_edge_inconsistent): Update.
25933 (correct_negative_edge_counts): Update.
25934 (is_inconsistent): Update.
25935 (set_bb_counts): Update.
25936 (read_profile_edge_counts): Update.
25937 (compute_frequency_overlap): Update.
25938 (compute_branch_probabilities): Update; Initialize and deinitialize
25940 (branch_prob): Update.
25941 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
25942 (edge_gcov_count): New.
25943 (bb_gcov_count): New.
25944 * shrink-wrap.c (try_shrink_wrapping): Update.
25945 * tracer.c (better_p): Update.
25946 * trans-mem.c (expand_transaction): Update.
25947 (ipa_tm_insert_irr_call): Update.
25948 (ipa_tm_insert_gettmclone_call): Update.
25949 * tree-call-cdce.c: Update.
25950 * tree-cfg.c (gimple_duplicate_sese_region): Update.
25951 (gimple_duplicate_sese_tail): Update.
25952 (gimple_account_profile_record): Update.
25953 (execute_fixup_cfg): Update.
25954 * tree-inline.c (copy_bb): Update.
25955 (copy_edges_for_bb): Update.
25956 (initialize_cfun): Update.
25957 (freqs_to_counts): Update.
25958 (copy_cfg_body): Update.
25959 (expand_call_inline): Update.
25960 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25961 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
25962 (try_unroll_loop_completely): Update.
25963 (try_peel_loop): Update.
25964 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25965 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
25966 * tree-ssa-loop-split.c (connect_loops): Update.
25967 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
25968 * tree-ssa-reassoc.c (branch_fixup): Update.
25969 * tree-ssa-tail-merge.c (replace_block_by): Update.
25970 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
25971 (compute_path_counts): Update.
25972 (update_profile): Update.
25973 (recompute_probabilities): Update.
25974 (update_joiner_offpath_counts): Update.
25975 (estimated_freqs_path): Update.
25976 (freqs_to_counts_path): Update.
25977 (clear_counts_path): Update.
25978 (ssa_fix_duplicate_block_edges): Update.
25979 (duplicate_thread_path): Update.
25980 * tree-switch-conversion.c (case_bit_test_cmp): Update.
25981 (struct switch_conv_info): Update.
25982 * tree-tailcall.c (decrease_profile): Update.
25983 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25984 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25985 * value-prof.c (check_counter): Update.
25986 (gimple_divmod_fixed_value): Update.
25987 (gimple_mod_pow2): Update.
25988 (gimple_mod_subtract): Update.
25989 (gimple_ic_transform): Update.
25990 (gimple_stringop_fixed_value): Update.
25991 * value-prof.h (gimple_ic): Update.
25993 2017-06-02 Carl Love <cel@us.ibm.com>
25995 * config/rs6000/rs6000-c: Add support for built-in functions
25996 vector double vec_doublee (vector signed int);
25997 vector double vec_doublee (vector unsigned int);
25998 vector double vec_doublee (vector float);
25999 vector double vec_doubleh (vector signed int);
26000 vector double vec_doubleh (vector unsigned int);
26001 vector double vec_doubleh (vector float);
26002 vector double vec_doublel (vector signed int);
26003 vector double vec_doublel (vector unsigned int);
26004 vector double vec_doublel (vector float);
26005 vector double vec_doubleo (vector signed int);
26006 vector double vec_doubleo (vector unsigned int);
26007 vector double vec_doubleo (vector float);.
26008 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26009 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26011 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26012 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26013 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26015 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26016 vec_doublel, vec_doubleh.
26017 * doc/extend.texi: Update the built-in documentation file for the
26018 new built-in functions.
26020 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26023 * ipa-inline-analysis.c (free_growth_caches): Set
26024 edge_removal_hook_holder to NULL after removing it.
26026 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26028 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26029 comparision with zero.
26031 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26032 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26033 for early expansion of vec_min and vec_max builtins.
26034 (builtin_function_type): Add min/max unsigned variants to those
26035 identified as having unsigned arguments.
26037 2017-06-02 Olivier Hainque <hainque@adacore.com>
26039 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26041 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26043 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26044 Use VALL_F16 iterator rather than VALL.
26046 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26048 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26049 Emit CBNZ inside loop when doing a strong exchange and comparing
26050 against zero. Generate the CC flags after the loop.
26052 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26054 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26055 (dl_section_ref): New.
26056 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26057 On AIX, append an expression to subtract the size of the
26058 section length to dl_section_ref.
26060 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26062 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26063 for early expansion of vector absolute builtins.
26065 2017-06-02 Richard Biener <rguenther@suse.de>
26067 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26068 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26070 2017-06-02 Richard Biener <rguenther@suse.de>
26072 PR tree-optimization/80948
26073 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26074 stmt order as well.
26076 2017-06-02 Richard Biener <rguenther@suse.de>
26078 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26080 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26081 for inductions as relevant.
26083 2017-06-02 Richard Biener <rguenther@suse.de>
26085 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26086 (vectorizable_induction): ... this. Remove dead code.
26088 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26090 * builtins. (expand_builtin_alloca): Remove second parameter and
26091 infer its value from the first parameter instead.
26092 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26094 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26096 PR rtl-optimization/80903
26097 * loop-doloop.c (add_test): Unshare sequence.
26099 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26101 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26103 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26105 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26107 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26108 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26109 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26110 (xlogue_layout::get_stub_rtx): Make static.
26111 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26112 (xlogue_layout::compute_stub_managed_regs): Rename to...
26113 (xlogue_layout::count_stub_managed_regs): ...this.
26114 (xlogue_layout::is_stub_managed_reg): New function.
26115 (xlogue_layout::m_stub_names): Rename to...
26116 (xlogue_layout::s_stub_names): ...this, make static.
26117 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26118 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26119 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26120 xlogue_layout::s_stub_names): Instantiate statics.
26121 (stub_managed_regs): Remove.
26122 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26123 (disable_call_ms2sysv_xlogues): Rename to...
26124 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26125 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26127 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26128 change after reload_completed.
26129 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26131 (ix86_expand_prologue): Likewise.
26132 (ix86_expand_epilogue): Likewise.
26133 (ix86_expand_split_stack_prologue): Likewise.
26134 (ix86_compute_frame_layout): Remove frame parameter ...
26135 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26136 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26138 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26139 (ix86_frame): Move from here ...
26140 * config/i386/i386.h (ix86_frame): ... to here.
26141 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26142 complete ix86_frame data structure instead. Remove some_ld_name.
26144 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26146 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26147 symbols that hold a DECL_VALUE_EXPR.
26149 2017-06-01 Martin Jambor <mjambor@suse.cz>
26151 PR tree-optimization/80898
26152 * tree-sra.c (process_subtree_disqualification): Removed.
26153 (disqualify_candidate): Do not acll
26154 process_subtree_disqualification.
26155 (subtree_mark_written_and_enqueue): New function.
26156 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26157 RHS has been disqualified and re-queue LHS if necessary. Apart
26158 from that, ignore disqualified RHS.
26160 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26162 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26163 address fetch for z10 or later.
26165 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26167 * config/arc/arc.md (tst_movb): Add guard when splitting.
26169 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26171 * config/arc/arc.c (arc_can_eliminate): Test against
26172 arc_frame_pointer_needed.
26174 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26176 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26177 to prevent store reordering.
26178 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26179 (type): Add block type.
26180 (stack_tie): Define special instruction to be used in
26183 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26185 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26186 constraint. It is not valid for the pattern.
26187 (noncommutative_binary_comparison): Likewise.
26189 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26191 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26194 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26196 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26197 be used by the reg-alloc.
26199 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26201 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26202 reg-alloc when having mul64 or mul32x16 instructions.
26203 (mulsidi3): Likewise.
26204 (umulsidi3): Likewise.
26205 (mulsi32x16): New pattern.
26206 (mulsi64): Likewise.
26207 (mulsidi64): Likewise.
26208 (umulsidi64): Likewise.
26209 (MUL32x16_REG): Define.
26210 (mul64_600): Use MUL32x16_REG.
26211 (mac64_600): Likewise.
26212 (umul64_600): Likewise.
26213 (umac64_600): Likewise.
26215 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26217 * config/arc/arc.md (mulsi3_700): Make it commutative.
26219 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26221 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26223 (*sign_extendsidi2_insn): Likewise for movstosw.
26225 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26227 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26228 the type of the input discriminant value. Convert the
26229 discriminant value of signedness vary.
26231 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26233 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26234 Add to -Wall section.
26236 2017-06-01 Richard Biener <rguenther@suse.de>
26238 PR middle-end/66313
26239 * fold-const.c (fold_plusminus_mult_expr): If the factored
26240 factor may be zero use a wrapping type for the inner operation.
26241 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26242 and handle moved defs.
26243 (process_assignment): Properly guard the unary op case. Return a
26244 tri-state indicating that moving the stmt before the call may allow
26245 to continue. Pass through to_move.
26246 (find_tail_calls): Handle moving unrelated defs before
26249 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26252 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26253 splitter result in the canonical way.
26255 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26257 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26258 also for 32bit target. Update insn attributes.
26259 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26261 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26263 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26264 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26265 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26266 (_mm_maskz_min_ss): New intrinsics.
26268 2017-05-31 Martin Liska <mliska@suse.cz>
26270 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26271 Change comment style to one we normally use.
26272 (vectorizable_reduction): Likewise.
26273 (vectorizable_induction): Likewise.
26274 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26275 (vectorizable_call): Likewise.
26276 (vectorizable_simd_clone_call): Likewise.
26277 (vectorizable_conversion): Likewise.
26278 (vectorizable_assignment): Likewise.
26279 (vectorizable_shift): Likewise.
26280 (vectorizable_operation): Likewise.
26281 (vectorizable_store): Likewise.
26282 (vectorizable_load): Likewise.
26283 * tree-vectorizer.h: Likewise.
26285 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26287 * passes.c (emergency_dump_function): New.
26288 * tree-pass.h (emergency_dump_function): Declare.
26289 * plugin.c (plugins_internal_error_function): Remove.
26290 * plugin.h (plugins_internal_error_function): Remove declaration.
26291 * toplev.c (internal_error_function): New static function. Use it...
26292 (general_init): ...here.
26294 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26296 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26297 (arc_rtx_costs): Add costs for new patterns.
26298 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26299 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26301 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26303 * tree-ssa-strlen.c (get_next_strinfo): New function.
26304 (get_stridx_plus_constant): Use it.
26305 (zero_length_string): Likewise.
26306 (adjust_related_strinfos): Likewise.
26307 (adjust_last_stmt): Likewise.
26309 2017-05-31 Richard Biener <rguenther@suse.de>
26312 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26313 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26315 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26317 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26318 loop_vinfo argument and use of dependence distance vectors.
26319 Check instead whether the two references differ only in their
26320 initial value and assume that they have the same alignment if the
26321 difference is a multiple of the vector alignment.
26322 (vect_analyze_data_refs_alignment): Update call accordingly.
26324 2017-05-31 Martin Liska <mliska@suse.cz>
26327 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26329 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26331 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26332 (create_intersect_range_checks): Move from ...
26333 * tree-data-ref.c (create_intersect_range_checks_index)
26334 (create_intersect_range_checks): ... to here.
26335 (create_runtime_alias_checks): New function factored from ...
26336 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26337 here. Call above function.
26338 * tree-data-ref.h (create_runtime_alias_checks): New function.
26340 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26342 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26343 segment length for dr_b and compute it in wide_int.
26345 2017-05-31 Richard Biener <rguenther@suse.de>
26347 PR tree-optimization/80906
26348 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26349 and pass through iv_map.
26350 (copy_bb_and_scalar_dependences): Adjust.
26351 (translate_pending_phi_nodes): Likewise.
26352 (copy_loop_close_phi_args): Handle code-generating IVs instead
26355 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26357 * diagnostic-color.c (color_dict): Add "type-diff".
26358 (parse_gcc_colors): Update comment.
26359 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26360 -fdiagnostics-show-template-tree and -fno-elide-type.
26361 (GCC_COLORS): Add type-diff to example.
26363 (-fdiagnostics-show-template-tree): New.
26364 (-fno-elide-type): New.
26365 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26366 the pp_format_decoder callback. Call any m_format_postprocessor's
26368 (pretty_printer::pretty_printer): Initialize
26369 m_format_postprocessor.
26370 (pretty_printer::~pretty_printer): Delete any
26371 m_format_postprocessor.
26372 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26373 (class format_postprocessor): New class.
26374 (struct pretty_printer::format_decoder): Document the new parameters.
26375 (struct pretty_printer::m_format_postprocessor): New field.
26376 * tree-diagnostic.c (default_tree_printer): Update for new
26377 bool and const char ** params.
26378 * tree-diagnostic.h (default_tree_printer): Likewise.
26380 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26382 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26383 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26384 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26385 rs6000_gen_cell_microcode code.
26386 (rs6000_final_prescan_insn): Delete.
26387 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26388 "warn-cell-microcode" entries.
26389 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26390 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26391 throughout. Change cc_reg_not_micro_cr0_operand to
26392 cc_reg_not_cr0_operand throughout.
26393 (*extendhi<mode>2_noload): Delete.
26394 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26395 (mwarn-cell-microcode): Delete.
26396 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26397 -mgen-cell-microcode and -mwarn-cell-microcode.
26399 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26402 * config/i386/constraints.md (Yd): New constraint.
26404 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26405 and (?Yd, r) alternatives. Update insn attributes.
26406 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26407 and (?*Yd, r) alternatives. Update insn attributes.
26408 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26410 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26412 * gimplify.c (gimplify_modify_expr): Don't create a
26413 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26416 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26418 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26420 2017-05-30 Richard Biener <rguenther@suse.de>
26422 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26423 and reduc_def fields.
26424 (STMT_VINFO_REDUC_TYPE): New define.
26425 (STMT_VINFO_REDUC_DEF): Likewise.
26426 (vect_force_simple_reduction): Adjust prototype.
26427 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26428 (vect_is_simple_reduction): Remove check_reduction argument.
26429 (vect_force_simple_reduction): Adjust and set
26430 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26431 (vectorizable_reduction): Do not re-do reduction analysis
26432 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26433 * tree-parloops.c (gather_scalar_reductions): Adjust.
26435 2017-05-30 Richard Biener <rguenther@suse.de>
26437 PR middle-end/80901
26438 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26441 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26443 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26444 Introduce unknown_misalignment parameter and remove vf.
26445 (vect_peeling_hash_get_lowest_cost):
26446 Pass unknown_misalignment parameter.
26447 (vect_enhance_data_refs_alignment):
26448 Fix unsupportable data ref treatment.
26450 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26452 * tree-vect-data-refs.c (vect_get_data_access_cost):
26453 Workaround for SLP handling.
26454 (vect_enhance_data_refs_alignment):
26455 Compute costs for doing no peeling at all, compare to the best
26456 peeling costs so far and avoid peeling if cheaper.
26458 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26460 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26461 Return peeling info and set costs to zero for unlimited cost
26463 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26464 with unknown misalignment. Compute and costs for unknown
26465 misalignment, compare them to the costs for known misalignment
26466 and choose the cheapest for peeling.
26468 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26470 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26471 (vect_get_peeling_costs_all_drs): Create function.
26472 (vect_peeling_hash_get_lowest_cost):
26473 Use vect_get_peeling_costs_all_drs.
26474 (vect_peeling_supportable): Create function.
26475 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26477 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26479 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26480 DR_HAS_NEGATIVE_STEP.
26481 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26482 (vect_enhance_data_refs_alignment): Use.
26483 (vect_duplicate_ssa_name_ptr_info): Use.
26484 * tree-vectorizer.h (dr_misalignment): Use.
26485 (known_alignment_for_access_p): Use.
26487 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26490 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26492 (has_section_name): New function.
26494 2017-05-30 Martin Liska <mliska@suse.cz>
26497 * auto-profile.c (get_function_decl_from_block): Fix
26500 2017-05-30 Richard Biener <rguenther@suse.de>
26502 PR middle-end/80876
26503 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26505 2017-05-30 Martin Liska <mliska@suse.cz>
26507 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26508 * dumpfile.h (struct dump_file_info): Remove ctors.
26510 2017-05-30 Martin Liska <mliska@suse.cz>
26512 * predict.def: Fix GNU coding style.
26514 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26516 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26517 Mark 'to' argument with ATTRIBUTE_UNUSED.
26519 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26521 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26522 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26523 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26526 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26528 * doc/install.texi (Options specification): Restore entry of
26529 --enable-sjlj-exceptions.
26531 2017-05-27 Michael Eager <eager@eagercon.com>
26534 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26536 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26538 * config/microblaze/microblaze.h
26539 (FIXED_REGISTERS): Update in macro.
26540 (CALL_USED_REGISTERS): Update in macro.
26542 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26544 * doc/install.texi: Add links to macOS binary distributions.
26546 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26550 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26552 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26554 2017-05-26 Martin Liska <mliska@suse.cz>
26556 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26558 2017-05-26 Martin Liska <mliska@suse.cz>
26560 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26561 always leading ';; '.
26562 (dump_bb_info): Likewise.
26563 (brief_dump_cfg): Likewise.
26564 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26565 * dumpfile.c: Remove usage of TDF_VERBOSE.
26566 * dumpfile.h (enum dump_kind): Likewise.
26567 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26568 * print-tree.c (debug_verbose): Remove.
26569 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26570 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26571 * tree-diagnostic.c (default_tree_printer): Replace
26572 TDF_DIAGNOSTIC with TDF_SLIM.
26574 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26576 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26577 in parameter loop, rather than loop_vinfo.
26578 (create_intersect_range_checks): Ditto.
26579 (vect_create_cond_for_alias_checks): Update call to above functions.
26581 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26583 PR tree-optimization/80815
26584 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26585 for merging runtime alias checks. Handle negative DR_STEPs.
26587 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26589 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26591 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26592 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26593 out code pruning runtime alias checks.
26594 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26595 factored out from above.
26596 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26598 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26600 (prune_runtime_alias_test_list): New decalaration.
26602 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26604 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26605 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26606 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26607 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26608 (operator==, comp_dr_with_seg_len_pair): Ditto.
26609 (vect_prune_runtime_alias_test_list): Ditto.
26611 2017-05-26 Martin Liska <mliska@suse.cz>
26614 * params.def: Bound partial-inlining-entry-probability param.
26616 2017-05-26 Marek Polacek <polacek@redhat.com>
26619 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26622 2017-05-26 Richard Biener <rguenther@suse.de>
26624 PR tree-optimization/80842
26625 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26628 2017-05-26 Richard Biener <rguenther@suse.de>
26630 PR tree-optimization/80844
26631 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26633 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26635 * doc/md.texi (Machine Constraints): Update x86 family
26636 machine constraints section to match 'config/i386/constraints.md'.
26638 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26640 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26642 2017-05-25 Nathan Sidwell <nathan@acm.org>
26644 * doc/invoke.texi (--enable-languages): Update documentation.
26646 2017-05-25 Martin Liska <mliska@suse.cz>
26648 * dumpfile.c: Add TDF_FOLDING.
26649 * dumpfile.h (enum dump_kind): Likewise.
26650 * genmatch.c (dt_simplify::gen_1): Use it.
26652 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26654 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26656 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26658 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26659 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26661 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26663 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26664 * match.pd (X == C): Rewrite it here.
26665 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26666 with_certain_nonzero_bits2): New predicates.
26667 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26669 2017-05-24 Nathan Sidwell <nathan@acm.org>
26671 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26674 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26677 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26679 * config/powerpcspe: New port. Files are copied from the rs6000
26680 port, with "rs6000" in filenames replaced by "powerpcspe".
26682 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26684 PR rtl-optimization/80754
26685 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26687 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26689 * config/sparc/sparc.md (length): Return the correct value for -mflat
26690 sibcalls to match output_sibcall.
26692 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26696 * config/rs6000/rs6000.c (struct machine_function): Add new field
26698 (rs6000_get_separate_components): Init that field, use it.
26699 (rs6000_components_for_bb): Use the field.
26701 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26703 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26705 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26707 PR middle-end/80823
26708 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26710 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26713 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26714 address against address_operand predicate.
26715 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26717 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26719 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26720 parameters passed indirectly.
26722 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26724 * config/i386/i386.md (*movdi_internal): Remove SSE4
26725 alternative 18 (?r, *v). Update insn attributes.
26726 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26727 Update insn attributes.
26728 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26729 Update insn attributes.
26730 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26731 alternative 1 (r, v). Remove isa attribute.
26732 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26733 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26734 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26736 2017-05-23 Tom de Vries <tom@codesourcery.com>
26738 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26741 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26743 * cgraphunit.c (symbol_table::process_new_functions): Update.
26744 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26745 (inline_generate_summary): Rename to ...
26746 (ipa_fn_summary_generate): ... this one.
26747 (inline_read_summary): Rename to ...
26748 (ipa_fn_summary_read): ... this one.
26749 (inline_write_summary): Rename to ...
26750 (ipa_fn_summary_write): ... this one.
26751 (inline_free_summary): Rename to ...
26752 (ipa_free_fn_summary): ... this one.
26753 (pass_data_local_fn_summary, pass_local_fn_summary,
26754 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26755 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26756 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26757 make_pass_ipa_fn_summary): New.
26758 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26759 inline_write_summary, inline_free_summary): Remove.
26760 (ipa_free_fn_summary) : New.
26761 * ipa-inline.c (ipa_inline): Update.
26762 (pass_ipa_inline): Do not generate summaries.
26763 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26765 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26766 and add pass_ipa_fn_summary.
26767 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26769 (make_pass_inline_parameters): Remove.
26771 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26773 * omp-low.c (struct omp_context): Remove "default_kind" member.
26776 * omp-offload.c (execute_oacc_device_lower): Remove the
26777 parallelism dimensions function attributes for unparallelized
26778 OpenACC kernels constructs.
26780 2017-05-23 Martin Liska <mliska@suse.cz>
26782 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26784 (cgraph_edge::make_speculative): Likewise.
26785 (cgraph_edge::resolve_speculation): Likewise.
26786 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26787 (cgraph_node::dump): Likewise.
26788 * cgraph.h: Likewise.
26789 * cgraphunit.c (analyze_functions): Likewise.
26790 (symbol_table::compile): Likewise.
26791 * ipa-cp.c (print_all_lattices): Likewise.
26792 (determine_versionability): Likewise.
26793 (initialize_node_lattices): Likewise.
26794 (ipcp_verify_propagated_values): Likewise.
26795 (estimate_local_effects): Likewise.
26796 (update_profiling_info): Likewise.
26797 (create_specialized_node): Likewise.
26798 (perhaps_add_new_callers): Likewise.
26799 (decide_about_value): Likewise.
26800 (decide_whether_version_node): Likewise.
26801 (identify_dead_nodes): Likewise.
26802 (ipcp_store_bits_results): Likewise.
26803 * ipa-devirt.c (dump_targets): Likewise.
26804 (ipa_devirt): Likewise.
26805 * ipa-icf.c (sem_item::dump): Likewise.
26806 (sem_function::equals): Likewise.
26807 (sem_variable::equals): Likewise.
26808 (sem_item_optimizer::read_section): Likewise.
26809 (sem_item_optimizer::execute): Likewise.
26810 (congruence_class::dump): Likewise.
26811 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26812 (dump_inline_summary): Likewise.
26813 (estimate_node_size_and_time): Likewise.
26814 (inline_analyze_function): Likewise.
26815 * ipa-inline-transform.c (inline_call): Likewise.
26816 * ipa-inline.c (report_inline_failed_reason): Likewise.
26817 (want_early_inline_function_p): Likewise.
26818 (edge_badness): Likewise.
26819 (update_edge_key): Likewise.
26820 (inline_small_functions): Likewise.
26821 * ipa-profile.c (ipa_profile): Likewise.
26822 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26823 (ipa_make_edge_direct_to_target): Likewise.
26824 (remove_described_reference): Likewise.
26825 (ipa_impossible_devirt_target): Likewise.
26826 (propagate_controlled_uses): Likewise.
26827 (ipa_print_node_params): Likewise.
26828 (ipcp_transform_function): Likewise.
26829 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26830 (propagate_pure_const): Likewise.
26831 * ipa-reference.c (generate_summary): Likewise.
26832 (read_write_all_from_decl): Likewise.
26833 (propagate): Likewise.
26834 (ipa_reference_read_optimization_summary): Likewise.
26835 * ipa-utils.c (ipa_merge_profiles): Likewise.
26836 * ipa.c (walk_polymorphic_call_targets): Likewise.
26837 (symbol_table::remove_unreachable_nodes): Likewise.
26838 (ipa_single_use): Likewise.
26839 * passes.c (execute_todo): Likewise.
26840 * predict.c (drop_profile): Likewise.
26841 * symtab.c (symtab_node::get_dump_name): New function.
26842 (symtab_node::dump_name): Likewise.
26843 (symtab_node::dump_asm_name): Likewise.
26844 (symtab_node::dump_references): Likewise.
26845 (symtab_node::dump_referring): Likewise.
26846 (symtab_node::dump_base): Likewise.
26847 (symtab_node::debug_symtab): Likewise.
26848 * tree-sra.c (convert_callers_for_node): Likewise.
26849 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26850 * value-prof.c (init_node_map): Likewise.
26852 2017-05-23 Martin Liska <mliska@suse.cz>
26854 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26855 and symtab_node::debug_symtab to symbol_table::debug.
26856 * cgraphunit.c (analyze_functions): Use the renamed function.
26857 (symbol_table::compile): Likewise.
26858 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26859 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26860 * passes.c (execute_todo): Likewise.
26861 * symtab.c (symbol_table::dump): New function.
26862 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26864 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26866 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26867 that nonconst implies exec.
26869 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26871 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26872 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26873 (inline_edge_summary_vec): Turn into ...
26874 (ipa_call_summaries): ... this one.
26875 (redirect_to_unreachable, edge_set_predicate,
26876 evaluate_properties_for_edge, inline_summary_alloc,
26877 reset_ipa_call_summary, reset_inline_summary,
26878 inline_summary_t::duplicate): Update.
26879 (inline_edge_duplication_hook): Turn to ...
26880 (ipa_call_summary_t::duplicate): ... this one.
26881 (inline_edge_removal_hook): Turn to ...
26882 (ipa_call_summary_t::remove): ... this one.
26883 (dump_inline_edge_summary): Turn to ...
26884 (dump_ipa_call_summary): ... this one.
26885 (estimate_function_body_sizes): Update.
26886 (inline_update_callee_summaries): Update.
26887 (remap_edge_change_prob): Update.
26888 (remap_edge_summaries): Update.
26889 (inline_merge_summary): Update.
26890 (do_estimate_edge_time): Update.
26891 (inline_generate_summary): Update.
26892 (inline_read_section): Update.
26893 (inline_read_summary): Update.
26894 (inline_free_summary): Update.
26895 * ipa-inline.c (can_inline_edge_p): Update.
26896 (compute_inlined_call_time): Update.
26897 (want_inline_small_function_p): Update.
26898 (edge_badness): Update.
26899 (early_inliner): Update.
26900 * ipa-inline.h (inline_edge_summary): Turn to ...
26901 (ipa_call_summary): ... this one.
26902 (ipa_call_summary_t): New class.
26903 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26904 (ipa_call_summaries): New.
26905 (inline_edge_summary): Remove.
26906 (estimate_edge_growth): Update.
26907 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26908 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26909 * ipa-split.c (execute_split_functions): Update.
26910 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26912 2017-05-23 Tom de Vries <tom@codesourcery.com>
26914 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26915 attributes): Document rdrand effective target.
26917 2017-05-23 Tom de Vries <tom@codesourcery.com>
26919 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26920 attributes): Sort alphabetically.
26922 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
26924 * config/avr/genmultilib.awk: Use gsub instead of gensub.
26926 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26929 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
26930 V2DF/V2DI splat into two separate patterns, one that handles
26931 registers, and the other that only handles memory. Drop support
26932 for splatting from a GPR on ISA 2.07 and then splitting the
26933 splat into direct move and splat.
26934 (vsx_splat_<mode>_reg): Likewise.
26935 (vsx_splat_<mode>_mem): Likewise.
26937 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
26939 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
26941 2017-05-22 Jakub Jelinek <jakub@redhat.com>
26943 PR middle-end/80809
26944 * omp-low.c (finish_taskreg_remap): New function.
26945 (finish_taskreg_scan): If unit size of ctx->record_type
26946 is non-constant, unshare the size expression and replace
26947 decls in it with possible outer var refs.
26949 PR middle-end/80809
26950 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
26951 GOVD_SHARED rather than GOVD_PRIVATE with it.
26952 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
26953 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
26955 PR middle-end/80853
26956 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
26957 as last argument to build_outer_var_ref for pointer bases of array
26958 section reductions.
26960 2017-05-19 Martin Sebor <msebor@redhat.com>
26962 * print-tree.c (print_node): Print DECL_READ_P flag.
26964 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26966 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
26967 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
26968 * cgraph.c: Likewise.
26969 * cgraphunit.c: Likewise.
26970 * gengtype.c: Likewise.
26971 * ipa-cp.c: Likewise.
26972 * ipa-devirt.c: Likewise.
26973 * ipa-icf.c: Likewise.
26974 * ipa-predicate.c: Likewise.
26975 * ipa-profile.c: Likewise.
26976 * ipa-prop.c: Likewise.
26977 * ipa-split.c: Likewise.
26979 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
26980 edge_predicate_pool, dump_inline_hints,
26981 inline_summary::account_size_time, redirect_to_unreachable,
26982 edge_set_predicate, set_hint_predicate,
26983 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
26984 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
26985 inline_summary_t::remove, remap_hint_predicate_after_duplication,
26986 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
26987 ipa_call_summary_t::remove, initialize_growth_caches,
26988 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
26989 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
26990 mark_modified, unmodified_parm_1, unmodified_parm,
26991 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
26992 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
26993 compute_bb_predicates, will_be_nonconstant_expr_predicate,
26994 will_be_nonconstant_predicate, record_modified_bb_info,
26995 get_minimal_bb, record_modified, param_change_prob,
26996 phi_result_unknown_predicate, predicate_for_phi_result,
26997 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
26998 estimate_function_body_sizes, compute_inline_parameters,
26999 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27000 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27001 inline_update_callee_summaries, remap_edge_change_prob,
27002 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27003 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27004 inline_analyze_function, inline_summary_t::insert,
27005 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27006 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27007 inline_free_summary): Move to ipa-fnsummary.h
27008 (predicate_t): Remove.
27009 * ipa-fnsummary.c: New file.
27010 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27011 (enum inline_hints_vals, inline_hints, agg_position_info,
27012 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27013 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27014 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27015 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27016 inline_read_summary, inline_write_summary, inline_free_summary,
27017 inline_analyze_function, initialize_inline_failed,
27018 inline_merge_summary, inline_update_overall_summary,
27019 compute_inline_parameters): Move to ipa-fnsummary.h
27020 * ipa-fnsummary.h: New file.
27021 * ipa-inline-transform.h: Include ipa-inline.h.
27022 * ipa-inline.c: LIkewise.
27024 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27026 * ipa-inline.c (edge_badness): Use inlined_time instead of
27027 inline_summaries->get.
27029 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27031 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27033 2017-05-22 Nathan Sidwell <nathan@acm.org>
27035 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27036 (fdump-lang): Document 'raw' option.
27037 * dumpfile.h (TDI_tu): Delete.
27038 * dumpfile.c (dump_files): Remove translation-unit.
27039 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27041 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27043 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27044 command option from $(AWK) call.
27045 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27046 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27047 [FORMAT]: Remove handling of variable.
27048 * config/avr/t-multilib: Regenerate.
27050 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27052 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27054 (dump_inline_summary): Do not print self_time.
27055 (estimate_function_body_sizes): Do not set self_time.
27056 (compute_inline_parameters): Likewise.
27057 (inline_read_section, inline_write_summary): Do not stream self_time.
27058 * ipa-inline.h (inline_summary): Drop self_time.
27060 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27062 * ipa-inline-analysis.c (account_size_time): Rename to ...
27063 (inline_summary::account_size_time): ... this one.
27064 (reset_ipa_call_summary): Turn to ...
27065 (ipa_call_summary::reset): ... this one.
27066 (reset_inline_summary): Turn to ...
27067 (inline_summary::reset): ... this one.
27068 (inline_summary_t::remove): Update.
27069 (inline_summary_t::duplicate): Update.
27070 (ipa_call_summary_t::remove): Update.
27071 (dump_inline_summary): Update.
27072 (estimate_function_body_sizes): Update.
27073 (compute_inline_parameters): Update.
27074 (estimate_node_size_and_time): Update.
27075 (inline_merge_summary): Update.
27076 (inline_update_overall_summary): Update.
27077 (inline_read_section): Update.
27078 (inline_write_summary): Update.
27079 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27080 add account_size_time and reset member functions.
27081 (ipa_call_summary): Add reset function.
27082 * ipa-predicate.h (predicate::operator &): Constify.
27084 2017-05-22 Richard Biener <rguenther@suse.de>
27086 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27088 2017-05-19 Jason Merrill <jason@redhat.com>
27090 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27092 2017-05-19 Marek Polacek <polacek@redhat.com>
27095 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27096 TYPE_OVERFLOW_WRAPS checks.
27098 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27100 * tree-core.h (enum omp_clause_default_kind): Add
27101 "OMP_CLAUSE_DEFAULT_PRESENT".
27102 * tree-pretty-print.c (dump_omp_clause): Handle it.
27103 * gimplify.c (enum gimplify_omp_var_data): Add
27104 "GOVD_MAP_FORCE_PRESENT".
27105 (gimplify_adjust_omp_clauses_1): Map it to
27106 "GOMP_MAP_FORCE_PRESENT".
27107 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27109 * gimplify.c (oacc_default_clause): Clarify.
27111 2017-05-19 Nathan Sidwell <nathan@acm.org>
27113 LANG_HOOK_REGISTER_DUMPS
27114 * toplev.c (general_init): Call register dump lang hook.
27115 * doc/invoke.texi: Document -fdump-lang option family.
27116 * dumpfile.c (dump_files): Remove class dump here.
27117 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27118 * dumpfile.h (tree_dump_index): Remove TDI_class.
27119 * langhooks-def.h (lhd_register_dumps): Declare.
27120 (LANG_HOOKS_REGISTER_DUMPS): Define.
27121 (LANG_HOOKS_INITIALIZER): Add it.
27122 * langhooks.c (lhd_register_dumps): Define.
27123 * langhooks.h (struct lang_hooks): Add register_dumps.
27125 2017-05-19 Nathan Sidwell <nathan@acm.org>
27127 * context.h (context::set_passes): New.
27128 * context.c (context::context): Do not create pass manager.
27129 * toplev.c (general_init): Create pass manager here.
27131 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27133 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27134 use this splitter if two add or or instructions would also work for
27135 the constant we want to generate.
27137 2017-05-19 Richard Biener <rguenther@suse.de>
27140 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27141 predicate evaluation.
27143 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27145 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27147 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27150 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27152 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27153 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27154 (inline_edge_summary_vec): Turn into ...
27155 (ipa_call_summaries): ... this one.
27156 (redirect_to_unreachable, edge_set_predicate,
27157 evaluate_properties_for_edge, inline_summary_alloc,
27158 reset_ipa_call_summary, reset_inline_summary,
27159 inline_summary_t::duplicate): Update.
27160 (inline_edge_duplication_hook): Turn to ...
27161 (ipa_call_summary_t::duplicate): ... this one.
27162 (inline_edge_removal_hook): Turn to ...
27163 (ipa_call_summary_t::remove): ... this one.
27164 (dump_inline_edge_summary): Turn to ...
27165 (dump_ipa_call_summary): ... this one.
27166 (estimate_function_body_sizes): Update.
27167 (inline_update_callee_summaries): Update.
27168 (remap_edge_change_prob): Update.
27169 (remap_edge_summaries): Update.
27170 (inline_merge_summary): Update.
27171 (do_estimate_edge_time): Update.
27172 (inline_generate_summary): Update.
27173 (inline_read_section): Update.
27174 (inline_read_summary): Update.
27175 (inline_free_summary): Update.
27176 * ipa-inline.c (can_inline_edge_p): Update.
27177 (compute_inlined_call_time): Update.
27178 (want_inline_small_function_p): Update.
27179 (edge_badness): Update.
27180 (early_inliner): Update.
27181 * ipa-inline.h (inline_edge_summary): Turn to ...
27182 (ipa_call_summary): ... this one.
27183 (ipa_call_summary_t): New class.
27184 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27185 (ipa_call_summaries): New.
27186 (inline_edge_summary): Remove.
27187 (estimate_edge_growth): Update.
27188 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27189 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27190 * ipa-split.c (execute_split_functions): Update.
27191 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27193 2017-05-19 Richard Biener <rguenther@suse.de>
27195 PR middle-end/80764
27196 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27198 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27200 * config/rs6000/rs6000.c (struct machine_function): Add field
27201 fpr_is_wrapped_separately.
27202 (rs6000_get_separate_components): Use 64 components. Handle the
27203 new FPR components.
27204 (rs6000_components_for_bb): Handle the FPR components.
27205 (rs6000_emit_prologue_components): Handle the FPR components.
27206 (rs6000_emit_epilogue_components): Handle the FPR components.
27207 (rs6000_set_handled_components): Handle the FPR components.
27208 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27209 that are already separately shrink-wrapped.
27210 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27211 that are already separately shrink-wrapped.
27213 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27216 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27219 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27220 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27221 cases where the register allocator uses a move and an offsettable
27222 memory operation to/from a FPR register on ISA 2.06/2.07.
27223 (define_peephole2 for Altivec d-form store): Likewise.
27225 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27228 * config/i386/mmx.md (*mov<mode>_internal): Enable
27229 alternatives 11, 12, 13 and 14 also for 32bit targets.
27230 Remove alternatives 15, 16, 17 and 18.
27231 * config/i386/sse.md (vec_concatv2di): Change
27232 alternative (!x, *y) to (x, ?!*Yn).
27234 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27236 * dumpfile.h (enum dump_kind): Remove stray comma.
27238 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27240 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27241 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27242 predicate::num_conditions
27243 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27244 (CHANGED): turn into predicate::changed.
27245 (agg_position_info): Move to ipa-predicate.h
27246 (add_condition, predicate::add_clause, predicate::operator &=,
27247 predicate::or_with, predicate::evaluate, predicate::probability,
27248 dump_condition, dump_clause, predicate::dump,
27249 predicate::remap_after_duplication, predicate::remap_after_inlining,
27250 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27251 (evaluate_conditions_for_known_args): Update.
27252 (set_cond_stmt_execution_predicate): Update.
27253 * ipa-inline.h: Include ipa-predicate.h
27254 (condition, inline_param_summary, conditions, agg_position_info,
27255 predicate): Move to ipa-predicate.h
27256 * ipa-predicate.c: New file.
27257 * ipa-predicate.h: New file.
27259 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27261 * final.c (leaf_function_p): Check we are not in a sequence.
27263 2017-05-18 Martin Liska <mliska@suse.cz>
27265 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27266 * dumpfile.c (dump_register): Use new enum dump_kind.
27267 (get_dump_file_name): Likewise.
27268 (dump_enable_all): Likewise.
27269 (dump_switch_p_1): Likewise.
27270 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27271 * dumpfile.h (enum dump_kind): New enum type.
27272 (struct dump_file_info): Create constructor and
27273 format fields and comments.
27274 * passes.c (pass_manager::register_one_dump_file):
27276 * statistics.c (statistics_early_init): Likewise.
27277 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27278 TDF_TREE with TDF_SLIM.
27279 (gather_memory_references_ref): Likewise.
27281 2017-05-18 Martin Liska <mliska@suse.cz>
27283 * vec.h (struct vnull): Use it.
27285 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27287 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27288 (true_predicate, false_predicate, true_predicate_p,
27289 false_predicate_p): Remove.
27290 (single_cond_predicate, not_inlined_predicate): Turn to member function
27292 (add_condition): Update.
27293 (add_clause): Turn to...
27294 (predicate::add_clause): ... this one; update; allow passing NULL
27296 (and_predicates): Turn to ...
27297 (predicate::operator &=): ... this one.
27298 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27299 (or_predicates): Turn to ...
27300 (predicate::or_with): ... this one.
27301 (evaluate_predicate): Turn to ...
27302 (predicate::evaluate): ... this one.
27303 (predicate_probability): Turn to ...
27304 (predicate::probability): ... this one.
27305 (dump_condition): Update.
27306 (dump_predicate): Turn to ...
27307 (predicate::dump): ... this one.
27308 (account_size_time): Update.
27309 (edge_set_predicate): Update.
27310 (set_hint_predicate): UPdate.
27311 (evaluate_conditions_for_known_args): Update.
27312 (evaluate_properties_for_edge): Update.
27313 (remap_predicate_after_duplication): Turn to...
27314 (predicate::remap_after_duplication): ... this one.
27315 (remap_hint_predicate_after_duplication): Update.
27316 (inline_summary_t::duplicate): UPdate.
27317 (dump_inline_edge_summary): Update.
27318 (dump_inline_summary): Update.
27319 (set_cond_stmt_execution_predicate): Update.
27320 (set_switch_stmt_execution_predicate): Update.
27321 (compute_bb_predicates): Update.
27322 (will_be_nonconstant_expr_predicate): Update.
27323 (will_be_nonconstant_predicate): Update.
27324 (phi_result_unknown_predicate): Update.
27325 (predicate_for_phi_result): Update.
27326 (array_index_predicate): Update.
27327 (estimate_function_body_sizes): Update.
27328 (estimate_node_size_and_time): Update.
27329 (estimate_ipcp_clone_size_and_time): Update.
27330 (remap_predicate): Rename to ...
27331 (predicate::remap_after_inlining): ... this one.
27332 (remap_hint_predicate): Update.
27333 (inline_merge_summary): Update.
27334 (inline_update_overall_summary): Update.
27335 (estimate_size_after_inlining): Update.
27336 (read_predicate): Rename to ...
27337 (predicate::stream_in): ... this one.
27338 (read_inline_edge_summary): Update.
27339 (write_predicate): Rename to ...
27340 (predicate::stream_out): ... this one.
27341 (write_inline_edge_summary): Update.
27342 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27343 (clause_t): Turn to uint32_t
27344 (predicate): Turn to class; implement constructor and operators
27346 (size_time_entry): Update.
27347 (inline_summary): Update.
27348 (inline_edge_summary): Update.
27350 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27352 * fold-const.c (fold_binary_loc): Move transformation...
27353 * match.pd (C - X CMP X): ... here.
27355 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27357 * config/sparc/sparc.c (sparc_option_override): Set function
27358 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27359 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27361 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27363 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27365 2017-05-18 Marek Polacek <polacek@redhat.com>
27368 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27369 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27370 gimple_assign_load_p.
27372 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27374 PR middle-end/80692
27375 * real.c (do_compare): Give decimal_do_compare preference over
27376 comparing just the signs.
27378 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27380 * doc/md.texi (Canonicalization of Instructions): Describe the
27381 canonical form of instructions that inherently set a condition
27384 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27386 PR middle-end/80775
27387 * tree-cfg.c: Move deletion of unreachable case statements to after
27388 the merging of consecutive case labels.
27390 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27392 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27393 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27394 restoring of callee-saved registers.
27396 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27398 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27399 * config/visium/visium.c (single_set_and_flags): Likewise.
27400 * config/visium/visium.md (Substitutions): Likewise.
27402 2017-05-17 Martin Liska <mliska@suse.cz>
27404 * cfg.c: Introduce dump_flags_t type and
27405 use it instead of int type.
27407 * cfghooks.c: Likewise.
27408 * cfghooks.h (struct cfg_hooks): Likewise.
27409 * cfgrtl.c: Likewise.
27410 * cfgrtl.h: Likewise.
27411 * cgraph.c (cgraph_node::get_body): Likewise.
27412 * coretypes.h: Likewise.
27413 * domwalk.c: Likewise.
27414 * domwalk.h: Likewise.
27415 * dumpfile.c (struct dump_option_value_info): Likewise.
27416 (dump_enable_all): Likewise.
27417 (dump_switch_p_1): Likewise.
27418 (opt_info_switch_p): Likewise.
27419 * dumpfile.h (enum tree_dump_index): Likewise.
27420 (struct dump_file_info): Likewise.
27421 * genemit.c: Likewise.
27422 * generic-match-head.c: Likewise.
27423 * gengtype.c (open_base_files): Likewise.
27424 * gimple-pretty-print.c: Likewise.
27425 * gimple-pretty-print.h: Likewise.
27426 * graph.c (print_graph_cfg): Likewise.
27427 * graphite-scop-detection.c (dot_all_sese): Likewise.
27428 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27429 * loop-unroll.c (report_unroll): Likewise.
27430 * passes.c (pass_manager::register_one_dump_file): Likewise.
27431 * print-tree.c: Likewise.
27432 * statistics.c: Likewise.
27433 * tree-cfg.c: Likewise.
27434 * tree-cfg.h: Likewise.
27435 * tree-dfa.c: Likewise.
27436 * tree-dfa.h: Likewise.
27437 * tree-dump.c (dump_function): Likewise.
27438 * tree-dump.h (struct dump_info): Likewise.
27439 * tree-pretty-print.c: Likewise.
27440 * tree-pretty-print.h: Likewise.
27441 * tree-ssa-live.c: Likewise.
27442 * tree-ssa-live.h: Likewise.
27443 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27444 * tree-vect-loop.c: Likewise.
27445 * tree-vect-slp.c: Likewise.
27447 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27448 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27450 PR tree-optimization/80457
27451 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27452 of all arguments to a statement as scalar_to_vec operations.
27453 (vectorizable_call): Adjust call to vect_model_simple_cost for
27455 (vectorizable_conversion): Likewise.
27456 (vectorizable_assignment): Likewise.
27457 (vectorizable_shift): Likewise.
27458 (vectorizable_operation): Likewise.
27459 (vectorizable_comparison): Likewise.
27460 (vect_is_simple_cond): Record the def types for operands.
27461 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27462 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27463 for statement argument count.
27465 2017-05-16 Carl Love <cel@us.ibm.com>
27467 * config/rs6000/rs6000-c: Add support for built-in functions
27468 vector unsigned long long vec_bperm (vector unsigned long long,
27469 vector unsigned char)
27470 vector signed long long vec_mule (vector signed int,
27472 vector unsigned long long vec_mule (vector unsigned int,
27473 vector unsigned int)
27474 vector signed long long vec_mulo (vector signed int,
27476 vector unsigned long long vec_mulo (vector unsigned int,
27477 vector unsigned int)
27478 vector signed char vec_sldw (vector signed char,
27479 vector signed char,
27481 vector unsigned char vec_sldw (vector unsigned char,
27482 vector unsigned char,
27484 vector signed short vec_sldw (vector signed short,
27485 vector signed short,
27487 vector unsigned short vec_sldw (vector unsigned short,
27488 vector unsigned short,
27490 vector signed int vec_sldw (vector signed int,
27493 vector unsigned int vec_sldw (vector unsigned int,
27494 vector unsigned int,
27496 vector signed long long vec_sldw (vector signed long long,
27497 vector signed long long,
27499 vector unsigned long long vec_sldw (vector unsigned long long,
27500 vector unsigned long long,
27502 * config/rs6000/rs6000-c: Add support for built-in functions
27503 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27504 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27505 * doc/extend.texi: Update the built-in documentation for the
27506 new built-in functions.
27508 2017-05-16 Marek Polacek <polacek@redhat.com>
27512 * tree.c (save_expr): Don't fold the expression.
27514 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27516 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27517 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27518 and (?*y,m). Update insn attributes.
27520 2017-05-16 Martin Liska <mliska@suse.cz>
27522 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27523 flags argument of print_gimple_stmt, print_gimple_expr,
27524 print_generic_stmt and print_generic_expr.
27525 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27526 * coretypes.h: Likewise.
27527 * except.c (dump_eh_tree): Likewise.
27528 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27529 * gimple-pretty-print.h: Likewise.
27530 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27531 (backprop::push_to_worklist): Likewise.
27532 (backprop::pop_from_worklist): Likewise.
27533 (backprop::process_use): Likewise.
27534 (backprop::intersect_uses): Likewise.
27535 (note_replacement): Likewise.
27536 * gimple-ssa-store-merging.c
27537 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27538 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27539 (pass_store_merging::execute): Likewise.
27540 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27541 (ssa_base_cand_dump_callback): Likewise.
27542 (dump_incr_vec): Likewise.
27543 (replace_refs): Likewise.
27544 (replace_mult_candidate): Likewise.
27545 (create_add_on_incoming_edge): Likewise.
27546 (create_phi_basis): Likewise.
27547 (insert_initializers): Likewise.
27548 (all_phi_incrs_profitable): Likewise.
27549 (introduce_cast_before_cand): Likewise.
27550 (replace_one_candidate): Likewise.
27551 * gimplify.c (gimplify_expr): Likewise.
27552 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27553 (set_rename): Likewise.
27554 (rename_uses): Likewise.
27555 (copy_loop_phi_nodes): Likewise.
27556 (add_close_phis_to_merge_points): Likewise.
27557 (copy_loop_close_phi_args): Likewise.
27558 (copy_cond_phi_args): Likewise.
27559 (graphite_copy_stmts_from_block): Likewise.
27560 (translate_pending_phi_nodes): Likewise.
27561 * graphite-poly.c (print_pdr): Likewise.
27562 (dump_gbb_cases): Likewise.
27563 (dump_gbb_conditions): Likewise.
27564 (print_scop_params): Likewise.
27565 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27566 (build_cross_bb_scalars_use): Likewise.
27567 (gather_bbs::before_dom_children): Likewise.
27568 * hsa-dump.c (dump_hsa_immed): Likewise.
27569 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27570 (get_replacement_map): Likewise.
27571 * ipa-inline-analysis.c (dump_condition): Likewise.
27572 (estimate_function_body_sizes): Likewise.
27573 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27574 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27575 * ipa-prop.c (ipa_dump_param): Likewise.
27576 (ipa_print_node_jump_functions_for_edge): Likewise.
27577 (ipa_modify_call_arguments): Likewise.
27578 (ipa_modify_expr): Likewise.
27579 (ipa_dump_param_adjustments): Likewise.
27580 (ipa_dump_agg_replacement_values): Likewise.
27581 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27582 * ipa-pure-const.c (check_stmt): Likewise.
27583 (pass_nothrow::execute): Likewise.
27584 * ipa-split.c (execute_split_functions): Likewise.
27585 * omp-offload.c (dump_oacc_loop_part): Likewise.
27586 (dump_oacc_loop): Likewise.
27587 * trans-mem.c (tm_log_emit): Likewise.
27588 (tm_memopt_accumulate_memops): Likewise.
27589 (dump_tm_memopt_set): Likewise.
27590 (dump_tm_memopt_transform): Likewise.
27591 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27592 (print_loop): Likewise.
27593 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27594 (chkp_gather_checks_info): Likewise.
27595 (chkp_get_check_result): Likewise.
27596 (chkp_remove_check_if_pass): Likewise.
27597 (chkp_use_outer_bounds_if_possible): Likewise.
27598 (chkp_reduce_bounds_lifetime): Likewise.
27599 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27600 (chkp_mark_completed_bounds): Likewise.
27601 (chkp_register_incomplete_bounds): Likewise.
27602 (chkp_mark_invalid_bounds): Likewise.
27603 (chkp_maybe_copy_and_register_bounds): Likewise.
27604 (chkp_build_returned_bound): Likewise.
27605 (chkp_get_bound_for_parm): Likewise.
27606 (chkp_build_bndldx): Likewise.
27607 (chkp_get_bounds_by_definition): Likewise.
27608 (chkp_generate_extern_var_bounds): Likewise.
27609 (chkp_get_bounds_for_decl_addr): Likewise.
27610 * tree-chrec.c (chrec_apply): Likewise.
27611 * tree-data-ref.c (dump_data_reference): Likewise.
27612 (dump_subscript): Likewise.
27613 (dump_data_dependence_relation): Likewise.
27614 (analyze_overlapping_iterations): Likewise.
27615 * tree-inline.c (expand_call_inline): Likewise.
27616 (tree_function_versioning): Likewise.
27617 * tree-into-ssa.c (dump_defs_stack): Likewise.
27618 (dump_currdefs): Likewise.
27619 (dump_names_replaced_by): Likewise.
27620 (dump_update_ssa): Likewise.
27621 (update_ssa): Likewise.
27622 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27623 * tree-parloops.c (build_new_reduction): Likewise.
27624 (try_create_reduction_list): Likewise.
27625 (ref_conflicts_with_region): Likewise.
27626 (oacc_entry_exit_ok_1): Likewise.
27627 (oacc_entry_exit_single_gang): Likewise.
27628 * tree-pretty-print.h: Likewise.
27629 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27630 (get_scalar_evolution): Likewise.
27631 (add_to_evolution): Likewise.
27632 (get_loop_exit_condition): Likewise.
27633 (analyze_evolution_in_loop): Likewise.
27634 (analyze_initial_condition): Likewise.
27635 (analyze_scalar_evolution): Likewise.
27636 (instantiate_scev): Likewise.
27637 (number_of_latch_executions): Likewise.
27638 (gather_chrec_stats): Likewise.
27639 (final_value_replacement_loop): Likewise.
27640 (scev_const_prop): Likewise.
27641 * tree-sra.c (dump_access): Likewise.
27642 (disqualify_candidate): Likewise.
27643 (create_access): Likewise.
27644 (reject): Likewise.
27645 (maybe_add_sra_candidate): Likewise.
27646 (create_access_replacement): Likewise.
27647 (analyze_access_subtree): Likewise.
27648 (analyze_all_variable_accesses): Likewise.
27649 (sra_modify_assign): Likewise.
27650 (initialize_constant_pool_replacements): Likewise.
27651 (find_param_candidates): Likewise.
27652 (decide_one_param_reduction): Likewise.
27653 (replace_removed_params_ssa_names): Likewise.
27654 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27655 * tree-ssa-copy.c (dump_copy_of): Likewise.
27656 (copy_prop_visit_cond_stmt): Likewise.
27657 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27658 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27659 (record_equivalences_from_stmt): Likewise.
27660 * tree-ssa-dse.c (compute_trims): Likewise.
27661 (delete_dead_call): Likewise.
27662 (delete_dead_assignment): Likewise.
27663 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27664 (forward_propagate_into_cond): Likewise.
27665 (pass_forwprop::execute): Likewise.
27666 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27667 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27669 (move_computations_worker): Likewise.
27670 (execute_sm): Likewise.
27671 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27672 (remove_exits_and_undefined_stmts): Likewise.
27673 (remove_redundant_iv_tests): Likewise.
27674 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27675 (adjust_iv_update_pos): Likewise.
27676 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27677 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27678 (value_replacement): Likewise.
27679 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27680 * tree-ssa-pre.c (print_pre_expr): Likewise.
27681 (get_representative_for): Likewise.
27682 (create_expression_by_pieces): Likewise.
27683 (insert_into_preds_of_block): Likewise.
27684 (eliminate_insert): Likewise.
27685 (eliminate_dom_walker::before_dom_children): Likewise.
27686 (eliminate): Likewise.
27687 (remove_dead_inserted_code): Likewise.
27688 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27689 * tree-ssa-reassoc.c (get_rank): Likewise.
27690 (eliminate_duplicate_pair): Likewise.
27691 (eliminate_plus_minus_pair): Likewise.
27692 (eliminate_not_pairs): Likewise.
27693 (undistribute_ops_list): Likewise.
27694 (eliminate_redundant_comparison): Likewise.
27695 (update_range_test): Likewise.
27696 (optimize_range_tests_var_bound): Likewise.
27697 (optimize_vec_cond_expr): Likewise.
27698 (rewrite_expr_tree): Likewise.
27699 (rewrite_expr_tree_parallel): Likewise.
27700 (linearize_expr): Likewise.
27701 (break_up_subtract): Likewise.
27702 (linearize_expr_tree): Likewise.
27703 (attempt_builtin_powi): Likewise.
27704 (attempt_builtin_copysign): Likewise.
27705 (transform_stmt_to_copy): Likewise.
27706 (transform_stmt_to_multiply): Likewise.
27707 (dump_ops_vector): Likewise.
27708 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27709 (print_scc): Likewise.
27710 (set_ssa_val_to): Likewise.
27711 (visit_reference_op_store): Likewise.
27712 (visit_use): Likewise.
27713 (sccvn_dom_walker::before_dom_children): Likewise.
27714 (run_scc_vn): Likewise.
27715 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27717 (expr_hash_elt::print): Likewise.
27718 (const_and_copies::pop_to_marker): Likewise.
27719 (const_and_copies::record_const_or_copy_raw): Likewise.
27720 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27721 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27722 (dump_predicates): Likewise.
27723 (find_uninit_use): Likewise.
27724 (warn_uninitialized_phi): Likewise.
27725 (pass_late_warn_uninitialized::execute): Likewise.
27726 * tree-ssa.c (verify_vssa): Likewise.
27727 (verify_ssa): Likewise.
27728 (maybe_optimize_var): Likewise.
27729 * tree-vrp.c (dump_value_range): Likewise.
27730 (dump_all_value_ranges): Likewise.
27731 (dump_asserts_for): Likewise.
27732 (register_edge_assert_for_2): Likewise.
27733 (vrp_visit_cond_stmt): Likewise.
27734 (vrp_visit_switch_stmt): Likewise.
27735 (vrp_visit_stmt): Likewise.
27736 (vrp_visit_phi_node): Likewise.
27737 (simplify_cond_using_ranges_1): Likewise.
27738 (fold_predicate_in): Likewise.
27739 (evrp_dom_walker::before_dom_children): Likewise.
27740 (evrp_dom_walker::push_value_range): Likewise.
27741 (evrp_dom_walker::pop_value_range): Likewise.
27742 (execute_early_vrp): Likewise.
27744 2017-05-16 Richard Biener <rguenther@suse.de>
27746 * dwarf2out.c (loc_list_from_tree_1): Do not create
27747 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27749 2017-05-16 Richard Biener <rguenther@suse.de>
27751 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27753 (note_variable_value_in_expr): If we resolved the decl ref
27754 do not push to the stack.
27756 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27758 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27759 operations in fast-math mode.
27760 (vaddq_f16): Likewise.
27761 (vmul_f16): Likewise.
27762 (vmulq_f16): Likewise.
27763 (vsub_f16): Likewise.
27764 (vsubq_f16): Likewise.
27765 * config/arm/neon.md (add<mode>3): New.
27767 (fma:<VH:mode>3): New. Also remove outdated comment.
27770 2017-05-16 Martin Liska <mliska@suse.cz>
27774 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27775 (odr_types_equivalent_p): Likewise.
27777 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27779 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27781 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27784 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27785 non-interunit SSE move alternatives with '?'.
27786 (zero-extendsidi peephole2): New peephole to skip intermediate
27787 general register in SSE zero-extend sequence.
27789 2017-05-15 Jeff Law <law@redhat.com>
27791 * reorg.c (relax_delay_slots): Create a new variable to hold
27792 the temporary target rather than clobbering TARGET_LABEL.
27794 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27795 missing argument to extract_bit_field call.
27796 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27798 2017-05-15 Martin Liska <mliska@suse.cz>
27801 * gcc.c (process_command): Do not allow empty argument of -o option.
27803 2017-05-15 Renlin Li <renlin.li@arm.com>
27805 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27806 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27807 * config/aarch64/constraints.md (Usf): Add long call check.
27808 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27809 (call_value): Likewise.
27810 (sibcall): Likewise.
27811 (sibcall_value): Likewise.
27813 (call_value_insn): New.
27814 (sibcall_insn): Update rtx pattern.
27815 (sibcall_value_insn): Likewise.
27816 (call_internal): Remove.
27817 (call_value_internal): Likewise.
27818 (sibcall_internal): Likewise.
27819 (sibcall_value_internal): Likewise.
27820 (call_reg): Likewise.
27821 (call_symbol): Likewise.
27822 (call_value_reg): Likewise.
27823 (call_value_symbol): Likewise.
27825 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27828 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27830 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27832 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27833 compatible with CCGOCmode and with CCZmode.
27835 2017-05-14 Martin Sebor <msebor@redhat.com>
27837 PR middle-end/77671
27838 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27839 (gimple_fold_builtin_snprintf): Same.
27840 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27841 (gimple_fold_builtin_snprintf): Same.
27842 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27843 of character types.
27844 (is_call_safe): New function.
27845 (try_substitute_return_value): Call it.
27846 (try_simplify_call): New function.
27847 (pass_sprintf_length::handle_gimple_call): Call it.
27849 2017-05-14 Martin Sebor <msebor@redhat.com>
27851 PR middle-end/80669
27852 * builtins.c (expand_builtin_stpncpy): Simplify.
27854 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27856 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27857 * config/i386/i386.h
27858 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27859 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27860 (struct machine_function): Add new members call_ms2sysv,
27861 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27862 (struct machine_frame_state): New fields sp_realigned and
27863 sp_realigned_offset.
27864 * config/i386/i386.c
27865 (enum xlogue_stub): New enum.
27866 (enum xlogue_stub_sets): New enum.
27867 (class xlogue_layout): New class.
27868 (struct ix86_frame): New fields stack_realign_allocate_offset,
27869 stack_realign_offset and outlined_save_offset. Modify comments to
27870 detail stack layout when using out-of-line stubs.
27871 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27872 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27873 -mcall-ms2sysv-xlogues.
27874 (stub_managed_regs): New static variable.
27875 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27876 registers managed by out-of-line stub.
27877 (disable_call_ms2sysv_xlogues): New function.
27878 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27879 m->call_ms2sysv when appropriate and compute frame layout for
27881 (sp_valid_at, fp_valid_at): New inline functions.
27882 (choose_basereg): New function.
27883 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27885 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27886 Use align parameter of choose_baseaddr to generated aligned SSE movs
27888 (pro_epilogue_adjust_stack): Modify to track
27889 machine_frame_state::sp_realigned.
27890 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27891 (ix86_nsaved_sseregs): Likewise.
27892 (ix86_emit_save_regs): Likewise.
27893 (ix86_emit_save_regs_using_mov): Likewise.
27894 (ix86_emit_save_sse_regs_using_mov): Likewise.
27895 (get_scratch_register_on_entry): Likewise.
27896 (gen_frame_set): New function.
27897 (gen_frame_load): Likewise.
27898 (gen_frame_store): Likewise.
27899 (emit_outlined_ms2sysv_save): Likewise.
27900 (emit_outlined_ms2sysv_restore): Likewise.
27901 (ix86_expand_prologue): Modify stack re-alignment code and call
27902 emit_outlined_ms2sysv_save when appropriate.
27903 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27904 parameter rtx_insn *insn, which allows the function to be used to only
27905 generate the notes.
27906 (ix86_expand_epilogue): Modify validity checks of frame and stack
27907 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27908 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27909 * config/i386/predicates.md
27910 (save_multiple): New predicate.
27911 (restore_multiple): Likewise.
27912 * config/i386/sse.md
27913 (save_multiple<mode>): New pattern.
27914 (save_multiple_realign<mode>): Likewise.
27915 (restore_multiple<mode>): Likewise.
27916 (restore_multiple_and_return<mode>): Likewise.
27917 (restore_multiple_leave_return<mode>): Likewise.
27918 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27920 2017-05-14 Julia Koval <julia.koval@intel.com>
27922 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
27923 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
27924 (__builtin_ia32_xsetbv): New builtins.
27925 * config/i386/i386.c (ix86_expand_special_args_builtin):
27927 (ix86_expand_builtin): Special expand for new intrinsics.
27928 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
27929 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
27930 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
27932 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27934 * cfganal.c (inverted_post_order_compute): Change argument type
27936 * cfganal.h (inverted_post_order_compute): Adjust prototype.
27937 * df-core.c (rest_of_handle_df_initialize): Adjust.
27938 (rest_of_handle_df_finish): Likewise.
27939 (df_analyze_1): Likewise.
27940 (df_analyze): Likewise.
27941 (loop_inverted_post_order_compute): Change argument to be a vec *.
27942 (df_analyze_loop): Adjust.
27943 (df_get_n_blocks): Likewise.
27944 (df_get_postorder): Likewise.
27945 * df.h (struct df_d): Change field to be a vec.
27946 * lcm.c (compute_laterin): Adjust.
27947 (compute_available): Likewise.
27948 * lra-lives.c (lra_create_live_ranges_1): Likewise.
27949 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
27950 * tree-ssa-pre.c (compute_antic): Likewise.
27952 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27954 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
27955 (depth_first_search::depth_first_search): Change structure init
27956 function to this constructor.
27957 (depth_first_search::add_bb): Rename function to this member.
27958 (depth_first_search::execute): Likewise.
27959 (flow_dfs_compute_reverse_finish): Adjust.
27961 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27963 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
27964 (longest_simple_path): Likewise.
27965 * shrink-wrap.c (spread_components): Likewise.
27966 (disqualify_problematic_components): Likewise.
27967 (emit_common_heads_for_components): Likewise.
27968 (emit_common_tails_for_components): Likewise.
27969 (insert_prologue_epilogue_for_components): Likewise.
27971 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27973 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
27976 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27978 * df-core.c (df_set_blocks): Start using auto_bitmap.
27979 (df_compact_blocks): Likewise.
27980 * df-problems.c (df_rd_confluence_n): Likewise.
27981 * df-scan.c (df_insn_rescan_all): Likewise.
27982 (df_process_deferred_rescans): Likewise.
27983 (df_update_entry_block_defs): Likewise.
27984 (df_update_exit_block_uses): Likewise.
27985 (df_entry_block_bitmap_verify): Likewise.
27986 (df_exit_block_bitmap_verify): Likewise.
27987 (df_scan_verify): Likewise.
27988 * lra-constraints.c (lra_constraints): Likewise.
27989 (undo_optional_reloads): Likewise.
27990 (lra_undo_inheritance): Likewise.
27991 * lra-remat.c (calculate_gen_cands): Likewise.
27992 (do_remat): Likewise.
27993 * lra-spills.c (assign_spill_hard_regs): Likewise.
27994 (spill_pseudos): Likewise.
27995 * tree-ssa-pre.c (bitmap_set_and): Likewise.
27996 (bitmap_set_subtract_values): Likewise.
27998 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28000 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28001 management with auto_bitmap.
28002 (fix_inter_tick): Likewise.
28003 (fix_recovery_deps): Likewise.
28004 * ira.c (add_store_equivs): Likewise.
28005 (find_moveable_pseudos): Likewise.
28006 (split_live_ranges_for_shrink_wrap): Likewise.
28007 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28008 (rtx_reuse_manager::seen_def_p): Likewise.
28009 (rtx_reuse_manager::set_seen_def): Likewise.
28010 * print-rtl.h (class rtx_reuse_manager): Likewise.
28012 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28014 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28016 (migrate_btr_def): Likewise.
28017 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28018 * df-core.c (loop_post_order_compute): Likewise.
28019 (loop_inverted_post_order_compute): Likewise.
28020 * hsa-common.h: Likewise.
28021 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28022 * init-regs.c (initialize_uninitialized_regs): Likewise.
28023 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28024 (inline_small_functions): Likewise.
28025 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28026 * ira.c (combine_and_move_insns): Likewise.
28027 (build_insn_chain): Likewise.
28028 * loop-invariant.c (find_invariants): Likewise.
28029 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28030 * predict.c (tree_predict_by_opcode): Likewise.
28031 (predict_paths_leading_to): Likewise.
28032 (predict_paths_leading_to_edge): Likewise.
28033 (estimate_loops_at_level): Likewise.
28034 (estimate_loops): Likewise.
28035 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28036 (spread_components): Likewise.
28037 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28038 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28039 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28040 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28041 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28042 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28043 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28044 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28045 (mark_threaded_blocks): Likewise.
28046 (thread_through_all_blocks): Likewise.
28047 * tree-ssa.c (verify_ssa): Likewise.
28048 (execute_update_addresses_taken): Likewise.
28049 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28051 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28053 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28055 (post_order_compute): Likewise.
28056 (inverted_post_order_compute): Likewise.
28057 (pre_and_rev_post_order_compute_fn): Likewise.
28059 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28061 * genrecog.c (int_set::int_set): Explicitly construct our
28062 auto_vec base class.
28063 * vec.h (auto_vec::auto_vec): New constructor.
28065 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28067 * bitmap.h (class auto_bitmap): New constructor taking
28068 bitmap_obstack * argument.
28070 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28072 * bitmap.h (class auto_bitmap): Change type of m_bits to
28073 bitmap_head, and adjust ctor / dtor and member operators.
28075 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28077 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28078 when returned register mode doesn't match original mode.
28080 2017-05-12 Jeff Law <law@redhat.com>
28081 Jakub Jelinek <jakub@redhat.com>
28083 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28084 we look for cc setter after the compare-elim changes.
28085 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28086 within the vector to match what compare-elim now expects.
28087 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28088 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28090 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28091 after the compare-elim changes.
28092 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28093 the vector to match what compare-elim now expects.
28094 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28095 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28096 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28097 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28098 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28100 * config/visium/visium.c (single_set_and_flags): Fix where
28101 we look for cc setter after the compare-elim changes.
28102 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28103 with the vector to match what compare-elim now expects.
28104 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28105 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28106 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28107 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28108 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28109 (neg<mode>2_insn_set_overflow): Likewise.
28111 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28113 PR middle-end/79794
28114 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28115 maybe_expand_insn call, set ops[0].target. If still set after call,
28116 set alt_rtl. Add extra arg to recursive calls.
28117 (extract_bit_field): Add alt_rtl argument. Pass to
28119 * expmed.h (extract_bit_field): Fix prototype.
28120 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28121 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28122 to extract_bit_field_calls.
28123 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28124 Pass alt_rtl to extract_bit_field calls.
28125 * calls.c (store_unaligned_arguments_into_psuedos)
28126 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28127 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28129 * optabs.h (struct expand_operand): Add target bitfield.
28131 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28133 * compare-elim.c (try_eliminate_compare): Canonicalize
28134 operation with embedded compare to
28135 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28136 (set (reg) (operation)].
28138 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28140 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28143 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28144 cost of adding a carry flag for ADC instruction.
28145 [case MINUS]: Ignore the cost of subtracting a carry flag
28146 for SBB instruction.
28148 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28150 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28152 * config/rs6000/bmiintrin.h: New file.
28153 * config/rs6000/bmi2intrin.h: New file.
28154 * config/rs6000/x86intrin.h: New file.
28156 2017-05-12 Jeff Law <law@redhat.com>
28158 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28161 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28163 PR middle-end/80707
28164 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28166 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28168 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28169 early expansion of vector divide builtins.
28170 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28171 builtins identified as having unsigned arguments.
28173 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28175 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28176 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28177 expansion of vector logical operations (and, andc, or, xor,
28180 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28182 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28183 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28185 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28187 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28188 early GIMPLE expansion of vector multiplies.
28190 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28192 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28193 TARGET_HAVE_MOVT conditional.
28194 (movt splitter): Likewise.
28196 2017-05-12 Richard Biener <rguenther@suse.de>
28198 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28199 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28200 Fold all stmts not inplace.
28202 2017-05-12 Richard Biener <rguenther@suse.de>
28204 PR tree-optimization/80713
28205 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28206 inserted_exprs bit for not removed stmts.
28208 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28210 PR middle-end/69921
28211 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28212 parallelized" attribute for parallelized OpenACC kernels.
28213 * omp-offload.c (execute_oacc_device_lower): Use it.
28215 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28216 Set "oacc kernels" attribute.
28217 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28218 parameter. Adjust all users.
28219 (oacc_fn_attrib_kernels_p): Remove function.
28220 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28221 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28222 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28223 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28224 assert "oacc kernels" attribute is set.
28226 2017-05-11 Carl Love <cel@us.ibm.com>
28228 * config/rs6000/rs6000-c: Add support for built-in functions
28229 vector unsigned char vec_popcnt (vector signed char)
28230 vector unsigned char vec_popcnt (vector unsigned char)
28231 vector unsigned short vec_popcnt (vector signed short)
28232 vector unsigned short vec_popcnt (vector unsigned short)
28233 vector unsigned int vec_popcnt (vector signed int)
28234 vector unsigned int vec_popcnt (vector unsigned int)
28235 vector unsigned long long vec_popcnt (vector signed long long)
28236 vector unsigned long long vec_popcnt (vector unsigned long long)
28237 vector signed long long vec_slo (vector signed long long,
28238 vector signed char)
28239 vector signed long long vec_slo (vector signed long long,
28240 vector unsigned char)
28241 vector unsigned long long vec_slo (vector unsigned long long,
28242 vector signed char)
28243 vector unsigned long long vec_slo (vector unsigned long long,
28244 vector unsigned char)
28245 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28246 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28247 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28248 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28249 * doc/extend.texi: Update the built-in documentation file for the
28250 new built-in functions.
28252 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28254 * attribs.h (sorted_attr_string): Move machine independent
28255 functions for target clone support from the i386 port to common
28256 code. Rename ix86_function_versions to common_function_versions.
28257 Rename make_name to make_unique_name.
28258 (common_function_versions): Likewise.
28259 (make_unique_name): Likewise.
28260 (make_dispatcher_decl): Likewise.
28261 (is_function_default_version): Likewise.
28262 * attribs.c (attr_strcmp): Likewise.
28263 (sorted_attr_string): Likewise.
28264 (common_function_versions): Likewise.
28265 (make_unique_name): Likewise.
28266 (make_dispatcher_decl): Likewise.
28267 (is_function_default_version): Likewise.
28268 * config/i386/i386.c (attr_strcmp): Likewise.
28269 (sorted_attr_string): Likewise.
28270 (ix86_function_versions): Likewise.
28271 (make_name): Likewise.
28272 (make_dispatcher_decl): Likewise.
28273 (is_function_default_version): Likewise.
28274 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28276 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28279 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28280 Account for direct move costs for vec_construct of integer
28283 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28286 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28287 (UNSPEC_STX_ATOMIC): Ditto.
28288 (loaddi_via_sse): New insn.
28289 (storedi_via_sse): Ditto.
28290 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28291 Update corresponding peephole2 patterns.
28292 (atomic_storedi_fpu): Ditto.
28294 2017-05-11 Julia Koval <julia.koval@intel.com>
28296 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28297 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28299 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28300 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28301 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28303 2017-05-11 Nathan Sidwell <nathan@acm.org>
28305 * graphite-poly.c: Include dumpfile.h.
28307 * dumpfle.h (dump_function): Declare here ...
28308 * tree-dump.h (dump_function): ... not here.
28309 * dumpfile.c: #include tree-cfg.h.
28310 (dump_function): Move here from ...
28311 * tree-dump.c (dump_function): ... here.
28312 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28313 * graphite-poly.c: Don't include tree-dump.h.
28314 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28315 * print-tree.c: Likewise.
28316 * stor-layout.c: Likewise.
28317 * tree-nested.c: Likewise.
28319 * dumpfile.c (dump_start): Use TDF_FLAGS.
28320 (dump_enable_all): Fix TDF_KIND check thinko.
28322 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28324 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28325 array entries to represent two legal parameterizations of the
28326 overloaded __builtin_cmpb function, as represented by the
28327 P6_OV_BUILTIN_CMPB constant.
28328 (altivec_resolve_overloaded_builtin): Add special case handling
28329 for the __builtin_cmpb function, as represented by the
28330 P6_OV_BUILTIN_CMPB constant.
28331 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28332 (BU_P6_64BIT_2): New macro.
28333 (BU_P6_OVERLOAD_2): New macro
28334 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28335 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28336 (CMPB): Add overload support to represent both 32-bit and 64-bit
28337 compare-bytes function.
28338 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28339 support for TARGET_CMPB.
28340 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28341 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28342 documentation of the __builtin_cmpb overloaded built-in function.
28344 2017-05-11 Richard Biener <rguenther@suse.de>
28346 PR tree-optimization/80705
28347 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28348 bases are not vectorizable.
28350 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28352 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28353 when counting register pressure.
28355 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28357 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28358 (iv_ca_more_deps): Renamed to ...
28359 (iv_ca_compare_deps): ... this.
28360 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28362 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28364 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28366 (determine_group_iv_costs): ... here.
28367 (find_inv_vars_cb): Record inv var if it's not recorded before.
28369 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28371 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28372 (get_shiftadd_cost): Ditto.
28374 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28376 * tree-ssa-address.c: Include header file.
28377 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28379 (add_to_parts): Refactor.
28380 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28381 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28384 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28386 PR tree-optimization/53090
28387 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28389 (extract_cond_operands): Detect condition with IV on both sides
28390 and return COMP_IV_EXPR_2.
28391 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28392 (rewrite_use_compare): Simplify by removing call to function
28393 extract_cond_operands.
28395 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28397 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28398 (extract_cond_operands): Detect condition comparing against non-
28399 invariant bound and return appropriate enum value.
28400 (find_interesting_uses_cond): Update use of extract_cond_operands.
28401 Handle its return value accordingly.
28402 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28404 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28406 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28407 nonlinear iv_use computation in loop invariant sensitive way.
28409 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28411 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28412 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28414 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28416 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28417 (dump_cand): Support iv_cand.inv_exprs.
28418 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28420 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28423 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28425 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28427 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28428 as local function. Include necessary header files.
28429 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28431 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28433 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28435 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28437 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28438 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28439 RSHIFT_EXPR and BIT_NOT_EXPR.
28441 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28443 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28444 (adjust_setup_cost): New parameter supporting round up adjustment.
28445 (struct address_cost_data): Delete.
28446 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28447 (split_address_cost, ptr_difference_cost): Delete.
28448 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28449 (struct ainc_cost_data): New struct.
28450 (get_address_cost_ainc): New function.
28451 (get_address_cost, get_computation_cost): Reimplement.
28452 (determine_group_iv_cost_address): Record inv_expr for all uses of
28454 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28455 (iv_ca_has_deps): Reimplemented to ...
28456 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28458 (iv_ca_extend): Call iv_ca_more_deps.
28460 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28462 * tree-ssa-address.c (struct mem_address): Move to header file.
28463 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28464 * tree-ssa-address.h (struct mem_address): Move from C file.
28465 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28467 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28469 * tree-affine.h (aff_combination_type): New interface.
28470 (aff_combination_zero_p): Remove static.
28471 (aff_combination_const_p): New interface.
28472 (aff_combination_singleton_var_p): New interfaces.
28474 2017-05-11 Richard Biener <rguenther@suse.de>
28476 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28477 Skip unreachable blocks and destinations.
28478 (eliminate): Move stmt removal and fixup ...
28479 (fini_eliminate): ... here. Skip inserted exprs.
28480 (pass_pre::execute): Move fini_pre after fini_eliminate.
28481 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28482 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28483 PRE to get rid of dead code that has invalid SSA form and
28484 split critical edges again.
28486 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28488 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28490 2017-05-11 Richard Biener <rguenther@suse.de>
28492 * passes.c (execute_function_todo): Verify loops if they are
28493 said to be up-to-date.
28494 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28495 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28497 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28500 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28501 handle calling assemble_external ourself.
28504 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28505 modes with zero size. Enhance comment.
28507 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28509 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28510 built-ins for vec_xl and vec_xst with short and char pointer
28513 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28515 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28516 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28517 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28518 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28519 (_mm_maskz_min_round_ss): New intrinsics.
28520 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28521 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28522 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28523 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28524 (__builtin_ia32_minss_mask_round): New builtins.
28525 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28526 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28527 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28529 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28530 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28532 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28535 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28537 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28538 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28539 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28540 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28541 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28542 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28543 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28544 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28545 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28546 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28547 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28548 (__builtin_ia32_mulss_mask_round): New builtins.
28549 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28550 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28551 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28553 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28554 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28556 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28559 2017-05-10 Julia Koval <julia.koval@intel.com>
28561 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28562 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28563 (_mm256_setr_m128i): New intrinsics.
28565 2017-05-10 Julia Koval <julia.koval@intel.com>
28567 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28568 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28569 (_mm_maskz_rcp14_ss): New intrinsics.
28570 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28571 (__builtin_ia32_rcp14ss_mask): New builtins.
28572 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28574 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28576 PR tree-optimization/51513
28577 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28578 (assert_unreachable_fallthru_edge_p): Use it.
28579 (group_case_labels_stmt): Likewise.
28580 * tree-cfg.h: Prototype it.
28581 * stmt.c: Include cfghooks.h and tree-cfg.h.
28582 (emit_case_dispatch_table) <gap_label>: New local variable.
28583 Use it to fill dispatch table gaps.
28584 Test for default_label before updating probabilities.
28585 (expand_case) <default_label>: Remove unneeded initialization.
28586 Test for unreachable default case statement and remove its edge.
28587 Set default_label accordingly.
28588 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28590 2017-05-10 Carl Love <cel@us.ibm.com>
28592 * config/rs6000/rs6000-c: Add support for built-in functions
28593 vector signed char vec_neg (vector signed char)
28594 vector signed short int vec_neg (vector short int)
28595 vector signed int vec_neg (vector signed int)
28596 vector signed long long vec_neg (vector signed long long)
28597 vector float vec_neg (vector float)
28598 vector double vec_neg (vector double)
28599 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28601 * config/rs6000/altivec.h: Add define for vec_neg
28602 * doc/extend.texi: Update the built-in documentation for the
28603 new built-in functions.
28605 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28607 PR tree-optimization/77644
28608 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28610 2017-05-10 Nathan Sidwell <nathan@acm.org>
28612 * dumpfile.h (TDI_lang_all): New.
28613 (TDF_KIND): New. Renumber others
28614 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28616 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28618 (get_dump_file_name): Adjust suffix generation.
28619 (dump_enable_all): Use TDF_KIND.
28620 * doc/invoke.texi (-fdump-lang-all): Document.
28622 * dumpfile.h: Tabify.
28624 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28627 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28628 Move member access before delete.
28630 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28632 * tree-inline.c (expand_call_inline): Split block at stmt
28635 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28638 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28639 are now unused after splitting mov{sf,sd}_hardfloat.
28640 (f32_lr2): Likewise.
28641 (f32_lm): Likewise.
28642 (f32_lm2): Likewise.
28643 (f32_li): Likewise.
28644 (f32_li2): Likewise.
28645 (f32_lv): Likewise.
28646 (f32_sr): Likewise.
28647 (f32_sr2): Likewise.
28648 (f32_sm): Likewise.
28649 (f32_sm2): Likewise.
28650 (f32_si): Likewise.
28651 (f32_si2): Likewise.
28652 (f32_sv): Likewise.
28653 (f32_dm): Likewise.
28654 (f32_vsx): Likewise.
28655 (f32_av): Likewise.
28656 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28657 For movsf, order stores so the VSX stores occur before the GPR
28658 store which encourages the register allocator to use a traditional
28659 FPR instead of a GPR. For movsd, order the stores so that the GPR
28660 store comes before the VSX stores to allow the power6 to work.
28661 This is due to the power6 not having a 32-bit integer store
28662 instruction from a FPR.
28663 (movsf_hardfloat): Likewise.
28664 (movsd_hardfloat): Likewise.
28666 2017-05-09 Martin Sebor <msebor@redhat.com>
28668 PR translation/80280
28669 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28672 PR translation/80280
28673 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28674 data member added in r247778.
28675 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28677 2017-05-09 Nathan Sidwell <nathan@acm.org>
28679 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28681 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28684 2017-05-09 Marek Polacek <polacek@redhat.com>
28686 * doc/invoke.texi: Fix typo.
28688 2017-05-09 Richard Biener <rguenther@suse.de>
28690 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28691 (vrp_val_is_min): Likewise.
28692 (set_value_range_to_value): Likewise.
28693 (set_value_range_to_nonnegative): Likewise.
28694 (gimple_assign_nonzero_p): Likewise.
28695 (gimple_stmt_nonzero_p): Likewise.
28696 (vrp_int_const_binop): Likewise. Remove unreachable case.
28697 (adjust_range_with_scev): Adjust comments.
28698 (compare_range_with_value): Likewise.
28699 (extract_range_from_phi_node): Likewise.
28700 (test_for_singularity): Likewise.
28702 2017-05-09 Richard Biener <rguenther@suse.de>
28704 * tree-vrp.c (get_single_symbol): Add assert that we don't
28705 get overflowed constants as invariant part.
28706 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28707 checks. Use wi::cmp instead of recursing for integer constants.
28708 (compare_values): Just ignore whether we assumed undefined
28709 overflow instead of failing the compare.
28710 (extract_range_for_var_from_comparison_expr): Add comment before the
28711 TREE_NO_WARNING sets.
28712 (test_for_singularity): Likewise.
28713 (extract_range_from_comparison): Do not disable optimization
28714 when we assumed undefined overflow.
28715 (extract_range_basic): Remove init of unused var.
28717 2017-05-09 Richard Biener <rguenther@suse.de>
28719 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28720 (extract_range_from_multiplicative_op_1): Adjust.
28721 (extract_range_from_binary_expr_1): Use int_const_binop.
28723 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28726 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28727 rs6000_store_data_bypass_p in seven define_bypass directives and
28728 in several comments.
28729 * config/rs6000/rs6000-protos.h: Add prototype for
28730 rs6000_store_data_bypass_p function.
28731 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28732 function implements slightly different (rs6000-specific) semantics
28733 than store_data_bypass_p, returning false rather than aborting
28734 with assertion error when arguments do not satisfy the
28735 requirements of store data bypass.
28736 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28737 rs6000_store_data_bypass_p.
28739 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28741 * config/xtensa/xtensa-protos.h
28742 (xtensa_initial_elimination_offset): New declaration.
28743 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28744 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28745 macro definition, add case for FRAME_POINTER_REGNUM when
28746 FRAME_GROWS_DOWNWARD.
28747 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28748 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28749 xtensa_initial_elimination_offset.
28751 2017-05-08 Nathan Sidwell <nathan@acm.org>
28753 * doc/invoke.texi: Alphabetize -fdump options.
28755 2017-05-08 Martin Sebor <msebor@redhat.com>
28757 PR translation/80280
28758 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28760 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28762 * target.def (compute_frame_layout): New optional target hook.
28763 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28764 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28765 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28767 * reload1.c (verify_initial_elim_offsets): Likewise.
28768 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28769 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28770 (arm_get_frame_offsets): Split up into this ...
28771 (arm_compute_frame_layout): ... and this function.
28773 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28775 * config/aarch64/constraints.md (Usa): New constraint.
28776 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28778 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28780 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28781 with_multilib_list after it has been checked.
28783 2017-05-08 Richard Biener <rguenther@suse.de>
28785 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28786 (bitmap_set_subtract_values): Likewise.
28788 2017-05-08 Richard Biener <rguenther@suse.de>
28790 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28791 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28793 (gimple_stmt_nonzero_warnv_p): Rename to ...
28794 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28796 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28797 (extract_range_basic): Adjust, do not disable propagation on
28798 strict overflow sensitive simplification.
28799 (vrp_visit_cond_stmt): Likewise.
28801 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28803 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28804 body size unconditionally.
28806 2017-05-07 Jeff Law <law@redhat.com>
28809 2017-05-06 Jeff Law <law@redhat.com>
28810 PR tree-optimization/78496
28811 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28814 PR tree-optimization/78496
28815 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28816 (simplify_stmt_using_ranges): Call it.
28817 (vrp_dom_walker::before_dom_children): Extract equivalences
28818 from an ASSERT_EXPR with an equality comparison against a
28821 2017-05-06 Jeff Law <law@redhat.com>
28823 PR tree-optimization/78496
28824 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28827 PR tree-optimization/78496
28828 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28829 (simplify_stmt_using_ranges): Call it.
28830 (vrp_dom_walker::before_dom_children): Extract equivalences
28831 from an ASSERT_EXPR with an equality comparison against a
28834 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28836 * lra-constraints.c (lra_copy_reg_equiv): New function.
28837 (split_reg): Use it to copy equivalence information from the
28838 original register to the spill register.
28840 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28842 PR rtl-optimization/75964
28843 * simplify-rtx.c (simplify_const_relational_operation): Remove
28844 invalid handling of comparisons of integer ABS.
28846 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28848 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28849 initialize to zero.
28850 (init_regs): Remove declaration.
28851 (function_arg_advance_32): Initialize error_p as boolean variable.
28853 2017-05-05 Nathan Sidwell <nathan@acm.org>
28855 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28856 lines. Use for (;;).
28858 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28860 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28861 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28862 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28863 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28864 VF=2 that require versioning.
28866 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28868 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28871 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28873 * diagnostic.h (diagnostic_override_option_index): Convert from
28874 macro to inline function.
28876 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28878 * diagnostic.c (last_module_changed_p): New function.
28879 (set_last_module): New function.
28880 (diagnostic_report_current_module): Convert macro usage to
28881 the above functions.
28882 * diagnostic.h (diagnostic_context::last_module): Strengthen
28883 from const line_map * to const line_map_ordinary *.
28884 (diagnostic_last_module_changed): Delete macro.
28885 (diagnostic_set_last_module): Delete macro.
28887 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28889 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28890 with diagnostic_report_diagnostic.
28891 (diagnostic_n_impl_richloc): Likewise.
28892 * diagnostic.h (report_diagnostic): Delete macro.
28893 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28894 with diagnostic_report_diagnostic.
28895 * substring-locations.c (format_warning_va): Likewise.
28897 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28899 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28900 save/restor of format_spec. Move option-printing code to...
28901 (print_option_information): ...this new function, and
28902 reimplement by simply printing to the pretty_printer,
28903 rather than appending to the format string.
28905 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28907 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28908 handling logic into...
28909 (update_effective_level_from_pragmas): ...this new function.
28911 2017-05-04 Andrew Waterman <andrew@sifive.com>
28913 * config/riscv/riscv.opt (mstrict-align): New option.
28914 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28915 (SLOW_UNALIGNED_ACCESS): Define.
28916 (riscv_slow_unaligned_access): Declare.
28917 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28919 (riscv_slow_unaligned_access): New variable.
28920 (rocket_tune_info): Set slow_unaligned_access to true.
28921 (optimize_size_tune_info): Set slow_unaligned_access to false.
28922 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
28923 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
28924 (riscv_option_override): Set riscv_slow_unaligned_access.
28925 * doc/invoke.texi: Add -mstrict-align to RISC-V.
28927 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
28929 * config/riscv/riscv.md: Unify indentation.
28931 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28936 * config/rs6000/rs6000.md (u code attribute): Add FIX and
28938 (extendsi<mode>2): Add support for doing sign extension via
28939 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
28940 don't have ISA 3.0 instructions.
28941 (extendsi<mode>2 splitter): Likewise.
28942 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
28943 generate the normal insns since SImode can now go in vector
28944 registers. Disallow the special UNSPECs needed for previous
28945 machines to hide SImode being used. Add new insns
28946 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
28947 (fix_trunc<mode>si2_stfiwx): Likewise.
28948 (fix_trunc<mode>si2_internal): Likewise.
28949 (fixuns_trunc<mode>si2): Likewise.
28950 (fixuns_trunc<mode>si2_stfiwx): Likewise.
28951 (fctiw<u>z_<mode>_smallint): Likewise.
28952 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
28953 of floating point to 32-bit integer from doing a direct move to
28954 the GPR registers to do a store.
28955 (fctiwz_<mode>): Break long line.
28957 2017-05-05 Bin Cheng <bin.cheng@arm.com>
28959 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
28960 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
28961 (addr_list, addr_offset_valid_p): New.
28962 (split_address_groups): Check offset validity with above function.
28963 (gt-tree-ssa-loop-ivopts.h): Include header file.
28965 2017-05-05 Nathan Sidwell <nathan@acm.org>
28967 * config.gcc (arm*-*-*): Add missing 'fi'.
28969 2017-05-05 Steve Ellcey <sellcey@cavium.com>
28971 * doc/invoke.texi (-fopt-info): Explicitly say order of options
28972 included in -fopt-info does not matter.
28973 * doc/optinfo.texi (-fopt-info): Fix description of default
28974 behavour. Explicitly say order of options included in -fopt-info
28977 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28979 * config.gcc: Allow combinations of aprofile and rmprofile values for
28980 --with-multilib-list.
28981 * config/arm/t-multilib: New file.
28982 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
28983 variables. Remove setting of ISA and floating-point ABI in
28984 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
28985 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
28986 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
28987 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
28989 * config/arm/t-rmprofile: Likewise except for the matches changes.
28990 * doc/install.texi (--with-multilib-list): Document the combination of
28991 aprofile and rmprofile values and warn about pitfalls in doing that.
28993 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28995 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
28996 (movdi_aarch64): Likewise.
28998 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29000 PR tree-optimization/80632
29001 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29003 (build_arrays): Initialize it for virtual phis.
29004 (fix_phi_nodes): Use it for virtual phis.
29006 PR tree-optimization/80558
29007 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29008 [x, y] op z into [x op, y op z] for op & or | if conditions
29011 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29012 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29015 * config/arm/arm.md (use_literal_pool): Remove.
29016 (64-bit immediate split): No longer takes cost into consideration
29017 if arm_disable_literal_pool is enabled.
29018 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29019 used when arm_disable_literal_pool is enabled.
29020 (arm_max_const_double_inline_cost): Remove use of
29021 arm_disable_literal_pool.
29022 (push_minipool_fix): Add assert.
29023 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29024 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29025 (no_literal_pool_sf_immediate): New.
29027 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29029 PR tree-optimization/80613
29030 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29031 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29033 2017-05-05 Richard Biener <rguenther@suse.de>
29035 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29037 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29039 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29040 of this flag from insn conditions due to removal from r247495.
29042 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29044 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29046 (arm_early_store_addr_dep_ptr): Likewise.
29047 * config/arm/aarch-common-protos.h
29048 (arm_early_load_addr_dep_ptr): Add prototype.
29049 (arm_early_store_addr_dep_ptr): Likewise.
29050 * config/arm/cortex-a53.md: Add new bypasses.
29052 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29054 * tree.c (next_type_uid): Change type to unsigned.
29055 (type_hash_canon): Decrement back next_type_uid if
29056 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29057 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29060 2017-05-04 Martin Sebor <msebor@redhat.com>
29062 * builtins.c: Fix a trivial typo in a comment.
29064 PR middle-end/79234
29065 * builtins.c (check_sizes): Adjust to handle reading past the end.
29066 Avoid printing excessive upper bound of ranges. Use %E to print
29067 tree nodes instead of converting them to %wu.
29068 (expand_builtin_memchr): New function.
29069 (compute_dest_size): Rename...
29070 (compute_objsize): ...to this.
29071 (expand_builtin_memcpy): Adjust.
29072 (expand_builtin_mempcpy): Adjust.
29073 (expand_builtin_strcat): Adjust.
29074 (expand_builtin_strcpy): Adjust.
29075 (check_strncat_sizes): Adjust.
29076 (expand_builtin_strncat): Adjust.
29077 (expand_builtin_strncpy): Adjust and simplify.
29078 (expand_builtin_memset): Adjust.
29079 (expand_builtin_bzero): Adjust.
29080 (expand_builtin_memcmp): Adjust.
29081 (expand_builtin): Handle memcmp.
29082 (maybe_emit_chk_warning): Check strncat just once.
29084 2017-05-04 Martin Sebor <msebor@redhat.com>
29086 PR preprocessor/79214
29087 PR middle-end/79222
29088 PR middle-end/79223
29089 * builtins.c (check_sizes): Add inlining context and issue
29090 warnings even when -Wno-system-headers is set.
29091 (check_strncat_sizes): Same.
29092 (expand_builtin_strncat): Same.
29093 (expand_builtin_memmove): New function.
29094 (expand_builtin_stpncpy): Same.
29095 (expand_builtin): Handle memmove and stpncpy.
29097 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29099 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29100 which is not used any more.
29102 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29104 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29106 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29108 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29109 (cortexa53_tunings): Likewise.
29110 (cortexa57_tunings): Likewise.
29111 (cortexa72_tunings): Likewise.
29112 (cortexa73_tunings): Likewise.
29114 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29116 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29117 Set loop alignment to 8.
29119 2017-05-04 Martin Sebor <msebor@redhat.com>
29121 PR translation/80280
29122 * builtins.c (expand_builtin_object_size): Add missing quoting to
29123 %D and like directives.
29124 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29125 (hsa_type_for_tree_type): Same.
29126 (verify_function_arguments): Same.
29127 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29128 * varasm.c (get_section): Same.
29131 2017-05-04 Martin Sebor <msebor@redhat.com>
29133 PR translation/80280
29134 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29136 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29138 * config/aarch64/aarch64.c (generic_addrcost_table):
29139 Change HI/TI mode setting.
29141 2017-05-04 Martin Jambor <mjambor@suse.cz>
29143 PR tree-optimization/80622
29144 * tree-sra.c (comes_initialized_p): New function.
29145 (build_accesses_from_assign): Only set write lazily when
29146 comes_initialized_p is false.
29147 (analyze_access_subtree): Use comes_initialized_p.
29148 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29149 instead of testing for PARM_DECL.
29151 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29153 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29154 constraint on operand 0 to allow more general addressing modes.
29155 Adjust output template.
29156 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29158 * config/aarch64/aarch64-protos.h
29159 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29160 * config/aarch64/constraints.md (Dp): New address constraint.
29161 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29164 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29166 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29167 update use of estimate_ipcp_clone_size_and_time.
29168 (estimate_local_effects): Update use of
29169 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29170 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29171 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29172 Return nonspecialized time.
29174 2017-05-04 Richard Biener <rguenther@suse.de>
29176 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29177 for the last VUSE which def dominates the PHI. Directly call
29179 (get_continuation_for_phi_1): Remove.
29181 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29183 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29184 to explain the use of truncating division. Cap the number of
29185 iterations to the maximum given by nb_iterations_upper_bound,
29188 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29190 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29191 * configure: Regenerate.
29192 * config.in: Regenerate.
29193 * config/i386/driver-mingw32.c: new file.
29194 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29195 * config.host: Link driver-mingw32.o on MinGW host.
29196 * doc/install.texi: Document new --enable-mingw-wildcard configure
29199 2017-05-04 Marek Polacek <polacek@redhat.com>
29201 PR tree-optimization/80612
29202 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29204 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29205 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29207 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29208 (movt splitter): Likewise.
29209 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29210 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29211 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29212 block for Thumb-1 with MOVT.
29213 (thumb2_legitimate_address_p): Move code block ...
29214 (can_avoid_literal_pool_for_label_p): ... into this new function.
29215 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29217 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29218 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29219 "M-profile targets with the MOVT instruction".
29221 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29223 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29224 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29225 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29227 2017-05-04 Martin Liska <mliska@suse.cz>
29229 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29230 variable cond_code.
29232 2017-05-04 Richard Biener <rguenther@suse.de>
29234 * tree.c (array_at_struct_end_p): Handle arrays at struct
29235 end with flexarrays more conservatively. Refactor and treat
29236 arrays of arrays or aggregates more strict. Fix
29237 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29238 * tree.c (array_at_struct_end_p): Adjust prototype.
29239 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29240 * gimple-fold.c (get_range_strlen): Likewise.
29241 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29243 2017-05-04 Richard Biener <rguenther@suse.de>
29245 PR tree-optimization/31130
29246 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29248 (supports_overflow_infinity): Likewise.
29249 (is_negative_overflow_infinity): Likewise.
29250 (is_positive_overflow_infinity): Likewise.
29251 (is_overflow_infinity): Likewise.
29252 (stmt_overflow_infinity): Likewise.
29253 (overflow_infinity_range_p): Likewise.
29254 (usable_range_p): Remove as always returning true.
29255 (make_overflow_infinity): Remove.
29256 (negative_overflow_infinity): Likewise.
29257 (positive_overflow_infinity): Likewise.
29258 (avoid_overflow_infinity): Likewise.
29259 (set_value_range): Adjust accordingly.
29260 (set_value_range_to_nonnegative): Likewise, remove now unused
29261 overflow_infinity arg.
29262 (vrp_operand_equal_p): Adjust.
29263 (update_value_range): Likewise.
29264 (range_int_cst_singleton_p): Likewise.
29265 (operand_less_p): Likewise.
29266 (compare_values_warnv): Likewise.
29267 (extract_range_for_var_from_comparison_expr): Likewise.
29268 (vrp_int_const_binop): Likewise.
29269 (zero_nonzero_bits_from_vr): Likewise.
29270 (extract_range_from_multiplicative_op_1): Likewise.
29271 (extract_range_from_binary_expr_1): Likewise.
29272 (extract_range_from_unary_expr): Likewise.
29273 (extract_range_from_comparison): Likewise.
29274 (extract_range_basic): Likewise.
29275 (adjust_range_with_scev): Likewise.
29276 (compare_ranges): Likewise.
29277 (compare_range_with_value): Likewise.
29278 (dump_value_range): Likewise.
29279 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29281 (simplify_cond_using_ranges): Adjust.
29283 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29285 * brig-builtins.def: Added a builtin for class_f64.
29286 * builtin-types.def: Added a builtin type needed by class_f64.
29288 2017-05-03 Jason Merrill <jason@redhat.com>
29290 * timevar.def: Add TV_CONSTEXPR.
29292 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29294 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29296 2017-05-03 Martin Jambor <mjambor@suse.cz>
29298 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29299 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29300 * ipa-cp.c (ipcp_propagate_stage): Do not call
29301 ipa_update_after_lto_read.
29302 * ipa-inline.c (ipa_inline): Likewise.
29304 2017-05-03 Martin Jambor <mjambor@suse.cz>
29306 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29307 tag. Added a default constructor and a destructor.
29308 (ipa_edge_args_sum_t): New class;
29309 (ipa_edge_args_sum): Declare.
29310 (ipa_edge_args_vector): Remove declaration.
29311 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29312 (ipa_free_edge_args_substructures): Remove declaration.
29313 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29314 (ipa_edge_args_info_available_for_edge_p): Likewise.
29315 * ipa-prop.c (ipa_edge_args_vector): Removed.
29316 (edge_removal_hook_holder): Likewise.
29317 (edge_duplication_hook_holder): Likewise.
29318 (ipa_edge_args_sum): New variable.
29319 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29320 ipa_edge_args_vector.
29321 (ipa_free_edge_args_substructures): Likewise.
29322 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29323 ipa_edge_args_vector.
29324 (ipa_edge_removal_hook): Turned into method
29325 ipa_edge_args_sum_t::remove.
29326 (ipa_edge_duplication_hook): Turned into method
29327 ipa_edge_args_sum_t::duplicate.
29328 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29329 registering edge hooks.
29330 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29331 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29332 ipa_edge_args_sum instead of ipa_edge_args_vector.
29333 * ipa-profile.c (ipa_profile): Likewise.
29335 2017-05-03 Martin Jambor <mjambor@suse.cz>
29337 * symbol-summary.h (function_summary): New method exists.
29338 (function_summary::symtab_removal): Deallocate through release.
29339 (call_summary): New class.
29340 (gt_ggc_mx): New overload.
29341 (gt_pch_nx): Likewise.
29342 (gt_pch_nx): Likewise.
29344 2017-05-03 Jeff Law <law@redhat.com>
29346 PR tree-optimization/78496
29347 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29348 from simplify_cond_using_ranges. Split off code to walk
29349 backwards through casts into ...
29350 (simplify_cond_using_ranges_2): New function.
29351 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29352 (execute_vrp): After identifying jump threads, call
29353 simplify_cond_using_ranges_2.
29355 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29358 * ipa-inline.h (inline_summary): Add ctor.
29359 (create_ggc): Do not use ggc_cleared_alloc.
29361 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29362 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29364 * gcc.c (handle_braces): Support escaping in switch matching
29366 * doc/invoke.texi (Spec Files): Document it.
29367 Remove superfluous @code markup in items.
29369 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29371 * diagnostic-show-locus.c (struct column_range): New struct.
29372 (get_affected_columns): New function.
29373 (get_printed_columns): New function.
29374 (struct correction): New struct.
29375 (correction::ensure_capacity): New function.
29376 (correction::ensure_terminated): New function.
29377 (struct line_corrections): New struct.
29378 (line_corrections::~line_corrections): New dtor.
29379 (line_corrections::add_hint): New function.
29380 (layout::print_trailing_fixits): Reimplement in terms of the new
29382 (selftest::test_overlapped_fixit_printing): New function.
29383 (selftest::diagnostic_show_locus_c_tests): Call it.
29385 2017-05-03 Nathan Sidwell <nathan@acm.org>
29387 Canonicalize canonical type hashing
29388 * tree.h (type_hash_canon_hash): Declare.
29389 * tree.c (type_hash_list, attribute_hash_list): Move into
29390 type_hash_canon_hash.
29391 (build_type_attribute_qual_variant): Break out hash code calc into
29392 type_hash_canon_hash.
29393 (type_hash_canon_hash): New. Generic type hash computation.
29394 (build_range_type_1, build_array_type_1, build_function_type,
29395 build_method_type_directly, build_offset_type, build_complex_type,
29396 make_vector_type): Call it.
29398 2017-05-03 Richard Biener <rguenther@suse.de>
29400 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29401 When all DRs have unknown misaligned do not always peel
29402 when there is a store but apply the same costing model as if
29403 there were only loads.
29405 2017-05-03 Richard Biener <rguenther@suse.de>
29408 PR tree-optimization/80492
29409 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29410 compare_base_decls returning dont-know properly.
29412 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29414 * config/arm/iterators.md (CCSI): New mode iterator.
29415 (arch): New mode attribute.
29416 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29417 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29418 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29419 code iterator for success result mode.
29420 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29421 the corresponding new insn generators.
29423 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29426 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29427 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29429 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29431 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29432 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29433 (DDR_A): Wrap DDR argument in brackets.
29434 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29435 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29436 (DDR_REVERSED_P): Likewise.
29438 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29440 PR tree-optimization/79472
29441 * tree-switch-conversion.c (struct switch_conv_info): Add
29442 contiguous_range and default_case_nonstandard fields.
29443 (collect_switch_conv_info): Compute contiguous_range and
29444 default_case_nonstandard fields, don't clear final_bb if
29445 contiguous_range and only the default case doesn't have the required
29447 (check_all_empty_except_final): Set default_case_nonstandard instead
29448 of failing if contiguous_range and the default case doesn't have empty
29450 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29451 and only the default case doesn't have the required constants. Skip
29453 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29454 if default_case_nonstandard.
29455 (build_constructors): Build constant 1 just once. Assert that default
29456 values aren't inserted in between cases if contiguous_range. Skip
29458 (build_arrays): Skip virtual phis.
29459 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29460 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29461 Handle virtual phis.
29462 (gen_inbound_check): Handle default_case_nonstandard case.
29463 (process_switch): Adjust check_final_bb caller. Call
29464 gather_default_values with the first non-default case instead of
29465 default case if default_case_nonstandard.
29467 2017-05-02 Nathan Sidwell <nathan@acm.org>
29469 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29470 check. Fix formatting.
29472 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29474 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29475 errors when comparing specialized and unspecialized times.
29477 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29479 * diagnostic-show-locus.c
29480 (layout::should_print_annotation_line_p): Make private.
29481 (layout::print_annotation_line): Make private.
29482 (layout::annotation_line_showed_range_p): Make private.
29483 (layout::show_ruler): Make private.
29484 (layout::print_source_line): Make private. Pass in line and
29485 line_width, rather than calling location_get_source_line. Drop
29487 (layout::print_leading_fixits): New method.
29488 (layout::print_any_fixits): Rename to...
29489 (layout::print_trailing_fixits): ...this, and make private.
29490 Don't print newline fixits.
29491 (diagnostic_show_locus): Move logic for printing one row into...
29492 (layout::print_line): ...this new function. Move the
29493 location_get_source_line call and error-handling from
29494 print_source_line to here. Call print_leading_fixits, and rename
29495 print_any_fixits to print_trailing_fixits.
29496 (selftest::test_fixit_insert_containing_newline): Update now that
29497 newlines are partially supported.
29498 (selftest::test_fixit_insert_containing_newline_2): New test.
29499 (selftest::test_fixit_replace_containing_newline): Update comments.
29500 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29501 * edit-context.c (class added_line): New class.
29502 (class edited_line): Describe newline handling in comment.
29503 (edited_line::actually_edited_p): New method.
29504 (edited_line::print_content): Delete redundant decl.
29505 (edited_line::m_predecessors): New field.
29506 (edited_file::print_content): Call edited_line::print_content.
29507 (edited_file::print_diff): Update to support newlines.
29508 (edited_file::print_diff_hunk): Likewise.
29509 (edited_file::print_run_of_changed_lines): New function.
29510 (edited_file::print_diff_line): Convert to...
29511 (print_diff_line): ...this.
29512 (edited_file::get_effective_line_count): New function.
29513 (edited_line::edited_line): Initialize new field m_predecessors.
29514 (edited_line::~edited_line): Clean up m_predecessors.
29515 (edited_line::apply_fixit): Handle newlines.
29516 (edited_line::get_effective_line_count): New function.
29517 (edited_line::print_content): New function.
29518 (edited_line::print_diff_lines): New function.
29519 (selftest::test_applying_fixits_insert_containing_newline): New
29521 (selftest::test_applying_fixits_replace_containing_newline): New
29523 (selftest::insert_line): New function.
29524 (selftest::test_applying_fixits_multiple_lines): Add example of
29526 (selftest::edit_context_c_tests): Call the new tests.
29528 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29530 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29531 parameter cand. Update dump information.
29532 (get_computation_cost): Update uses.
29534 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29536 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29537 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29538 (get_computation_at, rewrite_use_address): Update use of
29539 get_computation_aff.
29541 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29543 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29544 (get_computation): Delete.
29545 (get_computation_cost): Implement like get_computation_cost_at.
29546 Use get_computation_at.
29547 (get_computation_cost_at): Delete.
29548 (rewrite_use_nonlinear_expr): Use get_computation_at.
29549 (rewrite_use_compare, remove_unused_ivs): Ditto.
29551 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29553 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29555 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29557 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29558 (ivopts_global_cost_for_size): Rename parameter and update uses.
29559 (iv_ca_recount_cost): Update uses.
29560 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29561 candidates seperately in n_invs and n_cands.
29562 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29564 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29566 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29567 (find_inv_vars_cb): New.
29568 (find_depends): Renamed to ...
29569 (find_inv_vars): ... this.
29570 (add_candidate_1, force_var_cost): Call find_inv_vars.
29571 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29573 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29575 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29576 inv_vars. Add inv_exprs.
29577 (struct iv_cand): Rename depends_on to inv_vars.
29578 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29579 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29580 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29581 (dump_cand): Dump inv_vars.
29582 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29583 (record_invariant, find_depends, add_candidate_1): Ditto.
29584 (set_group_iv_cost, force_var_cost): Ditto.
29585 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29586 (get_computation_cost_at, get_computation_cost): Ditto.
29587 (determine_group_iv_cost_generic): Ditto.
29588 (determine_group_iv_cost_address): Ditto.
29589 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29590 (determine_group_iv_costs): Ditto.
29591 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29592 (iv_ca_set_remove_invariants): Renamed to ...
29593 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29594 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29595 (iv_ca_set_add_invariants): Renamed to ...
29596 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29597 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29598 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29599 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29600 (create_new_ivs): Remove useless dump.
29602 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29604 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29606 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29607 (iv_ca_set_no_cp, create_new_iv): Ditto.
29609 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29611 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29613 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29615 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29616 function tree_check2.
29618 2017-05-02 Martin Liska <mliska@suse.cz>
29620 * doc/gcov.texi: Add missing preposition.
29621 * gcov.c (function_info::function_info): Properly fill up
29622 all member variables.
29624 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29626 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29628 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29630 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29632 2017-05-02 Martin Liska <mliska@suse.cz>
29635 * lto-streamer-in.c (lto_read_tree_1): Remove
29636 LTO_STREAMER_DEBUG.
29637 * lto-streamer.c (struct tree_hash_entry): Likewise.
29638 (struct tree_entry_hasher): Likewise.
29639 (tree_entry_hasher::hash): Likewise.
29640 (tree_entry_hasher::equal): Likewise.
29641 (lto_streamer_init): Likewise.
29642 (lto_orig_address_map): Likewise.
29643 (lto_orig_address_get): Likewise.
29644 (lto_orig_address_remove): Likewise.
29645 * lto-streamer.h: Likewise.
29646 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29647 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29649 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29651 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29652 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29653 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29654 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29655 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29656 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29657 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29658 (mm_maskz_sub_ss): New intrinsics.
29659 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29660 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29661 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29662 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29663 (__builtin_ia32_subss_mask_round): New builtins.
29664 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29665 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29666 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29668 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29669 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29671 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29674 2017-05-02 Martin Jambor <mjambor@suse.cz>
29676 PR tree-optimization/78687
29677 * tree-sra.c (access): New field parent.
29678 (process_subtree_disqualification): New function.
29679 (disqualify_candidate): Call it.
29680 (build_accesses_from_assign): Reset write flag if creating an
29682 (build_access_subtree): Fill in parent field and also prpagate
29683 down grp_write flag.
29684 (create_artificial_child_access): New parameter set_grp_write, set
29685 grp_write to its value.
29686 (propagate_subaccesses_across_link): Also propagate grp_write flag
29688 (propagate_all_subaccesses): Push the closest parent back to work
29689 queue if add_access_to_work_queue returned true.
29691 2017-05-02 Richard Biener <rguenther@suse.de>
29693 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29694 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29695 -fstrict-overflow documentation.
29696 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29697 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29698 flag_strict_overflow.
29699 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29700 * lto-opts.c (lto_write_options): Do not stream it.
29701 * lto-wrapper.c (merge_and_complain): Do not handle it.
29702 * opts.c (default_options_table): Do not set -fstrict-overflow.
29703 (finish_options): Likewise do not clear it when sanitizing.
29704 * simplify-rtx.c (simplify_const_relational_operation): Do not
29705 test flag_strict_overflow.
29707 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29709 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29710 using enabled attribute.
29711 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29712 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29713 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29714 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29715 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29716 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29717 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29718 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29719 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29720 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29722 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29724 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29726 2017-05-02 Richard Biener <rguenther@suse.de>
29728 PR tree-optimization/80591
29730 2017-04-10 Richard Biener <rguenther@suse.de>
29732 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29735 2017-05-02 Richard Biener <rguenther@suse.de>
29737 PR tree-optimization/80549
29738 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29739 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29740 headers if they do not have a preheader.
29742 2017-05-02 Martin Liska <mliska@suse.cz>
29745 * common.opt: Fix typo.
29746 * doc/invoke.texi: Likewise.
29748 2017-05-01 Jan Beulich <jbeulich@suse.com>
29750 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29751 swapping, add (x,x,m,x,n) alternative.
29753 2017-05-01 Nathan Sidwell <nathan@acm.org>
29755 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29756 unnecessary unadjusted_alignment check.
29758 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29761 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29763 * gimplify.c (gimplify_cilk_detach): New function.
29764 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29765 * tree-core.h: Document EXPR_CILK_SPAWN.
29766 * tree.h (EXPR_CILK_SPAWN): Define.
29768 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29770 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29771 to use new fixit_hint representation, using the "replace" logic.
29772 (get_line_span_for_fixit_hint): Likewise.
29773 (layout::print_any_fixits): Likewise.
29774 (selftest::test_one_liner_many_fixits): Rename to...
29775 (selftest::test_one_liner_many_fixits_1): ...this, and update
29776 comment and expected output to reflect that the multiple fix-it
29777 hints are now consolidated into one insertion.
29778 (selftest::test_one_liner_many_fixits_2): New test.
29779 (selftest::test_diagnostic_show_locus_one_liner): Update for
29781 (selftest::test_fixit_consolidation): Update for fix-it API
29783 * diagnostic.c (print_parseable_fixits): Likewise.
29784 * edit-context.c (edited_line::m_line_events): Convert from
29785 auto_vec <line_event *> to auto_vec <line_event>.
29786 (class line_event): Convert from abstract base class to a concrete
29787 class, taking over the role of replace_event.
29788 (class insert_event): Delete.
29789 (class replace_event): Rename to class line_event. Convert to
29791 (edit_context::add_fixits): Reimplement.
29792 (edit_context::apply_insert): Delete.
29793 (edit_context::apply_replace): Rename to...
29794 (edit_context::apply_fixit): ...this. Convert to half-open range.
29795 (edited_file::apply_insert): Delete.
29796 (edited_file::apply_replace): Rename to...
29797 (edited_file::apply_fixit): ...this.
29798 (edited_line::~edited_line): Drop deletion of events.
29799 (edited_line::apply_insert): Delete.
29800 (edited_line::apply_replace): Rename to...
29801 (edited_line::apply_fixit): ...this. Convert to half-open range.
29802 Update for change to type of m_line_events.
29803 * edit-context.h (edit_context::apply_insert): Delete.
29804 (edit_context::apply_replace): Rename to...
29805 (edit_context::apply_fixit): ...this.
29807 2017-05-01 Martin Sebor <msebor@redhat.com>
29809 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29812 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29815 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29816 __get_cpuid_max returns 0.
29817 (__get_cpuid_count): Ditto.
29819 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29821 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29822 replacement expression is another instance of one of its arguments.
29824 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29827 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29828 check for stack push/pop autoinc.
29829 * config/i386/i386.c (ix86_agi_dependent): Return false
29830 if the only reason why modified_in_p returned true is that
29831 addr is SP based and set_insn is a push or pop.
29833 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29835 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29838 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29841 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29842 (account_size_time): Use two predicates - exec_pred and
29844 (evaluate_conditions_for_known_args): Compute both clause and
29846 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29847 (inline_summary_t::duplicate): Update.
29848 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29850 (compute_inline_parameters): Likewise.
29851 (estimate_edge_size_and_time): Update caluclation of time.
29852 (estimate_node_size_and_time): Compute both time and nonspecialized
29854 (estimate_ipcp_clone_size_and_time): Update.
29855 (inline_merge_summary): Update.
29856 (do_estimate_edge_time): Update.
29857 (do_estimate_edge_size): Update.
29858 (do_estimate_edge_hints): Update.
29859 (inline_read_section, inline_write_summary): Stream both new predicates.
29860 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29862 (compute_inlined_call_time): Cleanup.
29863 (big_speedup_p): Update.
29864 (edge_badness): Update.
29865 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29866 (size_time_entry): Replace predicate by exec_predicate and
29867 nonconst_predicate.
29868 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29869 (estimate_edge_time): Return also nonspec_time.
29870 (reset_edge_growth_cache): Update.
29872 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29874 PR rtl-optimization/80491
29875 * ifcvt.c (noce_process_if_block): When looking for x setter
29876 with missing else_bb, don't check only the insn right before
29877 cond_earliest, but look for the last insn that x is modified in
29878 within the same bb.
29880 PR rtl-optimization/80491
29881 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29883 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29885 PR tree-optimization/80487
29886 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29888 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29890 PR tree-optimization/79697
29891 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29892 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29893 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29895 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29896 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29898 2017-04-28 Martin Sebor <msebor@redhat.com>
29900 PR tree-optimization/80523
29901 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29902 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29904 (maybe_warn, format_directive, parse_directive): Use new functions.
29905 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29907 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29909 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29911 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29913 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29914 target_header_dir): Set correctly.
29915 * configure: Regenerated.
29916 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29917 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29918 instead of SYSTEM_HEADER_DIR.
29920 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
29922 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
29923 (estimate_local_effects): Likewise.
29924 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
29925 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
29926 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
29927 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
29928 do_estimate_edge_time, estimate_edge_time): Likewise.
29929 * ipa-inline-analysis.c (estimate_node_size_and_time,
29930 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
29931 (estimate_time_after_inlining): Remove.
29933 2017-04-28 Martin Liska <mliska@suse.cz>
29935 * doc/gcov.texi: Enhance documentation of gcov.
29937 2017-04-28 Martin Liska <mliska@suse.cz>
29939 * doc/gcov.texi: Sort options in alphabetic order.
29940 * doc/gcov-dump.texi: Likewise.
29941 * doc/gcov-tool.texi: Likewise.
29942 * gcov.c (print_usage): Likewise.
29943 * gcov-dump.c (print_usage): Likewise.
29944 * gcov-tool.c (print_merge_usage_message): Likewise.
29945 (print_rewrite_usage_message): Likewise.
29946 (print_overlap_usage_message): Likewise.
29948 2017-04-28 Martin Liska <mliska@suse.cz>
29950 PR gcov-profile/53915
29951 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
29953 2017-04-28 Martin Liska <mliska@suse.cz>
29955 PR gcov-profile/79891
29956 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
29957 is marked by compiler as living on a line.
29958 (get_cycles_count): Remove usage of the union.
29959 (output_intermediate_file): Likewise.
29960 (find_source): Fix GNU coding style.
29961 (accumulate_line_counts): Remove old non-all block mode.
29962 (output_lines): Remove usage of the union.
29963 * profile.c (output_location): Include all BBs, even if
29964 belonging to a same line (and file) as a previous BB.
29966 2017-04-28 Martin Liska <mliska@suse.cz>
29968 * gcov.c (process_args): Handle new argument 'w'.
29969 (read_graph_file): Assign ID to BBs.
29970 (output_branch_count): Display BB # if verbose flag is set.
29971 (output_lines): Likewise for arcs.
29972 (print_usage): Add '--verbose' option help.
29973 * doc/gcov.texi: Document --verbose (-w) option.
29975 2017-04-28 Martin Liska <mliska@suse.cz>
29977 * gcov.c (struct block_location_info): New struct.
29978 (process_file): Fill up the new structure.
29979 (read_graph_file): Replace usage of encoding by the newly added
29981 (add_line_counts): Likewise.
29982 (accumulate_line_counts): Remove usage of the union.
29983 (function_info::function_info): New function.
29984 (function_info::~function_info): Likewise.
29985 (process_file): Call delete instead of release_function.
29986 (release_function): Release the function.
29987 (release_structures): Call delete instead of release_function.
29988 (solve_flow_graph): Replace usage of num_blocks.
29989 (find_exception_blocks): Likewise.
29990 (output_lines): Fix GNU coding style.
29992 2017-04-28 Martin Liska <mliska@suse.cz>
29995 * coverage.c (coverage_remove_note_file): New function.
29996 * coverage.h: Declare the function.
29997 * toplev.c (finalize): Clean if an error has been seen.
29999 2017-04-28 Martin Liska <mliska@suse.cz>
30001 PR gcov-profile/80031
30002 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30003 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30004 * gcov.c (read_graph_file): Read just number of blocks.
30005 * profile.c (branch_prob): Do not stream 0 flags per a basic
30008 2017-04-28 Martin Liska <mliska@suse.cz>
30010 * gcov-dump.c (tag_*): Add new argument to declarations.
30011 (dump_gcov_file): Likewise.
30012 (tag_blocks): Add and use new argument depth.
30013 (tag_arcs): Likewise.
30014 (tag_lines): Likewise.
30015 (tag_counters): Likewise.
30016 (tag_summary): Likewise.
30017 (dump_working_sets): Use depth to do a proper indentation.
30019 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30022 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30023 * symtab.c (symtab_node::debug_symtab): Move definition here.
30025 2017-04-28 Richard Biener <rguenther@suse.de>
30027 * lto-streamer.h (LTO_major_version): Bump to 7.
30029 2017-04-28 Richard Biener <rguenther@suse.de>
30031 * tree-vrp.c (assert_info): New struct.
30032 (add_assert_info): New helper.
30033 (register_edge_assert_for_2): Refactor to add asserts to a vector
30035 (register_edge_assert_for_1): Likewise.
30036 (register_edge_assert_for): Likewise.
30037 (finish_register_edge_assert_for): New helper actually registering
30038 asserts where live on edge.
30039 (find_conditional_asserts): Adjust.
30040 (find_switch_asserts): Likewise.
30041 (evrp_dom_walker::try_find_new_range): Generalize.
30042 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30044 2017-04-27 Marek Polacek <polacek@redhat.com>
30047 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30048 arg10 and arg11 to itype.
30050 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30052 * doc/extend.texi (Object Size Checking): Improve grammar.
30054 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30057 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30058 that the logic for permitting reciprocal estimates matches that
30061 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30064 * tree.c (type_cache_hasher::equal): Only compare
30065 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30066 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30067 non-aggregate element types.
30068 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30069 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30071 2017-04-27 Richard Biener <rguenther@suse.de>
30073 PR middle-end/80533
30074 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30075 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30076 keeping a reference to a trailing array.
30078 2017-04-27 Richard Biener <rguenther@suse.de>
30080 PR middle-end/80539
30081 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30082 being in loop-closed SSA form conservatively.
30083 (chrec_fold_multiply_poly_poly): Likewise.
30085 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30087 PR middle-end/79665
30088 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30089 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30091 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30094 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30095 (aarch64_function_arg_alignment): Return unsigned int again, but still
30096 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30097 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30098 Don't emit -Wpsabi note.
30099 (aarch64_function_arg_boundary): Likewise.
30100 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30103 2017-04-26 Nathan Sidwell <nathan@acm.org>
30105 * tree.h (crc32_unsigned_n): Declare.
30106 (crc32_unsigned, crc32_unsigned): Make inline.
30107 * tree.c (crc32_unsigned_bits): Replace with ...
30108 (crc32_unsigned_n): ... this.
30109 (crc32_unsigned, crc32_byte): Remove.
30110 (crc32_string): Remove unnecessary braces.
30112 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30114 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30115 * ipa-inline-analysis.c (MAX_TIME): Remove.
30116 (account_size_time): Use sreal for time.
30117 (dump_inline_summary): Update.
30118 (estimate_function_body_sizes): Update.
30119 (estimate_edge_size_and_time): Update.
30120 (estimate_calls_size_and_time): Update.
30121 (estimate_node_size_and_time): Update.
30122 (inline_merge_summary): Update.
30123 (inline_update_overall_summary): Update.
30124 (estimate_time_after_inlining): Update.
30125 (inline_read_section): Update.
30126 (inline_write_summary): Update.
30127 * ipa-inline.c (compute_uninlined_call_time): Update.
30128 (compute_inlined_call_time): Update.
30129 (recursive_inlining): Update.
30130 (inline_small_functions): Update.
30131 (dump_overall_stats): Update.
30132 * ipa-inline.h: Include sreal.h.
30133 (size_time_entry): Turn time to sreal.
30134 (inline_summary): Turn self_time nad time to sreal.
30136 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30138 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30140 (sreal::stream_out, sreal::stream_in): New.
30141 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30143 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30145 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30148 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30151 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30153 (dimode_scalar_chain::compute_convert_gain): Ditto.
30154 (dimode_scalar_chain::make_vector_copies): Ditto.
30155 (dimode_scalar_chain::convert_reg): Ditto.
30156 (dimode_scalar_chain::convert_insn): Ditto.
30157 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30158 (VI248_AVX512BW_1): New mode iterator.
30159 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30160 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30163 2017-04-25 Martin Sebor <msebor@redhat.com>
30165 PR tree-optimization/80497
30166 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30167 constants are representable in HOST_WIDE_INT.
30168 (parse_directive): Ditto.
30170 2017-04-25 Martin Sebor <msebor@redhat.com>
30173 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30174 (new_zero_array): Adjust signature.
30175 (dom_info::dom_init): Used unsigned rather that size_t.
30176 (dom_info::dom_info): Same.
30178 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30179 Jakub Jelinek <jakub@redhat.com>
30182 * config/arm/arm.c: Include gimple.h.
30183 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30184 returns negative, increment ncrn only if it returned positive.
30185 (arm_needs_doubleword_align): Return int instead of bool,
30186 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30187 members, but if there is any such non-FIELD_DECL
30188 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30189 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30190 returns negative, increment nregs only if it returned positive.
30191 (arm_setup_incoming_varargs): Likewise.
30192 (arm_function_arg_boundary): Emit -Wpsabi note if
30193 arm_needs_doubleword_align returns negative, return
30194 DOUBLEWORD_ALIGNMENT only if it returned positive.
30196 2017-04-25 Marek Polacek <polacek@redhat.com>
30199 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30200 first argument to type.
30202 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30205 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30206 type checks to test for compatibility instead of equality.
30208 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30209 Jakub Jelinek <jakub@redhat.com>
30212 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30214 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30215 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30216 the alignment computation, but return their maximum in warn_alignment.
30217 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30218 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30220 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30221 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30224 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30226 * config/arc/simdext.md (dmpyh): Fix typo.
30228 2017-04-25 Richard Biener <rguenther@suse.de>
30230 PR tree-optimization/80492
30231 * alias.c (compare_base_decls): Handle registers with asm
30232 specification conservatively.
30233 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30234 compare_base_decls returning dont-know properly.
30236 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30238 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30239 (legitimate_offset_address_p): New function.
30240 (arc_legitimate_address_p): Use above function.
30242 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30244 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30246 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30248 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30249 ACCH registers whenever they are available.
30251 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30253 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30254 double regs fix when not used.
30256 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30258 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30260 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30261 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30263 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30265 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30266 when emitting short ADD instructions.
30268 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30270 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30272 (cmpsi_cc_c_insn): Likewise.
30273 (cbranchsi4_scratch): Compute proper instruction length using
30274 compact_hreg_operand.
30275 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30277 2017-04-25 Richard Biener <rguenther@suse.de>
30279 PR middle-end/80509
30280 * passes.c (pass_manager::pass_manager): Initialize
30281 m_name_to_pass_map.
30283 2017-04-25 Richard Biener <rguenther@suse.de>
30285 PR tree-optimization/79201
30286 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30288 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30291 * config/s390/vector.md: Split MEM->GPR vector moves for
30292 non-s_operand addresses.
30294 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30297 * config/s390/predicates.md (reload_const_wide_int_operand): New
30299 * config/s390/s390.md ("movti"): Remove d/P alternative.
30300 ("movti_bigconst"): New pattern definition.
30302 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30305 * s390-protos.h (s390_expand_cs_hqi): Removed.
30306 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30307 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30308 modes as well as CCZ1mode and CCZmode.
30309 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30310 signature of s390_emit_compare_and_swap.
30311 (s390_expand_cs_hqi): Likewise, make static.
30312 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30313 compare-and-swap, in some cases.
30314 (s390_expand_cs): Wrapper function.
30315 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30317 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30318 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30319 patterns for small and large integers. Forbid symref memory operands.
30320 Move expander to s390.c. Require cc register.
30321 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30322 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30323 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30324 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30325 symref memory operands. Remove CC mode and call s390_match_ccmode
30327 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30329 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30331 * config/s390/s390.md (define_peephole2): New peephole to help
30332 combining the load-and-test pattern with volatile memory.
30334 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30336 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30337 with CCZmode for TARGET_Z196.
30339 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30341 PR rtl-optimization/80501
30342 * combine.c (make_compound_operation_int): Set subreg_code to SET
30343 even for AND with mask of the sign bit of mode.
30345 PR rtl-optimization/80500
30346 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30347 sum's initial value.
30349 2017-04-25 Julian Brown <julian@codesourcery.com>
30350 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30352 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30354 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30356 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30358 2017-04-25 Julian Brown <julian@codesourcery.com>
30359 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30361 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30362 (thunderx2t99_sha): New Reservation.
30364 2017-04-25 Julian Brown <julian@codesourcery.com>
30365 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30367 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30368 type for 1-element load.
30370 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30372 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30374 2017-04-24 Martin Jambor <mjambor@suse.cz>
30376 PR tree-optimization/80293
30377 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30378 char arrays not totally scalarizable if it is false.
30379 (analyze_all_variable_accesses): Pass correct value in the new
30380 parameter. Add a statistics counter.
30382 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30384 PR middle-end/79931
30385 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30387 2017-04-24 Richard Biener <rguenther@suse.de>
30389 PR tree-optimization/80494
30390 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30391 out for complex types.
30393 2017-04-24 Richard Biener <rguenther@suse.de>
30395 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30396 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30397 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30398 (DFS): Adjust and never fail.
30399 (sccvn_dom_walker::fail): Remove.
30400 (sccvn_dom_walker::before_dom_children): Adjust.
30401 (run_scc_vn): Likewise and never fail.
30402 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30403 (pass_fre::execute): Likewise.
30405 2017-04-24 Richard Biener <rguenther@suse.de>
30407 PR tree-optimization/79725
30408 * tree-ssa-sink.c (statement_sink_location): Return whether
30409 failure reason was zero uses. Move that check later.
30410 (sink_code_in_bb): Deal with zero uses by removing the stmt
30413 2017-04-24 Richard Biener <rguenther@suse.de>
30416 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30417 pointer-based references.
30419 2017-04-24 Richard Biener <rguenther@suse.de>
30422 * pass_manager.h (pass_manager::operator new): Remove.
30423 (pass_manager::operator delete): Likewise.
30424 * passes.c (pass_manager::operator new): Remove.
30425 (pass_manager::operator delete): Likewise.
30426 (pass_manager::pass_manager): Zero individual pass members.
30428 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30431 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30432 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30433 Check "XEXP (src, 1)" operand here.
30434 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30435 Check "XEXP (src, 1)" operand here.
30436 (dimode_scalar_chain::make_vector_copies): Detect count register
30437 of a shift instruction. Zero extend count register from QImode
30438 to DImode to satisfy vector shift pattern count operand predicate.
30439 Substitute vector shift count operand with a DImode copy.
30440 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30443 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30445 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30446 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30447 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30448 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30449 (UNSPEC_NOREX_MEM): Remove definition.
30451 2017-04-21 Richard Biener <rguenther@suse.de>
30453 PR tree-optimization/79547
30454 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30455 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30456 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30457 without any constraints.
30459 2017-04-21 Richard Biener <rguenther@suse.de>
30461 PR tree-optimization/78847
30462 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30464 2017-04-21 Richard Biener <rguenther@suse.de>
30466 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30467 (build_distinct_type_copy): Likewise.
30468 (build_variant_type_copy): Likewise.
30469 * tree.c (build_qualified_type): Pass down mem-stat info.
30470 (build_distinct_type_copy): Likewise.
30471 (build_variant_type_copy): Likewise.
30473 2017-04-21 Richard Biener <rguenther@suse.de>
30475 PR tree-optimization/80237
30476 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30478 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30479 for a simplified result.
30481 2016-04-21 Richard Biener <rguenther@suse.de>
30483 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30484 sth as strict as a simple_iv but a chrec without symbols and an
30485 operand defined in the loop we are peeling (and not some subloop).
30486 (propagate_constants_for_unrolling): Propagate all constants.
30488 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30491 * config/i386/i386.c (print_reg): Remove assert for disalowed
30492 regno values, call output_operand_lossage instead.
30494 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30497 * config/i386/constraints.md (Yc): New register constraint.
30498 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30499 Use Yc constraint for alternative 2 of operand 0. Remove
30500 preferred_for_speed attribute.
30502 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30504 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30505 lastprivate clauses in SIMT case.
30507 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30509 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30511 2017-04-20 Richard Biener <rguenther@suse.de>
30513 PR tree-optimization/57796
30514 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30515 as N scalar stores.
30516 (vect_model_load_cost): Cost gathers as N scalar loads.
30518 2017-04-20 Richard Biener <rguenther@suse.de>
30520 * ggc-page.c (ggc_allocated_p): Rename to ...
30521 (safe_lookup_page_table_entry): ... this and return the lookup
30523 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30525 2017-04-20 Richard Biener <rguenther@suse.de>
30527 PR tree-optimization/80453
30528 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30529 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30530 from the conditions.
30531 (vn_phi_eq): Pass them down.
30532 (vn_phi_lookup): Record them.
30533 (vn_phi_insert): Likewise.
30535 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30537 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30538 uninitialized variable warning to avoid buffer overrun.
30540 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30543 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30544 is suppressed for '{ 0 }' in C.
30546 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30548 * BASE-VER: Set to 8.0.0.
30550 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30552 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30553 priority .init_array and .fini_array section with SECTION_NOTYPE
30556 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30558 PR middle-end/80423
30559 * tree.h (build_array_type): Add typeless_storage default argument.
30560 * tree.c (type_cache_hasher::equal): Also compare
30561 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30562 (build_array_type): Add typeless_storage argument, set
30563 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30565 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30566 (build_array_type): Likewise. Add typeless_storage argument.
30568 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30569 Jakub Jelinek <jakub@redhat.com>
30571 PR tree-optimization/80426
30572 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30573 operation on symbolic operands, also compute the overflow for the
30574 invariant part when the operation degenerates into a negation.
30576 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30579 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30580 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30583 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30585 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30588 * config/avr/avr.c (tree.h): Include it.
30589 (cgraph.h): Include it.
30590 (avr_encode_section_info): Don't warn for uninitialized progmem
30591 variable if it's just an alias.
30593 2017-04-19 Richard Biener <rguenther@suse.de>
30596 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30597 when needed by AutoPGO.
30599 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30602 * doc/lto.texi: Remove an extra 'that'.
30604 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30606 PR rtl-optimization/80429
30607 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30608 are only used in debug insns.
30610 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30611 Vladimir Makarov <vmakarov@redhat.com>
30613 * config/sparc/predicates.md (input_operand): Add comment. Return
30614 true for any memory operand when LRA is in progress.
30615 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30617 2017-04-18 Jeff Law <law@redhat.com>
30620 * mips.md ({return,simple_return}_internal): Do not overwrite
30623 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30625 PR tree-optimization/80443
30626 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30627 instead of adding 1, subtract -1 and similarly instead of subtracting
30630 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30632 PR rtl-optimization/80357
30633 * haifa-sched.c (tmp_bitmap): New variable.
30634 (model_recompute): Handle duplicate use records.
30635 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30636 (free_global_sched_pressure_data): Free it.
30638 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30641 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30642 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30643 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30644 instead of SYSTEM_HEADER_DIR.
30646 2017-04-18 Jeff Law <law@redhat.com>
30648 PR middle-end/80422
30649 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30650 predecessors after walking up the insn chain.
30652 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30655 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30656 sizetype type into debug info.
30658 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30661 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30662 unneeded test for TARGET_UPPER_REGS_SF.
30663 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30665 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30668 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30669 instead of gsi_after_labels.
30671 2017-04-18 Jeff Law <law@redhat.com>
30673 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30677 2017-04-13 Jeff Law <law@redhat.com>
30678 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30679 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30681 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30684 * config/avr/avr.c (intl.h): Include it.
30685 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30687 2017-04-18 Martin Liska <mliska@suse.cz>
30689 PR gcov-profile/78783
30690 * gcov-tool.c (gcov_output_files): Validate that destination
30691 file is either removed by the tool or by a user.
30693 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30694 Guy Benyei <guybe@mellanox.com>
30696 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30697 block, and do not negate it, the stored id is already negative.
30699 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30701 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30703 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30706 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30707 masks of options that should be turned off if the VSX vector
30708 options are turned off.
30709 (OTHER_P8_VECTOR_MASKS): Likewise.
30710 (OTHER_VSX_VECTOR_MASKS): Likewise.
30711 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30712 rs6000_disable_incompatible_switches to validate no type switches
30714 (rs6000_incompatible_switch): New function to disallow turning on
30715 other vector options if -mno-vsx, -mno-power8-vector, or
30716 -mno-power9-vector are specified.
30718 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30720 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30722 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30724 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30725 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30726 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30727 (ARG_POINTER_CFA_OFFSET): Likewise.
30729 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30731 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30732 conditions to take advantage of various optimizations.
30734 2017-04-13 Jeff Law <law@redhat.com>
30736 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30737 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30738 (zero_extendsidi2_dext): Likewise.
30740 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30743 * fold-const.c (fold_ternary_loc): Revert
30744 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30747 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30749 PR rtl-optimization/80343
30750 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30751 new scratch pseudo.
30753 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30756 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30757 to ubsan_encode_value.
30759 2017-04-13 Jeff Law <law@redhat.com>
30761 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30762 appearing in DEBUG_INSNs.
30764 2017-04-13 Martin Liska <mliska@suse.cz>
30766 PR gcov-profile/80413
30767 * gcov-io.c (gcov_write_string): Copy to buffer just when
30768 allocated size is greater than zero.
30770 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30773 * dwarf2out.c (decls_for_scope): Ignore declarations of
30774 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30776 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30779 * ipa-visibility.c (non_local_p): Fix typos.
30780 (localize_node): When localizing symbol in same comdat group,
30781 dissolve the group only when we know external symbols are going
30783 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30785 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30787 PR tree-optimization/79390
30788 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30789 order does not result in usable sequence, retry with reversed operand
30795 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30796 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30797 op0 instead of fold_convert_loc (loc, type, arg0).
30799 2017-04-12 Jeff Law <law@redhat.com>
30801 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30802 has a delay slot in the generated code.
30804 * config/cris/cris.md (cris_preferred_reload_class): Return
30805 GENNONACR_REGS rather than GENERAL_REGS.
30807 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30810 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30811 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30812 signedness of the result type.
30814 2017-04-12 Richard Biener <rguenther@suse.de>
30815 Jeff Law <law@redhat.com>
30817 PR tree-optimization/80359
30818 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30819 trim stores to TARGET_MEM_REFs.
30821 2017-04-12 Richard Biener <rguenther@suse.de>
30823 PR tree-optimization/79390
30824 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30825 threading case even more.
30827 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30830 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30831 for quad_address_p for TImode, instead of just not indexed_address.
30833 2017-04-12 Richard Biener <rguenther@suse.de>
30834 Bernd Edlinger <bernd.edlinger@hotmail.de>
30836 PR middle-end/79671
30837 * alias.c (component_uses_parent_alias_set_from): Handle
30838 TYPE_TYPELESS_STORAGE.
30839 (get_alias_set): Likewise.
30840 * tree-core.h (tree_type_common): Add typeless_storage flag.
30841 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30842 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30843 for types containing members with TYPE_TYPELESS_STORAGE.
30844 (place_field): Likewise.
30845 (layout_type): Likewise for ARRAY_TYPE.
30846 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30847 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30848 TYPE_TYPELESS_STORAGE.
30849 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30851 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30854 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30855 first argument to type.
30857 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30861 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30862 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30863 (rs6000_expand_binop_builtin): Likewise.
30864 (rs6000_expand_ternop_builtin): Likewise; also add missing
30865 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30866 vshasigma built-ins.
30867 * doc/extend.texi: Document that vec_xxpermdi's third argument
30868 must be a constant.
30870 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30872 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30873 Use shift_const cost parameter when calculating gain of STV shifts.
30875 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30877 PR rtl-optimization/70478
30878 * lra-constraints.c (process_alt_operands): Check memory for
30879 disfavoring memory insn operand.
30881 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30883 PR middle-end/80100
30884 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30885 left shift in unsigned HOST_WIDE_INT type.
30887 PR rtl-optimization/80385
30888 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30889 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30892 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30893 if they have any depend clauses.
30895 2017-04-11 Martin Liska <mliska@suse.cz>
30898 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30899 * ipa-split.c (split_function): Create a local comdat symbol
30900 if caller is in a comdat group.
30902 2017-04-11 Martin Liska <mliska@suse.cz>
30905 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30908 2017-04-11 Martin Sebor <msebor@redhat.com>
30910 PR middle-end/80364
30911 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30912 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30914 (directive::set_width, directive::set_precision, format_character):
30916 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30919 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
30922 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
30923 conflict, set target->arch_name instead of target->cpu_name.
30925 2017-04-11 Richard Biener <rguenther@suse.de>
30927 PR tree-optimization/80374
30928 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
30929 build_zero_cst, remove fold_convertible_p check again.
30931 2017-04-11 Martin Liska <mliska@suse.cz>
30934 * ubsan.c (instrument_object_size): Do not instrument register
30937 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30940 * config/i386/i386-builtin-types.def
30941 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
30942 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
30943 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
30944 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
30945 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
30946 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
30947 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
30948 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
30949 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
30950 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
30951 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
30952 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
30953 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
30954 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
30955 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
30956 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
30957 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
30958 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
30959 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
30960 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
30961 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
30962 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
30963 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
30964 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
30965 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
30966 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
30967 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
30968 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
30969 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
30970 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
30971 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
30972 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
30973 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
30974 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
30975 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
30976 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
30977 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
30978 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
30979 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
30980 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
30981 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
30982 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
30983 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
30984 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
30985 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
30986 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
30988 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
30989 flag to second_arg_count, handle 4 argument function type _COUNT
30990 aliases, handle second_arg_count on second argument rather than last.
30992 2017-04-10 Jeff Law <law@redhat.com>
30994 PR tree-optimization/80374
30995 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
30996 record anything if we can not convert integer_zero_node to the
30999 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31002 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31003 Enhance special handling given to the TARGET_P9_MINMAX option in
31004 relation to certain other options.
31006 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31008 PR tree-optimization/80153
31009 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31010 remove POINTER_PLUS_EXPR's base part directly, rather than through
31013 2017-04-10 Richard Biener <rguenther@suse.de>
31014 Bin Cheng <bin.cheng@arm.com>
31016 PR tree-optimization/80153
31017 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31018 the first element of pointer type aff_tree. Build result expr in
31020 (add_elt_to_tree): Convert to type unconditionally. Remove other
31021 fold_convert calls.
31022 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31023 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31025 2017-04-10 Richard Biener <rguenther@suse.de>
31027 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31030 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31032 PR rtl-optimization/70478
31033 * lra-constraints.c (curr_small_class_check): New.
31034 (update_and_check_small_class_inputs): New.
31035 (process_alt_operands): Update curr_small_class_check. Disfavor
31036 alternative insn memory operands. Check available regs for small
31039 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31042 * config/mips/mips.opt (-mvirt): Update description.
31043 * doc/invoke.texi (-mvirt): Likewise.
31045 2017-04-10 Richard Biener <rguenther@suse.de>
31047 PR middle-end/80362
31048 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31049 looking for NEGATE_EXPR in -A / -B to A / B folding.
31051 2017-04-10 Martin Liska <mliska@suse.cz>
31053 PR gcov-profile/80224
31054 * gcov.c (print_usage): Fix usage string.
31055 (get_gcov_intermediate_filename): Remove.
31056 (output_gcov_file): Use both for normal and intermediate format.
31057 (generate_results): Do not initialize special file for
31058 intermediate format.
31060 2017-04-10 Richard Biener <rguenther@suse.de>
31062 PR tree-optimization/80304
31063 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31066 2017-04-10 Nathan Sidwell <nathan@acm.org>
31069 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31070 (rs6000_init_builtins): Use it.
31072 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31074 * config/arm/arm.md (<mrc>): Add mode to SET source.
31075 (<mrrc>): Likewise.
31077 2017-04-10 Richard Biener <rguenther@suse.de>
31079 PR middle-end/80344
31080 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31082 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31085 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31086 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31087 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31088 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31089 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31090 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31091 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31092 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31093 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31094 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31095 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31096 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31097 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31098 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31099 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31100 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31101 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31102 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31103 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31104 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31105 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31106 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31107 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31109 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31111 PR rtl-optimization/70478
31112 * lra-constraints.c: Reverse the last patch.
31114 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31116 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31117 Add comment for WCHAR_T.
31119 2017-04-08 Martin Liska <mliska@suse.cz>
31122 2017-04-07 Martin Liska <mliska@suse.cz>
31125 * ipa-split.c (split_function): Add function part to a same comdat
31128 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31131 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31133 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31135 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31136 (vec_pairing): ...to this.
31137 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31138 (rs6000_sched_init): Adjust for name change.
31139 (struct rs6000_sched_context): Likewise.
31140 (rs6000_init_sched_context): Likewise.
31141 (rs6000_set_sched_context): Likewise.
31143 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31149 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31151 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31152 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31153 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31155 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31157 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31159 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31161 PR rtl-optimization/70703
31162 * ira-color.c (update_conflict_hard_regno_costs): Use
31163 int64_t instead of HOST_WIDE_INT.
31165 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31167 PR rtl-optimization/70478
31168 * lra-constraints.c (process_alt_operands): Disfavor alternative
31169 insn memory operands.
31171 2017-04-07 Jeff Law <law@redhat.com>
31173 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31174 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31176 2017-04-07 Martin Liska <mliska@suse.cz>
31179 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31180 Show error message instead of an ICE.
31182 2017-04-07 Martin Liska <mliska@suse.cz>
31185 * ipa-split.c (split_function): Add function part to a same comdat
31188 2017-04-07 Richard Biener <rguenther@suse.de>
31190 PR middle-end/80341
31191 * tree.c (get_unwidened): Also handle ! for_type case for
31193 * convert.c (do_narrow): Split out from ...
31194 (convert_to_integer_1): ... here. Do not pass final truncation
31195 type to get_unwidened for TRUNC_DIV_EXPR.
31197 2017-04-07 Richard Biener <rguenther@suse.de>
31199 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31201 (aff_combination_const): Adjust.
31202 (aff_combination_scale): Likewise.
31203 (aff_combination_add_elt): Likewise.
31204 (aff_combination_add_cst): Likewise.
31205 (aff_combination_convert): Likewise.
31206 (add_elt_to_tree): Likewise. Remove unused argument.
31207 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31209 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31211 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31213 * config/arm/arm.c (arm_default_short_enums): Use
31214 ARM_DEFAULT_SHORT_ENUMS.
31215 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31217 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31220 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31221 members with redundant out-of-class redeclaration.
31223 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31226 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31227 * config/i386/i386.md (*zero_extendsidi2):
31228 Add (?*x,*x) and (?*v,*v) alternatives.
31230 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31233 * config/i386/i386.c (ix86_expand_builtin)
31234 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31235 mode from insn data. Convert operands to insn operand mode.
31236 Copy operands that don't satisfy insn predicate to a register.
31238 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31240 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31243 2017-04-06 Richard Biener <rguenther@suse.de>
31245 PR tree-optimization/80334
31246 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31247 preserve alignment of accesses.
31249 2017-04-06 Richard Biener <rguenther@suse.de>
31251 PR tree-optimization/80262
31252 * tree-sra.c (build_ref_for_offset): Preserve address-space
31254 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31255 Drop useless address-space information on MEM_REF offsets.
31257 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31259 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31261 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31263 PR rtl-optimization/70703
31264 * ira-color.c (update_conflict_hard_regno_costs): Use
31265 HOST_WIDE_INT instead of long.
31267 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31270 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31271 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31273 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31274 for x86_64 target. Handle -m3dnowa option.
31276 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31278 PR rtl-optimization/70703
31279 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31280 (update_conflict_hard_regno_costs): Use long instead of unsigned
31281 arithmetic for cost calculation.
31283 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31284 Bernd Edlinger <bernd.edlinger@hotmail.de>
31287 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31290 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31293 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31294 ptr_mode with Pmode throughout.
31295 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31296 into probe_stack_range and use DImode.
31298 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31301 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31302 call_eh_return is true.
31304 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31306 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31307 Initialize last_match_fntype_index.
31309 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31312 * tree-nvr.c: Include internal-fn.h.
31313 (pass_return_slot::execute): Ignore internal calls without
31316 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31317 Richard Biener <rguenther@suse.de>
31320 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31321 captures used multiple times, except for the last use.
31322 * generic-match-head.c: Include gimplify.h.
31324 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31326 PR tree-optimization/79390
31327 * target.h (struct noce_if_info): Declare.
31328 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31329 * target.def (noce_conversion_profitable_p): New target hook.
31330 * ifcvt.h (struct noce_if_info): New type, moved from ...
31331 * ifcvt.c (struct noce_if_info): ... here.
31332 (noce_conversion_profitable_p): Renamed to ...
31333 (default_noce_conversion_profitable_p): ... this. No longer
31335 (noce_try_store_flag_constants, noce_try_addcc,
31336 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31337 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31338 instead of noce_conversion_profitable_p.
31339 * config/i386/i386.c: Include ifcvt.h.
31340 (ix86_option_override_internal): Don't override
31341 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31342 (ix86_noce_conversion_profitable_p): New function.
31343 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31344 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31345 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31346 * doc/tm.texi: Regenerated.
31348 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31350 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31353 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31356 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31357 instructions for small multiply cores.
31359 2017-04-04 Jeff Law <law@redhat.com>
31361 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31363 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31364 that are not set by the loop over the elements.
31366 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31369 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31370 int mode, convert_modes it to mode as unsigned, otherwise use
31371 lowpart_subreg to mode rather than SImode.
31372 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31373 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31374 Use DImode instead of SImode for the shift count operand.
31375 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31378 2017-04-04 Richard Biener <rguenther@suse.de>
31380 PR middle-end/80281
31381 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31382 arithmetic done for the negate or the plus. Simplify.
31383 (A - (-B) -> A + B): Likewise.
31384 * fold-const.c (split_tree): Make sure to not negate pointers.
31386 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31388 PR rtl-optimization/60818
31389 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31390 a compare of comparisons with the thing compared if this results
31391 in a different machine mode.
31393 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31395 * alias.c (base_alias_check): Fix typo in comment.
31396 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31397 * cgraphunit.c (symbol_table::compile): Likewise.
31398 * collect2.c (maybe_run_lto_and_relink): Likewise.
31399 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31400 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31401 * config/avr/avr.c (avr_map_op_t): Likewise.
31402 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31403 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31404 * config/epiphany/epiphany.md (movcc): Likewise.
31405 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31406 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31408 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31409 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31410 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31411 * config/sh/sh.c (sh_rtx_costs): Likewise.
31412 * fold-const.c (fold_truth_andor): Likewise.
31413 * genautomata.c (collapse_flag): Likewise.
31414 * gengtype.h (struct type::u::s): Likewise.
31415 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31416 * input.c (FORMAT_AMOUNT): Likewise.
31417 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31418 (known_aggs_to_agg_replacement_list): Likewise.
31419 * ipa-inline-analysis.c: Likewise.
31420 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31421 * ipa-polymorphic-call.c
31422 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31423 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31424 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31426 * modulo-sched.c (apply_reg_moves): Likewise.
31427 * omp-expand.c (build_omp_regions_1): Likewise.
31428 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31429 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31430 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31431 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31432 * value-prof.c: Likewise.
31433 * var-tracking.c (val_reset): Likewise.
31435 2017-04-03 Richard Biener <rguenther@suse.de>
31437 PR tree-optimization/80275
31438 * fold-const.c (split_address_to_core_and_offset): Handle
31441 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31443 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31444 descriptors is at least equal to that of functions.
31446 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31448 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31450 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31453 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31454 (mov<IMOD4:mode>): New expander.
31455 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31457 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31459 PR rtl-optimization/79405
31460 * fwprop.c (propagations_left): New variable.
31461 (forward_propagate_into): Decrement it.
31462 (fwprop_init): Initialize it.
31463 (fw_prop): If the variable has reached zero, stop propagating.
31464 (fwprop_addr): Ditto.
31466 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31469 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31470 a FUNCTION_DECL, pass it as decl instead of origin to
31473 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31475 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31478 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31481 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31482 TARGET_VSX_SMALL_INTEGER.
31484 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31486 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31487 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31489 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31491 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31492 extraction from odd-numbered MSA register.
31494 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31496 PR middle-end/80173
31497 * expmed.c (store_bit_field_1): Don't attempt to create
31498 a word subreg out of hard registers wider than word if they
31499 have HARD_REGNO_NREGS of 1 for their mode.
31501 PR middle-end/80163
31502 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31503 conversions to integer types wider than word and pointer.
31506 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31507 (rtx_equal_for_cselib_p): Pass 0 to it.
31508 * cselib.c (cselib_hasher::equal): Likewise.
31509 (rtx_equal_for_cselib_1): Add depth argument. If depth
31510 is 128, don't look up VALUE locs and punt. Increment
31511 depth in recursive calls when walking VALUE locs.
31513 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31515 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31516 (make_gcov_file_name): Use the canonical path name for generating
31518 (read_line): Fix handling of files with ascii null bytes.
31520 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31522 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31523 to initialise a vector register instead
31524 of using a const_int.
31526 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31528 PR translation/80189
31529 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31530 diagnostic messages.
31532 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31535 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31536 (dfp_diex_<mode>): Update mode of operand 1.
31537 * doc/extend.texi (dxex, dxexq): Document change to return type.
31538 (diex, diexq): Document change to argument type.
31540 2017-03-30 Martin Jambor <mjambor@suse.cz>
31543 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31544 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31545 it reflects the signature changes performed at the callee side.
31546 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31547 to cgraph_build_function_type_skip_args.
31548 (build_function_decl_skip_args): Adjust call to the above function.
31550 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31553 * config/i386/sse.md
31554 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31555 register as dest whenever it is a MEM not rtx_equal_p to the
31556 corresponding dup operand, and when forcing into reg move the
31557 reg into the memory afterwards.
31558 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31559 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31560 for the force_reg mode.
31561 (avx512vl_vextractf128<mode>): Use register as dest either
31562 always when a MEM, or when it is a MEM not rtx_equal_p to the
31563 corresponding dup operand, or even not when it is a CONST_VECTOR
31564 depending on the mode and lo vs. hi.
31565 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31567 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31568 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31569 Likewise. Require that operands[2] is even.
31570 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31571 Remove extraneous parens. Require that operands[2] is a multiple
31573 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31574 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31575 disallow memory then.
31577 2017-03-30 Richard Biener <rguenther@suse.de>
31579 PR tree-optimization/77498
31580 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31581 to non-constants over backedges.
31583 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31585 PR rtl-optimization/80233
31586 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31587 as last_combined_insn. Do not test for BARRIER_P separately.
31589 2017-03-29 Andreas Schwab <schwab@suse.de>
31592 * calls.c (prepare_call_address): Convert funexp to Pmode before
31593 copying to temp reg.
31595 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31597 PR tree-optimization/80158
31598 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31599 Handle possible future case of more than one alternate
31601 (replace_rhs_if_not_dup): Likewise.
31602 (replace_one_candidate): Likewise.
31604 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31606 PR rtl-optimization/80193
31607 * ira.c (ira): Do not check allocation for LRA.
31609 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31611 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31612 (nvptx_output_simt_exit): Declare.
31613 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31614 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31615 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31616 (nvptx_declare_function_name): ...here. Emit declaration of local
31617 memory space buffer for omp_simt_enter insn.
31618 (nvptx_output_unisimt_switch): New.
31619 (nvptx_output_softstack_switch): New.
31620 (nvptx_output_simt_enter): New.
31621 (nvptx_output_simt_exit): New.
31622 * config/nvptx/nvptx.h (struct machine_function): New fields
31623 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31624 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31625 (UNSPECV_SIMT_EXIT): Ditto.
31626 (omp_simt_enter_insn): New insn.
31627 (omp_simt_enter): New expansion.
31628 (omp_simt_exit): New insn.
31629 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31631 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31632 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31633 (expand_GOMP_SIMT_EXIT): New.
31634 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31635 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31636 (GOMP_SIMT_EXIT): Ditto.
31637 * target-insns.def (omp_simt_enter): New insn.
31638 (omp_simt_exit): Ditto.
31639 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31641 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31642 (lower_rec_input_clauses): Likewise.
31643 (lower_lastprivate_clauses): Handle SIMT privatization.
31645 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31646 (ompdevlow_adjust_simt_enter): New.
31647 (find_simtpriv_var_op): New.
31648 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31649 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31651 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31652 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31653 (copy_decl_for_dup_finish): Ditto.
31655 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31657 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31660 * config/i386/i386.c (ix86_option_override_internal): Always
31661 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31663 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31665 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31667 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31669 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31670 mark new edge's irreducible flag accordign to it.
31671 (vect_do_peeling): Check loop preheader edge's irreducible flag
31672 and pass it to function slpeel_add_loop_guard.
31674 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31676 PR tree-optimization/80218
31677 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31678 Update block frequencies and counts.
31680 2017-03-28 Richard Biener <rguenther@suse.de>
31682 PR tree-optimization/78644
31683 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31684 of a simplification result we may not use it at all.
31686 2017-03-28 Richard Biener <rguenther@suse.de>
31689 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31690 without arguments, generate default definition of a SSA name.
31692 2017-03-28 Richard Biener <rguenther@suse.de>
31694 PR middle-end/80222
31695 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31696 TYPE_REF_CAN_ALIAS_ALL references.
31697 * fold-const.c (fold_indirect_ref_1): Likewise.
31699 2017-03-28 Martin Liska <mliska@suse.cz>
31702 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31703 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31705 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31706 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31708 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31709 (EXTRA_SPECS): Define.
31710 (SUBTARGET_EXTRA_SPECS): Likewise.
31711 (SUBTARGET_CPP_SPEC): Likewise.
31712 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31713 SUBTARGET_EXTRA_SPECS.
31714 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31716 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31718 * config/arc/simdext.md (vst64_insn): Update pattern.
31719 (vld32wh_insn): Likewise.
31720 (vld32wl_insn): Likewise.
31721 (vld64_insn): Likewise.
31722 (vld32_insn): Likewise.
31724 2017-03-28 Marek Polacek <polacek@redhat.com>
31727 * fold-const.c (fold_comparison): Use protected_set_expr_location
31728 instead of SET_EXPR_LOCATION.
31730 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31732 * tree.c (add_expr): Avoid name lookup warning.
31734 2017-03-27 Jeff Law <law@redhat.com>
31736 PR tree-optimization/80216
31737 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31738 function name. Limit recursion depth.
31739 (record_temporary_equivalences): Corresponding changes.
31741 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31743 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31746 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31749 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31751 * cfgcleanup.c (reg_note_cfa_p): New array.
31752 (insns_have_identical_cfa_notes): New function.
31753 (old_insns_match_p): Don't cross-jump in between /f
31754 and non-/f instructions. If both i1 and i2 are frame related,
31755 verify all CFA notes, their order and content.
31757 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31760 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31761 HImode and SImode with zero extend to DImode to one insn.
31762 (bswap<mode>2_extenddi): Likewise.
31763 (bswapsi2_extenddi): Likewise.
31764 (bswaphi2_extendsi): Likewise.
31765 (bswaphi2): Combine bswap HImode and SImode into one insn.
31766 Separate memory insns from swapping register.
31767 (bswapsi2): Likewise.
31768 (bswap<mode>2): Likewise.
31769 (bswaphi2_internal): Delete, no longer used.
31770 (bswapsi2_internal): Likewise.
31771 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31772 store, and gpr<-gpr swap insns.
31773 (bswap<mode>2_store): Likewise.
31774 (bswaphi2_reg): Register only splitter, combine with the splitter.
31775 (bswaphi2 splitter): Likewise.
31776 (bswapsi2_reg): Likewise.
31777 (bswapsi2 splitter): Likewise.
31778 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31779 the insns into load, store, and register/register insns.
31780 (bswapdi2_ldbrx): Likewise.
31781 (bswapdi2_load): Likewise.
31782 (bswapdi2_store): Likewise.
31783 (bswapdi2_reg): Likewise.
31785 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31787 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31788 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31790 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31793 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31795 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31796 special handling for target option conflicts between dform
31797 options (-mpower9-dform, -mpower9-dform-vector,
31798 -mpower9-dform-scalar) and -mno-direct-move.
31800 2017-03-27 Richard Biener <rguenther@suse.de>
31802 PR tree-optimization/80181
31803 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31805 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31807 * config/arc/predicates.md (move_double_src_operand): Replace the
31808 call to move_double_src_operand with a call to address_operand.
31810 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31812 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31813 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31814 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31816 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31818 * config/arc/predicates.md (long_immediate_loadstore_operand):
31819 Consider scaled addresses cases.
31821 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31823 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31824 restored when in interrupt.
31825 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31826 doesn't have delay slot.
31828 2017-03-27 Richard Biener <rguenther@suse.de>
31831 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31832 inlined thunk clones.
31834 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31837 * asan.c (instrument_derefs): Copy over last operand from
31838 original COMPONENT_REF to the new COMPONENT_REF with
31839 DECL_BIT_FIELD_REPRESENTATIVE.
31840 * ubsan.c (instrument_object_size): Likewise.
31842 2017-03-27 Richard Biener <rguenther@suse.de>
31844 PR tree-optimization/80170
31845 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31846 sure DR/SCEV didnt fold in constants we do not see when looking
31847 at the reference base alignment.
31849 2017-03-27 Richard Biener <rguenther@suse.de>
31851 PR middle-end/80171
31852 * gimple-fold.c (fold_ctor_reference): Properly guard against
31853 NULL return value from canonicalize_constructor_val.
31855 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31858 * config/i386/i386.c (ix86_expand_builtin)
31859 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31860 flags reg setting and flags reg using instructions.
31861 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31862 clobbering instructions to zero extend op2.
31864 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31866 * doc/install.texi (Configuration) <--with-aix-soname>:
31867 Update link to AIX ld.
31869 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31871 PR rtl-optimization/80160
31872 PR rtl-optimization/80159
31873 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31874 reg_alternate_class into account.
31876 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31879 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31880 to consider in curr_insn_transform.
31882 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31884 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31885 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31886 and emit_mode_inner.
31888 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31890 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31891 argument to the overloaded builtin variants. Use the new flag to
31892 deprecate certain builtin variants.
31893 * config/s390/s390-builtin-types.def: Add new builtin types.
31894 * config/s390/s390-builtins.h: Support new flags field for
31895 overloaded builtins.
31896 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31897 (s390_macro_to_expand): Enable vector float data type.
31898 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31899 builtins by incrementing the __VEC__ version number.
31900 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31902 (s390_resolve_overloaded_builtin): Emit error messages depending
31903 on the builtin flags.
31904 * config/s390/s390.c (s390_expand_builtin): Support additional
31905 flags argument. Change error message to match the messages
31906 emitted in s390-c.c.
31907 * config/s390/s390.md: New UNSPEC_* constants.
31908 (op_type): Add new instruction types.
31909 * config/s390/vecintrin.h: Add new builtins and test data class
31911 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31912 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31913 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31914 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31915 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31916 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31918 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31919 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31920 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
31921 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
31923 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
31924 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
31925 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
31926 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
31928 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
31929 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
31930 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
31931 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
31932 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
31933 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
31934 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
31936 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
31937 ("vec_scatter_element<V_HW_4:mode>_DI")
31938 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
31939 ("vec_fpint<mode>", "vflls")
31940 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
31941 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
31942 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
31943 ("*vec_cmphe<mode>_cc"): ... these.
31945 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
31946 mode constant instead of magic value.
31948 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31950 * config/s390/s390.c (s390_expand_vec_compare): Support other
31951 vector floating point modes than just V2DF.
31952 (s390_expand_vcond): Likewise.
31953 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
31954 (s390_cannot_change_mode_class): Prevent mode changes between TF
31955 and V1TF in vector registers.
31956 * config/s390/s390.md (DF, SF): New mode attributes.
31957 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
31958 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
31959 SFmode support for VRs.
31960 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
31962 (VFT, VF_HW): New mode iterators.
31963 (vw, sdx): New mode attributes.
31964 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
31965 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
31966 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
31967 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
31968 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
31969 also the new vector floating point modes. Renaming to ...
31971 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
31972 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
31973 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
31974 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
31975 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
31976 ("vec_unordered<mode>"): ... these.
31978 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
31979 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
31980 ("*vec_extendv2df"): New insn definitions.
31982 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31984 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
31985 ("mulditi3_2", "*muldi3_sign"): New patterns.
31986 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
31987 rename the pattern definition.
31989 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31991 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
31993 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
31995 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31997 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
31998 instruction if possible.
31999 * config/s390/vector.md (vec_halfnumelts): New mode
32001 ("*vec_vllezlf<mode>"): New pattern.
32003 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32005 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32006 ("popcountv4si2", "popcountv2di2"): Rename to ...
32007 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32008 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32010 ("popcount<mode>2_vxe"): New pattern.
32012 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32014 * common/config/s390/s390-common.c (processor_flags_table): Add
32016 * config.gcc: Add arch12.
32017 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32018 Default to arch12 for unknown CPU model numbers.
32019 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32020 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32021 PROCESSOR_max sanity check.
32022 * config/s390/s390-opts.h (enum processor_type): Add
32024 * config/s390/s390.c (processor_table): Add arch12.
32025 (s390_expand_builtin): Add check for B_VXE flag.
32026 (s390_issue_rate): Add PROCESSOR_ARCH12.
32027 (s390_get_sched_attrmask): Likewise.
32028 (s390_get_unit_mask): Likewise.
32029 (s390_sched_score): Enable z13 scheduling for arch12.
32030 (s390_sched_reorder): Likewise.
32031 (s390_sched_variable_issue): Likewise.
32032 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32034 (s390_tune_attr): Use z13 scheduling also for arch12.
32035 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32036 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32037 (TARGET_VXE_P): New macros.
32038 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32039 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32040 * config/s390/s390.opt: Add arch12 as processor_type.
32042 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32044 * config/s390/s390.md
32045 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32046 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32047 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32049 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32050 Rename expanders to ...
32052 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32053 ("fixuns_truncdddi2_emu"): ... these.
32055 ("fixuns_trunc<mode>si2_emu"): New expander.
32057 ("*fixuns_truncdfdi2_z13"): Rename to ...
32058 ("*fixuns_truncdfdi2_vx"): ... this.
32060 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32062 * config/s390/2964.md: Remove the single element vector compare
32063 instructions which are no longer used.
32064 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32066 (s390_canonicalize_comparison): Remove handling of DFmode
32068 (s390_expand_vec_compare_scalar): Remove function.
32069 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32070 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32072 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32074 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32076 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32078 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32079 will anyway by matched by mov<mode>_64dfp.
32081 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32083 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32084 vlef/vstef. Add missing operand to vleif.
32086 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32088 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32089 pair for all vector types with 64 bit elements.
32090 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32091 * config/s390/vector.md (V_HW_64): ... here.
32092 (V_128_NOSINGLE): New mode iterator.
32093 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32094 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32095 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32096 ("*vec_load_pairv2di"): Change to ...
32097 ("*vec_load_pair<mode>"): ... this one.
32099 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32101 * config/s390/constraints.md: Add comments.
32102 (jKK): Reject element sizes > 8 bytes.
32103 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32105 * config/s390/s390.md: Add the s_operand checks formerly in
32106 s390_split_ok_p to various splitters where they are still
32108 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32109 for 128 bit vectors. Plus two splitters.
32111 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32113 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32116 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32119 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32120 error if the boundary argument is not constant.
32122 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32124 PR rtl-optimization/80112
32125 * loop-doloop.c (doloop_condition_get): Don't check condition
32126 if cmp isn't SET with IF_THEN_ELSE src.
32128 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32130 PR tree-optimization/80158
32131 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32132 replacing a candidate statement, also replace it for the
32133 candidate's alternate interpretation.
32134 (replace_rhs_if_not_dup): Likewise.
32135 (replace_one_candidate): Likewise.
32137 2017-03-24 Richard Biener <rguenther@suse.de>
32139 PR tree-optimization/80167
32140 * graphite-isl-ast-to-gimple.c
32141 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32143 (translate_isl_ast_to_gimple::get_rename): Likewise.
32145 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32147 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32148 handling of certain combinations of target options, including the
32149 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32150 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32152 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32155 * config/arm/arm.md (*load_multiple): Add reload_completed to
32156 matching condition.
32158 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32159 Richard Biener <rguenth@suse.de>
32161 PR tree-optimization/79908
32162 PR tree-optimization/80136
32163 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32164 been cast away, gimplify_and_add suffices.
32166 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32168 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32170 2017-03-23 Richard Biener <rguenther@suse.de>
32172 PR tree-optimization/80032
32173 * gimplify.c (gimple_push_cleanup): Forced unconditional
32174 cleanups still have to go to the conditional_cleanups
32177 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32179 PR tree-optimization/80072
32180 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32182 (next_operand_entry_id): Change type to unsigned int.
32183 (sort_by_operand_rank): Make sure to return the right return value
32184 even if unsigned fields are bigger than INT_MAX.
32185 (struct oecount): Change cnt and id type to unsigned int.
32186 (oecount_hasher::equal): Formatting fix.
32187 (oecount_cmp): Make sure to return the right return value
32188 even if unsigned fields are bigger than INT_MAX.
32189 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32192 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32193 TREE_READONLY on result if writing it more than once.
32196 * doc/invoke.texi (-fsanitize=thread): Document that with
32197 -fnon-call-exceptions atomics are not able to throw
32201 * tsan.c: Include tree-eh.h.
32202 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32203 maybe_clean_or_replace_eh_stmt where needed.
32204 (instrument_memory_accesses): Add cfg_changed argument.
32205 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32206 if it returned true.
32207 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32209 PR rtl-optimization/63191
32210 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32211 wrapper function, moved the whole old content into ...
32212 (ix86_delegitimize_address_1): ... this. New inline function.
32213 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32214 true as last argument instead of ix86_delegitimize_address.
32216 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32218 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32219 cortexa57_branch_cost.
32221 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32223 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32225 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32228 * doc/md.texi (Constraints): Document wA constraint.
32229 * config/rs6000/constraints.md (wA): New.
32230 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32231 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32232 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32233 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32235 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32238 * gimplify.c (is_oacc_declared): New function.
32239 (oacc_default_clause): Use it to set default flags for acc declared
32240 variables inside parallel regions.
32241 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32242 declared variables.
32243 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32244 declare attribute to any decl as necessary.
32246 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32249 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32250 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32251 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32252 (arm_arch_lpae): This.
32253 * config/arm/arm.c (arm_arch7ve): Rename into ...
32254 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32255 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32258 2017-03-22 Martin Liska <mliska@suse.cz>
32261 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32262 error message instead of an ICE.
32264 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32266 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32268 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32270 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32272 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32275 2017-03-21 Martin Sebor <msebor@redhat.com>
32277 * doc/extend.texi: Use "cannot" instead of "can't."
32278 * doc/hostconfig.texi: Same.
32279 * doc/install.texi: Same.
32280 * doc/invoke.texi: Same.
32281 * doc/loop.texi: Same.
32282 * doc/md.texi: Same.
32283 * doc/objc.texi: Same.
32284 * doc/rtl.texi: Same.
32285 * doc/tm.texi: Same.
32286 * doc/tm.texi.in: Same.
32287 * doc/trouble.texi: Same.
32289 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32292 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32293 (collect_checksum_attributes): Set it.
32294 (die_checksum_ordered): Use it.
32296 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32298 PR tree-optimization/79908
32299 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32300 change: For a VA_ARG whose LHS has been cast away, use
32301 force_gimple_operand to construct the side effects.
32303 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32305 PR translation/80001
32306 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32307 more amenable to translation.
32308 (oacc_loop_auto_partitions): Likewise.
32310 2017-03-21 Marek Polacek <polacek@redhat.com>
32311 Martin Sebor <msebor@redhat.com>
32313 PR tree-optimization/80109
32314 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32315 on INTEGRAL_TYPE_P.
32317 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32318 Segher Boessenkool <segher@kernel.crashing.org>
32321 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32322 check reg_used_between_p between insn and one of succ or succ2
32323 depending on if succ is artificial insn not inserted into insn
32326 2017-03-21 Martin Liska <mliska@suse.cz>
32328 PR gcov-profile/80081
32329 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32330 * doc/gcc.texi: Include gcov-dump stuff.
32331 * doc/gcov-dump.texi: New file.
32333 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32335 PR rtl-optimization/79150
32336 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32337 conditional jump, if the jump is the last insn of the loop.
32339 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32340 Richard Biener <rguenth@suse.de>
32342 PR tree-optimization/79908
32343 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32344 been cast away, use force_gimple_operand to construct the side
32347 2017-03-21 Martin Liska <mliska@suse.cz>
32349 PR libfortran/79956
32350 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32353 2017-03-21 Brad Spengler <spender@grsecurity.net>
32356 * plugin.c (htab_hash_plugin): New function.
32357 (add_new_plugin): Use it and adjust.
32358 (parse_plugin_arg_opt): Adjust.
32359 (init_one_plugin): Likewise.
32361 2017-03-21 Richard Biener <rguenther@suse.de>
32363 PR tree-optimization/80032
32364 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32365 if set force the cleanup to happen unconditionally.
32366 (gimplify_target_expr): Push inserted clobbers with force_uncond
32367 to avoid them being removed by control-dependent DCE.
32369 2017-03-21 Richard Biener <rguenther@suse.de>
32371 PR tree-optimization/80122
32372 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32373 va_arg_pack_len when the inlined call stmt requires pack
32375 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32377 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32380 * tsan.c (instrument_builtin_call): If the memory model argument
32381 is not a constant, assume it is valid.
32384 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32387 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32389 PR rtl-optimization/79910
32390 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32391 if its dest is used by an insn before I2 (other than the combined
32392 insns themselves, which are properly handled already).
32394 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32397 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32399 * combine.c (record_used_regs): New static function.
32400 (try_combine): Handle situations where there is an additional
32401 instruction between I2 and I3 which needs to have a LOG_LINK
32405 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32407 * combine.c (try_combine): Delete redundant i1 test. Call
32408 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32410 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32413 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32414 alternatives 13/14.
32416 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32418 PR tree-optimization/80054
32419 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32420 the optimization if a PHI or any of its arguments is not dominated
32421 by the candidate's basis. Use gphi* rather than gimple* as
32423 (replace_profitable_candidates): Clean up a gimple* variable that
32424 should be a gphi* variable.
32426 2017-03-20 Martin Sebor <msebor@redhat.com>
32429 * doc/extend.texi (attribute constructor): Document present limitation.
32431 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32434 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32435 __POWER9_VECTOR__ #ifdef control, change template definition to
32436 use Power9-specific built-in function.
32437 (vec_any_eq): Likewise.
32438 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32439 to control outcomes from this test.
32440 (vector_ae_<mode>p): For VEC_F modes, likewise.
32442 2017-03-20 Ian Lance Taylor <iant@google.com>
32444 * config/i386/i386.c (ix86_function_regparm): Save an extra
32445 register for -fsplit-stack with DECL_STATIC_CHAIN.
32447 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32450 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32451 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32453 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32455 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32457 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32460 2017-03-20 Marek Polacek <polacek@redhat.com>
32463 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32465 2017-03-20 Richard Biener <rguenther@suse.de>
32467 PR tree-optimization/80113
32468 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32469 allocate extra SSA name for PHI def.
32470 (add_close_phis_to_outer_loops): Likewise.
32471 (add_close_phis_to_merge_points): Likewise.
32472 (copy_loop_close_phi_args): Likewise.
32473 (copy_cond_phi_nodes): Likewise.
32475 2017-03-20 Martin Liska <mliska@suse.cz>
32477 PR middle-end/79753
32478 * tree-chkp.c (chkp_build_returned_bound): Do not build
32479 returned bounds for a LHS that's not a BOUNDED_P type.
32481 2017-03-20 Martin Liska <mliska@suse.cz>
32485 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32486 COMPLEX_CST and VECTOR_CST.
32488 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32491 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32492 target operand. A new splitter adds the clobber statement in case
32493 the target operand is dead anyway.
32495 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32497 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32498 to age-old versions of binutils and glibc.
32500 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32502 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32504 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32506 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32508 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32510 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32511 requirement for binutils 2.13.
32513 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32515 * combine.c (try_combine): Delete redundant i1 test. Call
32516 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32518 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32520 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32521 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32523 <riscv64-*-elf>: Re-arrange section
32524 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32525 <riscv32-*-linux>: Likewise.
32526 <riscv64-*-elf>: Likewise
32527 <riscv64-*-linux>: Likewise.
32529 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32532 * aarch64.opt(verbose-cost-dump): Fix typo.
32534 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32537 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32538 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32540 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32542 * reload.c (find_reloads): When reloading a nonoffsettable address,
32543 use RELOAD_OTHER for it and its address reloads.
32545 PR rtl-optimization/79910
32546 * combine.c (record_used_regs): New static function.
32547 (try_combine): Handle situations where there is an additional
32548 instruction between I2 and I3 which needs to have a LOG_LINK
32551 2017-03-17 Jeff Law <law@redhat.com>
32553 PR tree-optimization/71437
32554 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32555 conditional in the hash table first.
32556 (vrp_dom_walker::before_dom_children): Extract condition from
32557 ASSERT_EXPR. Record condition, its inverion and any implied
32558 conditions as well.
32560 2017-03-17 Marek Polacek <polacek@redhat.com>
32561 Markus Trippelsdorf <markus@trippelsdorf.de>
32563 PR tree-optimization/80079
32564 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32567 2017-03-17 Richard Biener <rguenther@suse.de>
32569 PR middle-end/80075
32570 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32571 Properly verify the LHS before the RHS possibly claims to be
32573 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32576 2017-03-17 Martin Jambor <mjambor@suse.cz>
32578 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32579 (List of -O2 options): Likewise.
32580 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32583 2017-03-17 Tom de Vries <tom@codesourcery.com>
32585 * gcov-dump.c (print_usage): Print bug_report_url.
32587 2017-03-17 Richard Biener <rguenther@suse.de>
32589 PR middle-end/80050
32590 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32591 (parser::peek): Likewise.
32593 2017-03-17 Richard Biener <rguenther@suse.de>
32595 PR tree-optimization/80048
32596 * sese.c (free_sese_info): Properly release rename_map and
32597 copied_bb_map elements.
32599 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32601 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32602 Add linked-list forward and backlinks. Insert on
32603 construction, remove on destruction.
32604 (class pass_store_merging): Add m_stores_head field.
32605 (pass_store_merging::terminate_and_process_all_chains):
32606 Iterate over m_stores_head list.
32607 (pass_store_merging::terminate_all_aliasing_chains):
32609 (pass_store_merging::execute): Check for debug stmts first.
32610 Push new chains onto the m_stores_head stack.
32612 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32615 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32616 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32617 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32619 2017-03-16 Jeff Law <law@redhat.com>
32621 PR tree-optimization/71437
32622 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32623 member function. Implementation moved into after_dom_children
32624 member function and into the threader's thread_outgoing_edges
32626 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32627 some code into new thread_outgoing_edges.
32628 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32629 definition. Simplify marker handling (do it here). Assume we always
32630 have the available expression and the const/copies tables.
32631 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32633 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32634 * tree-vrp.c (equiv_stack): No longer file scoped.
32635 (vrp_dom_walker): New class.
32636 (vrp_dom_walker::before_dom_children): New member function.
32637 (vrp_dom_walker::after_dom_children): Likewise.
32638 (identify_jump_threads): Setup domwalker. Use it rather than
32639 walking edges in a random order by hand. Simplify setup/finalization.
32640 (finalize_jump_threads): Remove.
32641 (vrp_finalize): Do not call identify_jump_threads here.
32642 (execute_vrp): Do it here instead and call thread_through_all_blocks
32645 PR tree-optimization/71437
32646 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32648 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32650 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32651 (dom_opt_dom_walker::thread_across_edge): Remove
32652 handle_dominating_asserts argument. All callers changed.
32653 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32654 changes. Remove calls to lhs_of_dominating_assert. Other
32655 uses of handle_dominating_asserts turn into unconditional code
32656 (simplify_control_stmt_condition_1): Likewise.
32657 (simplify_control_stmt_condition): Likewise.
32658 (thread_through_normal_block, thread_across_edge): Likewise.
32659 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32660 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32661 object if it is not an SSA_NAME.
32662 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32663 before calling into the VRP specific simplifiers.
32664 (identify_jump_threads): Remove handle_dominating_asserts
32667 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32670 * tree-diagnostic.c (default_tree_printer): No longer static.
32671 * tree-diagnostic.h (default_tree_printer): New prototype.
32673 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32675 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32676 Change ins into fmov.
32678 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32680 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32681 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32682 Use h_con constraint for operand 1.
32683 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32684 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32686 2017-03-15 Jeff Law <law@redhat.com>
32688 PR tree-optimization/71437
32689 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32690 (record_temporary_equivalences): Use it.
32692 PR tree-optimization/71437
32693 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32694 tree-ssa-scopedtables.
32695 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32696 (record_conditions, record_cond, vuse_eq): Likewise.
32697 (record_edge_info): Adjust to API tweak of record_conditions.
32698 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32699 (record_temporary_equivalences, optimize_stmt): Likewise.
32700 (eliminate_redundant_computations): Likewise.
32701 (record_equivalences_from_stmt): Likewise.
32702 * tree-ssa-scopedtables.c: Include options.h and params.h.
32703 (vuse_eq): New function, moved from tree-ssa-dom.c
32704 (build_and_record_new_cond): Likewise.
32705 (record_conditions): Likewise. Accept vector of conditions rather
32706 than edge_equivalence structure for first argument.
32707 for the first argument.
32708 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32709 from tree-ssa-dom.c.
32710 (avail_exprs_stack::record_cond): Likewise.
32711 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32712 from tree-ssa-dom.c.
32713 (avail_exprs_stack): Add new member functions lookup_avail_expr
32715 (record_conditions): Declare.
32717 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32720 * lra-constraints.c (process_alt_operands): Increase reject for
32721 reloading an input/output operand.
32723 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32726 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32727 insns to convert from signed/unsigned char/short to IEEE 128-bit
32729 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32731 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32734 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32735 subreg of inner mode for values already in registers.
32737 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32739 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32740 iteration reg is used after the loop.
32742 2017-03-14 Martin Sebor <msebor@redhat.com>
32744 PR tree-optimization/79800
32745 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32746 precision in negative-positive range.
32747 (format_floating): Call non-const overload with adjusted precision.
32749 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32752 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32755 2017-03-14 Martin Sebor <msebor@redhat.com>
32757 PR middle-end/80020
32758 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32759 * builtins.def (aligned_alloc): Use it.
32762 * Makefile.in (GTFILES): Add calls.c.
32763 * calls.c: Include "gt-calls.h".
32765 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32767 PR rtl-optimization/79728
32768 * regs.h (struct target_regs): New field
32769 x_contains_allocatable_regs_of_mode.
32770 (contains_allocatable_regs_of_mode): New macro.
32771 * reginfo.c (init_reg_sets_1): Initialize it, and change
32772 contains_reg_of_mode so it includes global regs as well.
32773 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32774 rather than contains_regs_of_mode.
32776 2017-03-14 Martin Liska <mliska@suse.cz>
32778 * doc/invoke.texi: Document options that can't be combined with
32779 -fcheck-pointer-bounds.
32781 2017-03-14 Martin Liska <mliska@suse.cz>
32783 PR middle-end/79831
32784 * doc/invoke.texi (-Wchkp): Document the option.
32786 2017-03-14 Martin Liska <mliska@suse.cz>
32788 * Makefile.in: Install gcov-dump.
32790 2017-03-14 Martin Liska <mliska@suse.cz>
32792 * multiple_target.c (expand_target_clones): Bail out for
32793 an invalid attribute.
32795 2017-03-14 Richard Biener <rguenther@suse.de>
32797 * alias.c (struct alias_set_entry): Pack properly.
32798 * cfgloop.h (struct loop): Likewise.
32799 * cse.c (struct set): Likewise.
32800 * ipa-utils.c (struct searchc_env): Likewise.
32801 * loop-invariant.c (struct invariant): Likewise.
32802 * lra-remat.c (struct cand): Likewise.
32803 * recog.c (struct change_t): Likewise.
32804 * rtl.h (struct address_info): Likewise.
32805 * symbol-summary.h (function_summary): Likewise.
32806 * tree-loop-distribution.c (struct partition): Likewise.
32807 * tree-object-size.c (struct object_size_info): Likewise.
32808 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32809 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32810 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32811 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32812 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32813 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32814 (struct _stmt_vec_info): Likewise.
32816 2017-03-14 Martin Liska <mliska@suse.cz>
32819 * multiple_target.c (create_dispatcher_calls): Check that
32820 a target can create a function dispatcher.
32822 2017-03-14 Martin Liska <mliska@suse.cz>
32825 * multiple_target.c (expand_target_clones): Drop local.local
32826 flag for default implementation.
32828 2017-03-14 Richard Biener <rguenther@suse.de>
32830 PR tree-optimization/80030
32831 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32833 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32835 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32836 gcc_fallthrough() instead of __attribute__((fallthrough));
32838 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32840 * doc/gcc.texi: Remove "up" link to (DIR).
32841 * doc/gccint.texi: Ditto.
32843 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32845 * doc/install.texi (Specific) <avr>: Remove reference to
32848 2017-03-13 Jeff Law <law@redhat.com>
32850 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32851 attribute rather than comments.
32853 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32854 match_scratch operand is highest.
32856 2017-03-13 Martin Liska <mliska@suse.cz>
32858 PR middle-end/78339
32859 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32860 is a CHKP clone, use original declaration.
32862 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32864 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32865 (arc_conditional_register_usage): Use a different allocation order
32866 when optimizing for size.
32867 * common/config/arc/arc-common.c (arc_option_optimization_table):
32868 Section anchors default on when optimizing for size.
32870 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32872 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32874 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32876 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32877 * config/arc/arc.md (cpu_facility): Add cd variant.
32878 (*movqi_insn): Add code density variant.
32879 (*movhi_insn): Likewise.
32880 (*movqi_insn): Likewise.
32881 (*addsi3_mixed): Likewise.
32882 (subsi3_insn): Likewise.
32884 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32886 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32888 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32890 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32891 expressions with MINUS and UNARY ops.
32893 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32896 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32898 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32899 between vec_select and vector argument.
32900 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32901 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32902 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32903 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32904 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32905 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32907 2017-03-13 Richard Biener <rguenther@suse.de>
32910 * params.def (vect-max-peeling-for-alignment): Fix typo.
32912 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32914 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32915 issue that only occurred with binutils below 2.18.
32917 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32919 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32920 refer to binutils 2.11/2.12 minimum.
32922 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32924 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
32925 ftp.kernel.org and simplify binutils requirement.
32927 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
32929 * doc/invoke.texi (Warning Options): Fix spelling of link-time
32931 (Optimize Options): Ditto. Also remove redundancy.
32933 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32935 PR translation/79848
32936 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
32938 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
32939 to G_ to avoid double translation.
32941 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32943 PR translation/79923
32944 * auto-profile.c (get_combined_location): Convert leading
32945 character of diagnostics to lower case and remove trailing period.
32946 (read_profile): Likewise for various diagnostics.
32947 * config/arm/arm.c (arm_option_override): Remove trailing period
32948 from various diagnostics.
32949 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
32950 (msp430_expand_delay_cycles): Likewise.
32952 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32955 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
32956 full command-line argument, rather than just "str".
32957 (aarch64_validate_march): Likewise.
32958 (aarch64_validate_mtune): Likewise.
32960 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
32962 PR rtl-optimization/78911
32963 * lra-assigns.c (must_not_spill_p): New function.
32964 (spill_for): Use it.
32966 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32968 PR tree-optimization/79981
32969 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
32970 ATOMIC_COMPARE_EXCHANGE ifn result.
32971 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
32972 IFN_ATOMIC_COMPARE_EXCHANGE.
32974 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32977 * opts.c (parse_sanitizer_options): Add missing question mark to
32978 "did you mean" message.
32980 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32982 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
32984 (VMULEUH_UNS): Likewise.
32985 (VMULOUB_UNS): Likewise.
32986 (VMULOUH_UNS): Likewise.
32987 * config/rs6000/rs6000.c (builtin_function_type): Remove
32988 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
32990 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32993 * read-rtl-function.c (function_reader::read_rtx_operand): Update
32994 x with result of extra_parsing_for_operand_code_0.
32995 (function_reader::extra_parsing_for_operand_code_0): Convert
32996 return type from void to rtx, returning x. When reading
32997 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
32998 larger size containing struct block_symbol.
33000 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33002 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33003 -mfloat128-hardware without -m64.
33005 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33008 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33009 entries to the case statement that marks unsigned arguments to
33010 overloaded functions.
33012 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33014 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33015 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33017 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33020 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33021 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33023 2017-03-10 Martin Liska <mliska@suse.cz>
33027 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33028 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33031 2017-03-10 Olivier Hainque <hainque@adacore.com>
33033 * tree-switch-conversion (array_value_type): Start by resetting
33034 candidate type to it's main variant.
33036 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33038 PR rtl-optimization/79909
33039 * combine.c (try_combine): Use simplify_replace_rtx on individual
33040 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33041 of the whole CALL_INSN_FUNCTION_USAGE.
33043 PR tree-optimization/79972
33044 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33045 get_range_info on SSA_NAMEs. Formatting fixes.
33047 2017-03-10 Richard Biener <rguenther@suse.de>
33048 Jakub Jelinek <jakub@redhat.com>
33050 PR tree-optimization/77975
33051 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33052 edge to be constant.
33053 (get_val_for): For constant x return it. Formatting fix.
33054 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33055 would use the same bases as the current one.
33057 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33059 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33060 instead of vec_select for V1TImode.
33061 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33063 (VSX_LE_128): Add V1TI to this mode iterator.
33064 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33065 (*vsx_le_perm_store_<mode>): Likewise.
33066 (pre-reload splitter for VSX stores): Likewise.
33067 (post-reload splitter for VSX stores): Likewise.
33068 (*vsx_xxpermdi2_le_<mode>): Likewise.
33069 (*vsx_lxvd2x2_le_<mode>): Likewise.
33070 (*vsx_stxvd2x2_le_<mode>): Likewise.
33072 2017-03-09 Michael Eager <eager@eagercon.com>
33074 Correct failures with --enable-checking=yes,rtl.
33076 * config/microblaze/microblaze.c (microblaze_expand_shift):
33077 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33078 test for const0_rtx.
33079 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33080 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33082 2017-03-09 Richard Biener <rguenther@suse.de>
33084 PR tree-optimization/79977
33085 * graphite-scop-detection.c (scop_detection::merge_sese):
33086 Handle the case of extra exits to blocks dominating the entry.
33088 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33090 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33093 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33095 PR rtl-optimization/79949
33096 * lra-constraints.c (process_alt_operands): Check memory when
33097 trying to predict a cycle. Print about the overall increase.
33099 2017-03-09 Richard Biener <rguenther@suse.de>
33101 PR middle-end/79971
33102 * gimple-expr.c (useless_type_conversion_p): Preserve
33103 TYPE_SATURATING for fixed-point types.
33105 2017-03-09 Richard Biener <rguenther@suse.de>
33108 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33109 alignment of BLKmode params.
33111 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33114 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33115 (VALL_NO_V2Q): Likewise.
33117 * config/aarch64/aarch64-simd.md
33118 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33120 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33121 VALL_NO_V2Q mode iterator.
33122 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33124 2017-03-09 Martin Liska <mliska@suse.cz>
33126 PR tree-optimization/79631
33127 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33128 tree_int_cst_sign_bit just for INTEGER constants.
33130 2017-03-09 Martin Liska <mliska@suse.cz>
33134 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33137 2017-03-09 Marek Polacek <polacek@redhat.com>
33140 * tree.c (inchash::add_expr): Handle TREE_VEC.
33142 2017-03-09 Martin Liska <mliska@suse.cz>
33145 (chkp_narrow_size_and_offset): New function.
33146 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33147 (void chkp_parse_bit_field_ref): New function.
33148 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33149 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33151 2017-03-09 Martin Liska <mliska@suse.cz>
33154 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33155 (chkp_find_bounds_1): Remove gcc_unreachable.
33157 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33160 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33161 BUILT_IN_SYNC*, determine the access type from the size suffix and
33162 always build a MEM_REF with that type. Handle forgotten
33163 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33166 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33167 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33168 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33169 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33170 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33171 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33172 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33173 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33174 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33175 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33176 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33177 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33178 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33179 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33180 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33181 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33182 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33183 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33184 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33185 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33186 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33187 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33188 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33189 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33190 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33191 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33192 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33193 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33194 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33195 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33196 definitions outside of __OPTIMIZE__ guarded section.
33199 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33200 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33201 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33202 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33205 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33207 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33208 ("vfenez<mode>"): Add missing constraints.
33210 2017-03-08 Martin Sebor <msebor@redhat.com>
33213 * config/nds32/nds32.c (nds32_option_override):
33214 Fix misspelled diagnostic.
33216 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33219 * gimplify.c (gimplify_omp_for): Replace index var in outer
33220 taskloop statement with an artificial variable and add
33221 OMP_CLAUSE_PRIVATE clause for it.
33223 2017-03-08 Richard Biener <rguenther@suse.de>
33225 PR tree-optimization/79955
33226 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33227 for accesses that are completely outside of the variable.
33229 2017-03-08 Andrew Haley <aph@redhat.com>
33231 PR tree-optimization/79943
33232 * tree-ssa-loop-split.c (compute_new_first_bound): When
33233 calculating the new upper bound, (END-BEG) should be added, not
33236 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33238 * config/avr/avr.md (setmemhi): Make sure match_dup
33239 operand number comes before match_scratch.
33241 2017-03-08 Richard Biener <rguenther@suse.de>
33243 PR tree-optimization/79920
33244 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33245 with ncopies == 1 to ...
33246 (vect_transform_slp_perm_load): ... here. Properly compute
33247 all element loads by iterating VF times over the group. Do
33248 not handle ncopies (computed in a broken way) in
33249 vect_create_mask_and_perm.
33251 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33254 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33255 is a uniform vector, use uniform_vector_p return value instead of
33256 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33258 2017-03-07 Marek Polacek <polacek@redhat.com>
33260 PR middle-end/79809
33261 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33262 (alloca_call_type): Likewise.
33264 2017-03-07 Martin Liska <mliska@suse.cz>
33266 * gcov.c (process_args): Put comment to correct location.
33268 2017-03-07 Martin Liska <mliska@suse.cz>
33270 PR middle-end/68270
33271 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33272 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33273 (chkp_narrow_bounds_for_field): Likewise.
33274 (chkp_parse_array_and_component_ref): Pass one more argument to
33277 2017-03-07 Richard Biener <rguenther@suse.de>
33279 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33282 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33284 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33285 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33287 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33290 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33291 to end of description.
33292 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33294 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33296 PR rtl-optimization/79901
33297 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33299 (*avx512f_<code><mode>3<mask_name>): ... this.
33300 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33301 iterator instead of VI8_AVX2_AVX512BW.
33303 PR rtl-optimization/79901
33304 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33305 min/max expander, expand it using expand_vec_cond_expr.
33308 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33311 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33314 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33315 to void * for PCH reasons.
33316 * dwarf2out.c (output_loc_operands, output_die): Cast
33317 v.val_vec.array to unsigned char *.
33319 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33322 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33325 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33327 PR rtl-optimization/79571
33328 * lra-constraints.c (process_alt_operands): Calculate static
33329 reject and subtract it from overall when only addresses will be
33332 2017-03-06 Julia Koval <julia.koval@intel.com>
33335 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33336 incoming stack boundary to 128 for 64-bit targets.
33338 2017-03-06 Richard Biener <rguenther@suse.de>
33340 PR tree-optimization/79894
33341 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33342 to NULL after folding it.
33344 2017-03-06 Richard Biener <rguenther@suse.de>
33346 PR tree-optimization/79824
33347 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33348 check disabling peeling for gaps.
33350 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33352 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33353 attributes): Document gettimeofday.
33355 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33357 * config/s390/s390.c (s390_option_override_internal): Set
33358 PARAM_MIN_VECT_LOOP_BOUND
33360 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33362 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33363 * config/s390/s390.md: Likewise.
33365 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33368 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33369 (<avx2_avx512>_perm<mode>): Rename to ...
33370 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33372 (<avx512>_perm<mode>_mask): Rename to ...
33373 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33375 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33376 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33377 instead of VI8F_256_512.
33378 (avx512f_perm<mode>): New define_expand.
33379 (avx512f_perm<mode>_mask): Likewise.
33380 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33381 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33383 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33385 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33386 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33388 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33390 2017-03-06 Martin Liska <mliska@suse.cz>
33393 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33394 when having a SSA NAME w/o VAR_DECL assigned to it.
33396 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33398 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33399 msa_dpsub_<su>_d): Fix MODE for vec_select.
33401 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33403 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33405 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33407 2017-03-06 Richard Biener <rguenther@suse.de>
33409 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33410 * plugin.c (register_plugin_info): Likewise.
33411 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33413 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33415 * config/i386/sse.md (sse_storehps, sse_storelps,
33416 avx_<castmode><avxsizesuffix>_<castmode>,
33417 avx512f_<castmode><avxsizesuffix>_<castmode>,
33418 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33419 in condition that at least one operand is not a MEM.
33421 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33423 PR middle-end/79805
33424 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33425 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33427 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33428 gimple_call_nothrow_p flag based on whether original builtin can throw.
33429 If it can, emit following stmts on the fallthrough edge.
33430 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33431 don't create new bb if inserting just debug stmts on the edge, try to
33432 insert them on the fallthru bb or just reset debug stmts.
33434 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33437 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33438 restore recog_data (including the operand rtxes inside it) around
33439 the call to get_insn_template.
33441 2017-03-03 Martin Sebor <msebor@redhat.com>
33443 PR tree-optimization/79699
33444 * context.c (context::~context): Free MPFR caches to avoid
33445 a memory leak on program exit.
33447 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33449 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33450 Use wide_int::ulow () instead of .elt (0).
33452 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33454 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33455 (*pushxf): Limit oF constraint to 32bit targets and add oC
33456 constraint for 64bit targets.
33457 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33458 (*pushdf): Change rmF constraint to rmC.
33460 2017-03-03 Martin Liska <mliska@suse.cz>
33462 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33463 Remove unused variable.
33465 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33468 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33469 is a memory operand, increase num_memory.
33470 (ix86_expand_args_builtin): Likewise.
33472 2017-03-03 Jan Hubicka <jh@suse.cz>
33475 * ipa-devirt.c (maybe_record_node): Properly handle
33476 __cxa_pure_virtual visibility.
33478 2017-03-03 Martin Liska <mliska@suse.cz>
33480 PR tree-optimization/79803
33481 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33483 (pass_loop_prefetch::execute): Disabled optimization if an
33484 assumption about L1 cache size is not met.
33486 2017-03-03 Martin Liska <mliska@suse.cz>
33488 PR rtl-optimization/79574
33489 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33490 (hash_scan_set): Likewise.
33491 (dump_hash_table): Likewise.
33492 (hoist_code): Likewise.
33494 2017-03-03 Richard Biener <rguenther@suse.de>
33496 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33497 in place of uhigh ().
33498 (fixed_convert_from_real): Likewise.
33500 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33503 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33505 2017-03-03 Richard Biener <rguenther@suse.de>
33507 PR middle-end/79818
33508 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33509 TYPE_OVERFLOW_UNDEFINED check.
33511 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33513 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33515 (vector_ae_<mode>_p): Likewise.
33516 (vector_nez_<mode>_p): Likewise.
33517 (vector_ne_v2di_p): Likewise.
33518 (vector_ae_v2di_p): Likewise.
33519 (vector_ne_<mode>_p): Likewise.
33520 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33522 (vsx_tsqrt<mode>2_fe): Likewise.
33524 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33527 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33529 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33531 PR rtl-optimization/79780
33532 * cprop.c (one_cprop_pass): When second and further conditional trap
33533 in a single basic block is turned into an unconditional trap, turn it
33534 into a deleted note to avoid RTL verification failures.
33536 2017-03-02 Richard Biener <rguenther@suse.de>
33538 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33540 2017-03-02 Richard Biener <rguenther@suse.de>
33542 PR tree-optimization/79345
33544 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33545 param and abort the walk, returning -1 if it is hit.
33546 (walk_aliased_vdefs): Take a limit param and pass it on.
33547 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33548 defaulting to 0 and return a signed int.
33549 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33550 (check_defs): New helper.
33551 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33552 about uninitialized memory.
33553 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33554 bogus uninitialized warning.
33555 (fixed_convert_from_real): Likewise.
33557 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33559 PR tree-optimization/66768
33560 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33561 iv_use if base object can't be determined.
33563 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33565 PR tree-optimization/79345
33566 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33567 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33568 (get_pattern_stats): Initialize it.
33569 * genemit.c (gen_expand): Verify match_scratch numbers come after
33570 match_operand/match_dup numbers.
33571 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33572 match_scratch numbers.
33573 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33575 * config/s390/s390.md (trunctdsd2): Likewise.
33577 2017-03-02 Richard Biener <rguenther@suse.de>
33579 * wide-int.h (wide_int_storage::operator=): Implement in terms
33582 2017-03-02 Richard Biener <rguenther@suse.de>
33584 PR tree-optimization/79777
33585 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33586 the to insert expression to sth existing.
33588 2017-03-01 Martin Sebor <msebor@redhat.com>
33590 PR middle-end/79692
33591 * gimple-ssa-sprintf.c
33592 (directive::known_width_and_precision): New function.
33593 (format_integer): Use it.
33594 (get_mpfr_format_length): Consider the full range of precision
33595 when computing %g output with the # flag. Set the likely byte
33596 count to 3 rather than 1 when precision is indeterminate.
33597 (format_floating): Correct the lower bound of precision.
33599 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33601 * doc/invoke.texi: Document default code model for 64-bit Linux.
33603 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33606 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33607 udiv rather than div since input pattern is unsigned.
33609 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33611 * config/i386/i386.c (print_reg): Warn for values of
33612 unsupported size in integer register.
33614 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33617 * config/rs6000/predicates.md (current_file_function_operand): Do
33618 not allow self calls to be local if the function is replaceable.
33620 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33623 * config/rs6000/altivec.h (vec_ctz and others): Change the
33624 preprocessor macro that controls conditional compilation from
33625 _ARCH_PWR9 to __POWER9_VECTOR__.
33626 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33627 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33628 control (instead of _ARCH_PWR9 control) so that template
33629 definition uses power9-specific function.
33630 (vec_any_eq): Likewise.
33631 (vec_all_ne): Change macro definition to use a power9-specific
33632 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33633 _ARCH_PWR9 control).
33634 (vec_any_eq) Likewise.
33635 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33636 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33637 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33638 support for xvcmpnedp instruction.
33639 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33640 macro expansion so that Power9 implementation of vec_all_ne does
33641 not use the AltiVec predicate framework.
33642 (VCMPNEH_P): Likewise.
33643 (VCMPNEW_P): Likewise.
33644 (VCMPNED_P): Likewise.
33645 (VCMPNEFP_P): Likewise.
33646 (VCMPNEDP_P): Likewise.
33647 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33648 implementation of vec_any_eq to not use AltiVec predicate
33650 (VCMPAEH_P): Likewise.
33651 (VCMPAEW_P): Likewise.
33652 (VCMPAED_P): Likewise.
33653 (VCMPAEFP_P): Likewise.
33654 (VCMPAEDP_P): Likewise.
33655 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33656 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33657 not use the AltiVec predicate framework.
33658 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33659 of vec_any_eq to not use AltiVec predicate framework.
33660 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33661 support for predefined __POWER9_VECTOR__ macro to indicate that
33662 Power9 instruction selection is enabled.
33663 (altivec_overloaded_builtins): Remove extraneous
33664 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33665 function argument types RS6000_BTI_bool_V16QI and
33666 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33667 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33668 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33669 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33670 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33671 Power9 for implementations of vec_cmpne. Change the signature for
33672 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33673 (representing vec_all_ne) to remove the previously described first
33674 argument of type RS6000_BTI_INTSI, as this was an artifact of
33675 reliance on the AltiVec predicate framework, which is no longer
33676 used in the implementation of these functions. Add
33677 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33678 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33679 since, unlike the AltiVec predicate framework implementation, we
33680 do not share function descriptors between vec_alle and vec_anyeq.
33681 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33682 set of modes that receive special treatment even when
33683 TARGET_P9_VECTOR is true. The special treatment emits code that
33684 does not depend on Power9 instructions.
33685 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33686 define_expand to not rely on AltiVec predicate framework.
33687 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33689 (vector_ne_v2di_p): Change this define_expand to not rely on
33690 AltiVec predicate framework.
33691 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33693 (vector_ne_<mode>_p): Change this define_expand to not rely on
33694 AltiVec predicate framework.
33695 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33697 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33698 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33699 define_insn pattern.
33700 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33701 define_insn pattern because the xvcmpne<VSs>. instruction is not
33703 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33704 instruction is not supported.
33706 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33708 * config/nvptx/nvptx.c: Include intl.h.
33710 2017-03-01 Martin Jambor <mjambor@suse.cz>
33713 * ipa-prop.h (ipa_bits): Removed field known.
33714 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33715 to pointers. Adjusted their comments to warn about their sharing.
33716 (ipcp_transformation_summary): Change bits to a vector of pointers.
33717 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33718 (ipa_get_ipa_bits_for_value): Declare.
33719 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33720 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33721 (ipa_bits_hash_table): Likewise.
33722 (ipa_vr_ggc_hash_traits): Likewise.
33723 (ipa_vr_hash_table): Likewise.
33724 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33725 being pointers and vr_known being removed.
33726 (ipa_set_jf_unknown): Likewise.
33727 (ipa_get_ipa_bits_for_value): New function.
33728 (ipa_set_jfunc_bits): Likewise.
33729 (ipa_get_value_range): New overloaded functions.
33730 (ipa_set_jfunc_vr): Likewise.
33731 (ipa_compute_jump_functions_for_edge): Use the above functions to
33732 construct bits and vr parts of jump functions.
33733 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33734 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33736 (ipcp_grow_transformations_if_necessary): Also allocate
33737 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33739 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33740 them. Fix too long lines.
33741 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33742 vr_known being removed.
33743 (ipa_read_jump_function): Use new setter functions to construct bits
33744 and vr parts of jump functions or set them to NULL.
33745 (write_ipcp_transformation_info): Adjust for bits being pointers.
33746 (read_ipcp_transformation_info): Likewise.
33747 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33749 Include gt-ipa-prop.h.
33750 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33752 (ipcp_store_bits_results): Likewise.
33753 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33754 Do not write to existing jump functions but use a temporary instead.
33756 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33759 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33760 attempt to use its first operand as BIT_FIELD_REF base.
33762 2017-03-01 Richard Biener <rguenther@suse.de>
33764 PR middle-end/79721
33765 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33766 interpolating formula in wrapping arithmetic.
33767 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33769 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33771 PR tree-optimization/79734
33772 * tree-vect-generic.c (expand_vector_condition): Optimize
33773 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33774 Handle VEC_COND_EXPR where comparison has different inner width from
33775 type's inner width.
33777 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33779 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33780 markup, and similar issues. Remove @opindex entries for things
33781 that aren't options. Add missing -mmpy-option entries.
33783 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33785 PR tree-optimization/79737
33786 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33787 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33788 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33789 instead of byte_size. Formatting fix.
33790 (shift_bytes_in_array_right): Formatting fix.
33792 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33795 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33796 condition on optimize for the leaf function test.
33798 2017-02-28 Martin Liska <mliska@suse.cz>
33801 * read-rtl-function.c (function_reader::handle_unknown_directive):
33802 Bail out when one uses -flto.
33804 2017-02-28 Martin Liska <mliska@suse.cz>
33806 * common.opt: Replace space with tabular for options of <number>
33808 * config/i386/i386.opt: Show <number> value for
33809 -mlarge-data-threshold.
33810 * opts.c (print_filtered_help): Do not display number in hexadecimal
33813 2017-02-28 Martin Liska <mliska@suse.cz>
33815 * common.opt: Fix --help=option -Q for options which are of
33818 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33820 * config/i386/i386.c (print_reg): Error out for values
33821 of 8-bit size in invalid integer register.
33823 2017-02-28 Martin Sebor <msebor@redhat.com>
33825 PR tree-optimization/79691
33826 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33828 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33831 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33832 gcc_unreachable with output_operand_lossage.
33834 2017-02-28 Richard Biener <rguenther@suse.de>
33836 PR tree-optimization/79740
33837 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33839 (visit_nary_op): Insert the nary into the hashtable if we
33840 pattern-matched sth.
33841 * tree-ssa-pre.c (eliminate_insert): Robustify.
33843 2017-02-28 Richard Biener <rguenther@suse.de>
33845 PR middle-end/79731
33846 * fold-const.c (decode_field_reference): Reject out-of-bound
33849 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33851 * config/i386/i386.c: Include intl.h.
33852 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33853 instead of just cond ? "..." : "...".
33854 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33855 * coverage.c (read_counts_file): Likewise.
33856 * omp-offload.c: Include intl.h.
33857 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33858 of just cond ? "..." : "...".
33859 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33860 of just cond ? "..." : "...".
33862 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33865 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33867 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33868 'tune for' CPU name.
33869 * config/arm/arm-cpu-data.h: Regenerated.
33871 2017-02-28 Richard Biener <rguenther@suse.de>
33873 PR tree-optimization/79732
33874 * tree-inline.c (expand_call_inline): Do not shadow var.
33876 2017-02-28 Richard Biener <rguenther@suse.de>
33878 PR tree-optimization/79723
33879 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33880 address-space properly.
33882 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33884 * doc/optinfo.texi (Optimization groups): Fix option used for
33886 * doc/invoke.texi (-fopt-info): Document "omp".
33887 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33888 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33889 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33890 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33891 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33893 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33895 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33896 OPTGROUP_OMP with "omp".
33898 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33901 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33902 for arithmetic shift of unsigned V2DI.
33904 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33906 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33907 arc/linux.h headers.
33908 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33909 (LINK_SPEC): Likewise.
33910 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33911 (EXTRA_SPECS): Likewise.
33912 (STARTFILE_SPEC): Likewise.
33913 (ENDFILE_SPEC): Likewise.
33914 (LIB_SPEC): Likewise.
33915 (TARGET_SDATA_DEFAULT): Likewise.
33916 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33917 (MULTILIB_DEFAULTS): Likewise.
33918 (DWARF2_UNWIND_INFO): Likewise.
33919 * config/arc/big.h: New file.
33920 * config/arc/elf.h: Likewise.
33921 * config/arc/linux.h: Likewise.
33922 * config/arc/t-uClibc: Remove.
33924 2017-02-27 Bin Cheng <bin.cheng@arm.com>
33926 PR tree-optimization/77536
33927 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
33928 (tree_transform_and_unroll_loop): Use above function to compute the
33929 estimated niter of unrolled loop and use it when scaling profile.
33930 Also use count info rather than frequency if it's non-zero.
33931 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
33932 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
33933 (vect_transform_loop): Call above function.
33935 2017-02-27 Richard Biener <rguenther@suse.de>
33937 PR tree-optimization/45397
33938 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
33939 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
33940 (visit_nary_op): Add pattern matching for CSEing sign-changed
33941 or truncated operations with wider ones.
33943 2017-02-27 Richard Biener <rguenther@suse.de>
33945 PR tree-optimization/79690
33946 * tree-vect-stmts.c (vectorizable_store): Use vector type
33947 built from the DR with address-space.
33949 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
33951 * doc/invoke.texi (Optimize Options): Refine the description
33952 of asan-use-after-return.
33954 2017-02-25 Alan Modra <amodra@gmail.com>
33956 PR rtl-optimization/79584
33957 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
33958 base, not ad->base_term, the reg within base. Remove assertion
33959 that ad->base == ad->base_term. Replace gen_int_mode using
33960 bogus mode with const0_rtx.
33962 2017-02-25 Jakub Jelinek <jakub@redhat.com>
33964 PR middle-end/79396
33965 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
33966 FMA_EXPR like tcc_binary or tcc_unary.
33968 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
33971 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
33973 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
33974 (output_loc_operands): Handle DW_OP_call_ref and
33975 DW_OP_GNU_variable_value.
33976 (struct variable_value_struct): New type.
33977 (struct variable_value_hasher): Likewise.
33978 (variable_value_hash): New variable.
33979 (string_types): Remove.
33980 (copy_loc_descr): New function.
33981 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
33982 (prepend_loc_descr_to_each): New function.
33983 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
33984 instead of add_loc_descr_to_each if the first argument is single
33985 location list and the second has multiple.
33986 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
33987 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
33988 when looking for variable value which doesn't have other location info.
33989 (loc_list_from_tree): Formatting fix.
33990 (gen_array_type_die): Simplify DW_AT_string_length handling.
33991 (adjust_string_types): Remove.
33992 (gen_subprogram_die): Don't call adjust_string_types nor test/set
33993 string_types. Call resolve_variable_values.
33994 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
33995 (resolve_addr_in_expr): Likewise. Add A argument.
33996 (copy_deref_exprloc): Remove deref argument. Adjust for the
33997 original expression being DW_OP_GNU_variable_value with optionally
33998 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
33999 optionally after it.
34000 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34001 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34002 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34003 (variable_value_hasher::hash, variable_value_hasher::equal): New
34005 (resolve_variable_value_in_expr, resolve_variable_value,
34006 resolve_variable_values, note_variable_value_in_expr,
34007 note_variable_value): New functions.
34008 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34011 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34014 * opts.h (handle_generated_option): Add GENERATED_P argument.
34015 * opts-common.c (handle_option): Adjust function comment.
34016 (handle_generated_option): Add GENERATED_P argument, pass it to
34018 (control_warning_option): Pass false to handle_generated_option
34020 * opts.c (maybe_default_option): Pass true to handle_generated_option
34022 * optc-gen.awk: Likewise.
34024 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34026 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34027 a REG, look at the REG it is a SUBREG of.
34028 (splitter for cmpeqsi_t): Ditto.
34030 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34032 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34033 the special USEs with the pattern of the insn, not the insn itself.
34035 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34038 * doc/invoke.texi: Document -mload-store-pairs.
34040 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34041 Sandra Loosemore <sandra@codesourcery.com>
34043 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34044 argument isn't a CONST_INT.
34045 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34046 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34047 (nios2_validate_compare): Bypass alternate compare logic if *op2
34048 is not a CONST_INT.
34049 (ldstwm_operation_p): Return false if first_base is not a REG or
34050 if first_offset is not a CONST_INT.
34052 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34054 * config/cris/cris.md: Use correct operand in a define_peephole2.
34056 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34058 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34060 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34062 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34063 this_insn if it is an INSN or JUMP_INSN.
34064 (force_offsettable): Look at base, not at addr.
34065 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34066 on things that aren't necessarily CONST_INTs.
34068 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34070 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34071 -mfpmath=sse is the default also for x86-32 targets with SSE2
34072 instruction set when @option{-ffast-math} is enabled
34074 2017-02-24 Jeff Law <law@redhat.com>
34076 PR rtl-optimizatoin/79286
34077 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34080 2017-02-24 Richard Biener <rguenther@suse.de>
34082 PR tree-optimization/79389
34083 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34086 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34088 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34089 function comment to reflect reality.
34090 (loop_exits_before_overflow): Fix typo in function description.
34092 2017-02-24 Richard Biener <rguenther@suse.de>
34094 PR tree-optimization/79389
34095 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34096 properly that a threading opportunity exists. Detect conditional
34097 copy/constant propagation opportunities.
34099 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34101 * config/visium/visium.md (type): Add trap.
34102 (b): New mode attribute.
34103 (*btst): Rename into...
34104 (*btst<mode>): ...this and adjust.
34105 (*cbranchsi4_btst_insn): Rename into...
34106 (*cbranch<mode>4_btst_insn): ...this and adjust.
34107 (trap): New define_insn.
34109 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34111 PR tree-optimization/79389
34112 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34113 (noce_reversed_cond_code): New function.
34114 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34115 reversed_comparison_code. Formatting fix.
34116 (noce_try_store_flag): Test rev_cond != NULL in addition to
34117 reversed_comparison_code.
34118 (noce_try_store_flag_constants): Likewise.
34119 (noce_try_store_flag_mask): Likewise.
34120 (noce_try_addcc): Use rev_cond if non-NULL instead of
34121 reversed_comparison_code.
34122 (noce_try_cmove_arith): Likewise. Formatting fixes.
34123 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34124 (noce_find_if_block): Initialize rev_cond.
34125 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34126 instead of false as last argument never attempt to reverse it
34129 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34131 PR tree-optimization/79663
34132 * tree-predcom.c (combine_chains): Process refs in reverse order
34133 only for ZERO length chains, and add explaining comment.
34135 2017-02-23 Jeff Law <law@redhat.com>
34137 PR tree-optimization/79578
34138 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34139 in call to operand_equal_p.
34141 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34144 * config/i386/cpuid.h: Fix another undefined behavior.
34146 2017-02-23 Richard Biener <rguenther@suse.de>
34148 PR tree-optimization/79683
34149 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34150 vector types for data-refs.
34152 2017-02-23 Martin Liska <mliska@suse.cz>
34154 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34156 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34158 PR middle-end/79665
34159 * internal-fn.c (get_range_pos_neg): Moved to ...
34160 * tree.c (get_range_pos_neg): ... here. No longer static.
34161 * tree.h (get_range_pos_neg): New prototype.
34162 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34163 are known to be in between 0 and signed maximum inclusive, try to
34164 expand both unsigned and signed divmod and use the cheaper one from
34167 2017-02-22 Jeff Law <law@redhat.com>
34169 PR tree-optimization/79578
34170 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34171 to compare base operands.
34173 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34176 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34177 gpc_reg_operand instead of fpr_reg_operand.
34179 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34181 * config/mips/mips.c (mips_return_in_memory): Force FP
34182 vector types to be returned in memory for o32 ABI.
34184 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34186 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34187 instead of DW_TAG_member for static data member declarations and don't
34188 set no_linkage_name for static inline data members.
34189 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34192 2017-02-22 Martin Liska <mliska@suse.cz>
34194 * doc/invoke.texi: Replace inequality signs with square brackets
34197 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34200 * lra-constraints.c (simplify_operand_subreg): Handle
34201 WORD_REGISTER_OPERATIONS targets.
34203 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34206 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34207 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34208 elimination by swapping fld*.
34210 2017-02-22 Richard Biener <rguenther@suse.de>
34212 PR tree-optimization/79673
34213 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34214 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34215 irrelevant address-space qualifiers and avoiding a
34216 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34218 2017-02-22 Richard Biener <rguenther@suse.de>
34220 PR tree-optimization/79666
34221 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34222 to not symbolically negate if that may introduce undefined
34225 2017-02-22 Martin Liska <mliska@suse.cz>
34228 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34229 * data-streamer-out.c (streamer_write_gcov_count_stream):
34231 * value-prof.c (stream_out_histogram_value): Make assert more
34232 precise based on type of counter.
34234 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34237 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34238 Use nonimmediate_operand instead of memory_operand for operand 1.
34239 (float-extend standard_x87sse_constant_load splitter): Ditto.
34241 2017-02-21 Jeff Law <law@redhat.com>
34243 PR tree-optimization/79621
34244 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34245 blocks with edges to themselves.
34247 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34250 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34251 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34252 Use gimple_call_builtin_p.
34255 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34256 on temporarily removed DEBUG_INSNs.
34258 PR tree-optimization/79649
34259 * tree-loop-distribution.c (classify_partition): Give up on
34260 non-generic address space loads/stores.
34262 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34264 * doc/loop.texi (Loop manipulation): Remove nonexistent
34265 tree_ssa_loop_version from the documentation.
34266 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34268 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34271 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34272 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34273 * config/rs6000/rs6000.c: Include except.h.
34274 (rs6000_expand_split_stack_prologue): Call
34275 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34277 2017-02-21 Martin Jambor <mjambor@suse.cz>
34280 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34281 have been analyzed.
34283 2017-02-21 Martin Jambor <mjambor@suse.cz>
34285 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34286 for backward compatibility only.
34287 * doc/invoke.texi (Option Summary): Remove all references to
34288 -fipa-cp-alignment.
34290 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34294 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34296 * lra-constraints.c (curr_insn_transform): Handle
34297 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34299 2017-02-21 Martin Liska <mliska@suse.cz>
34301 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34303 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34305 PR translation/79638
34306 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34308 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34311 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34312 (arm_function_ok_for_sibcall): Return false for an indirect call by
34313 descriptor if all the argument registers are used.
34314 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34315 alignment of the function.
34317 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34319 PR tree-optimization/61441
34320 * simplify-rtx.c (simplify_const_unary_operation): For
34321 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34322 the sNaN unmodified.
34324 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34326 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34327 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34328 instead of SYSTEM_HEADER_DIR.
34330 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34331 Martin Liška <mliska@suse.cz>
34333 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34334 Fix typos and grammar, use active voice, and clarify.
34336 2017-02-20 Marek Polacek <polacek@redhat.com>
34338 PR middle-end/79537
34339 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34342 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34344 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34347 * config/i386/i386.c (ix86_expand_builtin): Handle
34348 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34349 ix86_builtins_isa[fcode].isa as a requirement of those
34350 flags and any other flag in the bitmask.
34351 (ix86_init_mmx_sse_builtins): Use 0 instead of
34352 ~OPTION_MASK_ISA_64BIT as mask.
34353 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34354 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34355 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34356 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34358 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34361 * lra-constraints.c (split_reg): Check requested split mode
34362 is supported by the register.
34364 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34366 * lra-constraints.c (simplify_operand_subreg): Remove early
34369 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34372 * lra-constraints.c (curr_insn_transform): Tighten condition
34373 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34375 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34378 * lra-constraints.c (curr_insn_transform): Handle
34379 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34381 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34384 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34386 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34388 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34391 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34394 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34396 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34399 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34401 * final.c (last_columnnum, override_columnnum): New variables.
34402 (final_start_function): Set last_columnnum, pass it to begin_prologue
34403 hook and pass 0 to dwarf2out_begin_prologue.
34404 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34405 to source_line debug hook.
34406 (notice_source_line): Compute last_columnnum and for debug_column_info
34407 return true on column changes.
34408 * debug.h (struct gcc_debug_hooks): Add column argument to
34409 source_line and begin_prologue hooks.
34410 (debug_nothing_int_charstar_int_bool): Remove prototype.
34411 (debug_nothing_int_int_charstar,
34412 debug_nothing_int_int_charstar_int_bool): New prototypes.
34413 (dwarf2out_begin_prologue): Add column argument.
34414 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34415 begin_prologue hooks.
34416 (debug_nothing_int_charstar_int_bool): Remove.
34417 (debug_nothing_int_int_charstar,
34418 debug_nothing_int_int_charstar_int_bool): New functions.
34419 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34420 through to dwarf2out_source_line.
34421 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34422 (dwarf2out_source_line): Add column argument, emit it if requested.
34423 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34425 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34426 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34427 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34428 through to dwarf2out_begin_prologue.
34429 (vmsdbgout_source_line): Add column argument, pass it through to
34430 dwarf2out_source_line.
34431 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34432 dbxout_source_line caller.
34433 (dbxout_source_line): Add column argument.
34435 * common.opt (gno-column-info, gcolumn-info): New options.
34436 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34437 (check_die): Also test for multiple DW_AT_decl_column attributes.
34438 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34439 DW_AT_decl_column if requested.
34440 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34442 (gen_variable_die): Likewise.
34443 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34444 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34447 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34448 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34449 (ix86_handle_option): Handle OPT_m3dnowa.
34450 * doc/invoke.texi (-m3dnowa): Document.
34451 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34452 -m3dnowa instead of -m3dnow -march=athlon.
34455 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34456 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34458 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34461 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34462 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34463 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34464 generator for vsx_xxpermdi_<mode>_be.
34465 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34466 force big-endian semantics.
34467 (vsx_xxpermdi_<mode>_be): New define_expand with same
34468 implementation as previous version of vsx_xxpermdi_<mode>.
34470 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34472 PR tree-optimization/79327
34473 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34474 variable, its initialization and use.
34476 2017-02-17 Julia Koval <julia.koval@intel.com>
34478 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34479 (OPTION_MASK_ISA_PKU_UNSET): New.
34480 (ix86_handle_option): Handle -mrdpid.
34481 * config/i386/cpuid.h (bit_RDPID): New.
34482 * config/i386/driver-i386.c (host_detect_local_cpu):
34483 Detect RDPID feature.
34484 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34485 * config/i386/i386-c.c (ix86_target_macros_internal):
34487 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34488 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34489 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34490 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34491 * config/i386/i386.md (define_insn "rdpid"): New.
34492 * config/i386/i386.opt Add -mrdpid.
34493 * config/i386/immintrin.h (_rdpid_u32): New.
34495 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34497 PR rtl-optimization/79541
34498 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34499 instead of transforming it into USE.
34501 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34503 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34504 If HONOR_SNANS (SFmode) force the input to a register.
34505 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34506 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34507 an frsp or similar insn.
34509 2017-02-17 Martin Liska <mliska@suse.cz>
34511 PR rtl-optimization/79577
34512 * params.def (selsched-max-sched-times): Increase minimum to 1.
34514 2017-02-17 Martin Liska <mliska@suse.cz>
34516 PR rtl-optimization/79574
34517 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34519 2017-02-17 Martin Liska <mliska@suse.cz>
34521 PR tree-optimization/79529
34522 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34523 ssa_defined_default_def_p to handle cases which are implicitly
34525 * tree-ssa.c (ssa_defined_default_def_p): New function.
34526 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34527 which are implicitly defined.
34528 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34530 2017-02-17 Richard Biener <rguenther@suse.de>
34532 PR middle-end/79576
34533 * params.def (max-ssa-name-query-depth): Limit to 10.
34535 2017-02-17 Richard Biener <rguenther@suse.de>
34537 PR tree-optimization/79552
34538 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34541 2017-02-17 Richard Biener <rguenther@suse.de>
34544 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34546 2017-02-17 Marek Polacek <polacek@redhat.com>
34548 PR middle-end/79536
34549 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34550 (fold_negate_expr): New wrapper.
34552 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34554 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34555 Correct terminology and de-emphasize pre-standard behavior.
34557 2017-02-16 Alan Modra <amodra@gmail.com>
34559 PR rtl-optimization/79286
34560 * ira.c (def_dominates_uses): New function.
34561 (update_equiv_regs): Don't create an equivalence for insns that
34562 may trap where the register def does not dominate the use.
34564 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34566 PR rtl-optimization/78127
34567 * lra.c (lra): Call lra_eliminate before finish the loop after
34570 2017-02-16 Richard Biener <rguenther@suse.de>
34572 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34574 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34575 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34576 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34577 (isl_val_int_from_wi): New function.
34578 (extract_affine_gmp): Rename to ...
34579 (extract_affine_wi): ... this, take a widest_int.
34580 (extract_affine_int): Just wrap extract_affine_wi.
34581 (add_param_constraints): Use isl_val_int_from_wi.
34582 (add_loop_constraints): Likewise, and extract_affine_wi.
34584 2017-02-15 Jeff Law <law@redhat.com>
34586 PR middle-end/79521
34587 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34588 ira_init_register_move_cost_if_necessary.
34590 2017-02-15 Martin Sebor <msebor@redhat.com>
34592 PR middle-end/32003
34593 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34594 removed in a prior commit.
34596 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34598 PR tree-optimization/79347
34599 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34600 counters during peeling.
34602 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34604 * Makefile.in (site.exp): Remove "set ISLVER".
34606 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34609 * real.c (real_from_integer): Call real_convert even for decimal.
34611 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34614 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34616 2017-02-14 Andrew Pinski <apinski@cavium.com>
34618 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34619 cores and change the partno/implementer to be correct.
34620 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34621 the 'B" as the implementer.
34622 * config/aarch64/aarch64-tune.md: Regenerate.
34624 2017-02-14 Carl Love <cel@us.ibm.com>
34626 * config/rs6000/rs6000.c: Add case statement entry to make the
34627 xvcvuxdsp built-in argument unsigned.
34628 * config/rs6000/vsx.md: Fix the source and return operand types so they
34629 match the instruction definitions from the ISA document. Fix typo
34630 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34633 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34636 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34637 member early_clobber_alts.
34638 * lra-lives.c (reg_early_clobber_p): New.
34639 (process_bb_lives): Use it.
34640 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34641 (debug_operand_data): Initialize early_clobber_alts.
34642 (setup_operand_alternative): Set up early_clobber_alts.
34643 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34644 alternatives to new_insn_reg.
34645 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34647 (lra_update_insn_regno_info): Pass the new arg.
34649 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34651 PR middle-end/79505
34652 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34653 (new_oacc_loop_raw): Don't clear already cleared fields.
34656 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34657 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34658 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34660 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34663 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34665 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34668 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34669 the extra instruction to the right place to store 128-bit constant
34672 2017-02-14 Martin Sebor <msebor@redhat.com>
34674 PR middle-end/79448
34675 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34676 warning for strings of unknown length.
34678 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34680 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34682 2017-02-14 Jeff Law <law@redhat.com>
34685 * ira-costs.c (scan_one_insn): Initialize register move costs
34686 for pseudos seen in USE/CLOBBER insns.
34688 PR tree-optimization/79095
34689 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34690 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34691 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34692 if the operands are known to be not equal, then the resulting range
34694 (intersect_ranges): If the new range is ~[0,0] and the old range is
34695 wide, then prefer ~[0,0].
34696 * tree-vrp.c (overflow_comparison_p_1): New function.
34697 (overflow_comparison_p): New function.
34698 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34699 if NAME is used in an overflow test.
34700 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34701 overflow check that can be expressed as an equality test, then adjust
34702 ops to be that equality test.
34704 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34706 * config/s390/s390-builtin-types.def: Remove flags argument.
34707 * config/s390/s390.c (s390_init_builtins): Likewise.
34709 2017-02-14 Martin Liska <mliska@suse.cz>
34711 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34712 vector. Fix trailing white spaces.
34714 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34716 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34719 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34721 PR rtl-optimization/68664
34722 * config/arm/arm.c (arm_sched_can_speculate_insn):
34723 New function. Declare prototype.
34724 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34726 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34728 PR rtl-optimization/68664
34729 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34731 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34733 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34735 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34736 max skip bytes for function, loop and jump.
34738 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34740 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34741 ABS_EXPR for gimple dump.
34743 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34746 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34748 PR tree-optimization/79408
34749 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34750 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34751 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34752 also if rhs1 is INTEGER_CST.
34754 2017-02-14 Richard Biener <rguenther@suse.de>
34756 PR middle-end/79432
34757 * tree-into-ssa.c (insert_phi_nodes): When the function can
34758 have abnormal edges rewrite SSA names with broken use-def
34759 dominance out of SSA and register them for PHI insertion.
34761 2017-02-13 Martin Sebor <msebor@redhat.com>
34763 PR middle-end/79496
34764 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34765 clearing info.nowrite flag when snprintf size argument is a range.
34767 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34769 * cprop.c (cprop_jump): Add missing space in string literal.
34770 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34771 (get_constraint_for_component_ref): Likewise.
34772 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34773 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34774 * lra-constraints.c (process_alt_operands): Likewise.
34775 * ipa-inline.c (inline_small_functions): Likewise.
34776 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34777 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34778 * trans-mem.c (diagnose_tm_1_op): Likewise.
34779 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34780 (grid_parallel_clauses_gridifiable): Likewise.
34782 * config/nvptx/mkoffload.c (process): Add space in between
34785 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34786 "MOD4_SSE_REGS" and "ALL_REGS".
34788 * spellcheck.c (test_data): Add , in between "foo" and "food".
34790 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34793 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34794 boundary crossing check and subsequent code generation agree.
34796 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34798 * config/aarch64/aarch64.c (has_memory_op): Delete.
34799 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34802 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34804 PR rtl-optimization/79388
34805 PR rtl-optimization/79450
34806 * combine.c (distribute_notes): When removing TEM_INSN for which
34807 corresponding dest has last value recorded, invalidate that last
34810 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34812 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34813 of explicit '@'. Add missing assembly comment marker on branch costs
34816 2017-02-13 Nathan Sidwell <nathan@acm.org>
34818 * gengtype-lex.l (<in_struct>): Add '/'.
34820 2017-02-13 Martin Liska <mliska@suse.cz>
34823 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34825 2017-02-13 Richard Biener <rguenther@suse.de>
34827 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34829 * configure: Re-generate.
34830 * config.in: Likewise.
34831 * graphite-dependences.c: Simplify as if
34832 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34833 * graphite-isl-ast-to-gimple.c: Likewise.
34834 * graphite-optimize-isl.c: Likewise.
34835 * graphite-poly.c: Likewise.
34836 * graphite-sese-to-poly.c: Likewise.
34837 * graphite.h: Likewise.
34838 * toplev.c: Include isl/version.h and use isl_version () for
34839 printing the ISL version.
34840 * doc/install.texi: Update ISL requirement.
34842 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34844 * doc/standards.texi (Standards): Update reference to
34847 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34849 * doc/extend.texi (Named Address Spaces): sourceware.org now
34851 * doc/install.texi (Binaries): Ditto.
34854 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34856 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34857 standard terminology "stringize"/"stringizing" throughout.
34858 * doc/cppinternals.texi: Likewise.
34860 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34862 * doc/extend.texi: Fix some spelling mistakes and typos.
34863 * doc/invoke.texi: Likewise.
34865 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34868 * params.def (inline-min-speedup) Change from 10 to 8.
34870 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34872 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34875 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34878 * ipa-inline-analysis.c (get_minimal_bb): New function.
34879 (record_modified): Use it.
34880 (remap_edge_change_prob): Handle also ancestor functions.
34882 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34884 * doc/contrib.texi (Contributors): Remove broken link into
34885 the Mauve CVS repository.
34887 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34889 PR middle-end/79454
34890 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34891 result computation whenever lhs doesn't have vector mode, not
34892 just when it has BLKmode.
34894 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34896 * doc/makefile.texi (profiledbootstrap): Refer to the
34897 installation instructions only in textual form.
34899 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34902 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34904 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34906 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34907 (Specific): Update mingw-w64 reference.
34909 (Specific): Remove broken link to Renesas RX processor.
34911 2017-02-10 Richard Biener <rguenther@suse.de>
34913 * toplev.c (process_options): Do not mention obsolete graphite
34914 options when printing sorry message about missing graphite support.
34915 Mention -floop-nest-optimize.
34917 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34919 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34920 (vtst_p16): Likewise.
34921 (vtstq_p8): Likewise.
34922 (vtstq_p16): Likewise.
34924 (vtstq_p64): Likewise.
34925 * config/arm/arm_neon.h (vgetq_lane_p64): New.
34926 (vset_lane_p64): New.
34927 (vsetq_lane_p64): New.
34929 2017-02-10 Jakub Jelinek <jakub@redhat.com>
34931 PR tree-optimization/79411
34932 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
34933 stmt operands are SSA_NAMEs used in abnormal phis.
34934 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
34937 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34940 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
34943 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34945 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
34947 2017-02-09 Jakub Jelinek <jakub@redhat.com>
34949 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
34953 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
34954 not arbitrary TREE_CONSTANT.
34957 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
34958 "omp declare target link" attribute unless is_global_var.
34959 * omp-offload.c (find_link_var_op): Likewise.
34961 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34962 Chung-Lin Tang <cltang@codesourcery.com>
34964 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
34966 (gimplify_adjust_omp_clauses): Don't delete TILE.
34967 (gimplify_omp_for): Deal with TILE.
34968 * internal-fn.c (expand_GOACC_TILE): New function.
34969 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
34971 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
34972 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
34974 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
34975 avoid DIV for outermost collapse var.
34976 (expand_oacc_for): Insert tile element loop as needed. Adjust.
34977 Remove out of date comments, fix whitespace.
34978 * omp-general.c (omp_extract_for_data): Deal with tiling.
34979 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
34980 adjust OLF_DIM_BASE value.
34981 (struct omp_for_data): Add tiling field.
34982 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
34983 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
34984 for auto loops. Remove default auto determining, moved to
34985 oacc_loop_fixed_partitions.
34986 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
34987 stmts, add e_mask field.
34988 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
34989 (oacc_thread_numbers): Use oacc_dim_call.
34990 (oacc_xform_tile): New.
34991 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
34992 (finish_oacc_loop): Adjust for ifns vector.
34993 (oacc_loop_discover_walk): Append loop abstraction sites to list,
34994 add case for GOACC_TILE fns.
34995 (oacc_loop_xform_loop): Delete.
34996 (oacc_loop_process): Iterate over call list directly, and add
34997 handling for GOACC_TILE fns.
34998 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35000 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35001 vector partitioning to outer loops. Assign 2 partitions to loops
35002 when available. Add TILE handling.
35003 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35004 (execite_oacc_device_lower): Process GOACC_TILE fns,
35005 ignore unknown specs.
35006 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35007 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35008 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35010 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35012 * configure.ac (ACX_BUGURL): Update.
35013 * configure: Regenerate.
35015 2017-02-09 Richard Biener <rguenther@suse.de>
35017 PR tree-optimization/69823
35018 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35019 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35021 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35023 * config/arc/arc-c.def: Add __NPS400__ definition.
35024 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35025 (TARGET_NPS400): Define.
35027 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35029 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35031 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35032 pointer, arch_info.
35033 (arc_cpu_types): Fill the arch_info field with a pointer into the
35034 arc_arch_types table.
35035 (arc_selected_cpu): Declare.
35036 * config/arc/arc.c (arc_selected_cpu): Make global.
35037 (arc_selected_arch): Delete.
35038 (arc_base_cpu): Delete.
35039 (arc_override_options): Remove references to deleted variables,
35040 update access to arch information.
35041 (ARC_OPT): Update access to arch information.
35042 (ARC_OPTX): Likewise.
35043 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35044 (TARGET_ARC600): Update access to arch information.
35045 (TARGET_ARC601): Likewise.
35046 (TARGET_ARC700): Likewise.
35047 (TARGET_EM): Likewise.
35048 (TARGET_HS): Likewise.
35049 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35052 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35055 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35056 condition/operands for integer GE/LE/GEU/LEU operations.
35058 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35060 PR translation/79397
35061 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35064 2017-02-08 Martin Jambor <mjambor@suse.cz>
35067 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35068 whether allocation happened.
35069 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35070 nothing was allocated.
35072 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35074 PR tree-optimization/79408
35075 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35076 constant, but SSA_NAME with a known integer range, use the minimum
35077 of that range instead of op1 to determine if modulo can be replaced
35078 with its first operand.
35080 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35082 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35084 2017-02-08 Richard Biener <rguenther@suse.de>
35086 PR tree-optimization/71824
35087 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35088 Check all loops contained in the merged region.
35090 2017-02-07 Andrew Pinski <apinski@cavium.com>
35092 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35094 2017-02-07 Andrew Pinski <apinski@cavium.com>
35096 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35097 (thunderxt88): Likewise.
35098 (thunderxt81): Disable LSE and change v8.1 to v8.
35099 (thunderxt83): Likewise.
35101 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35102 Richard Biener <rguenther@suse.de>
35104 PR middle-end/79399
35105 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35106 type from int to size_t.
35107 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35109 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35111 PR rtl-optimization/79386
35112 * cprop.c (bypass_conditional_jumps): Initialize
35113 bypass_last_basic_block already before splitting bbs after
35114 unconditional traps...
35115 (bypass_conditional_jumps): ... rather than here.
35118 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35119 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35120 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35121 fix -masm=intel patterns.
35123 2017-02-07 Richard Biener <rguenther@suse.de>
35125 PR tree-optimization/79256
35126 PR middle-end/79278
35127 * builtins.c (get_object_alignment_2): Use min_align_of_type
35128 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35129 and ADJUST_FIELD_ALIGN.
35131 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35133 * doc/tm.texi: Regenerate.
35134 * stor-layout.c (layout_decl): Adjust.
35135 (update_alignment_for_field): Likewise.
35136 (place_field): Likewise.
35137 (min_align_of_type): Likewise.
35138 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35139 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35140 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35141 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35142 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35143 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35144 * config/i386/i386.c (x86_field_alignment): Likewise.
35145 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35146 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35147 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35148 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35149 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35150 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35154 2017-01-30 Richard Biener <rguenther@suse.de>
35156 PR tree-optimization/79256
35157 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35158 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35161 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35163 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35164 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35165 builtins to SImode and emit a zero-extend, if necessary.
35167 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35169 * docs/invoke.texi (RISC-V Options): Alphabetize.
35171 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35173 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35176 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35178 * config/riscv/riscv.c: New file.
35179 * common/config/riscv/riscv-common.c: Likewise.
35180 * config.gcc: Likewise.
35181 * config/riscv/constraints.md: Likewise.
35182 * config/riscv/elf.h: Likewise.
35183 * config/riscv/generic.md: Likewise.
35184 * config/riscv/linux.h: Likewise.
35185 * config/riscv/multilib-generator: Likewise.
35186 * config/riscv/peephole.md: Likewise.
35187 * config/riscv/pic.md: Likewise.
35188 * config/riscv/predicates.md: Likewise.
35189 * config/riscv/riscv-builtins.c: Likewise.
35190 * config/riscv/riscv-c.c: Likewise.
35191 * config/riscv/riscv-ftypes.def: Likewise.
35192 * config/riscv/riscv-modes.def: Likewise.
35193 * config/riscv/riscv-opts.h: Likewise.
35194 * config/riscv/riscv-protos.h: Likewise.
35195 * config/riscv/riscv.h: Likewise.
35196 * config/riscv/riscv.md: Likewise.
35197 * config/riscv/riscv.opt: Likewise.
35198 * config/riscv/sync.md: Likewise.
35199 * config/riscv/t-elf-multilib: Likewise.
35200 * config/riscv/t-linux: Likewise.
35201 * config/riscv/t-linux-multilib: Likewise.
35202 * config/riscv/t-riscv: Likewise.
35203 * configure.ac: Likewise.
35204 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35205 Waterman as RISC-V maintainers.
35206 * doc/install.texi: Add RISC-V entries.
35207 * doc/invoke.texi: Add RISC-V options section.
35208 * doc/md.texi: Add RISC-V constraints section.
35209 * configure: Regenerated.
35211 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35214 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35215 false values to be constant vectors with all 0 or all 1 bits set.
35216 (vcondu<mode><mode>): Likewise.
35217 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35219 (fpmask_comparison_operator): Update comment.
35220 (vecint_comparison_operator): New predicate.
35221 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35222 vector conditionals when the true and false values are constant
35223 vectors with all 0 bits or all 1 bits set.
35225 2017-02-06 Martin Sebor <msebor@redhat.com>
35227 PR tree-optimization/79376
35228 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35230 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35232 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35233 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35234 to simplify split condition.
35236 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35238 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35239 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35242 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35244 PR rtl-optimization/68664
35245 * target.def (can_speculate_insn): New hook.
35246 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35247 * doc/tm.texi: Regenerate.
35248 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35249 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35250 (rs6000_sched_can_speculate_insn): New function.
35252 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35254 PR tree-optimization/79284
35255 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35256 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35257 vectorizable_mask_load_store, vectorizable_operation,
35258 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35259 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35260 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35261 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35262 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35263 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35264 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35265 is_gimple_assign (stmt). Replace another such test with
35266 is_gimple_assign (stmt).
35268 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35271 * config/avr/avr.c (rtl-iter.h): Include it.
35272 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35273 (avr_legitimate_combined_insn): ...and implementation.
35275 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35277 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35278 * config/s390/s390.c (s390_const_operand_ok)
35279 (s390_canonicalize_comparison, s390_extract_part)
35280 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35281 (s390_contiguous_bitmask_p, s390_rtx_costs)
35282 (legitimize_pic_address): Likewise.
35283 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35284 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35285 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35286 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35287 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35289 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35291 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35292 REGNO($0) == REGNO($1).
35294 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35296 * config/s390/linux.h(SIZE_TYPE): Add comment.
35298 2017-02-06 Julian Brown <julian@codesourcery.com>
35299 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35300 Virendra Pathak <virendra.pathak@broadcom.com>
35302 * config/aarch64/aarch64-cores.def: Change the scheduler
35304 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35305 * config/aarch64/thunderx2t99.md: New file.
35307 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35309 * doc/standards.texi (Go Language): Update link to language
35312 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35314 * tree-eh.c (lower_resx): Sanitize profile.
35315 (cleanup_empty_eh_move_lp): Likewise.
35317 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35320 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35322 * cfgloopmanip.h (loop_version): Update prototype.
35323 * modulo-sched.c (sms_schedule): Update call of loop_version.
35324 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35325 * tree-parloops.c (gen_parallel_loop): Likewise.
35326 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35327 * tree-ssa-loop-split.c (split_loop): Likewise.
35328 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35329 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35331 2017-02-05 Martin Liska <mliska@suse.cz>
35334 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35336 (print_operand_address): Initialize a struct to zero.
35338 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35340 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35341 garbage collector only in textual form.
35343 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35345 * doc/extend.texi (x86 specific memory model extensions for
35346 transactional memory): Simplify a phrase.
35348 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35351 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35352 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35353 (atomic_storedi_1): Likewise.
35355 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35357 PR tree-optimization/79338
35358 * tree-parloops.c (gather_scalar_reductions): Don't call
35359 vect_analyze_loop_form for loop->inner before destroying loop's
35362 2017-02-03 Martin Sebor <msebor@redhat.com>
35364 PR tree-optimization/79327
35365 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35366 when precision has resulted in leading zeros.
35367 (format_integer): Adjust the likely counter to assume an unknown
35368 argument that may be zero is non-zero.
35370 2017-02-03 Jason Merrill <jason@redhat.com>
35373 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35374 avoid copying non-taken branch.
35376 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35378 PR tree-optimization/79340
35379 * tree-vect-loop.c (vectorizable_reduction): Release
35380 vec_defs elements after safe_splicing them into other vectors.
35383 PR tree-optimization/79327
35384 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35385 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35387 (format_integer): Use wide_int_to_tree instead of build_int_cst
35388 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35389 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35390 of shortest and longest sequence.
35392 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35394 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35395 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35397 2017-02-03 Walter Lee <walt@tilera.com>
35400 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35401 after initial stackframe link reg save.
35402 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35404 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35407 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35408 wu for stxssp alternative.
35410 2017-02-03 Martin Sebor <msebor@redhat.com>
35412 PR tree-optimization/79352
35413 * gimple-fold.c (get_range_strlen): Add argument.
35414 (get_range_strlen): Change return type to bool.
35415 (get_maxval_strlen): Pass in a dummy argument.
35416 * gimple-fold.h (get_range_strlen): Change return type to bool.
35417 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35418 * tree.h (array_at_struct_end_p): Add argument.
35419 * tree.c (array_at_struct_end_p): Handle it.
35421 2017-02-03 Martin Liska <mliska@suse.cz>
35424 * multiple_target.c (create_dispatcher_calls): Redirect edge
35425 from a caller of a dispatcher.
35426 (expand_target_clones): Make the clones local.
35427 (ipa_target_clone): Do both target clones and resolvers.
35428 (ipa_dispatcher_calls): Remove the pass.
35429 (pass_dispatcher_calls::gate): Likewise.
35430 (make_pass_dispatcher_calls): Likewise.
35431 * passes.def (pass_target_clone): Put as very first IPA early
35434 2017-02-03 Martin Liska <mliska@suse.cz>
35436 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35437 in case of a function with ifunc attribute.
35439 2017-02-03 Martin Liska <mliska@suse.cz>
35441 * cgraph.c (cgraph_node::dump): Dump function version info.
35442 * symtab.c (symtab_node::dump_base): Add missing new line.
35444 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35446 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35447 (ifcombine_ifandif): Use it.
35449 2017-02-03 Martin Liska <mliska@suse.cz>
35451 * doc/invoke.texi: Document default value for
35452 use-after-scope-direct-emission-threshold.
35454 2017-02-03 Martin Liska <mliska@suse.cz>
35456 PR tree-optimization/79339
35457 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35458 (format_floating): Likewise.
35460 2017-02-03 Martin Liska <mliska@suse.cz>
35463 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35465 (ipa_node_params_t::remove): Likewise.
35466 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35467 initialization from removed ipa_node_params_t::insert.
35468 (ipa_node_params::~ipa_node_params): Move from removed
35469 ipa_node_params_t::release.
35470 * symbol-summary.h (symbol_summary::m_released): New member.
35471 Do not release a summary twice. Do not allow to call finalizer
35472 for types of a summary that live in GGC memory.
35474 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35476 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35479 2017-02-02 Martin Sebor <msebor@redhat.com>
35481 PR middle-end/79275
35482 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35483 (format_string): Tighten up the range of output for non-constant
35484 strings and correct the expected range for wide non-constant strings.
35486 2017-02-02 Martin Sebor <msebor@redhat.com>
35488 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35490 PR middle-end/32003
35491 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35493 (-fdump-tree-@var): Add to index and document how to come up
35494 with pass-specific option and dump file names.
35495 (-fdump-passes): Clarify where to look for output.
35497 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35499 PR middle-end/77445
35500 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35501 statistics of the analyzed path; allow threading for speed when
35502 any of BBs along the path are optimized for speed.
35504 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35506 PR middle-end/78468
35507 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35508 settings of the virtual registers.
35511 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35513 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35514 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35517 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35519 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35520 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35522 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35524 * config/s390/s390.md: Add missing comments with the expanded
35526 * config/s390/vector.md: Likewise.
35527 * config/s390/vx-builtins.md: Likewise.
35529 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35532 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35533 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35534 conditions on a single line.
35536 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35538 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35539 __S390_VX__ to __VX__.
35541 2017-02-01 Andrew Pinski <apinski@cavium.com>
35543 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35544 stmt_info to record_stmt_cost.
35545 (vect_get_known_peeling_cost): Pass stmt_info if known to
35547 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35548 cpu_vector_cost field into
35549 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35550 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35551 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35552 splitting of scalar_stmt_cost and vec_stmt_cost.
35553 (thunderx_vector_cost): Likewise.
35554 (cortexa57_vector_cost): LIkewise.
35555 (exynosm1_vector_cost): Likewise.
35556 (xgene1_vector_cost): Likewise.
35557 (thunderx2t99_vector_cost): Improve after the splitting of the two
35559 (aarch64_builtin_vectorization_cost): Update for the splitting of
35560 scalar_stmt_cost and vec_stmt_cost.
35562 2017-02-01 Torvald Riegel <triegel@redhat.com>
35563 Richard Henderson <rth@redhat.com>
35565 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35566 conditional on existance of a fast atomic load.
35567 * optabs-query.c (can_atomic_load_p): New function.
35568 * optabs-query.h (can_atomic_load_p): Declare it.
35569 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35570 no fast atomic load is available for the particular size of access.
35571 (expand_atomic_compare_and_swap): Likewise.
35572 (expand_atomic_load): Likewise.
35573 (expand_atomic_store): Likewise.
35574 (expand_atomic_fetch_op): Likewise.
35575 * testsuite/lib/target-supports.exp
35576 (check_effective_target_sync_int_128): Remove x86 because it provides
35577 no fast atomic load.
35578 (check_effective_target_sync_int_128_runtime): Likewise.
35580 2017-02-01 Richard Biener <rguenther@suse.de>
35582 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35583 (graphite_transform_loops): Provide opt-info for optimized nests.
35584 * tree-parloop.c (parallelize_loops): Provide opt-info for
35585 parallelized loops.
35587 2017-02-01 Richard Biener <rguenther@suse.de>
35589 PR middle-end/79315
35590 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35591 was not set before.
35593 2017-02-01 Richard Biener <rguenther@suse.de>
35595 PR tree-optimization/71824
35596 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35597 Verify the loops are valid in the merged SESE region.
35598 (scop_detection::can_represent_loop_1): Check analyzing the
35599 evolution of the number of iterations in the region succeeds.
35601 2017-01-31 Ian Lance Taylor <iant@golang.org>
35603 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35604 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35606 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35608 PR preprocessor/79210
35609 * input.c (get_substring_ranges_for_loc): Replace line_width
35610 assertion with error-handling.
35612 2017-01-31 Richard Biener <rguenther@suse.de>
35614 PR tree-optimization/77318
35615 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35616 (create_pw_aff_from_tree): Take loop parameter.
35617 (add_condition_to_pbb): Pass loop of the condition to
35618 create_pw_aff_from_tree.
35620 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35622 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35623 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35625 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35629 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35631 (convert_int_to_float128): Likewise.
35632 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35633 (convert_int_to_float128): Likewise.
35634 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35635 (UNSPEC_IEEE128_CONVERT): Likewise.
35636 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35637 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35638 Use local variables for IBM extended format.
35639 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35640 (fix_trunc<mode>si2_fprs): Likewise.
35641 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35642 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35643 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35644 to know that we can now have integers of all sizes in vector
35646 (fix<uns>_<mode>di2_hw): Likewise.
35647 (float<uns>_<mode>si2_hw): Likewise.
35648 (fix_<mode>si2_hw): Likewise.
35649 (fixuns_<mode>si2_hw): Likewise.
35650 (float<uns>_<mode>di2_hw): Likewise.
35651 (float_<mode>di2_hw): Likewise.
35652 (float_<mode>si2_hw): Likewise.
35653 (floatuns_<mode>di2_hw): Likewise.
35654 (floatuns_<mode>si2_hw): Likewise.
35655 (xscvqp<su>wz_<mode>): Delete, no longer used.
35656 (xscvqp<su>dz_<mode>): Likewise.
35657 (xscv<su>dqp_<mode>): Likewise.
35658 (ieee128_mfvsrd_64bit): Likewise.
35659 (ieee128_mfvsrd_32bit): Likewise.
35660 (ieee128_mfvsrwz): Likewise.
35661 (ieee128_mtvsrw): Likewise.
35662 (ieee128_mtvsrd_64bit): Likewise.
35663 (ieee128_mtvsrd_32bit): Likewise.
35665 2017-01-31 Martin Liska <mliska@suse.cz>
35668 * ipa-prop.c (ipa_free_all_node_params): Call release method
35669 instead of ~sumbol_summary to not to trigger double times
35672 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35674 PR tree-optimization/71691
35675 * bitmap.h (class auto_bitmap): New.
35676 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35677 is_maybe_undefined instead of ssa_undefined_value_p.
35679 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35681 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35682 __S390_ARCH_LEVEL__ to __ARCH__.
35684 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35686 PR tree-optimization/79267
35687 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35688 if should_remove_lhs_p is true.
35690 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35693 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35694 (add_alignment_attribute): New.
35695 (base_type_die): Add alignment attribute.
35696 (subrange_type_die): Likewise.
35697 (modified_type_die): Likewise.
35698 (gen_array_type_die): Likewise.
35699 (gen_descr_array_type_die: Likewise.
35700 (gen_enumeration_type_die): Likewise.
35701 (gen_subprogram_die): Likewise.
35702 (gen_variable_die): Likewise.
35703 (gen_field_die): Likewise.
35704 (gen_ptr_to_mbr_type_die): Likewise.
35705 (gen_struct_or_union_type_die): Likewise.
35706 (gen_subroutine_type_die): Likewise.
35707 (gen_typedef_die): Likewise.
35708 (base_type_cmp): Compare alignment attribute.
35710 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35713 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35714 (setb_unsigned) New pattern for setb with CCUNS.
35715 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35716 subfc./subfe sequence to avoid overflow problems. Generate a
35717 shorter sequence with cmpld/setb for power9.
35718 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35719 for generating subfc. instruction.
35720 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35721 now uses this instruction.
35723 2017-01-30 Ian Lance Taylor <iant@google.com>
35726 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35727 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35729 2017-01-30 Martin Sebor <msebor@redhat.com>
35731 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35732 Move constant to the right of a relational operator.
35733 (get_mpfr_format_length, format_character, format_string): Ditto.
35734 (should_warn_p, maybe_warn): Same.
35736 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35738 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35741 * asan.c (get_translation_unit_decl): Remove function.
35742 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35744 2017-01-30 Martin Liska <mliska@suse.cz>
35746 PR gcov-profile/79259
35747 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35748 -fprofile-generate.
35750 2017-01-30 Martin Liska <mliska@suse.cz>
35753 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35754 Initialize variables with NULL value.
35756 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35759 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35761 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35763 2017-01-30 Richard Biener <rguenther@suse.de>
35765 PR tree-optimization/79276
35766 * tree-vrp.c (process_assert_insertions): Properly adjust common
35767 when removing a duplicate.
35769 2017-01-30 Richard Biener <rguenther@suse.de>
35771 PR tree-optimization/79256
35772 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35773 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35775 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35777 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35780 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35781 ("*r<noxa>sbg_<mode>_sll_bitmask")
35782 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35783 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35784 Use contiguous_bitmask_nowrap_operand.
35786 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35789 * config/rs6000/altivec.h (vec_xl): Revise #define.
35790 (vec_xst): Likewise.
35792 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35794 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35796 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35798 PR rtl-optimization/79194
35799 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35800 traps before call to bypass_conditional_jumps.
35802 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35804 PR tree-optimization/71374
35805 * lra-constraints.c (check_conflict_input_operands): New.
35806 (match_reload): Use it.
35808 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35811 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35812 account to calculate conflict_set.
35814 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35816 PR rtl-optimization/78559
35817 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35818 other_insn in combine.
35820 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35822 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35823 uint16_type_node for BT_UINT16.
35825 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35827 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35828 "RTL Tests" to menu.
35829 (GIMPLE Tests): New node.
35830 (RTL Tests): New node.
35832 2017-01-27 Richard Biener <rguenther@suse.de>
35834 PR tree-optimization/79245
35835 * tree-loop-distribution.c (distribute_loop): Apply cost
35836 modeling also to detected patterns.
35838 2017-01-27 Richard Biener <rguenther@suse.de>
35840 PR tree-optimization/71433
35841 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35842 (compare_assert_loc): New function.
35843 (process_assert_insertions): Sort and optimize assert locations
35844 to remove duplicates and push down identical assertions on
35845 edges to their destination block.
35847 2017-01-27 Richard Biener <rguenther@suse.de>
35849 PR tree-optimization/79244
35850 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35851 out SSA names even if abnormal.
35853 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35855 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35856 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35857 instead of MPFR_RNDN.
35859 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35862 * arm.c (arm_option_override): Don't call build_target_option_node
35863 until after doing all option overrides.
35864 (arm_valid_target_attribute_tree): Likewise.
35866 2017-01-27 Martin Liska <mliska@suse.cz>
35868 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35869 for {cd}tors and C++ {cd}tors.
35871 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35873 * config/s390/s390.md ("*setmem_long_and")
35874 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35876 2017-01-26 Martin Sebor <msebor@redhat.com>
35878 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35881 2017-01-26 Martin Sebor <msebor@redhat.com>
35883 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35884 HAVE_DFmode before using XFmode or DFmode.
35885 (parse_directive): Avoid using the z length modifier to avoid
35886 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35888 PR middle-end/78703
35889 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35890 to accept adjustment as an array.
35891 (get_int_range): New function.
35892 (struct directive): Make width and prec arrays.
35893 (directive::set_width, directive::set_precision): Call get_int_range.
35894 (format_integer, format_floating): Handle width and precision ranges.
35895 (format_string, parse_directive): Same.
35897 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35900 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35901 comdat_type_p set, just clone them, but keep the children in the
35905 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35906 which have direct callers with -fvar-tracking-assignments enabled
35908 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35909 inside of type units.
35911 2017-01-26 Martin Sebor <msebor@redhat.com>
35913 PR middle-end/78703
35914 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35916 (struct format_result): Replace number_chars, number_chars_min,
35917 and number_chars_max with a single member of struct result_range.
35919 (format_result::operator+=): Adjust.
35920 (struct fmtresult): Remove bounded. Handle likely and unlikely
35922 (fmtresult::adjust_for_width_or_precision): New function.
35923 (fmtresult:type_max_digits): New function.
35924 (bytes_remaining): Handle likely and unlikely counters.
35925 (min_bytes_remaining): Remove.
35926 (format_percent): Simplify.
35927 (format_integer, format_floating): Set likely and unlikely counters.
35928 (get_string_length, format_character, format_string): Same.
35929 (format_plain, should_warn_p): New function.
35930 (maybe_warn): Call should_warn_p. Update diagnostic messages
35931 and handle those for all directives, including plain strings.
35932 (format_directive): Handle likely and unlikely counters.
35933 Remove unnecessary quoting from diagnostics. Add an informational
35935 (add_bytes): Remove.
35936 (pass_sprintf_length::compute_format_length): Simplify.
35937 (try_substitute_return_value): Handle likely and unlikely counters.
35939 2017-01-26 Carl Love <cel@us.ibm.com>
35941 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
35942 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
35944 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
35947 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
35948 endianess for subregs into account.
35949 * lra-constraints.c (lra_constraints): Do risky transformations
35950 always on the first iteration.
35951 * lra-lives.c (check_pseudos_live_through_calls): Add arg
35952 last_call_used_reg_set.
35953 (process_bb_lives): Define and use last_call_used_reg_set.
35954 * lra.c (lra): Always continue after lra_constraints on the first
35957 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
35959 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
35961 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
35963 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35965 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
35966 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
35967 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
35968 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
35969 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
35970 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
35971 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
35972 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
35973 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
35975 2017-01-26 Marek Polacek <polacek@redhat.com>
35978 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
35979 for the third operand.
35981 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35983 PR middle-end/79236
35984 * omp-low.c (struct omp_context): Add simt_stmt field.
35985 (scan_omp_for): Return omp_context *.
35986 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
35987 context to the _simt_ SIMD stmt.
35988 (lower_omp_for): For combined SIMD with sibling _simt_
35989 SIMD, make sure to use the same decls in _looptemp_
35990 clauses as in the sibling.
35992 2017-01-26 David Sherwood <david.sherwood@arm.com>
35994 PR middle-end/79212
35995 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
35998 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36001 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36002 emit fld b; fld a; if possible.
36004 * brig-builtins.def: Update copyright years.
36005 * config/arm/arm_acle_builtins.def: Update copyright years.
36007 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36010 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36011 constraint instead of o for the stxsd instruction.
36013 2017-01-25 Carl Love <cel@us.ibm.com>
36015 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36016 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36018 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36020 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36022 2017-01-25 Richard Biener <rguenther@suse.de>
36024 PR tree-optimization/69264
36025 * target.def (vector_alignment_reachable): Improve documentation.
36026 * doc/tm.texi: Regenerate.
36027 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36029 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36030 earlier changes with respect to TYPE_USER_ALIGN.
36031 (vector_alignment_reachable_p): Likewise. Improve dumping.
36033 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36036 * config/arm/arm.md (xordi3): Force constant operand into a register
36039 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36041 * doc/invoke.texi (-fstore-merging): Correct default optimization
36042 levels at which it is enabled.
36043 (-O): Move -fstore-merging from list to...
36046 2017-01-25 Richard Biener <rguenther@suse.de>
36049 * omp-expand.c: Include debug.h.
36050 (expand_omp_taskreg): Make sure to generate early debug before
36051 outlining anything from a function.
36052 (expand_omp_target): Likewise.
36053 (grid_expand_target_grid_body): Likewise.
36055 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36058 * asan.c (get_translation_unit_decl): New function.
36059 (asan_add_global): Extract modules file name from globals
36060 TRANSLATION_UNIT_DECL name.
36062 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36065 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36066 for long calls with APCS frame and VFP.
36068 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36070 * cfg.c (original_copy_tables_initialized_p): New function.
36071 * cfg.h (original_copy_tables_initialized_p): New decl.
36072 * cfgrtl.c (relink_block_chain): Guard the call to
36073 free_original_copy_tables with a call to
36074 original_copy_tables_initialized_p.
36075 * cgraph.h (symtab_node::native_rtl_p): New decl.
36076 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36077 (symtab_node::needed_p): Don't assert for early assembly output
36078 for __RTL functions.
36079 (cgraph_node::finalize_function): Set "force_output" for __RTL
36081 (cgraph_node::analyze): Bail out early for __RTL functions.
36082 (analyze_functions): Update assertion to support __RTL functions.
36083 (cgraph_node::expand): Bail out early for __RTL functions.
36084 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36086 * function.h (struct function): Update comment for field
36088 * gimple-expr.c: Include "tree-pass.h".
36089 (gimple_has_body_p): Return false for __RTL functions.
36090 * Makefile.in (OBJS): Add run-rtl-passes.o.
36091 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36093 (gcc::pass_manager::get_clean_slate): New accessor.
36094 * passes.c: Include "insn-addr.h".
36095 (should_skip_pass_p): Add logging. Update logic for running
36096 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36097 property-provider override so it is only done for gimple passes.
36099 (skip_pass): New function.
36100 (execute_one_pass): Call skip_pass when skipping passes.
36101 * read-md.c (md_reader::read_char): Support filtering
36102 the input to a subset of line numbers.
36103 (md_reader::md_reader): Initialize fields
36104 m_first_line and m_last_line.
36105 (md_reader::read_file_fragment): New function.
36106 * read-md.h (md_reader::read_file_fragment): New decl.
36107 (md_reader::m_first_line): New field.
36108 (md_reader::m_last_line): New field.
36109 * read-rtl-function.c (function_reader::create_function): Only
36110 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36111 curr_properties. Set DECL_INITIAL to a dummy block.
36112 (read_rtl_function_body_from_file_range): New function.
36113 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36115 * run-rtl-passes.c: New file.
36116 * run-rtl-passes.h: New file.
36118 2017-01-24 Jeff Law <law@redhat.com>
36120 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36123 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36125 PR tree-optimization/79159
36126 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36127 (record_nonwrapping_iv): Improve boundary using above function if no
36128 value range information.
36130 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36131 Martin Jambor <mjambor@suse.cz>
36133 * brig-builtins.def: New file.
36134 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36135 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36136 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36137 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36138 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36139 * builtin-types.def (BT_INT8): New.
36140 (BT_INT16): Likewise.
36141 (BT_UINT8): Likewise.
36142 (BT_UINT16): Likewise.
36143 (BT_FN_ULONG): Likewise.
36144 (BT_FN_UINT_INT): Likewise.
36145 (BT_FN_UINT_ULONG): Likewise.
36146 (BT_FN_UINT_LONG): Likewise.
36147 (BT_FN_UINT_PTR): Likewise.
36148 (BT_FN_ULONG_PTR): Likewise.
36149 (BT_FN_INT8_FLOAT): Likewise.
36150 (BT_FN_INT16_FLOAT): Likewise.
36151 (BT_FN_UINT32_FLOAT): Likewise.
36152 (BT_FN_UINT16_FLOAT): Likewise.
36153 (BT_FN_UINT8_FLOAT): Likewise.
36154 (BT_FN_UINT64_FLOAT): Likewise.
36155 (BT_FN_UINT16_UINT32): Likewise.
36156 (BT_FN_UINT32_UINT16): Likewise.
36157 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36158 (BT_FN_INT_PTR_INT): Likewise.
36159 (BT_FN_UINT_PTR_UINT): Likewise.
36160 (BT_FN_LONG_PTR_LONG): Likewise.
36161 (BT_FN_ULONG_PTR_ULONG): Likewise.
36162 (BT_FN_VOID_UINT64_UINT64): Likewise.
36163 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36164 (BT_FN_INT8_INT8_INT8): Likewise.
36165 (BT_FN_INT16_INT16_INT16): Likewise.
36166 (BT_FN_INT_INT_INT): Likewise.
36167 (BT_FN_UINT_FLOAT_UINT): Likewise.
36168 (BT_FN_FLOAT_UINT_UINT): Likewise.
36169 (BT_FN_ULONG_UINT_UINT): Likewise.
36170 (BT_FN_ULONG_UINT_PTR): Likewise.
36171 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36172 (BT_FN_UINT_UINT_UINT): Likewise.
36173 (BT_FN_VOID_UINT_PTR): Likewise.
36174 (BT_FN_UINT_UINT_PTR: Likewise.
36175 (BT_FN_UINT32_UINT64_PTR): Likewise.
36176 (BT_FN_INT_INT_UINT_UINT): Likewise.
36177 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36178 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36179 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36180 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36181 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36182 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36183 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36184 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36185 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36186 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36187 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36188 * doc/frontends.texi: List BRIG FE.
36189 * doc/install.texi (Testing): Add BRIG tesring requirements.
36190 * doc/invoke.texi (Overall Options): Mention BRIG.
36191 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36193 2017-01-24 Richard Biener <rguenther@suse.de>
36195 PR translation/79208
36196 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36198 2017-01-24 Martin Jambor <mjambor@suse.cz>
36201 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36202 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36203 and known_contexts.
36205 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36207 PR middle-end/79123
36208 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36209 casts from signed to unsigned really don't have a range.
36211 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36213 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36214 GMP_RNDx for compatiblity.
36216 2017-01-24 Martin Liska <mliska@suse.cz>
36219 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36220 that would prevent us to call alloca with -1 as argument.
36222 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36224 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36225 Avoid -Wformat-security warning.
36227 2017-01-23 Andrew Pinski <apinski@cavium.com>
36229 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36232 2017-01-23 Martin Sebor <msebor@redhat.com>
36234 PR middle-end/78703
36235 * gimple-ssa-sprintf.c (warn_level): New global.
36236 (format_integer): Use it here and throughout the rest of the file.
36237 Use the same switch to compute sign as base.
36238 (maybe_warn): New function.
36239 (format_directive): Factor out warnings into maybe_warn.
36240 Add debugging output. Use warn_level.
36241 (add_bytes): Use warn_level.
36242 (pass_sprintf_length::compute_format_length): Add debugging output.
36243 (try_substitute_return_value): Same.
36244 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36246 PR middle-end/78703
36247 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36248 (struct fmtresult, format_integer, format_floating): Adjust.
36249 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36250 (get_string_length, format_string,format_directive): Same.
36251 (pass_sprintf_length::compute_format_length): Same.
36252 (try_substitute_return_value): Simplify slightly.
36254 PR middle-end/78703
36255 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36256 (fmtresult::operator+=): Outlined.
36257 (struct fmtresult): Add ctors.
36258 (struct conversion_spec): Rename...
36259 (struct directive): ...to this. Add and remove data members.
36260 (directive::set_width, directive::set_precision): New functions.
36261 (format_percent): Use fmtresult ctor.
36262 (get_width_and_precision): Remove.
36263 (format_integer): Make naming changes. Avoid computing width and
36265 (format_floating): Same. Adjust indentation.
36266 (format_character, format_none): New functions.
36267 (format_string): Moved character handling to format_character.
36268 (format_directive): Remove arguments, change return type.
36269 (parse_directive): New function.
36270 (pass_sprintf_length::compute_format_length): Move directive
36271 parsing to parse_directive.
36273 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36275 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36276 (assign_assembler_name_if_needed): ... this.
36277 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36278 (assign_assembler_name_if_needed): ... this.
36279 (free_lang_data_in_cgraph): Adjust callers.
36280 * cgraphunit.c (cgraph_node::analyze): Likewise.
36281 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36284 2017-01-23 Richard Biener <rguenther@suse.de>
36286 PR tree-optimization/79088
36287 PR tree-optimization/79188
36288 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36289 resetting loop bounds after last path deletion. Reset loop
36290 bounds of the target loop, make code match the comments.
36291 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36292 Make sure loops need no fixups.
36294 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36296 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36297 exponent support with double type for first argument.
36298 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36299 type returned by __builtin_vec_extract_sig,
36300 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36301 functions from "vector int" to "vector unsigned int" or from
36302 "vector long long int" to "vector unsigned long long int".
36303 Changed type returned by __builtin_vec_extract_exp,
36304 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36305 functions from "vector int" to "vector unsigned int" or from
36306 "vector long long int" to "vector unsigned long long int".
36307 Changed return type of __builtin_vec_test_data_class,
36308 __builtin_vec_test_data_class_sp, and
36309 __builtin_vec_test_data_class_dp from "vector int" to
36310 "vector bool int" or from "vector long long int" to "vector bool
36311 long long int" and changed second argument type from "unsigned
36312 int" to "int". Added new overloaded function forms "vector float
36313 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36314 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36315 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36316 double, vector unsigned long long int)" and "vector double
36317 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36318 long int)". Changed return type of
36319 __builtin_scalar_test_data_class and
36320 __builtin_scalar_test_data_class_sp and
36321 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36322 int" and changed second argument from "unsigned int" to "int".
36323 Changed type returned by __builtin_scalar_test_neg,
36324 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36325 from "int" to "bool int". Added new overloaded function form
36326 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36327 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36328 exponent double-precision with floating point first argument.
36329 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36330 documentation of scalar_test_data_class, scalar_test_neg,
36331 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36332 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36333 vec_test_data_class built-in functions to reflect refinements in
36334 their type signatures.
36336 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36338 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36340 (aarch64_elf_asm_destructor): Likewise.
36342 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36344 PR rtl-optimization/78634
36345 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36346 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36347 * ifcvt.c (noce_try_cmove): Add missing cost check.
36349 PR rtl-optimization/71724
36350 * combine.c (if_then_else_cond): Look for situations where it is
36351 beneficial to undo the work of one of the recursive calls.
36353 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36355 PR tree-optimization/70754
36356 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36357 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36358 combined stmt before it if not NULL.
36359 (combine_chains): Process refs reversely and compute dominance point
36362 2017-01-23 Martin Liska <mliska@suse.cz>
36364 PR tree-optimization/79196
36365 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36366 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36368 (strlen_optimize_stmt): Call the renamed function.
36370 2017-01-23 Michael Matz <matz@suse.de>
36372 PR tree-optimization/78384
36373 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36375 2017-01-23 Richard Biener <rguenther@suse.de>
36377 PR tree-optimization/79186
36378 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36379 both incoming edges before moving an assert.
36381 2017-01-23 Martin Jambor <mjambor@suse.cz>
36383 * ipa-prop.c (load_from_param_1): Removed.
36384 (load_from_unmodified_param): Bits from load_from_param_1 put back
36386 (load_from_param): Removed.
36387 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36388 with stmt. Reverted back to use of load_from_unmodified_param.
36390 2017-01-23 Martin Jambor <mjambor@suse.cz>
36393 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36394 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36395 field a pointer to garbage collected vector, mark lattices and
36396 ipcp_orig_node with GTY((skip)).
36397 (ipa_get_param_count): Adjust to descriptors being a pointer.
36398 (ipa_get_param): Likewise.
36399 (ipa_get_type): Likewise.
36400 (ipa_get_param_move_cost): Likewise.
36401 (ipa_set_param_used): Likewise.
36402 (ipa_get_controlled_uses): Likewise.
36403 (ipa_set_controlled_uses): Likewise.
36404 (ipa_is_param_used): Likewise.
36405 (ipa_node_params_t): Move into garbage collector. New methods insert
36407 (ipa_node_params_sum): Annotate wth GTY(()).
36408 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36410 (ipa_load_from_parm_agg): Adjust declaration.
36411 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36412 * ipa-profile.c (ipa_profile): Likewise.
36413 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36414 (ipa_populate_param_decls): Make descriptors parameter garbage
36416 (ipa_dump_param): Adjust to descriptors being a pointer.
36417 (ipa_alloc_node_params): Likewise.
36418 (ipa_initialize_node_params): Likewise.
36419 (load_from_param_1): Make descriptors parameter garbage collected.
36420 (load_from_unmodified_param): Likewise.
36421 (load_from_param): Likewise.
36422 (ipa_load_from_parm_agg): Likewise.
36423 (ipa_node_params::~ipa_node_params): Removed.
36424 (ipa_free_all_node_params): Remove call to delete operator.
36425 (ipa_node_params_t::insert): New.
36426 (ipa_node_params_t::remove): Likewise.
36427 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36428 copy known_csts and known_contexts vectors.
36429 (ipa_read_node_info): Adjust to descriptors being a pointer.
36430 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36432 (ipcp_transform_function): Make descriptors variable garbage
36435 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36437 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36438 * config/i386/avx512dqintrin.h: Ditto.
36439 * config/i386/avx512fintrin.h: Ditto.
36440 * config/i386/i386.c: Handle new builtins.
36441 * config/i386/i386-builtin.def: Add new builtins.
36442 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36443 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36445 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36446 Martin Liska <mliska@suse.cz>
36448 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36449 * asan.c (asan_expand_poison_ifn): Support stores and use
36450 appropriate ASAN report function.
36451 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36452 * internal-fn.def (ASAN_POISON_USE): Declare.
36453 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36454 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36455 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36456 ASAN_POISON calls w/o LHS.
36457 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36458 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36459 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36460 * gimplify.c (asan_poison_variables): Add attribute
36461 use_after_scope_memory to variables that really needs to live
36463 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36464 having the attribute.
36466 2017-01-23 Martin Liska <mliska@suse.cz>
36468 * asan.c (create_asan_shadow_var): New function.
36469 (asan_expand_poison_ifn): Likewise.
36470 * asan.h (asan_expand_poison_ifn): New declaration.
36471 * internal-fn.c (expand_ASAN_POISON): Likewise.
36472 * internal-fn.def (ASAN_POISON): New builtin.
36473 * sanopt.c (pass_sanopt::execute): Expand
36474 asan_expand_poison_ifn.
36475 * tree-inline.c (copy_decl_for_dup_finish): Make function
36477 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36478 * tree-ssa.c (is_asan_mark_p): New function.
36479 (execute_update_addresses_taken): Rewrite local variables
36480 (identified just by use-after-scope as addressable) into SSA.
36482 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36484 * doc/install.texi (Specific): opensource.apple.com uses https
36485 now. Remove trailing slash.
36487 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36489 * README.Portability: Remove note on an Irix compatibility issue.
36491 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36493 * gcov.c (INCLUDE_ALGORITHM): Define.
36494 (INCLUDE_VECTOR): Define.
36495 No longer include <vector> and <algorithm> directly.
36497 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36499 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36501 * doc/invoke.texi (Code Gen Options): Ditto.
36503 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36506 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36508 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36510 rtl-optimization/79125
36511 * cprop.c (local_cprop_pass): Handle cases where we make an
36512 unconditional trap.
36514 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36518 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36519 read from, for big endian.
36521 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36523 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36524 register pauth builtins for LP64 only.
36526 2017-01-20 Marek Polacek <polacek@redhat.com>
36529 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36532 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36534 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36536 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36537 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36538 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36540 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36543 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36544 in big-endian mode when they are not a single duplicated value.
36546 2017-01-20 Richard Biener <rguenther@suse.de>
36548 * BASE-VER: Bump to 7.0.1.
36550 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36552 * omp-low.c (omplow_simd_context): New struct. Use it...
36553 (lower_rec_simd_input_clauses): ...here and...
36554 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36555 references to idx, lane, max_vf, is_simt.
36557 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36559 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36562 2017-01-20 Martin Jambor <mjambor@suse.cz>
36564 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36565 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36567 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36568 (GTFILES): Rename hsa.c to hsa-common.c.
36569 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36570 * hsa-dump.c: Likewise.
36571 * hsa-gen.c: Likewise.
36572 * hsa-regalloc.c: Likewise.
36573 * ipa-hsa.c: Likewise.
36574 * omp-expand.c: Likewise.
36575 * omp-low.c: Likewise.
36576 * toplev.c: Likewise.
36578 2017-01-20 Marek Polacek <polacek@redhat.com>
36581 * doc/invoke.texi: Document -Wduplicated-branches.
36582 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36583 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36584 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36585 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36586 return 0 only when not OEP_LEXICOGRAPHIC.
36587 (fold_build_cleanup_point_expr): Use the expression
36588 location when building CLEANUP_POINT_EXPR.
36589 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36590 * tree.c (add_expr): Handle error_mark_node.
36592 2017-01-20 Martin Liska <mliska@suse.cz>
36595 * tree-profile.c (init_ic_make_global_vars): Do not call
36597 (gimple_init_gcov_profiler): Likewise.
36599 2017-01-20 Martin Liska <mliska@suse.cz>
36602 * cgraph.h (maybe_create_reference): Remove argument and
36604 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36606 * ipa-cp.c (create_specialized_node): Likewise.
36607 * symtab.c (symtab_node::maybe_create_reference): Handle
36608 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36610 2017-01-20 Martin Liska <mliska@suse.cz>
36612 * read-rtl-function.c (function_reader::create_function): Use
36613 build_decl instread of build_decl_stat.
36615 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36617 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36618 * config/i386/avx512dqintrin.h: Ditto.
36619 * config/i386/avx512fintrin.h: Ditto.
36620 * config/i386/i386-builtin-types.def: Add new types.
36621 * config/i386/i386.c: Handle new types.
36622 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36623 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36624 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36625 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36626 (__builtin_ia32_kshiftridi): New.
36627 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36629 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36633 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36634 define to rs6000_init_stack_protect_guard.
36635 (rs6000_init_stack_protect_guard): New function.
36637 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36638 Yunqiang Su <yunqiang.su@imgtec.com>
36640 * config.gcc (supported_defaults): Add madd4.
36641 (with_madd4): Add validation.
36642 (all_defaults): Add madd4.
36643 * config/mips/mips.opt (mmadd4): New option.
36644 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36646 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36648 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36649 (ISA_HAS_FUSED_MADD4): Likewise.
36650 * doc/invoke.texi (-mmadd4): Document the new option.
36651 * doc/install.texi (--with-madd4): Document the new option.
36653 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36655 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36656 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36657 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36658 (aarch64_init_pauth_hint_builtins): New.
36659 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36660 (aarch64_expand_builtin): Expand new builtins.
36662 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36664 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36665 * combine-stack-adj.c (no_unhandled_cfa): Handle
36666 REG_CFA_TOGGLE_RA_MANGLE.
36667 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36668 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36669 info for return address signing.
36670 (aarch64_expand_epilogue): Likewise.
36672 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36674 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36675 * config/aarch64/aarch64-protos.h
36676 (aarch64_return_address_signing_enabled): New declaration.
36677 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36679 (aarch64_expand_prologue): Sign return address before it's pushed onto
36681 (aarch64_expand_epilogue): Authenticate return address fetched from
36683 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36684 (aarch64_attributes): New function attributes for "sign-return-address".
36685 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36686 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36687 ("*do_return"): Generate combined instructions according to key index.
36688 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36689 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36691 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36692 * config/aarch64/aarch64.opt (msign-return-address=): New.
36693 * doc/extend.texi (AArch64 Function Attributes): Documents
36694 "sign-return-address=".
36695 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36697 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36699 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36700 overall option summary.
36702 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36704 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36705 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36706 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36707 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36709 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36711 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36712 -mpower9-minmax by default for -mcpu=power9.
36713 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36714 128-bit floating point.
36716 2017-01-20 Alan Modra <amodra@gmail.com>
36718 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36719 optimizing for size.
36721 2017-01-20 Alan Modra <amodra@gmail.com>
36724 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36725 for strcmp and strncmp from corresponding builtin decl.
36727 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36729 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36730 instead of i386/rtems-64.h.
36731 * config/i386/rtems-64.h: Remove.
36733 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36737 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36739 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36741 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36743 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36744 Change int to HOST_WIDE_INT.
36745 * config/aarch64/aarch64-protos.h
36746 (aarch64_simd_gen_const_vector_dup): Likewise.
36747 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36749 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36751 * langhooks-def.h (lhd_type_for_size): New decl.
36752 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36753 * langhooks.c (lhd_type_for_size): New function, taken from
36756 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36758 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36759 define_bypass for CR latency.
36760 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36761 (power9-alu2): Add define_bypass for CR latency.
36763 (power9-mul): Update insn latency.
36764 (power9-mul-compare): Update insn latency, bypass latency and remove
36767 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36769 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36771 * config/aarch64/aarch64.md
36772 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36773 aarch64_nopcrelative_literal_loads.
36774 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36776 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36778 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36779 TARGET_LOONGSON_3A.
36780 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36782 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36785 * config.gcc (supported_defaults): Add lxc1-sxc1.
36786 (with_lxc1_sxc1): Add validation.
36787 (all_defaults): Add lxc1-sxc1.
36788 * config/mips/mips.opt (mlxc1-sxc1): New option.
36789 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36791 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36792 __mips_no_lxc1_sxc1.
36793 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36794 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36795 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36797 2017-01-19 Richard Biener <rguenther@suse.de>
36799 PR tree-optimization/72488
36800 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36801 sure to restore SSA info.
36802 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36804 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36806 PR rtl-optimization/79121
36807 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36808 of the inner type when shifting an extended value.
36810 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36813 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36814 interposable aliases.
36816 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36819 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36820 Use the evmergelohi instruction.
36821 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36822 (mov_sitf_e500_subreg8_2_be): Likewise.
36823 (mov_sitf_e500_subreg12_2_le): Likewise.
36824 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36825 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36826 (mov_sitf_e500_subreg8_2_le): Likewise.
36827 (mov_sitf_e500_subreg12_2_be): Likewise.
36829 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36831 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36832 attribute from vecsimple to vecperm.
36833 (altivec_vbpermq2): Likewise.
36835 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36838 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36840 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36841 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36842 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36843 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36844 case where N arg is SIZE_MAX.
36845 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36846 (cmpstrsi): Add pattern.
36848 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36850 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36851 __builtin_vec_revb builtins.
36852 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36853 built-in functions to support generation of the ISA 3.0 XXBR<x>
36854 vector byte reverse instructions.
36855 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36856 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36857 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36858 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36859 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36860 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36861 (P9V_BUILTIN_VEC_REVB): Likewise.
36862 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36863 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36864 (p9_xxbrq_v16qi): Likewise.
36865 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36866 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36867 (p9_xxbrh_v8hi): Likewise.
36868 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36869 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36870 vec_revb built-in functions.
36872 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36874 PR rtl-optimization/78952
36875 * config/i386/i386.md (any_extract): New code iterator.
36876 (*insvqi_2): Use any_extract for source operand.
36877 (*insvqi_3): Use any_shiftrt for source operand.
36879 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36881 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36883 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36885 2017-01-18 Matthias Klose <doko@ubuntu.com>
36887 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36889 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36891 * config/rs6000/altivec.h (vec_bperm): Change #define.
36892 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36893 (altivec_vbpermq2): New define_insn.
36894 (altivec_vbpermd): Likewise.
36895 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36896 function interface.
36897 (VBPERMD): Likewise.
36898 (VBPERM): New polymorphic function interface.
36899 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36900 Add entries for P9V_BUILTIN_VEC_VBPERM.
36901 * doc/extend.texi: Add interfaces for vec_bperm.
36903 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36905 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36906 first letter of error messages.
36907 (s390_resolve_overloaded_builtin): Likewise.
36908 * config/s390/s390.c (s390_expand_builtin): Likewise.
36909 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36910 (s390_valid_target_attribute_inner_p): Likewise.
36911 * config/s390/s390.md ("tabort"): Likewise.
36913 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36915 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36916 (ISA_AVOID_DIV_HILO): New macro.
36917 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36918 (ISA_HAS_DDIV): Likewise.
36920 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36922 * doc/invoke.texi (fabi-version): Correct number of occurrences.
36924 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36926 * doc/invoke.texi (fabi-version): Spelling fix.
36928 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36931 * doc/invoke.texi (fabi-version): Mention mangling fix for
36934 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36937 * doc/invoke.texi (fabi-version): Document discriminator mangling.
36939 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
36942 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
36943 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
36945 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
36947 (stack_protect_test): Ditto.
36948 * config/rs6000/rs6000.opt (mstack-protector-guard=,
36949 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
36951 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
36952 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
36953 -mstack-protector-guard-offset=.
36954 (RS/6000 and PowerPC Options): Ditto.
36956 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36958 * config/i386/i386.h (MASK_CLASS_P): New define.
36959 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
36960 there are no registers from different register sets also when
36961 mask registers are used. Update function comment.
36962 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
36963 to (*k/*r) and (*k/*km) alternatives.
36965 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
36967 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
36968 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
36969 (EH_RETURN_HANDLER_RTX): New define.
36970 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
36971 Force frame pointer in EH return functions.
36972 (aarch64_expand_epilogue): Add barrier for eh_return.
36973 (aarch64_final_eh_return_addr): Remove.
36974 (aarch64_eh_return_handler_rtx): New function.
36975 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
36977 (aarch64_eh_return_handler_rtx): New prototype.
36979 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36981 * config/rs6000/altivec.h (vec_rlmi): New #define.
36982 (vec_vrlnm): Likewise.
36983 (vec_rlnm): Likewise.
36984 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
36985 (UNSPEC_VRLNM): Likewise.
36986 (VIlong): New mode iterator.
36987 (altivec_vrl<VI_char>mi): New define_insn.
36988 (altivec_vrl<VI_char>nm): Likewise.
36989 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
36991 (VRLDNM): Likewise.
36992 (RLNM): New polymorphic function entry.
36993 (VRLWMI): New monomorphic function entry.
36994 (VRLDMI): Likewise.
36995 (RLMI): New polymorphic function entry.
36996 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
36997 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
36998 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37001 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37004 * dwarf2out.c (field_byte_offset): Restore the
37005 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37006 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37007 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37010 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37013 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37015 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37018 * dwarf2out.c (add_data_member_location_attribute): For constant
37019 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37020 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37021 DW_AT_byte_size attributes.
37023 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37025 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37026 after forcing to constant memory when the code model is medium.
37028 2017-01-17 Julia Koval <julia.koval@intel.com>
37031 * config/i386/avx512fintrin.h
37032 (_mm512_i32gather_ps): Change __addr type to void const*.
37033 (_mm512_mask_i32gather_ps): Ditto.
37034 (_mm512_i32gather_pd): Ditto.
37035 (_mm512_mask_i32gather_pd): Ditto.
37036 (_mm512_i64gather_ps): Ditto.
37037 (_mm512_mask_i64gather_ps): Ditto.
37038 (_mm512_i64gather_pd): Ditto.
37039 (_mm512_mask_i64gather_pd): Ditto.
37040 (_mm512_i32gather_epi32): Ditto.
37041 (_mm512_mask_i32gather_epi32): Ditto.
37042 (_mm512_i32gather_epi64): Ditto.
37043 (_mm512_mask_i32gather_epi64): Ditto.
37044 (_mm512_i64gather_epi32): Ditto.
37045 (_mm512_mask_i64gather_epi32): Ditto.
37046 (_mm512_i64gather_epi64): Ditto.
37047 (_mm512_mask_i64gather_epi64): Ditto.
37048 (_mm512_i32scatter_ps): Change __addr type to void*.
37049 (_mm512_mask_i32scatter_ps): Ditto.
37050 (_mm512_i32scatter_pd): Ditto.
37051 (_mm512_mask_i32scatter_pd): Ditto.
37052 (_mm512_i64scatter_ps): Ditto.
37053 (_mm512_mask_i64scatter_ps): Ditto.
37054 (_mm512_i64scatter_pd): Ditto.
37055 (_mm512_mask_i64scatter_pd): Ditto.
37056 (_mm512_i32scatter_epi32): Ditto.
37057 (_mm512_mask_i32scatter_epi32): Ditto.
37058 (_mm512_i32scatter_epi64): Ditto.
37059 (_mm512_mask_i32scatter_epi64): Ditto.
37060 (_mm512_i64scatter_epi32): Ditto.
37061 (_mm512_mask_i64scatter_epi32): Ditto.
37062 (_mm512_i64scatter_epi64): Ditto.
37063 (_mm512_mask_i64scatter_epi64): Ditto.
37064 * config/i386/avx512pfintrin.h
37065 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37066 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37067 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37068 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37069 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37070 (_mm512_prefetch_i32scatter_ps): Ditto.
37071 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37072 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37073 (_mm512_prefetch_i64scatter_pd): Ditto.
37074 (_mm512_prefetch_i64scatter_ps): Ditto.
37075 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37076 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37077 * config/i386/avx512vlintrin.h
37078 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37079 (_mm_mmask_i32gather_ps): Ditto.
37080 (_mm256_mmask_i32gather_pd): Ditto.
37081 (_mm_mmask_i32gather_pd): Ditto.
37082 (_mm256_mmask_i64gather_ps): Ditto.
37083 (_mm_mmask_i64gather_ps): Ditto.
37084 (_mm256_mmask_i64gather_pd): Ditto.
37085 (_mm_mmask_i64gather_pd): Ditto.
37086 (_mm256_mmask_i32gather_epi32): Ditto.
37087 (_mm_mmask_i32gather_epi32): Ditto.
37088 (_mm256_mmask_i32gather_epi64): Ditto.
37089 (_mm_mmask_i32gather_epi64): Ditto.
37090 (_mm256_mmask_i64gather_epi32): Ditto.
37091 (_mm_mmask_i64gather_epi32): Ditto.
37092 (_mm256_mmask_i64gather_epi64): Ditto.
37093 (_mm_mmask_i64gather_epi64): Ditto.
37094 (_mm256_i32scatter_ps): Change __addr type to void*.
37095 (_mm256_mask_i32scatter_ps): Ditto.
37096 (_mm_i32scatter_ps): Ditto.
37097 (_mm_mask_i32scatter_ps): Ditto.
37098 (_mm256_i32scatter_pd): Ditto.
37099 (_mm256_mask_i32scatter_pd): Ditto.
37100 (_mm_i32scatter_pd): Ditto.
37101 (_mm_mask_i32scatter_pd): Ditto.
37102 (_mm256_i64scatter_ps): Ditto.
37103 (_mm256_mask_i64scatter_ps): Ditto.
37104 (_mm_i64scatter_ps): Ditto.
37105 (_mm_mask_i64scatter_ps): Ditto.
37106 (_mm256_i64scatter_pd): Ditto.
37107 (_mm256_mask_i64scatter_pd): Ditto.
37108 (_mm_i64scatter_pd): Ditto.
37109 (_mm_mask_i64scatter_pd): Ditto.
37110 (_mm256_i32scatter_epi32): Ditto.
37111 (_mm256_mask_i32scatter_epi32): Ditto.
37112 (_mm_i32scatter_epi32): Ditto.
37113 (_mm_mask_i32scatter_epi32): Ditto.
37114 (_mm256_i32scatter_epi64): Ditto.
37115 (_mm256_mask_i32scatter_epi64): Ditto.
37116 (_mm_i32scatter_epi64): Ditto.
37117 (_mm_mask_i32scatter_epi64): Ditto.
37118 (_mm256_i64scatter_epi32): Ditto.
37119 (_mm256_mask_i64scatter_epi32): Ditto.
37120 (_mm_i64scatter_epi32): Ditto.
37121 (_mm_mask_i64scatter_epi32): Ditto.
37122 (_mm256_i64scatter_epi64): Ditto.
37123 (_mm256_mask_i64scatter_epi64): Ditto.
37124 (_mm_i64scatter_epi64): Ditto.
37125 (_mm_mask_i64scatter_epi64): Ditto.
37126 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37127 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37128 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37129 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37130 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37131 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37132 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37133 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37134 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37135 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37136 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37137 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37138 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37139 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37140 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37141 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37142 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37143 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37144 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37145 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37146 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37147 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37148 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37149 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37150 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37151 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37152 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37153 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37154 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37155 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37156 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37157 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37158 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37159 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37160 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37161 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37162 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37163 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37164 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37165 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37166 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37167 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37168 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37169 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37170 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37171 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37172 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37173 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37174 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37175 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37176 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37177 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37178 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37179 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37180 definitions accordingly.
37182 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37183 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37186 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37189 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37192 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37193 hard regs for inner regno.
37195 2017-01-17 Martin Liska <mliska@suse.cz>
37198 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37199 assumption and add comment.
37201 2017-01-17 Nathan Sidwell <nathan@acm.org>
37203 * ipa-visibility.c (localize_node): New function, broken out of ...
37204 (function_and_variable_visibility): ... here. Call it.
37206 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37208 PR middle-end/77445
37209 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37210 correctly set frequency of oudgoing edge.
37211 (duplicate_thread_path): Fix profile updating.
37213 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37216 * configure.ac: Add GCC_BASE_VER.
37217 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37218 version from BASE-VER file.
37219 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37220 (gcc.o): Depend on $(BASEVER).
37221 * common.opt (dumpfullversion): New option.
37222 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37223 * doc/invoke.texi: Document -dumpfullversion.
37224 * doc/install.texi: Document --with-gcc-major-version-only.
37225 * configure: Regenerated.
37227 2017-01-17 Richard Biener <rguenther@suse.de>
37229 PR tree-optimization/71433
37230 * tree-vrp.c (register_new_assert_for): Merge same asserts
37231 on all incoming edges.
37232 (process_assert_insertions_for): Handle insertions at the
37235 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37237 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37238 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37240 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37243 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37246 2017-01-17 Alan Modra <amodra@gmail.com>
37249 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37250 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37251 symbolic stack limit when pic.
37253 2017-01-16 Martin Sebor <msebor@redhat.com>
37255 PR tree-optimization/78608
37256 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37258 2017-01-16 Jeff Law <law@redhat.com>
37261 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37262 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37263 for several include directories that may be relative to sysroot.
37264 * config/i386/x-mingw32 (gplus_includedir): Define.
37265 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37266 (native_system_includedir): Likewise.
37267 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37268 override if TARGET_SYSTEM_ROOT is defined.
37269 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37271 PR tree-optimization/79090
37272 PR tree-optimization/33562
37273 PR tree-optimization/61912
37274 PR tree-optimization/77485
37275 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37276 and computed trims into the dump file.
37278 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37280 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37282 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37285 * gimplify.c (gimplify_init_constructor): If want_value and
37286 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37290 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37291 sequence. Formatting fixes.
37292 (doloop_optimize): Formatting fixes.
37295 * gcc.c (debug_level_greater_than_spec_func): New function.
37296 (static_spec_functions): Add debug-level-gt spec function.
37297 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37299 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37300 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37301 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37302 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37303 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37304 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37306 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37308 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37309 QImode fixups to general and mask registers only.
37311 2017-01-16 Carl Love <cel@us.ibm.com>
37313 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37314 for built-in functions
37315 vector signed char vec_nabs (vector signed char)
37316 vector signed short vec_nabs (vector signed short)
37317 vector signed int vec_nabs (vector signed int)
37318 vector signed long long vec_nabs (vector signed long long)
37319 vector float vec_nabs (vector float)
37320 vector double vec_nabs (vector double)
37321 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37323 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37324 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37325 * doc/extend.texi: Update the documentation file for the new built-in
37328 2017-01-16 Martin Sebor <msebor@redhat.com>
37330 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37333 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37335 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37336 UNSPEC_VSX__XXSPLTD to require special splat handling.
37338 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37341 * system.h: Poison strndup.
37343 2017-01-16 Alan Modra <amodra@gmail.com>
37346 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37349 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37351 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37353 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37355 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37356 call recog here. Assert that INSN_CODE (insn) is non-negative.
37358 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37361 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37363 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37364 in the currently scheduled RTL fragment.
37366 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37368 PR rtl-optimization/78751
37369 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37372 2017-01-14 Jeff Law <law@redhat.com>
37374 PR tree-optimization/79090
37375 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37376 variable length stores.
37377 (compute_trims): Delete dead assignment to *trim_tail.
37378 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37381 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37383 PR rtl-optimization/78626
37384 PR rtl-optimization/78727
37385 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37386 of a block, and split such blocks after everything else is finished.
37388 2017-01-14 Alan Modra <amodra@gmail.com>
37391 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37392 target legitimate_combined_insn.
37393 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37394 (rs6000_legitimate_combined_insn): New function.
37395 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37397 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37398 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37399 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37401 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37403 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37405 2017-01-13 Jeff Law <law@redhat.com>
37407 PR tree-optimization/33562
37408 PR tree-optimization/61912
37409 PR tree-optimization/77485
37410 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37412 (delete_dead_assignment): Likewise.
37413 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37414 statement to delete_dead_call and delete_dead_assignment.
37416 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37419 * substring-locations.c (format_warning_va): Strengthen case 1 so
37420 that both endpoints of the substring must be within the format
37421 range for just the substring to be printed.
37423 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37425 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37426 * config/i386/i386.c (ix86_target_string): Add missing options
37427 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37428 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37429 flags_other and ix86_target_other to flags2_other. Display unknown
37431 (ix86_valid_target_attribute_inner_p): Add missing options and
37432 reorder options by implied ISAs, as in ix86_target_string.
37434 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37436 * hash-table.h (hash_table::too_empty_p): New function.
37437 (hash_table::expand): Use it.
37438 (hash_table::traverse): Likewise.
37439 (hash_table::empty_slot): Use sizeof (value_type) instead of
37440 sizeof (PTR) to convert bytes to elements. Shrink the table
37441 if the current size is excessive for the current number of
37444 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37446 * ira-costs.c (record_reg_classes): Break from the inner loop
37447 early once alt_fail is known to be true. Update outer loop
37448 handling accordingly.
37450 2017-01-13 Jeff Law <law@redhat.com>
37452 * tree-ssa-dse.c (decrement_count): New function.
37453 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37454 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37455 when we know the partially dead statement is a mem* function.
37457 PR tree-optimization/61912
37458 PR tree-optimization/77485
37459 * tree-ssa-dse.c: Include expr.h.
37460 (maybe_trim_constructor_store): New function.
37461 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37463 PR tree-optimization/33562
37464 PR tree-optimization/61912
37465 PR tree-optimization/77485
37466 * doc/invoke.texi: Document new dse-max-object-size param.
37467 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37468 * tree-ssa-dse.c: Include params.h.
37469 (dse_store_status): New enum.
37470 (initialize_ao_ref_for_dse): New, partially extracted from
37472 (valid_ao_ref_for_dse, normalize_ref): New.
37473 (setup_live_bytes_from_ref, compute_trims): Likewise.
37474 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37475 (maybe_trim_partially_dead_store): Likewise.
37476 (maybe_trim_complex_store): Likewise.
37477 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37478 Track what bytes live from the original store. Return tri-state
37479 for dead, partially dead or live.
37480 (dse_dom_walker): Add constructor, destructor and new private members.
37481 (delete_dead_call, delete_dead_assignment): New extracted from
37483 (dse_optimize_stmt): Make a member of dse_dom_walker.
37484 Use initialize_ao_ref_for_dse.
37486 PR tree-optimization/33562
37487 PR tree-optimization/61912
37488 PR tree-optimization/77485
37489 * sbitmap.h (bitmap_count_bits): Prototype.
37490 (bitmap_clear_range, bitmap_set_range): Likewise.
37491 * sbitmap.c (bitmap_clear_range): New function.
37492 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37494 2017-01-13 Martin Liska <mliska@suse.cz>
37497 * function.c (set_cfun): Add new argument force.
37498 * function.h (set_cfun): Likewise.
37499 * ipa-inline-transform.c (inline_call): Use the function when
37500 strict alising from is dropped for function we inline to.
37502 2017-01-13 Richard Biener <rguenther@suse.de>
37504 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37505 for dumping GIMPLE INTEGER_CSTs.
37507 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37509 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37510 to 201112L since C++17.
37512 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37515 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37516 if -fsanitize=kernel-address is present.
37518 2017-01-13 Richard Biener <rguenther@suse.de>
37520 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37521 as _Literal ( type ) number in case usual suffixes do not
37522 preserve all information.
37524 2017-01-13 Richard Biener <rguenther@suse.de>
37526 PR tree-optimization/77283
37527 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37528 and ssa-iterators.h.
37529 (is_feasible_trace): Implement a cost model based on joiner
37532 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37535 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37536 char or short to __float128/_Float128 directly.
37538 2017-01-12 Martin Sebor <msebor@redhat.com>
37540 to -Wformat-overflow.
37541 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37542 (min_bytes_remaining): Same.
37543 (get_string_length): Same.
37544 (format_string): Same.
37545 (format_directive): Same.
37547 (pass_sprintf_length::handle_gimple_call): Same.
37549 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37551 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37552 info.nowrite calls with no lhs that can't throw. Return bool
37553 whether gsi_remove has been called or not.
37554 (pass_sprintf_length::handle_gimple_call): Return bool whether
37555 try_substitute_return_value called gsi_remove. Formatting fix.
37556 (pass_sprintf_length::execute): Don't use gsi_remove if
37557 handle_gimple_call returned true.
37560 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37561 be removed due to side-effects, don't remove following barrier nor
37562 turn the successor edge into fallthru edge.
37564 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37567 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37568 element-reversing loads and stores as not swappable.
37570 2017-01-12 Nathan Sidwell <nathan@acm.org>
37571 Nicolai Stange <nicstange@gmail.com>
37573 * combine.c (try_combine): Don't ignore result of overlap checking
37574 loop. Combine overlap & asm check into single loop.
37576 2017-01-12 Richard Biener <rguenther@suse.de>
37578 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37579 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37581 2017-01-12 Richard Biener <rguenther@suse.de>
37583 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37584 and TS_TARGET_OPTION directly derive from TS_BASE.
37585 * tree-core.h (tree_optimization_option): Derive from tree_base.
37586 (tree_target_option): Likewise.
37588 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37590 * config/i386/i386.c (memory_address_length): Increase len
37591 only when rip_relative_addr_p returns false.
37593 2017-01-11 Julia Koval <julia.koval@intel.com>
37595 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37596 (OPTION_MASK_ISA_SGX_SET): New.
37597 (ix86_handle_option): Handle OPT_msgx.
37598 * config.gcc: Added sgxintrin.h.
37599 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37600 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37601 * config/i386/i386.c (ix86_target_string): Add -msgx.
37603 (ix86_option_override_internal): Handle new options.
37604 (ix86_valid_target_attribute_inner_p): Add sgx.
37605 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37606 * config/i386/i386.opt: Add msgx.
37607 * config/i386/sgxintrin.h: New file.
37608 * config/i386/x86intrin.h: Add sgxintrin.h.
37610 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37613 * fold-const.c (maybe_nonzero_address): Return 1 for function
37615 (tree_single_nonzero_warnv_p): Don't handle function local objects
37619 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37622 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37625 * opts.c: Include "spellcheck.h"
37626 (struct string_fragment): New struct.
37627 (struct edit_distance_traits<const string_fragment &>): New
37629 (get_closest_sanitizer_option): New function.
37630 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37632 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37634 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37636 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37637 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37638 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37639 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37640 for initial die_offset if dwarf_split_debug_info.
37641 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37642 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37644 (output_skeleton_debug_sections): Formatting fix. Use
37645 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37646 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37648 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37650 * config/arm/cortex-a53.md: Add bypasses for
37651 cortex_a53_r2f_cvt.
37652 (cortex_a53_r2f): Only use for transfers.
37653 (cortex_a53_f2r): Likewise.
37654 (cortex_a53_r2f_cvt): Add reservation for conversions.
37655 (cortex_a53_f2r_cvt): Likewise.
37657 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37659 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37660 to all inlined functions, change static to extern.
37662 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37665 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37667 (arm_assemble_integer): Likewise.
37669 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37671 * config.gcc: Use new awk script to check CPU, FPU and architecture
37672 parameters for --with-... options.
37673 * config/arm/parsecpu.awk: New file
37674 * config/arm/arm-cpus.in: New file.
37675 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37677 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37679 * config/arm/t-arm: Update dependency rules.
37680 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37681 of processing .def files.
37682 * config/arm/genopt.sh: Deleted.
37683 * config/arm/gentune.sh: Deleted.
37684 * config/arm/arm-cores.def: Deleted.
37685 * config/arm/arm-arches.def: Deleted.
37686 * config/arm/arm-fpus.def: Deleted.
37687 * config/arm/arm-tune.md: Regenerated.
37688 * config/arm/arm-tables.opt: Regenerated.
37689 * config/arm/arm-cpu.h: New generated file.
37690 * config/arm/arm-cpu-data.h: New generated file.
37691 * config/arm/arm-cpu-cdata.h: New generated file.
37693 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37696 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37698 (input_varpool_node): Unpack dynamically_initialized bit.
37700 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37702 PR rtl-optimization/79032
37703 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37704 the alignment of the adjusted memory reference against that of MODE,
37705 instead of the alignment of the original memory reference.
37707 2017-01-11 Martin Jambor <mjambor@suse.cz>
37709 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37711 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37712 decorated functions.
37714 2017-01-11 Richard Biener <rguenther@suse.de>
37716 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37717 set range/nonnull info for PHI results. Do not set it on
37718 stmts marked for removal.
37720 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37722 * expr.c (store_field): In the bitfield case, fetch the return value
37723 from the registers before applying a single big-endian adjustment.
37724 Always do a final load for a BLKmode value not larger than a word.
37726 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37729 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37730 that we correctly handle column numbers greater than
37731 LINE_MAP_MAX_COLUMN_NUMBER.
37733 2017-01-10 Martin Sebor <msebor@redhat.com>
37735 PR middle-end/78245
37736 * gimple-ssa-sprintf.c (get_destination_size): Call
37737 {init,fini}object_sizes.
37738 * tree-object-size.c (addr_object_size): Adjust.
37739 (pass_through_call): Adjust.
37740 (pass_object_sizes::execute): Adjust.
37741 * tree-object-size.h (fini_object_sizes): Declare.
37743 2017-01-10 Martin Sebor <msebor@redhat.com>
37745 PR tree-optimization/78775
37746 * builtins.c (get_size_range): Move...
37747 * calls.c: ...to here.
37748 (alloc_max_size): Accept zero argument.
37749 (operand_signed_p): Remove.
37750 (maybe_warn_alloc_args_overflow): Call get_size_range.
37751 * calls.h (get_size_range): Declare.
37753 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37755 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37756 from TI's devices.csv file as of September 2016.
37757 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37759 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37761 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37762 * doc/invoke.texi: Likewise.
37763 * doc/md.texi: Likewise.
37764 * doc/objc.texi: Likewise.
37766 2017-01-10 Joshua Conner <joshconner@google.com>
37768 * config/arm/fuchsia-elf.h: New file.
37769 * config/fuchsia.h: New file.
37770 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37771 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37773 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37775 2016-01-10 Richard Biener <rguenther@suse.de>
37777 PR tree-optimization/79034
37778 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37779 Propagate out degenerate PHIs in the joiner.
37781 2017-01-10 Martin Liska <mliska@suse.cz>
37783 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37784 (sort_congruence_classes_by_decl_uid): Likewise.
37785 (sort_congruence_class_groups_by_decl_uid): Likewise.
37786 (sem_item_optimizer::merge_classes): Sort class, groups in these
37787 classes and members in the groups by DECL_UID of declarations.
37788 This would make merge operations stable.
37790 2017-01-10 Martin Liska <mliska@suse.cz>
37792 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37793 usage of m_classes_vec.
37794 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37795 (sem_item_optimizer::get_group_by_hash): Likewise.
37796 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37797 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37798 (sem_item_optimizer::verify_classes): Likewise.
37799 (sem_item_optimizer::process_cong_reduction): Likewise.
37800 (sem_item_optimizer::dump_cong_classes): Likewise.
37801 (sem_item_optimizer::merge_classes): Likewise.
37802 * ipa-icf.h (congruence_class_hash): Rename from
37803 congruence_class_group_hash. Remove declaration of m_classes_vec.
37805 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37807 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37808 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37809 * config.gcc: Add avx512vpopcntdqintrin.h.
37810 * config/i386/avx512vpopcntdqintrin.h: New.
37811 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37812 * config/i386/i386-builtin-types.def: Add new types.
37813 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37814 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37815 __builtin_ia32_vpopcountq_v8di_mask): New.
37816 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37817 __AVX512VPOPCNTDQ__.
37818 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37819 (PTA_AVX512VPOPCNTDQ): Define.
37820 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37821 TARGET_AVX512VPOPCNTDQ_P): Define.
37822 * config/i386/i386.opt: Add mavx512vpopcntdq.
37823 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37824 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37826 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37828 PR middle-end/77484
37829 * predict.def (PRED_CALL): Set to 67.
37831 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37833 * expr.c (store_field): In the bitfield case, if the value comes from
37834 a function call and is of an aggregate type returned in registers, do
37835 not modify the field mode; extract the value in all cases if the mode
37836 is BLKmode and the size is not larger than a word.
37838 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37841 * config/i386/cpuid.h: Fix undefined behavior.
37843 2017-01-04 Jeff Law <law@redhat.com>
37845 PR tree-optimization/79007
37846 PR tree-optimization/67955
37847 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37848 conservative for pt.null when flag_non_call_exceptions is on.
37850 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37852 PR translation/79019
37853 PR translation/79020
37854 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37855 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37856 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37858 * config/avr/avr.opt (maccumulate-args): Likewise.
37859 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37860 * common.opt (freport-bug): Likewise.
37861 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37862 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37863 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37864 translatable string.
37865 * config/i386/i386.c (function_value_32): Likewise.
37866 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37867 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37869 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37870 * common/config/msp430/msp430-common.c (msp430_handle_option):
37872 * symtab.c (symtab_node::verify_base): Likewise.
37873 * opts.c (set_debug_level): Likewise.
37874 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37875 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37876 missing whitespace to translatable strings.
37877 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37878 * config/sh/superh.h: Likewise.
37879 * config/i386/xopintrin.h: Likewise.
37880 * config/i386/znver1.md: Likewise.
37881 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37882 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37883 * double-int.h (struct double_int): Likewise.
37884 * double-int.c (div_and_round_double): Likewise.
37885 * wide-int.cc: Likewise.
37886 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37887 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37888 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37889 (crossjumps_occurred): ... this.
37890 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37893 PR tree-optimization/78899
37894 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37895 returning bool return struct loop *, NULL for failure and the new
37897 (versionable_outer_loop_p): Don't version outer loop if it has
37898 dont_vectorized bit set.
37899 (tree_if_conversion): When versioning outer loop, ensure
37900 tree_if_conversion is performed also on the inner loop of the
37901 non-vectorizable outer loop copy.
37902 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37903 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37904 prevent vectorization of it.
37905 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37906 the outer loop vectorization of the non-scalar version is attempted
37907 before vectorization of the inner loop in scalar version. If
37908 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37909 vectorization of its inner loop.
37910 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37911 has 2 inner loops, rename also on edges from bb whose single pred
37912 is outer_loop->header. Fix typo in function comment.
37914 2017-01-09 Martin Sebor <msebor@redhat.com>
37917 * asan.c (asan_emit_stack_protection): Increase local buffer size
37918 to avoid snprintf truncation warning.
37920 2017-01-09 Andrew Pinski <apinski@cavium.com>
37922 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
37923 to reference thunderx2t99 for the tuning structure
37924 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
37926 (thunderx2t99_extra_costs): This.
37927 * config/aarch64/aarch64-tune.md: Regenerate.
37928 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
37929 (vulcan_addrcost_table): This.
37930 (vulcan_regmove_cost): Rename to ...
37931 (thunderx2t99_regmove_cost): This.
37932 (vulcan_vector_cost): Rename to ...
37933 (thunderx2t99_vector_cost): this.
37934 (vulcan_branch_cost): Rename to ...
37935 (thunderx2t99_branch_cost): This.
37936 (vulcan_tunings): Rename to ...
37937 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
37938 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
37940 2017-01-09 Martin Jambor <mjambor@suse.cz>
37944 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
37945 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
37946 (propagate_vr_accross_jump_function): Use the above function for all
37947 value range computations for pass-through jump functions and type
37948 converasion from explicit value range values.
37949 (ipcp_propagate_stage): Do not attempt to deduce types of formal
37950 parameters from TYPE_ARG_TYPES.
37951 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
37952 (ipa_write_node_info): Stream type of the actual argument.
37953 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
37955 2017-01-09 Martin Liska <mliska@suse.cz>
37958 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
37959 (lookup_compiler): Do not show error message with have_E.
37961 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37963 PR tree-optimization/78938
37964 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
37965 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
37966 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
37967 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
37970 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37972 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
37975 2017-01-09 Richard Biener <rguenther@suse.de>
37977 PR tree-optimization/78997
37978 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
37979 name condition properly.
37981 2017-01-09 Richard Biener <rguenther@suse.de>
37984 * dwarf2out.c (is_cxx): New overload with context.
37985 (is_naming_typedef_decl): Use it.
37987 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37989 * invoke.texi (Option Summary): Correct spacing in option lists
37990 and add line breaks to fix over-long lines.
37992 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37994 PR middle-end/17660
37996 * extend.texi (Common Variable Attributes): Add xref to GCC
37997 Internals manual to explain mode attribute keywords.
37999 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38002 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38003 and Preprocessor Options.
38004 (Options for Linking): Document -pthread here....
38005 (RS/6000 and PowerPC Options): ...not here.
38006 (Solaris 2 Options): ...or here.
38007 * doc/cppopts.texi: Document -pthread.
38009 2017-01-08 Martin Sebor <msebor@redhat.com>
38011 PR middle-end/77708
38012 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38013 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38014 New member functions.
38015 (format_directive): Used them.
38017 (pass_sprintf_length::handle_gimple_call): Same.
38018 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38019 to avoid truncation for any argument.
38020 (extract_affine_mul): Same.
38021 * tree.c (get_file_function_name): Same.
38023 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38025 PR middle-end/77484
38026 * predict.def (PRED_INDIR_CALL): Set to 86.
38028 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38030 PR preprocessor/54124
38031 * doc/cppopts.texi: Reformat -d subtable to list the full name
38032 of the options. Add cross-reference to the docs for the general
38033 compiler -d options.
38034 * doc/invoke.texi (Developer Options): Add cross-reference to the
38035 preprocessor-specific -d option documentation.
38037 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38039 PR preprocessor/13498
38040 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38041 redudant material, and reflect new command-line options.
38042 (System Headers): Likewise.
38044 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38046 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38047 -isystem, and -idirafter. Copy-edit.
38048 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38049 default for -ftrack-macro-expansion. Delete obsolete and
38050 badly-formatted implementation details about -fdebug-cpp output.
38051 * doc/cppwarnopts.texi: Copy-edit.
38053 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38056 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38057 that the transition from a max line width >= 1<<10 to narrower
38058 lines works correctly.
38060 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38062 * doc/options.texi (PerFunction): New.
38063 * opt-functions.awk (switch_flags): Map both Optimization and
38064 PerFunction to CL_OPTIMIZATION.
38065 * opth-gen.awk: Test for PerFunction flag along with
38067 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38068 it only when the latter is present. Skip those that don't in
38069 the hash function generator.
38070 * common.opt (fvar-tracking): Mark as PerFunction instead of
38072 (fvar-tracking-assignments): Likewise.
38073 (fvar-tracking-assignments-toggle): Likewise.
38074 (fvar-tracking-uninit): Likewise.
38076 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38078 PR translation/79018
38079 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38082 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38085 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38086 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38087 TARGET_LONG_JUMP_TABLE_OFFSETS.
38088 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38089 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38090 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38091 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38092 * config/m68k/m68k.md (tablejump expander): Likewise.
38093 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38094 TARGET_LONG_JUMP_TABLE_OFFSETS.
38095 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38096 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38098 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38099 David Holsgrove <david.holsgrove@xilinx.com>
38101 * common/config/microblaze/microblaze-common.c
38102 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38103 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38105 * config/microblaze/microblaze.c (microblaze_must_save_register)
38106 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38108 (microblaze_eh_return): New function.
38109 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38110 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38111 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38112 * config/microblaze/microblaze.md (eh_return): New pattern.
38114 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38116 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38117 GCC_DIAGNOSTIC_STRINGIFY): Define.
38119 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38121 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38123 * config/arm/arm.md (<mcrr>): New.
38125 * config/arm/arm.c (arm_arch5te): New.
38126 (arm_option_override): Set arm_arch5te.
38127 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38129 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38130 (arm_mcrr_qualifiers): ... this. New.
38131 (MRRC_QUALIFIERS): Define to...
38132 (arm_mrrc_qualifiers): ... this. New.
38133 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38135 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38136 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38137 (MRRCI, mrrc, MRRC): New.
38138 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38139 VUNSPEC_MRRC2): New.
38141 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38143 * config/arm/arm.md (<mcr>): New.
38145 * config/arm/arm.c (arm_coproc_builtin_available): Add
38146 support for mcr, mrc, mcr2 and mrc2.
38147 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38148 (arm_mcr_qualifiers): ... this. New.
38149 (MRC_QUALIFIERS): Define to ...
38150 (arm_mrc_qualifiers): ... this. New.
38151 (MCR_QUALIFIERS): Define to ...
38152 (arm_mcr_qualifiers): ... this. New.
38153 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38155 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38156 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38157 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38158 VUNSPEC_MRC2): New.
38160 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38162 * config/arm/arm.md (*ldc): New.
38166 * config/arm/arm.c (arm_coproc_builtin_available): Add
38167 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38168 (arm_coproc_ldc_stc_legitimate_address): New.
38169 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38170 'qualifier_const_pointer'.
38171 (LDC_QUALIFIERS): Define to...
38172 (arm_ldc_qualifiers): ... this. New.
38173 (STC_QUALIFIERS): Define to...
38174 (arm_stc_qualifiers): ... this. New.
38175 * config/arm/arm-protos.h
38176 (arm_coproc_ldc_stc_legitimate_address): New.
38177 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38178 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38179 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38180 stc2, stcl, stc2l): New.
38181 * config/arm/constraints.md (Uz): New.
38182 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38183 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38184 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38185 VUNSPEC_STC2L): New.
38187 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38189 * config/arm/arm.md (<cdp>): New.
38190 * config/arm/arm.c (neon_const_bounds): Rename this ...
38191 (arm_const_bounds): ... this.
38192 (arm_coproc_builtin_available): New.
38193 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38194 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38195 (CDP_QUALIFIERS): Define to...
38196 (arm_cdp_qualifiers): ... this. New.
38198 (arm_expand_builtin_args): Add case for 6 arguments.
38199 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38200 (arm_const_bounds): ... this.
38201 (arm_coproc_builtin_available): New.
38202 * config/arm/arm_acle.h (__arm_cdp): New.
38204 * config/arm/arm_acle_builtins.def (cdp): New.
38206 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38207 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38208 'arm_const_bounds'.
38209 * config/arm/types.md (coproc): New.
38210 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38211 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38212 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38213 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38215 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38217 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38218 (UBINOP_QUALIFIERS): New.
38220 (acle_builtin_data): New. Change comment.
38221 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38222 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38223 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38224 arm_acle_builtins.def.
38225 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38226 (arm_init_acle_builtins): New.
38227 (CRC32_BUILTIN): Remove.
38228 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38229 crc32cb, crc32ch and crc32cw.
38230 (arm_init_crc32_builtins): Remove.
38231 (arm_init_builtins): Use arm_init_acle_builtins rather
38232 than arm_init_crc32_builtins.
38233 (arm_expand_acle_builtin): New.
38234 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38235 * config/arm/arm_acle_builtins.def: New.
38237 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38239 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38240 (arm_builtin_datum): ... this.
38241 (arm_init_neon_builtin): Rename to ...
38242 (arm_init_builtin): ... this. Add a new parameters PREFIX
38243 and USE_SIG_IN_NAME.
38244 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38245 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38246 'arm_builtin_datum'.
38247 (arm_init_vfp_builtins): Likewise.
38248 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38249 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38250 (arm_expand_neon_args): Rename to ...
38251 (arm_expand_builtin_args): ... this. Rename builtin_arg
38252 enum values and differentiate between ARG_BUILTIN_MEMORY
38253 and ARG_BUILTIN_NEON_MEMORY.
38254 (arm_expand_neon_builtin_1): Rename to ...
38255 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38256 values, arm_expand_builtin_args and add bool parameter NEON.
38257 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38258 (arm_expand_vfp_builtin): Likewise.
38259 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38261 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38263 PR middle-end/77484
38264 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38265 * predict.c (tree_estimate_probability_bb): Reverse direction of
38266 polymorphic call predictor.
38268 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38270 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38271 (determine_pass_name_match): ...this new function and...
38272 (should_skip_pass_p): ...this new function.
38274 2017-01-06 Nathan Sidwell <nathan@acm.org>
38276 * ipa-visibility.c (function_and_variable_visibility): Reformat
38277 comments and long lines. Remove extrneous if.
38278 * symtab.c (symtab_node::make_decl_local): Fix code format.
38279 (symtab_node::set_section_for_node): Fix comment typo.
38281 2017-01-06 Martin Liska <mliska@suse.cz>
38284 * lra-constraints.c: Rename invariant to lra_invariant.
38285 * predict.c (set_even_probabilities): Initialize e to NULL.
38287 2017-01-05 Martin Sebor <msebor@redhat.com>
38289 PR tree-optimization/78910
38290 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38291 (format_integer): Correct off-by-one error in the handling
38292 of precision with negative numbers in signed conversions..
38294 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38296 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38298 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38300 PR tree-optimization/71016
38301 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38302 factor_out_conditional_conversion. Formatting fix.
38303 (factor_out_conditional_conversion): Add cond_stmt argument.
38304 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38305 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38308 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38310 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38311 read-rtl-function.o, and selftest-rtl.o.
38312 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38313 (selftest::aarch64_test_loading_full_dump): New function.
38314 (selftest::aarch64_run_selftests): New function.
38315 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38316 selftest::aarch64_run_selftests.
38317 * config/i386/i386.c
38318 (selftest::ix86_test_loading_dump_fragment_1): New function.
38319 (selftest::ix86_test_loading_call_insn): New function.
38320 (selftest::ix86_test_loading_full_dump): New function.
38321 (selftest::ix86_test_loading_unspec): New function.
38322 (selftest::ix86_run_selftests): Call the new functions.
38323 * emit-rtl.c (maybe_set_max_label_num): New function.
38324 * emit-rtl.h (maybe_set_max_label_num): New decl.
38325 * function.c (instantiate_decls): Guard call to
38326 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38327 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38329 * gensupport.c (gen_reader::gen_reader): Pass "false"
38330 for new "compact" param of rtx_reader.
38331 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38332 rather than an empty string for NULL strings.
38333 * read-md.c: Potentially include config.h rather than bconfig.h.
38334 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38335 (have_error): New global, copied from errors.c.
38336 (md_reader::read_name): Rename to...
38337 (md_reader::read_name_1): ...this, adding "out_loc" param,
38338 and converting "missing name or number" to returning false, rather
38340 (md_reader::read_name): Reimplement in terms of read_name_1.
38341 (md_reader::read_name_or_nil): New function.
38342 (md_reader::read_string): Handle "(nil)" by returning NULL.
38343 (md_reader::md_reader): Add new param "compact".
38344 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38345 (md_reader::read_file): New method.
38346 * read-md.h (md_reader::md_reader): Add new param "compact".
38347 (md_reader::read_file): New method.
38348 (md_reader::is_compact): New accessor.
38349 (md_reader::read_name): Convert return type from void to file_location.
38350 (md_reader::read_name_or_nil): New decl.
38351 (md_reader::read_name_1): New decl.
38352 (md_reader::m_compact): New field.
38353 (noop_reader::noop_reader): Pass "false" for new "compact" param
38355 (rtx_reader::rtx_reader): Add new "compact" param.
38356 (rtx_reader::read_rtx_operand): Make virtual and convert return
38357 type from void to rtx.
38358 (rtx_reader::read_until): New decl.
38359 (rtx_reader::handle_any_trailing_information): New virtual function.
38360 (rtx_reader::postprocess): New virtual function.
38361 (rtx_reader::finalize_string): New virtual function.
38362 (rtx_reader::m_in_call_function_usage): New field.
38363 (rtx_reader::m_reuse_rtx_by_id): New field.
38364 * read-rtl-function.c: New file.
38365 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38366 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38367 (selftest::verify_three_block_rtl_cfg): New decl.
38368 * read-rtl-function.h: New file.
38369 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38370 For host, include function.h, memmodel.h, and emit-rtl.h.
38371 (one_time_initialization): New function.
38372 (struct compact_insn_name): New struct.
38373 (compact_insn_names): New array.
38374 (find_code): Handle insn codes in compact dumps.
38375 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38376 (bind_subst_iter_and_attr): Likewise.
38377 (add_condition_to_string): Likewise.
38378 (add_condition_to_rtx): Likewise.
38379 (apply_attribute_uses): Likewise.
38380 (add_current_iterators): Likewise.
38381 (apply_iterators): Likewise.
38382 (initialize_iterators): Guard usage of apply_subst_iterator with
38383 #ifdef GENERATOR_FILE.
38384 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38385 (md_reader::read_mapping): Likewise.
38386 (add_define_attr_for_define_subst): Likewise.
38387 (add_define_subst_attr): Likewise.
38388 (read_subst_mapping): Likewise.
38389 (check_code_iterator): Likewise.
38390 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38392 (one_time_initialization): New function.
38393 (rtx_reader::read_until): New method.
38394 (read_flags): New function.
38395 (parse_reg_note_name): New function.
38396 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38397 Handle reuse_rtx ids.
38398 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38399 Add parsing support for RTL dumps, mirroring the special-cases in
38400 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38401 values, and calling handle_any_trailing_information.
38402 (rtx_reader::read_rtx_operand): Convert return type from void
38403 to rtx, returning return_rtx. Handle case 'e'. Call
38404 finalize_string on XSTR and XTMPL fields.
38405 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38406 "(nil)" values were omitted. Call the postprocess vfunc on the
38408 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38409 class ctor. Initialize m_in_call_function_usage. Call
38410 one_time_initialization.
38411 * rtl-tests.c (selftest::test_uncond_jump): Call
38412 set_new_first_and_last_insn.
38413 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38414 * selftest-rtl.c: New file.
38415 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38416 (selftest::get_insn_by_uid): New decl.
38417 * selftest-run-tests.c (selftest::run_tests): Call
38418 read_rtl_function_c_tests.
38419 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38420 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38423 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38425 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38426 operands in a special way. Assert that pos+len <= mode precision.
38428 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38430 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38431 3 argument Alias with unlimited for the negative form.
38432 (fno-vect-cost-model): Removed.
38434 2017-01-05 Martin Liska <mliska@suse.cz>
38436 * hsa-gen.c (gen_hsa_divmod): New function.
38437 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38439 2017-01-05 Martin Liska <mliska@suse.cz>
38442 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38445 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38447 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38448 small constant length operands.
38450 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38452 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38453 between loop iterations.
38455 2017-01-05 Martin Liska <mliska@suse.cz>
38458 * gimplify.c (gimplify_decl_expr): Compare to
38459 asan_poisoned_variables instread of checking flags.
38460 (gimplify_target_expr): Likewise.
38461 (gimplify_expr): Likewise.
38462 (gimplify_function_tree): Conditionally initialize
38463 asan_poisoned_variables.
38465 2017-01-04 Jeff Law <law@redhat.com>
38467 PR tree-optimizatin/78812
38468 * rtl.h (contains_mem_rtx_p): Prototype.
38469 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38470 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38471 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38472 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38473 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38475 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38477 * input.c (assert_char_at_range): Default-initialize actual_range.
38479 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38481 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38482 to match the caller.
38484 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38486 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38487 insns after final jump in test to emit dummy move.
38489 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38491 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38492 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38494 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38496 * multiple_target.c (create_dispatcher_calls): Init e_next.
38497 * tree-ssa-loop-split.c (split_loop): Init border.
38498 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38501 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38506 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38507 (altivec_register_operand): Do not return true if the operand
38508 contains a SUBREG mixing SImode and SFmode.
38509 (vsx_register_operand): Likewise.
38510 (vsx_reg_sfsubreg_ok): New predicate.
38511 (vfloat_operand): Do not return true if the operand contains a
38512 SUBREG mixing SImode and SFmode.
38513 (vint_operand): Likewise.
38514 (vlogical_operand): Likewise.
38515 (gpc_reg_operand): Likewise.
38516 (int_reg_operand): Likewise.
38517 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38518 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38519 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38521 (rs6000_emit_move_si_sf_subreg): New helper function.
38522 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38523 fixup SUBREGs involving SImode and SFmode.
38524 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38525 numbers for the new peephole2 optimization.
38526 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38527 the GLIBC math library that do AND/IOR/XOR operations on single
38528 precision floating point.
38529 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38530 target macros to say whether we need to avoid SUBREGs mixing
38532 (TARGET_ALLOW_SF_SUBREG): Likewise.
38533 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38534 (UNSPEC_SI_FROM_SF): Likewise.
38535 (iorxor): Change spacing.
38536 (and_ior_xor): New iterator for AND, IOR, and XOR.
38537 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38538 (movdi_from_sf_zero_ext): Likewise.
38539 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38540 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38541 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38542 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38543 (fms<mode>4): Likewise.
38544 (fnma<mode>4): Likewise.
38545 (fnms<mode>4): Likewise.
38546 (nfma<mode>4): Likewise.
38547 (nfms<mode>4): Likewise.
38549 2017-01-04 Marek Polacek <polacek@redhat.com>
38552 * doc/invoke.texi: Document -Wpointer-compare.
38554 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38556 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38559 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38560 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38563 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38566 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38567 documentation of the powerpc_popcntb_ok attribute.
38568 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38569 code to issue warning messages if a requested CPU configuration is
38570 not supported by the binary (assembler and loader) toolchain.
38571 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38572 made to define a built-in function that has been disabled.
38573 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38574 made to define a built-in function that has been disabled.
38575 (altivec_init_builtins): Add comment explaining why definition
38576 of the DST built-in functions is not preceded by an assertion
38577 check. Add assertions to prevent ICE if attempts are made to
38578 define an altivec predicate or an abs* built-in function that has
38580 (htm_init_builtins): Add comment explaining why definition of the
38581 htm built-in functions is not preceded by an assertion check.
38583 2017-01-04 Jeff Law <law@redhat.com>
38585 PR tree-optimizatin/67955
38586 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38587 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38588 the points-to solution does not include pt_null. Use DECL_PT_UID
38591 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38593 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38594 Use gen_int_mode instead of gen_lopwart for const_int operands.
38596 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38598 PR tree-optimization/71563
38599 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38600 out of range value - has low bits known to be zero.
38602 2017-01-04 Alan Modra <amodra@gmail.com>
38604 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38605 * configure: Regenerate.
38606 * config.in: Regenerate.
38608 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38611 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38612 a substring of the message, but strcmp with the whole message. Ifdef
38613 ENABLE_NLS, translate the message first using dgettext.
38615 2017-01-03 Jeff Law <law@redhat.com>
38617 PR tree-optimizatin/78856
38618 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38619 (mark_threaded_blocks): Remove code to truncate thread paths that
38620 cross multiple loop headers. Instead invalidate the cached loop
38621 iteration information and handle case of a thread path walking
38622 into an irreducible region.
38624 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38627 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38628 assertions. Add support for doing the signbit if the IEEE 128-bit
38629 floating point value is in a GPR.
38630 * config/rs6000/rs6000.md (Fsignbit): Delete.
38631 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38632 Update the length attribute if the value is in a GPR.
38633 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38634 the sign or zero extension instruction, since the value is always 0/1.
38635 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38638 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38639 extracting SImode to a GPR register so that we can generate a
38640 store, limit the vector to be in a traditional Altivec register
38641 for the vextuwrx instruction.
38643 2017-01-03 Ian Lance Taylor <iant@google.com>
38645 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38647 2017-01-03 Martin Sebor <msebor@redhat.com>
38649 PR tree-optimization/78696
38650 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38651 precision. Use MPFR for %f for greater fidelity. Correct handling
38653 (pass_sprintf_length::compute_format_length): Set width and precision
38654 specified by asrerisk to void_node for vararg functions.
38655 (try_substitute_return_value): Adjust dump output.
38657 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38659 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38661 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38663 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38664 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38665 -mlra/-mno-lra was passed to the compiler.
38667 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38669 PR rtl-optimization/65618
38670 * emit-rtl.c (try_split): Move initialization of "before" and
38671 "after" to just before the call to emit_insn_after_setloc.
38673 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38675 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38677 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38679 * dwarf2out.c (gen_enumeration_type_die): When
38680 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38682 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38684 PR tree-optimization/78965
38685 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38686 Change first argument from const call_info & to call_info &. For %n
38687 set info.nowrite to false.
38689 PR middle-end/78901
38690 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38691 possibly throwing calls.
38693 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38694 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38695 and fns handling, rather than in a separate case SSA_NAME.
38697 2017-01-02 Jeff Law <law@redhat.com>
38699 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38700 fixes for first_period and second_period variables.
38702 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38705 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38706 (*insvqi_1): New insn pattern.
38707 (*insvqi_1_mem_rex64): Ditto.
38708 (*insvqi_2): Ditto.
38709 (*insvqi_3): Rename from *insvqi.
38711 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38713 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38715 * doc/cfg.texi (Edges): Remove reference to Java.
38716 (Maintaining the CFG): Ditto.
38718 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38720 PR middle-end/77674
38721 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38722 transparent aliases.
38724 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38726 PR middle-end/77484
38727 * predict.def (PRED_CALL): Update hitrate.
38728 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38729 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38730 into direct/indirect/polymorphic variants.
38732 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38734 Update copyright years.
38736 * gcc.c (process_command): Update copyright notice dates.
38737 * gcov-dump.c (print_version): Ditto.
38738 * gcov.c (print_version): Ditto.
38739 * gcov-tool.c (print_version): Ditto.
38740 * gengtype.c (create_file): Ditto.
38741 * doc/cpp.texi: Bump @copying's copyright year.
38742 * doc/cppinternals.texi: Ditto.
38743 * doc/gcc.texi: Ditto.
38744 * doc/gccint.texi: Ditto.
38745 * doc/gcov.texi: Ditto.
38746 * doc/install.texi: Ditto.
38747 * doc/invoke.texi: Ditto.
38749 Copyright (C) 2017 Free Software Foundation, Inc.
38751 Copying and distribution of this file, with or without modification,
38752 are permitted in any medium without royalty provided the copyright
38753 notice and this notice are preserved.