1 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
4 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
5 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
6 ATTR_MATHFN_FPROUNDING_ERRNO.
8 2017-12-12 Richard Biener <rguenther@suse.de>
10 PR tree-optimization/83385
11 * graphite-scop-detection.c (get_order, order): Remove.
12 (bb_to_rpo): New global.
14 (build_scops): Sort pbbs in RPO order.
16 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
18 * combine.c (simplify_set): Do not transform subregs to zero_extends
19 if the destination mode is a vector mode.
21 2017-12-12 Jakub Jelinek <jakub@redhat.com>
23 PR tree-optimization/83359
24 * tree-cfg.h (fold_loop_internal_call): Declare.
25 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
26 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
27 (find_loop_dist_alias): New function.
28 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
29 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
32 PR tree-optimization/80631
33 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
34 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
35 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
36 hardcoding zero as the value if COND_EXPR is never true. For
37 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
38 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
39 hardcoding MAX_EXPR as the reduction operation.
40 (is_nonwrapping_integer_induction): Allow negative step.
41 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
42 vect_create_epilog_for_reduction, if no value is suitable, don't
43 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
45 2017-12-12 Richard Biener <rguenther@suse.de>
47 PR tree-optimization/81889
48 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
49 range info from the non-wrapping IV instead of just the range
52 2017-12-12 Julia Koval <julia.koval@intel.com>
54 * config.gcc: Add vaesintrin.h.
55 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI): New type.
56 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
57 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi): New builtins.
58 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
59 * config/i386/immintrin.h: Include vaesintrin.h.
60 * config/i386/sse.md (vaesdec_<mode>): New pattern.
61 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
62 _mm_aesdec_epi128): New intrinsics.
64 2017-12-12 Julia Koval <julia.koval@intel.com>
66 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
67 OPTION_MASK_ISA_VAES_UNSET): New.
68 (ix86_handle_option): Handle -mvaes.
69 * config/i386/cpuid.h: Define bit_VAES.
70 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
71 * config/i386/i386-c.c (__VAES__): New.
72 * config/i386/i386.c (ix86_target_string): Add -mvaes.
73 (ix86_valid_target_attribute_inner_p): Ditto.
74 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
75 * config/i386/i386.opt: Add -mvaes.
76 * doc/invoke.texi: Ditto.
78 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
80 * debug.h (gcc_debug_hooks): Add inline_entry.
81 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
82 * debug.c (do_nothing_debug_hooks): Likewise.
83 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
84 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
85 (dwarf2_lineno_debug_hooks): Likewise.
87 * common.opt (gstatement-frontiers): New, setting
88 debug_nonbind_markers_p.
89 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
90 * toplev.c (process_options): Autodetect value for debug statement
92 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
93 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
95 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
96 markers. Integrate source bind into debug stmt expand loop.
97 (pass_expand::execute): Check debug marker limit. Avoid deep
98 TER and expand debug locations for debug bind insns only.
99 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
101 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
102 * final.c (reemit_insn_block_notes): Take current block from
103 nonbind markers. Declare note where it's first set.
104 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
105 begin stmt markers if enabled.
106 (notice_source_line): Handle nonbind markers. Fail if their
107 location is unknown or that of builtins.
108 (rest_of_handle_final): Convert begin stmt markers to notes if
109 var-tracking didn't run.
110 (rest_of_clean_state): Skip begin stmt markers.
111 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
113 * function.c (allocate_struct_function): Set begin_stmt_markers.
114 * function.h (struct function): Add debug_marker_count counter
115 and debug_nonbind_markers flag.
116 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
117 * gimple-low.c (lower_function_body): Adjust
118 debug_nonbind_markers.
119 (lower_stmt): Drop or skip gimple debug stmts.
120 (lower_try_catch): Skip debug stmts.
121 * gimple.c (gimple_build_debug_begin_stmt): New.
122 (gimple_copy): Increment debug_marker_count if copying one.
123 * gimple.h (gimple_build_debug_begin_stmt): Declare.
124 * gimplify.c (rexpr_location): New.
125 (rexpr_has_location): New.
126 (warn_switch_unreachable_r): Handle gimple debug stmts.
127 (shortcut_cond_r): Call expr_location.
129 (find_goto_label): New.
130 (shortcut_cond_expr): Call expr_has_location, expr_location, and
132 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
134 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
135 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
136 (LANG_HOOKS_INITIALIZER): ... this.
137 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
138 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
140 (update_ebb_live_info): Skip debug insn markers.
141 * lra.c (debug_insn_static_data): Rename to...
142 (debug_bind_static_data): ... this.
143 (debug_marker_static_data): New.
144 (lra_set_insn_recog_data): Select one of the above depending
146 (lra_update_isn_regno_info): Don't assume debug insns have
148 (push_insns): Skip debug insns.
149 * lto-streamer-in.c (input_function): Drop debug stmts
150 depending on active options. Adjust debug_nonbind_markers.
151 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
152 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
153 begin stmt marker notes.
154 (print_insn): Likewise.
155 * recog.c (extract_insn): Recognize rtl for debug markers.
156 * rtl.def (DEBUG_MARKER): New.
157 * tree-inline.c: Include params.h.
158 (remap_gimple_stmt): Handle nonbind markers.
159 (maybe_move_debug_stmts_to_successors): Likewise.
160 (copy_debug_stmt): Likewise.
161 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
162 markers regardless of no side effects.
163 (tsi_link_before): Don't update container's side effects when adding
165 (tsi_link_after): Likewise.
166 (expr_first): Skip begin stmt markers.
167 (expr_last): Likewise.
168 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
169 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
170 Disregard nonbind markers.
171 * tree.c (make_node_stat): Don't set side effects for begin stmt
173 (build1_stat): Likewise.
174 * tree.def (DEBUG_BEGIN_STMT): New.
175 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
176 * var-tracking.c (delete_debug_insns): Renamed to...
177 (delete_vta_debug_insns): ... this.
178 (reemit_marker_as_note): New.
179 (vt_initialize): Reemit markers.
180 (delete_vta_debug_insns): Likewise.
181 (vt_debug_insns_local): Reemit or delete markers.
182 (variable_tracking_main_1): Likewise.
183 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
184 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
185 (gimple_debug_nonbind_marker_p): New.
186 (gimple_build_debug_bind): Adjust.
187 (gimple_build_debug_begin_stmt): New.
188 * doc/invoke.texi (max-debug-marker-count): New param.
189 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
190 (debug_parameter_ref, debug_marker): New.
191 (NOTE_INSN_BEGIN_STMT): New.
192 (DEBUG_INSN): Describe begin stmt markers.
194 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
195 without debug insns, we wouldn't, but clean up debug insns
196 after a control flow insn nevertheless.
198 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
199 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
200 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
201 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
202 (rtl_tidy_fallthru_edge): Likewise.
203 (rtl_verify_fallthru): Likewise.
204 (rtl_verify_bb_layout): Likewise.
205 (skip_insns_after_block): Likewise.
206 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
207 * dwarf2out.c: Include print-rtl.h.
208 (dwarf2out_next_real_insn): New.
209 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
210 Dump debug binds in asm comments.
211 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
212 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
213 callers to use gsi_start_nondebug_bb instead.
214 (gsi_after_labels): Skip gimple debug stmts.
215 (gsi_start_nondebug): New.
216 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
217 (proper_loop_form_for_interchange): Adjust.
218 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
219 * gimple.h (gimple_seq_last_nondebug_stmt): New.
220 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
221 (collect_fallthrough_labels): Likewise.
222 (should_warn_for_implicit_fallthrough): Likewise.
223 (warn_implicit_fallthrough_r): Likewise.
224 (expand_FALLTHROUGH_r): Likewise.
225 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
226 (graphite_copy_stmts_from_block): Skip nonbind markers.
227 * haifa-sched.c (sched_extend_bb): Skip debug insns.
228 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
229 * jump.c (clean_barriers): Skip debug insns.
230 * omp-expand.c (expand_parallel_call): Skip debug insns.
231 (expand_task_call): Likewise.
232 (remove_exit_barrier): Likewise.
233 (expand_omp_taskreg): Likewise.
234 (expand_omp_for_init_counts): Likewise.
235 (expand_omp_for_generic): Likewise.
236 (expand_omp_for_static_nochunk): Likewise.
237 (expand_omp_for_static_chunk): Likewise.
238 (expand_omp_simd): Likewise.
239 (expand_omp_taskloop_for_outer): Likewise.
240 (expand_omp_taskloop_for_inner): Likewise.
241 (expand_oacc_for): Likewise.
242 (expand_omp_sections): Likewise.
243 (expand_omp_single): Likewise.
244 (expand_omp_synch): Likewise.
245 (expand_omp_atomic_load): Likewise.
246 (expand_omp_atomic_store): Likewise.
247 (expand_omp_atomic_fetch_op): Likewise.
248 (expand_omp_atomic_pipeline): Likewise.
249 (expand_omp_atomic_mutex): Likewise.
250 (expand_omp_target): Likewise.
251 (grid_expand_omp_for_loop): Likewise.
252 (grid_expand_target_grid_body): Likewise.
253 (build_omp_regions_1): Likewise.
254 * omp-low.c (check_combined_parallel): Skip debug stmts.
255 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
256 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
257 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
259 * tree-cfg.c (make_blobs_1): Skip debug stmts.
260 (make_edges): Likewise.
261 (cleanup_dead_labels): Likewise.
262 (gimple_can_merge_blocks_p): Likewise.
263 (stmt_starts_bb_p): Likewise.
264 (gimple_block_label): Likewise.
265 (gimple_redirect_edge_and_branch): Likewise.
266 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
268 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
270 * tree-pretty-print (print_declaration): Omit initializer in slim
272 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
274 (eliminate_unnecessary_stmts): Stabilize block removal order.
275 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
276 * var-tracking.c (get_first_insn): New.
277 (vt_emit_notes): Call it.
278 (vt_initialize): Walk any insns before the first BB.
279 (delete_debug_insns): Likewise.
281 * gimple.h (enum gimple_debug_subcode): Add
282 GIMPLE_DEBUG_BEGIN_STMT.
283 (gimple_debug_begin_stmt_p): New.
284 (gimple_debug_nonbind_marker_p): New.
285 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
286 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
287 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
288 * insn-notes.def (BEGIN_STMT): New.
289 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
290 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
291 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
292 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
293 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
294 (INSN_DEBUG_MARKER_KIND): New.
295 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
296 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
297 (INSN_VAR_LOCATION_PTR): New.
298 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
300 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
301 deps only if debug bind insns are enabled.
302 (pass_expand::execute): Avoid deep TER and expand
303 debug locations for debug bind insns only.
304 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
305 debug stmts special handling down to debug bind stmts.
306 * combine.c (try_combine): Narrow debug insns special handling
307 down to debug bind insns.
308 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
309 Narrow debug insns preexisting special handling down to debug
311 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
312 handling down to debug bind insns.
313 * function.c (instantiate_virtual_regs): Skip debug markers,
314 adjust handling of debug binds.
315 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
316 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
317 * haifa-sched.c (schedule_insn): Narrow special handling of debug
318 insns to debug bind insns.
319 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
320 special handling of debug stmts to debug bind stmts.
321 * ipa-split.c (split_function): Likewise.
322 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
323 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
325 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
326 * regrename.c (build_def_use): Likewise.
327 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
328 (pass_cprop_hardreg): Narrow special casing of debug insns to
330 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
331 * reload1.c (reload): Likewise.
332 * sese.c (sese_insert_phis_for_liveouts): Narrow special
333 casing of debug stmts to debug bind stmts.
334 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
335 * ssa-iterators.h (num_imm_uses): Likewise.
336 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
337 debug stmts to debug bind stmts.
338 * tree-inline.c (tree_function_versioning): Narrow special casing
339 of debug stmts to debug bind stmts.
340 * tree-loop-distribution.c (generate_loops_for_partition):
341 Narrow special casing of debug stmts to debug bind stmts.
342 * tree-sra.c (analyze_access_subtree): Narrow special casing
343 of debug stmts to debug bind stmts.
344 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
345 stmts to debug bind stmts.
346 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
347 casing of debug stmts to debug bind stmts.
348 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
349 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
350 casing of debug stmts to debug bind stmts.
351 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
353 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
354 debug stmts to debug bind stmts.
355 (gimple_replace_ssa_lhs): Likewise.
356 (insert_debug_temp_for_var_def): Likewise.
357 (insert_debug_temps_for_defs): Likewise.
358 (reset_debug_uses): Likewise.
359 * tree-ssanames.c (release_ssa_name_fn): Likewise.
360 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
361 (adjust_debug_stmts): Likewise.
362 (adjust_phi_and_debug_stmts): Likewise.
363 (vect_do_peeling): Likewise.
364 * tree-vect-loop.c (vect_transform_loop): Likewise.
365 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
366 * var-tracking.c (adjust_mems): Narrow special casing of debug
367 insns to debug bind insns.
368 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
369 (compute_bb_dataflow, vt_find_locations): Likewise.
370 (vt_expand_loc, emit_notes_for_changes): Likewise.
371 (vt_init_cfa_base): Likewise.
372 (vt_emit_notes): Likewise.
373 (vt_initialize): Likewise.
374 (vt_finalize): Likewise.
376 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
377 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
378 (next_nonnote_nondebug_insn_bb): New.
379 (prev_nonnote_nondebug_insn_bb): New.
380 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
381 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
382 (prev_nonnote_nondebug_insn_bb): Declare.
383 (next_nonnote_nondebug_insn_bb): Declare.
384 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
385 * cfgrtl.c (get_last_bb_insn): Likewise.
386 * lra.c (push_insns): Likewise.
388 2017-12-11 David Malcolm <dmalcolm@redhat.com>
391 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
393 * selftest.c (selftest::test_runner::test_runner): New ctor.
394 (selftest::test_runner::~test_runner): New dtor.
395 * selftest.h (class selftest::test_runner): New class.
397 2017-12-11 Carl Love <cel@us.ibm.com>
399 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
400 vec_extract_fp32_from_shortl]): Add #defines.
401 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
402 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
403 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
404 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
405 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
406 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
407 * doc/extend.texi: Add documentation for the added builtins.
409 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
411 PR rtl-optimization/80693
412 PR rtl-optimization/81019
413 PR rtl-optimization/81020
414 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
415 are not mentioned in i3. Place the REG_UNUSED note on i2,
416 possibly modified to REG_DEAD, if it did not originate in i3.
418 2017-12-11 Jakub Jelinek <jakub@redhat.com>
420 * recog.c (store_data_bypass_p_1): New function.
421 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
422 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
424 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
426 PR rtl-optimization/83361
427 * ifcvt.c (if_convert): Call fixup_partitions.
429 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
431 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
432 early folding of splat_u{8,16,32}.
434 2017-12-11 Jakub Jelinek <jakub@redhat.com>
436 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
437 output_operand_lossage first argument with capital letter.
438 (aarch64_override_options): Don't start error and sorry first argument
441 2017-12-11 Andi Kleen <ak@linux.intel.com>
443 PR gcov-profile/83355
444 * auto-profile.c (string_table::get_index_by_decl): Don't
445 recurse when abstract origin points to itself.
447 2017-12-11 Bin Cheng <bin.cheng@arm.com>
449 PR tree-optimization/83320
450 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
451 (prune_datarefs_not_in_loop): Ditto.
453 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
455 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
457 2017-12-10 Jakub Jelinek <jakub@redhat.com>
459 PR tree-optimization/83337
460 * gimple-loop-interchange.cc (compute_access_stride): Handle
461 bitfield DRs properly.
463 2017-12-09 Jakub Jelinek <jakub@redhat.com>
465 PR tree-optimization/83338
466 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
467 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
468 vector of unsigned integers to vector of signed integers.
470 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
472 PR rtl-optimization/83317
473 * lra-constraints.c (process_address_1): Add insn code check.
475 2017-12-08 Michael Matz <matz@suse.de>
477 Fix PR tree-optimization/83323
478 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
479 head-controlled loops and loop BBs.
480 * common.opt (funroll-and-jam): Remove, instead ...
481 (floop-unroll-and-jam): ... reuse this option.
482 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
483 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
484 (-floop-unroll-and-jam): ... this option.
486 2017-12-08 Jakub Jelinek <jakub@redhat.com>
488 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
489 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
490 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
492 2017-12-08 Richard Biener <rguenther@suse.de>
495 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
496 handle accesses outside of zero-sized vars.
498 2017-12-08 Martin Jambor <mjambor@suse.cz>
500 PR tree-optimization/83141
501 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
502 test for MEM_REFs implicitely changing types with padding. Remove
504 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
506 2017-12-08 Jakub Jelinek <jakub@redhat.com>
508 * config/arc/arc.c (arc_attribute_table): Add exclusions to
510 * config/avr/avr.c (avr_attribute_table): Likewise.
511 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
512 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
513 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
514 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
515 of struct attribute_spec.
516 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
518 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
521 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
523 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
526 * profile-count.c (profile_count::from_gcov_type): Move from
527 profile-count.h; handle overflow.
528 * profile-count.h (profile_count::from_gcov_type): Move offline.
530 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
532 PR rtl-optimization/83304
533 * combine.c (move_deaths): If we do not know where a register died,
536 2017-12-08 Richard Biener <rguenther@suse.de>
538 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
539 Provide -fopt-info-loop feedback when we interchange in a nest.
541 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
543 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
544 for armv6 ARM CPU IDs.
546 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
548 * common/config/arm/arm-common.c: Include <algorithm>.
549 (INCLUDE_VECTOR): Define.
550 (compare_opt_names): New function.
551 (arm_rewrite_selected_arch): Only strip out extensions that can be
552 expressed through -mfpu. Sort the remaining extensions
555 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
557 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
558 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
559 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
560 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
562 2017-06-08 Tristan Gingold <gindold@adacore.com>
565 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
566 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
568 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
570 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
571 of struct attribute_spec.
573 2017-12-08 Julia Koval <julia.koval@intel.com>
575 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
576 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
577 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
578 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
579 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
580 _mm_maskz_dpwssds_epi32): Ditto.
582 2017-12-08 Richard Biener <rguenther@suse.de>
584 PR tree-optimization/81303
585 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
586 conditions try to create a comparison vector type matching
587 the data vector type.
588 (vectorizable_condition): Adjust.
589 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
590 Leave invariant conditions alone in case we can vectorize those.
592 2017-12-08 Julia Koval <julia.koval@intel.com>
594 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
595 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
596 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
597 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
598 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
600 2017-12-08 Julia Koval <julia.koval@intel.com>
602 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
603 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
604 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
605 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
606 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
607 _mm_maskz_dpbusds_epi32): New intrinsics.
609 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
611 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
613 (ld<bh>io_signed): Likewise.
614 (st<bhw>io): Likewise.
615 * config/nios2/predicates.md (ldstio_memory_operand): Allow
616 SMALL_INT12 constant integer operand.
618 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
620 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
622 (nios2_large_constant_allowed): ...this. Adjust uses.
623 (nios2_plus_symbolic_constant_p): Rename to...
624 (nios2_plus_large_constant_p): ...this. Adjust uses.
625 (nios2_legitimate_address_p): Correct CONST_INT handling.
626 (nios2_symbolic_memory_operand_p): Rename to...
627 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
628 (nios2_large_constant_p): Check for large constant integers too.
629 (nios2_split_large_constant): Handle constant integers.
630 (nios2_split_symbolic_memory_operand): Rename to...
631 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
632 (nios2_legitimize_constant_address): Handle constant integers.
633 (r0rel_constant_p): Handle small constant integers.
634 (nios2_print_operand_address): Handle r0-relative integer addresses.
635 * config/nios2/nios2-protos.h: Adjust for renamed functions.
636 * config/nios2/nios2.md: Adjust for renamed functions.
638 2017-12-07 Andrew Waterman <andrew@sifive.com>
640 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
641 (TARGET_HAVE_SRODATA_SECTION): New define.
642 (riscv_select_section): New function.
644 2017-12-08 Joseph Myers <joseph@codesourcery.com>
645 Alexander Monakov <amonakov@ispras.ru>
646 Jakub Jelinek <jakub@redhat.com>
649 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
651 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
653 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
654 a bool success value. Don't call output_operand_lossage here.
655 (aarch64_print_ldpstp_address): Return a bool success value.
656 (aarch64_print_operand_address): Call output_addr_const if
657 aarch64_print_address_internal fails.
658 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
659 'y'; call output_operand_lossage instead. Call output_operand_lossage
660 if aarch64_print_ldpstp_address fails.
662 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
664 * tree-vector-builder.h
665 (tree_vector_builder::binary_encoded_nelts): Declare.
666 * tree-vector-builder.c
667 (tree_vector_builder::binary_encoded_nelts): New function.
668 * fold-const.c (negate_expr_p): Likewise.
669 (operand_equal_p, fold_checksum_tree): Likewise.
670 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
671 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
672 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
673 (uniform_vector_p): Likewise.
674 * varasm.c (const_hash_1, compare_constant): Likewise.
675 * tree-ssa-ccp.c: Include tree-vector-builder.h.
676 (valid_lattice_transition): Operate directly on the VECTOR_CST
678 * ipa-icf.c: Include tree-vector-builder.h.
679 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
680 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
682 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
684 * tree.c (build_vector): Delete.
685 * tree.h (build_vector): Make static and move into the self-testing
688 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
690 * vector-builder.h (vector_builder::derived): New const overload.
691 (vector_builder::elt): New function.
692 * tree-vector-builder.h (tree_vector_builder::type): New function.
693 (tree_vector_builder::apply_step): Declare.
694 * tree-vector-builder.c (tree_vector_builder::apply_step): New
696 * gimple-fold.h (tree_vector_builder): Declare.
697 (gimple_build_vector): Take a tree_vector_builder instead of a
698 type and vector of elements.
699 * gimple-fold.c (gimple_build_vector): Likewise.
700 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
702 (get_initial_defs_for_reduction): Likewise.
703 (vectorizable_induction): Likewise.
705 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
707 * tree-vector-builder.h
708 (tree_vector_builder::new_binary_operation): Declare.
709 * tree-vector-builder.c
710 (tree_vector_builder::new_binary_operation): New function.
711 * fold-const.c (fold_relational_const): Use it.
712 (const_binop): Likewise. Check that both input vectors have
713 the same number of elements, thus excluding things like WIDEN_SUM.
714 Check whether it is possible to operate directly on the encodings
717 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
719 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
720 new_unary_operation, operating only on the encoded elements.
721 (const_unop): Likewise.
722 (exact_inverse): Likewise.
723 (distributes_over_addition_p): New function.
724 (const_binop): Use tree_vector_builder and new_unary_operation
725 for combinations of VECTOR_CST and INTEGER_CST. Operate only
726 on the encoded elements unless the encoding is strided and the
727 operation does not distribute over addition.
728 (fold_convert_const): Use tree_vector_builder and
729 new_unary_operation. Operate only on the encoded elements
730 for truncating integer conversions, or for non-stepped encodings.
732 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
734 * config/sparc/sparc.c: Include tree-vector-builder.h.
735 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
736 * expmed.c: Include tree-vector-builder.h.
737 (make_tree): Use tree_vector_builder instead of build_vector.
738 * fold-const.c: Include tree-vector-builder.h.
739 (const_binop): Use tree_vector_builder instead of build_vector.
740 (const_unop): Likewise.
741 (native_interpret_vector): Likewise.
742 (fold_vec_perm): Likewise.
743 (fold_ternary_loc): Likewise.
744 * gimple-fold.c: Include tree-vector-builder.h.
745 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
747 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
748 (simplify_vector_constructor): Use tree_vector_builder instead
750 * tree-vect-generic.c: Include tree-vector-builder.h.
751 (add_rshift): Use tree_vector_builder instead of build_vector.
752 (expand_vector_divmod): Likewise.
753 (optimize_vector_constructor): Likewise.
754 * tree-vect-loop.c: Include tree-vector-builder.h.
755 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
756 of build_vector. Explicitly use a stepped encoding for
758 * tree-vect-slp.c: Include tree-vector-builder.h.
759 (vect_get_constant_vectors): Use tree_vector_builder instead
761 (vect_transform_slp_perm_load): Likewise.
762 (vect_schedule_slp_instance): Likewise.
763 * tree-vect-stmts.c: Include tree-vector-builder.h.
764 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
765 (vect_gen_perm_mask_any): Likewise.
766 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
767 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
769 (build_vector_from_val): Likewise. Explicitly use a duplicate
772 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
774 * doc/generic.texi (VECTOR_CST): Describe new representation of
776 * vector-builder.h: New file.
777 * tree-vector-builder.h: Likewise.
778 * tree-vector-builder.c: Likewise.
779 * Makefile.in (OBJS): Add tree-vector-builder.o.
780 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
781 * tree-core.h (tree_base): Add a vector_cst field to the u union.
782 (tree_vector): Change the number of elements to
783 vector_cst_encoded_nelts.
784 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
785 (VECTOR_CST_ELTS): Delete.
786 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
787 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
788 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
789 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
790 (VECTOR_CST_ENCODED_ELT): Likewise.
791 (vector_cst_encoded_nelts): New function.
792 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
793 VECTOR_CST_NELTS_PER_PATTERN as arguments.
794 (vector_cst_int_elt, vector_cst_elt): Declare.
795 * tree.c: Include tree-vector-builder.h.
796 (tree_code_size): Abort if passed VECTOR_CST.
797 (tree_size): Update for new VECTOR_CST layout.
798 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
799 VECTOR_CST_NELTS_PER_PATTERN as arguments.
800 (build_vector): Use tree_vector_builder.
801 (vector_cst_int_elt, vector_cst_elt): New functions.
802 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
803 encoded elements and then create the vector in the canonical form.
804 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
805 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
806 (tree_c_tests): Call test_vector_cst_patterns.
807 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
809 (hash_tree): Likewise.
810 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
811 (streamer_write_tree_header): Likewise.
812 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
813 (streamer_alloc_tree): Likewise. Update call to make_vector.
814 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
816 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
818 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
819 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
820 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
822 2017-12-07 Bin Cheng <bin.cheng@arm.com>
823 Richard Biener <rguenther@suse.de>
825 PR tree-optimization/81303
826 * Makefile.in (gimple-loop-interchange.o): New object file.
827 * common.opt (floop-interchange): Reuse the option from graphite.
828 * doc/invoke.texi (-floop-interchange): Ditto. New document for
829 -floop-interchange and mention it for -O3.
830 * opts.c (default_options_table): Enable -floop-interchange at -O3.
831 * gimple-loop-interchange.cc: New file.
832 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
833 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
834 * passes.def (pass_linterchange): New pass.
835 * timevar.def (TV_LINTERCHANGE): New time var.
836 * tree-pass.h (make_pass_linterchange): New declaration.
837 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
838 interchange. Record IV before/after increment in new parameters.
839 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
840 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
842 (check_reduction_path): ...New function here.
843 * tree-vectorizer.h (check_reduction_path): New declaration.
845 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
848 PR rtl-optimization/80818
849 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
850 always early clobbered.
851 * lra-lives.c (process_bb_lives): Check input hard regs for early
852 clobbered non-operand hard reg.
854 2017-12-07 Jakub Jelinek <jakub@redhat.com>
857 * tree-cfg.c (verify_gimple_assign_binary): Don't require
858 types_compatible_p, just that TYPE_MODE is the same.
860 2017-12-07 Martin Sebor <msebor@redhat.com>
863 * attribs.c (empty_attribute_table): Initialize new member of
864 struct attribute_spec.
865 (decl_attributes): Add argument. Handle mutually exclusive
866 combinations of attributes.
867 (selftests::test_attribute_exclusions): New function.
868 (selftests::attribute_c_tests): Ditto.
869 * attribs.h (decl_attributes): Add default argument.
870 * selftest.h (attribute_c_tests): Declare.
871 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
872 * tree-core.h (attribute_spec::exclusions, exclude): New type and
874 * doc/extend.texi (Common Function Attributes): Update const and pure.
875 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
876 of struct attribute_spec.
877 * config/arc/arc.c (arc_attribute_table): Same.
878 * config/arm/arm.c (arm_attribute_table): Same.
879 * config/avr/avr.c ( avr_attribute_table): Same.
880 * config/bfin/bfin.c (bfin_attribute_table): Same.
881 * config/cr16/cr16.c (cr16_attribute_table): Same.
882 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
883 * config/h8300/h8300.c (h8300_attribute_table): Same.
884 * config/i386/i386.c (ix86_attribute_table): Same.
885 * config/ia64/ia64.c (ia64_attribute_table): Same.
886 * config/m32c/m32c.c (m32c_attribute_table): Same.
887 * config/m32r/m32r.c (m32r_attribute_table): Same.
888 * config/m68k/m68k.c (m68k_attribute_table): Same.
889 * config/mcore/mcore.c (mcore_attribute_table): Same.
890 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
891 * config/mips/mips.c (mips_attribute_table): Same.
892 * config/msp430/msp430.c (msp430_attribute_table): Same.
893 * config/nds32/nds32.c (nds32_attribute_table): Same.
894 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
895 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
896 * config/rl78/rl78.c (rl78__attribute_table): Same.
897 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
898 * onfig/rx/rx.c (rx_attribute_table): Same.
899 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
900 * config/sh/sh.c (sh_attribute_table): Same.
901 * config/sparc/sparc.c (sparc_attribute_table): Same.
902 * config/spu/spu.c (spu_attribute_table): Same.
903 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
904 * config/v850/v850.c (v850_attribute_table): Same.
905 * config/visium/visium.c (visium_attribute_table): Same.
907 2017-12-07 Tamar Christina <tamar.christina@arm.com>
910 * config/arm/arm.c (INCLUDE_STRING): Define.
911 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
912 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
915 2017-12-07 Michael Matz <matz@suse.de>
917 Add unroll and jam pass
919 * gimple-loop-jam.c: New file.
920 * Makefile.in (OBJS): Add gimple-loop-jam.o.
921 * common.opt (funroll-and-jam): New option.
922 * opts.c (default_options_table): Add unroll-and-jam at -O3.
923 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
924 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
925 * passes.def: Add pass_loop_jam.
926 * timevar.def (TV_LOOP_JAM): Add.
927 * tree-pass.h (make_pass_loop_jam): Declare.
928 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
929 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
930 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
931 to flow_loop_tree_node_add.
932 (duplicate_subloops, copy_loops_to): Append to sibling list.
933 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
934 * doc/invoke.texi (-funroll-and-jam): Document new option.
935 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
937 2017-12-07 Richard Biener <rguenther@suse.de>
939 PR tree-optimization/83296
940 PR tree-optimization/67769
941 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
942 flow sensitive info in an unrelated BB.
943 (value_replacement): Use reset_flow_sensitive_info.
944 (minmax_replacement): Reset flow sensitive info on the def
945 we move. Do not reset flow sensitive info in the whole BB
947 (abs_replacement): Likewise.
949 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
952 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
953 rs6000_cpu to the given -mcpu=, or to the default processor.
955 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
957 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
958 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
959 instead of rs6000_cpu_attr.
960 (rs6000_adjust_cost): Ditto.
961 (is_microcoded_insn): Ditto.
962 (rs6000_adjust_priority): Ditto.
963 (rs6000_issue_rate): Ditto.
964 (rs6000_use_sched_lookahead): Ditto.
965 (rs6000_use_sched_lookahead_guard): Ditto.
966 (rs6000_sched_reorder): Ditto.
967 (force_new_group): Ditto.
968 * config/rs6000/rs6000.md (cpu attribute): Ditto.
969 (group_ending_nop): Ditto.
971 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
973 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
974 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
975 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
976 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
978 (rs6000_reassociation_width): Ditto.
979 (rs6000_emit_epilogue): Ditto.
980 (rs6000_adjust_cost): Ditto.
981 (is_microcoded_insn): Ditto.
982 (is_cracked_insn): Ditto.
983 (rs6000_adjust_priority): Ditto.
984 (rs6000_sched_reorder): Ditto.
985 (rs6000_sched_reorder2): Ditto.
986 (insn_must_be_first_in_group): Ditto.
987 (insn_must_be_last_in_group): Ditto.
988 (rs6000_register_move_cost): Ditto.
989 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
992 2017-12-07 Julia Koval <julia.koval@intel.com>
994 * config.gcc: Add vaesintrin.h.
995 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
997 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
998 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1000 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1001 * config/i386/immintrin.h: Include vaesintrin.h.
1002 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1003 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1004 _mm_aesdec_epi128): New intrinsics.
1006 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1008 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1009 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1010 c-family/c-spellcheck.cc.
1011 (best_macro_match::best_macro_match): Likewise.
1013 (struct edit_distance_traits<cpp_hashnode *>): Move to
1014 c-family/c-spellcheck.h.
1015 (class best_macro_match): Likewise.
1017 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1019 PR tree-optimization/83293
1020 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1021 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1022 might insert into empty bb.
1025 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1026 simplify for plus with :c added, and pointer_plus without that.
1027 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1028 with undefined overflow and the conversion is not widening,
1029 perform negation in utype and only convert to type afterwards.
1030 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1031 simplify for plus with :c added, and pointer_plus without that.
1032 If type is integral with undefined overflow and the conversion is
1033 not widening, perform minus in utype and only convert to type
1034 afterwards. Move the last pointer_diff_expr simplify into the
1037 2017-12-06 Martin Sebor <msebor@redhat.com>
1039 PR tree-optimization/82646
1040 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1041 strncpy, not maxlen.
1043 2017-12-06 Martin Sebor <msebor@redhat.com>
1045 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1048 PR tree-optimization/83075
1049 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1050 strncat/strncpy don't change length of source string.
1052 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1055 2017-11-29 Martin Aberg <maberg@gaisler.com>
1057 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1058 to prevent b2bst errata sequence.
1059 (sqrtdf2_fix): Likewise.
1061 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1063 PR tree-optimization/81945
1064 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1065 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1066 to dest_cfun has orig_loop_num set, either remap it to the new
1067 loop number if the loop got moved too, or clear it.
1069 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1071 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1073 (thunderx2t99_nothing): New insn reservation.
1074 (thunderx2t99_mrs): New insn reservation.
1075 (thunderx2t99_multiple): New insn reservation.
1076 (thunderx2t99_alu_basi): Add bfx to reservation.
1077 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1079 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1082 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1084 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1086 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1087 tree-ssa-pre.c::remove_dead_inserted_code.
1088 * tree-ssa-dce.h: New file.
1089 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1090 (remove_dead_inserted_code): Move and rename to function
1091 tree-ssa-dce.c::simple_dce_from_worklist.
1092 (pass_pre::execute): Update use.
1094 2017-12-05 Richard Biener <rguenther@suse.de>
1096 PR tree-optimization/83277
1097 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1098 to code-gen liveout vars.
1100 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1102 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1103 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1104 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1107 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1109 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1111 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1112 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1115 2017-12-05 Richard Biener <rguenther@suse.de>
1117 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1118 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1119 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1120 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1122 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1124 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1126 2017-12-05 Martin Liska <mliska@suse.cz>
1127 Jakub Jelinek <jakub@redhat.com>
1129 * doc/invoke.texi: Document the options.
1130 * flag-types.h (enum sanitize_code): Add
1131 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1132 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1133 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1134 * opts.c: Define new sanitizer options.
1135 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1136 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1138 2017-12-05 Julia Koval <julia.koval@intel.com>
1140 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1141 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1142 (ix86_handle_option): Handle -mavx512vnni.
1143 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1144 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1145 * config/i386/i386-c (__AVX512VNNI__): New.
1146 * config/i386/i386.c (ix86_target_string): Handle new option.
1147 (ix86_valid_target_attribute_inner_p): Handle new option.
1148 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1149 * config/i386/i386.opt (mavx512vnni): New option.
1151 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1154 * athlon.md: Disable for generic.
1155 * haswell.md: Enable for generic.
1156 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1157 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1159 (ix86_adjust_cost): Move generic to haswell path.
1161 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1163 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1164 instead of MEM_P in a couple more places. Fix formatting issues.
1166 2017-12-04 Jim Wilson <jimw@sifive.com>
1168 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1169 instead of GP_REG_LAST-1.
1170 (riscv_adjust_libcall_cfi_prologue): Likewise.
1171 (riscv_adjust_libcall_cri_epilogue): Likewise.
1172 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1175 2017-12-04 Luis Machado <luis.machado@linaro.org>
1177 * ipa-pure-const.c (check_decl): Add missing newline.
1178 (state_from_flags): Likewise.
1180 2017-12-04 Jeff Law <law@redhat.com>
1182 PR tree-optimizatin/78496
1183 * gimple-ssa-evrp-analyze.h
1184 (evrp_range_analyzer::get_vr_values): Simplify.
1185 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1186 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1187 and gimple-ssa-evrp-analyze.h.
1188 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1189 (simplify_stmt_for_jump_threading): Copy a blob of code from
1190 tree-vrp.c to use ranges to simplify statements.
1191 (dom_opt_dom_walker::before_dom_children): Call
1192 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1193 (dom_opt_dom_walker::after_dom_children): Similarly for
1194 evrp_range_analyzer::leave.
1195 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1198 * gimple-ssa-evrp-analyze.c
1199 (evrp_range_analyzer::extract_range_from_stmt): Always use
1200 vr_values::update_value_range so preexisting range info is
1201 medged with new range info, even if the new range is VR_VARYING.
1203 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1205 * combine.c: Adjust comment.
1206 (use_crosses_set_p): Delete.
1207 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1208 (try_combine): Ditto.
1210 2017-12-04 Richard Biener <rguenther@suse.de>
1212 PR tree-optimization/83255
1213 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1214 Re-add zero-iteration check.
1216 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1218 PR rtl-optimization/83245
1219 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1220 hard registers as earlyclobber, also if not in an asm.
1222 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1226 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1229 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1230 rs6000_cpu based on cpu_index, not tune_index.
1232 2017-12-04 Richard Biener <rguenther@suse.de>
1234 PR tree-optimization/83238
1235 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1236 code match comment, rejecting invalid SESE regions.
1238 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1240 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1241 require base operand is a REG_POINTER prior to reload on targets
1242 with non-equivalent space registers.
1244 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1246 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1247 (update_specialized_profile): Fix updating of counts.
1248 (perhaps_add_new_callers): Likewise.
1250 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1253 * x86-tune.def: Remove obsolette FIXMEs.
1254 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1255 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1256 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1258 (X86_TUNE_PAD_RETURNS): Disable for generic.
1260 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1262 PR tree-optimization/83170
1263 PR tree-optimization/83241
1264 * gimple-ssa-store-merging.c
1265 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1266 gimple_vuse (ins_stmt) in case it has changed.
1267 (imm_store_chain_info::output_merged_store): Likewise.
1269 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1273 * tree-cfg.c (pass_warn_function_return::execute): Handle
1274 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1275 with BUILTINS_LOCATION.
1279 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1280 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1281 instead of DECL_MODE.
1283 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1284 last argument to rtx pointer.
1285 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1286 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1287 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1288 depending on the chosen ISAs.
1289 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1290 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1291 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1293 * config/i386/sse.md (mov<mode>_internal): Likewise.
1294 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1296 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1298 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1300 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1302 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1303 parameters from prototype.
1304 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1305 parameters. Don't print a comment.
1306 (emit_fusion_gpr_load): Adjust.
1307 (emit_fusion_load_store): Adjust.
1308 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1309 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1310 comment on the second line.
1312 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1315 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1316 rs6000_cpu based on cpu_index, not tune_index.
1318 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1320 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1321 which_alternative instead of which_alternative + 1.
1322 (output_asm_insn): Print an extra tab if the template is short.
1324 2017-12-01 Jim Wilson <jimw@sifive.com>
1326 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1328 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1329 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1331 * doc/tm.texi: Regenerate.
1333 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1336 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1337 whether we can allocate pseudos before trying to fix an address.
1338 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1339 memory address is indexed or indirect.
1340 (floatuns_<mode>si2_hw2): Likewise.
1342 2017-12-01 Jason Merrill <jason@redhat.com>
1344 * Makefile.in (TAGS): Add c-family/*.cc.
1346 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1348 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1349 (call_value_insn): Likewise.
1350 (sibcall_insn): Likewise.
1351 (sibcall_value_insn): Likewise.
1352 (movsi_aarch64): Likewise.
1353 (movdi_aarch64): Likewise.
1354 (add_losym_): Likewise.
1355 (ldr_got_small_): Likewise.
1356 (ldr_got_small_sidi): Likewise.
1357 (ldr_got_small_28k_): Likewise.
1358 (ldr_got_small_28k_sidi): Likewise.
1359 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1360 Move output_addr_const to symbolic case. Add error check.
1362 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1364 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1365 * config/s390/s390.md: Disable MVC merging peephole if it would
1366 disable operand forwarding.
1367 (new peephole2): Split MVCs if it would turn them into up to 2
1370 2017-12-01 Richard Biener <rguenther@suse.de>
1372 PR tree-optimization/83232
1373 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1374 detection of same access. Instead of breaking the group here
1375 do not consider the duplicate. Add comment explaining real fix.
1377 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1379 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1381 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1383 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1385 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1387 * function.h (struct function): Remove cilk_frame_decl,
1388 is_cilk_function and calls_cilk_spawn fields.
1389 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1391 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1392 cilk_elemental field.
1393 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1394 * target.def: Adjust comment.
1395 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1396 Don't test cilk_elemental.
1398 PR tree-optimization/83233
1399 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1400 bswap_stat name for the struct.
1403 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1404 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1405 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1406 SWITCH_BREAK_LABEL_P set on the label.
1407 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1408 added for default case if it was missing and not all cases covered.
1409 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1410 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1412 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1413 to avoid -Wimplicit-fallthrough warning.
1414 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1415 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1417 PR tree-optimization/83221
1418 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1420 (init_reassoc): Formatting fix.
1423 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1424 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1428 * tree-cfg.c: Include asan.h.
1429 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1431 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1433 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1435 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1437 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1438 vpcmpeqd instruction.
1440 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1442 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1443 registers type for 128bit mode.
1445 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1447 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1449 * gengtype-state.c (read_state_pair): Likewise.
1450 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1451 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1452 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1453 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1454 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1455 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1456 * builtins.c (fold_builtin_3): Likewise.
1457 * graphite-scop-detection.c
1458 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1459 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1462 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1465 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1466 parameter. Return true if ignore_decl_rtl_set_p is true and other
1467 conditions are satisfied.
1468 * asan.h (asan_protect_global): Add new parameter.
1469 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1470 to asan_protect_global calls.
1472 2017-11-30 Jim Wilson <jimw@sifive.com>
1474 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1475 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1476 -mstrict-align, add info on default value. Delete redundant lines for
1477 -mabi. Add missing -mexplicit-relocs docs.
1479 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1481 * config/arc/arc.md (trap): New pattern.
1483 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1485 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1486 instruction to end into a delay slot.
1487 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1488 can be placed into a delay slot against reg_note.
1490 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1492 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1493 labels number of usages.
1495 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1497 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1499 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1501 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1504 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1505 multiplication by power of 2 constant into two shifts + comparison.
1507 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1510 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1512 2017-11-30 Richard Biener <rguenther@suse.de>
1514 PR tree-optimization/83202
1515 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1516 allow_peel argument and guard peeling.
1517 (canonicalize_loop_induction_variables): Likewise.
1518 (canonicalize_induction_variables): Pass false.
1519 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1520 peeling from cunrolli.
1522 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1524 * combine.c (try_combine): Print a message to dump file whenever
1525 I0, I1, or I2 cannot be combined into I3.
1527 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1529 PR rtl-optimization/83156
1530 PR rtl-optimization/82621
1531 * combine.c (try_combine): Don't split an I2 if one of the dests is
1532 set again before I3. Allow unused dests.
1534 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1536 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1538 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1540 PR rtl-optimization/80818
1541 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1542 recursively. Use insn code for clobber.
1543 (lra_set_insn_recog_data): Pass the new arg to
1544 collect_non_operand_hard_regs.
1545 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1547 (lra_update_insn_regno_info): Pass insn to
1548 add_regs_to_insn_regno_info.
1550 2017-11-29 Jim Wilson <jimw@sifive.com>
1551 Andrew Waterman <andrew@sifive.com>
1553 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1554 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1555 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1556 use SINGLE_SHIFT_COST.
1557 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1558 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1560 2017-11-29 Julia Koval <julia.koval@intel.com>
1562 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1563 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1564 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1565 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1566 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1567 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1568 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1569 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1570 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1571 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1572 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1573 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1574 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1575 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1576 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1577 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1578 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1579 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1580 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1581 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1582 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1583 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1584 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1585 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1586 __builtin_ia32_vpshldv_v2di_mask,
1587 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1588 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1589 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1591 2017-11-29 Julia Koval <julia.koval@intel.com>
1593 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1594 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1595 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1596 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1597 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1598 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1599 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1600 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1601 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1602 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1603 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1604 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1605 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1606 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1607 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1608 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1609 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1610 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1611 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1612 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1613 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1615 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1617 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1618 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1621 2017-11-29 Martin Aberg <maberg@gaisler.com>
1623 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1624 to prevent b2bst errata sequence.
1625 (sqrtdf2_fix): Likewise.
1627 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1629 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1630 (div_sqrt_insn_p): New function.
1631 (sparc_do_work_around_errata): Insert NOP instructions to
1632 prevent sequences that could trigger the TN-0013 errata for
1633 certain LEON3 processors.
1634 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1635 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1636 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1637 (in_branch_delay): Prevent div and sqrt in delay slot if
1639 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1641 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1643 * config/sparc/sparc.c (atomic_insn_p): New function.
1644 (sparc_do_work_around_errata): Insert NOP instructions to
1645 prevent sequences that could trigger the TN-0010 errata for
1647 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1648 instruction referable in atomic_insns_p.
1650 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1652 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1653 (atomic_compare_and_swap_leon3_1): Likewise.
1656 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1658 * config/sparc/sparc.c (fpop_insn_p): New function.
1659 (sparc_do_work_around_errata): Insert NOP instructions to
1660 prevent sequences that could trigger the TN-0012 errata for
1662 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1663 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1664 (in_branch_annul_delay): Prevent floating-point instructions
1665 in delay slot of annulled integer branch.
1667 2017-11-29 Richard Biener <rguenther@suse.de>
1669 PR tree-optimization/83202
1670 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1672 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1673 nodes vectorized to the same stmts multiple times.
1674 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1675 (vect_analyze_slp_instance): Adjust.
1676 (scalar_stmts_to_slp_tree_map_t): New typedef.
1677 (vect_schedule_slp_instance): Add a map recording the SLP node
1678 representing the vectorized stmts for a set of scalar stmts.
1679 Avoid code-generating redundancies.
1680 (vect_schedule_slp): Allocate map and pass it down.
1682 2017-11-29 Nathan Sidwell <nathan@acm.org>
1685 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1688 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1690 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1691 cases for printing LDP/STP memory addresses.
1692 (aarch64_print_address_internal): Renamed from
1693 aarch64_print_operand_address, added parameter, add Pmode check.
1694 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1695 (aarch64_print_operand_address): Indirect to
1696 aarch64_print_address_internal.
1697 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1698 'y' operand output specifier.
1700 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1703 * tree.c (build_simple_mem_ref_loc): Handle
1704 get_addr_base_and_unit_offset returning a MEM_REF.
1707 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1710 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1713 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1715 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1717 (layout::move_to_column): Likewise.
1719 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1722 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1723 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1725 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1726 Martin Jambor <mjambor@suse.cz>
1729 * tree.h (expr_type_first_operand_type_p): Declare
1730 * tree.c (expr_type_first_operand_type_p): New function.
1731 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1732 (ipa_value_from_jfunc): Adjust declaration.
1733 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1734 Use it as result type for arithmetics, unless it is NULL in which case
1735 be more conservative.
1736 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1737 ipa_get_jf_pass_through_result.
1738 (propagate_vals_across_pass_through): Likewise.
1739 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1740 is to propagate_vals_across_pass_through.
1741 (propagate_constants_across_call): Pass PARM_TYPE to
1742 propagate_scalar_across_jump_function.
1743 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1744 ipa_value_from_jfunc.
1745 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1746 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1747 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1748 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1749 target_type, pass it to ipa_value_from_jfunc.
1750 (update_indirect_edges_after_inlining): Pass parameter type to
1751 try_make_edge_direct_simple_call.
1753 2017-11-28 Jeff Law <law@redhat.com>
1755 * gimple-ssa-evrp-analyze.c
1756 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1757 refine ranges if scev_initialized_p returns true.
1758 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1760 2017-11-28 Julia Koval <julia.koval@intel.com>
1762 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1763 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1764 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1765 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1766 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1767 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1768 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1769 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1770 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1771 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1772 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1773 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1774 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1775 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1776 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1777 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1778 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1779 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1780 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1781 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1782 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1783 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1785 2017-11-28 Julia Koval <julia.koval@intel.com>
1787 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1788 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1789 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1790 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1791 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1792 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1793 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1794 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1795 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1796 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1797 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1798 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1799 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1800 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1801 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1802 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1803 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1804 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1805 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1806 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1807 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1808 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1809 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1810 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1811 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1812 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1813 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1814 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1815 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1816 __builtin_ia32_vpshld_v2di_mask): New builtins.
1817 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1818 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1820 2017-11-28 Richard Biener <rguenther@suse.de>
1822 PR tree-optimization/80776
1823 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1825 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1827 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1828 If the incoming edge is an effective fallthru because the other
1829 edge only reaches a __builtin_unreachable () then record ranges
1830 derived from the controlling condition in SSA info.
1831 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1832 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1834 2017-11-28 Olivier Hainque <hainque@adacore.com>
1836 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1837 on mingw build hosts.
1839 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1842 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1844 2017-11-28 Richard Biener <rguenther@suse.de>
1847 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1848 copies generated from memcpy use a character array as reference
1851 2017-11-28 Julia Koval <julia.koval@intel.com>
1852 Sebastian Peryt <sebastian.peryt@intel.com>
1854 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1855 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1856 cilk-common.o, cilk.h, cilk-common.c): Remove.
1858 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1859 * builtins.c (is_builtin_name): Remove cilkplus condition.
1860 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1861 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1862 cilk-builtins.def, cilkplus.def): Remove.
1863 * cif-code.def (CILK_SPAWN): Remove.
1864 * cilk-builtins.def: Delete.
1865 * cilk-common.c: Ditto.
1867 * cilkplus.def: Ditto.
1868 * config/darwin.h (fcilkplus): Delete.
1869 * cppbuiltin.c: Ditto.
1870 * doc/extend.texi: Remove cilkplus doc.
1871 * doc/generic.texi: Ditto.
1872 * doc/invoke.texi: Ditto.
1873 * doc/passes.texi: Ditto.
1874 * gcc.c (fcilkplus): Remove.
1875 * gengtype.c (cilk.h): Remove.
1876 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1878 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1880 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1882 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1883 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1884 cilkplus conditions.
1885 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1886 inline_read_section): Ditto.
1887 * ipa-inline-analysis.c (cilk.h): Remove.
1888 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1889 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1890 append_linker_options): Remove condition for fcilkplus.
1891 * lto/lto-lang.c (cilk.h): Remove.
1892 (lto_init): Remove condition for fcilkplus.
1893 * omp-expand.c (expand_cilk_for_call): Delete.
1894 (expand_omp_taskreg, expand_omp_for_static_chunk,
1895 expand_omp_for): Remove cilkplus
1897 (expand_cilk_for): Delete.
1898 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1899 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1900 execute_lower_omp, diagnose_sb_0): Ditto.
1901 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1902 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1903 * tree-nested.c: Ditto.
1904 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1905 (dump_generic_node): Ditto.
1906 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1907 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
1909 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
1911 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1913 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
1914 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
1915 for vector mode and !TARGET_SIMD.
1917 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1919 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
1921 * tree.h (SWITCH_LABELS): Remove.
1922 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
1923 assert SWITCH_BODY is non-NULL.
1924 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
1926 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
1928 PR tree-optimization/80788
1929 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
1930 has TREE_OVERFLOW set, call drop_tree_overflow.
1932 2017-11-28 Richard Biener <rguenther@suse.de>
1934 PR tree-optimization/83158
1935 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
1937 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
1940 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
1941 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
1943 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
1945 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
1946 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
1948 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1951 * varasm.c (compare_constant): Take the mode of the constants into
1952 account when comparing floating point constants.
1954 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
1956 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
1957 from explicit instantiation of debug_helper.
1958 * vec.h (DEFINE_DEBUG_VEC): Ditto.
1960 2017-11-27 Richard Biener <rguenther@suse.de>
1962 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
1965 2017-11-27 Richard Biener <rguenther@suse.de>
1967 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
1968 (type_hash_canon): Also clear int_cst_hash_table entry for
1970 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
1972 2017-11-27 Tamar Christina <tamar.christina@arm.com>
1974 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1975 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1976 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1978 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1980 * hash-map.h (gt_cleare_cache): Avoid UB.
1982 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
1984 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
1985 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
1986 (decide_unroll_runtime_iterations): Likewise.
1987 (decide_unroll_stupid): Likewise.
1989 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1992 * config/i386/i386.md: Add a loop with incssp.
1994 2017-11-27 Martin Jambor <mjambor@suse.cz>
1996 PR tree-optimization/81248
1997 * tree-sra.c (splice_param_accesses): Remove size check.
1998 (decide_one_param_reduction): Fix size check.
1999 * gimple-pretty-print.c (dump_profile): Silence warning.
2000 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2002 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2005 * dbxout.c (lastlineno): New variable.
2006 (dbx_debug_hooks): Use dbxout_switch_text_section as
2007 switch_text_section debug hook.
2008 (dbxout_function_end): Switch to current_function_section
2009 rather than function_section. If crtl->has_bb_partition,
2010 output just one N_FUN, depending on in_cold_section_p.
2011 (dbxout_source_line): Remember last lineno in lastlineno.
2012 (dbxout_switch_text_section): New function.
2013 (dbxout_function_decl): Adjust dbxout_block caller.
2014 (dbx_block_with_cold_children): New function.
2015 (dbxout_block): Return true if any LBRAC/RBRAC have been
2016 emitted. Use dbx_block_with_cold_children at depth == 0
2017 in second partition. Add PARENT_BLOCKNUM argument, pass
2018 it optionally adjusted to children. Output LBRAC/RBRAC
2019 around recursive call only if the block is in the current
2020 partition, if not and anything was output, emit empty
2022 * final.c (final_scan_insn): Compute cold_function_name
2023 before calling switch_text_section debug hook. Call
2024 that hook even if dwarf2out_do_frame if not emitting
2028 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2029 TREE_READONLY decls.
2031 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2033 PR rtl-optimization/82488
2034 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2036 2017-11-26 Julia Koval <julia.koval@intel.com>
2038 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2040 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2043 2017-11-26 Julia Koval <julia.koval@intel.com>
2045 * config/i386/driver-i386.c (host_detect_local_cpu):
2046 Detect skylake-avx512.
2048 2017-11-26 Julia Koval <julia.koval@intel.com>
2050 * config.gcc: Add -march=cannonlake.
2051 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2052 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2053 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2054 (PTA_CANNONLAKE): New.
2055 (processor_target_table): Add cannonlake.
2056 (ix86_option_override_internal): Ditto.
2057 (fold_builtin_cpu): Ditto.
2058 (get_builtin_code_for_version): Handle cannonlake.
2059 (M_INTEL_COREI7_CANNONLAKE): New.
2060 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2061 * doc/invoke.texi: Add -march=cannonlake.
2063 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2065 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2066 (try_init_one_plugin): Alternative implementation for MinGW.
2067 * Makefile.in (plugin_implib): New.
2068 (gengtype-lex.c): Fix broken AIX workaround.
2069 * configure: Regenerate.
2070 * doc/plugins.texi: Document support for MinGW.
2072 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2074 PR rtl-optimization/81553
2075 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2076 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2077 is a shift where C1 has different mode than the whole shift, use C1's
2078 mode for MULT rather than the shift's mode.
2081 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2082 builtins not enabled in the currently selected ISA.
2084 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2086 PR tree-optimization/71026
2087 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2088 (insert_reciprocals): Change to insert reciprocals before a division
2089 by a square and to insert the square of a reciprocal.
2090 (execute_cse_reciprocals_1): Change to consider division by a square.
2091 (register_division_in): Add importance parameter.
2093 2017-11-24 Richard Biener <rguenther@suse.de>
2095 PR tree-optimization/82402
2096 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2097 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2099 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2101 * match.pd (0-ptr): New transformation.
2103 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2106 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2107 when skipping checking.
2109 2017-11-24 Richard Biener <rguenther@suse.de>
2111 PR tree-optimization/83128
2112 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2113 (vn_reference_lookup_3): Likewise.
2115 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2118 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2119 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2122 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2123 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2124 check for the latter with gimple_call_builtin_p. Do not handle
2125 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2127 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2129 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2132 2017-11-23 Julia Koval <julia.koval@intel.com>
2134 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2135 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2136 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2137 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2138 _mm512_maskz_expandloadu_epi16): New intrinsics.
2139 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2140 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2141 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2142 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2143 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2144 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2145 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2146 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2147 _mm256_maskz_expandloadu_epi8): New intrinsics.
2148 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2149 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2150 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2151 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2152 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2153 * config/i386/sse.md (VI248_VLBW): New iterator.
2154 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2156 2017-11-23 Julia Koval <julia.koval@intel.com>
2158 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2159 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2160 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2161 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2162 _mm512_mask_compressstoreu_epi16): New.
2163 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2164 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2165 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2166 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2167 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2168 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2169 _mm256_mask_compressstoreu_epi8): New.
2170 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2171 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2172 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2173 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2174 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2175 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2176 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2177 __builtin_ia32_compresshi128_mask,
2178 __builtin_ia32_compressstoreuqi512_mask,
2179 __builtin_ia32_compressstoreuhi512_mask,
2180 __builtin_ia32_compressstoreuqi256_mask,
2181 __builtin_ia32_compressstoreuqi128_mask,
2182 __builtin_ia32_compressstoreuhi256_mask,
2183 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2184 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2186 (ix86_expand_special_args_builtin): Handle new types.
2187 (s4fma_expand): Handle new builtin array.
2188 * config/i386/immintrin.h: Include new headers.
2189 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2190 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2192 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2194 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2195 fixes. Declare temp and g variables at the top in order to avoid
2196 {} in most of the cases.
2198 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2200 * match.pd (ptr-0): New transformation.
2202 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2204 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2205 (struct addr_mode_cost_table): New.
2206 (struct tune_params): Add field addr_mode_costs.
2207 * config/arm/arm.c (generic_addr_mode_costs): New.
2208 (arm_slowmul_tune): Initialise addr_mode_costs field.
2209 (arm_fastmul_tune): Likewise.
2210 (arm_strongarm_tune): Likewise.
2211 (arm_xscale_tune): Likewise.
2212 (arm_9e_tune): Likewise.
2213 (arm_marvell_pj4_tune): Likewise.
2214 (arm_v6t2_tune): Likewise.
2215 (arm_cortex_tune): Likewise.
2216 (arm_cortex_a8_tune): Likewise.
2217 (arm_cortex_a7_tune): Likewise.
2218 (arm_cortex_a15_tune): Likewise.
2219 (arm_cortex_a35_tune): Likewise.
2220 (arm_cortex_a53_tune): Likewise.
2221 (arm_cortex_a57_tune): Likewise.
2222 (arm_exynosm1_tune): Likewise.
2223 (arm_xgene1_tune): Likewise.
2224 (arm_cortex_a5_tune): Likewise.
2225 (arm_cortex_a9_tune): Likewise.
2226 (arm_cortex_a12_tune): Likewise.
2227 (arm_cortex_a73_tune): Likewise.
2228 (arm_v7m_tune): Likewise.
2229 (arm_cortex_m7_tune): Likewise.
2230 (arm_v6m_tune): Likewise.
2231 (arm_fa726te_tune): Likewise.
2232 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2234 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2236 * config/arm/arm.c (arm_mem_costs): New function.
2237 (arm_rtx_costs_internal): Use arm_mem_costs.
2239 2017-11-23 Mark Wielaard <mark@klomp.org>
2241 * dwarf2out.c (init_sections_and_labels): Use generation to create
2242 unique ranges_section_label and ranges_base_label. Return generation.
2243 (output_rnglists): Add generation argument. Use generation to create
2244 unique ranges labels.
2245 (dwarf2out_finish): Get generation from init_sections_and_labels
2246 and pass generation to output_rnglists.
2248 2017-11-23 Mike Stump <mikestump@comcast.net>
2249 Eric Botcazou <ebotcazou@adacore.com>
2251 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2252 * cfgloop.h (struct loop): Add unroll field.
2253 * function.h (struct function): Add has_unroll bitfield.
2254 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2255 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2256 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2257 (pass_rtl_unroll_loops::gate): Likewise.
2258 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2259 for which loop->unroll==1.
2260 (decide_unroll_constant_iterations): Use note for consistency and
2261 take loop->unroll into account. Return early if loop->unroll is set.
2262 Fix thinko in existing test.
2263 (decide_unroll_runtime_iterations): Use note for consistency and
2264 take loop->unroll into account.
2265 (decide_unroll_stupid): Likewise.
2266 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2267 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2268 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2270 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2271 (print_loop): Print loop->unroll if set.
2272 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2273 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2274 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2276 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2277 loop->unroll is set and smaller than the trip count. Otherwise bypass
2278 entirely the heuristics if loop->unroll is set. Remove dead note.
2279 Fix off-by-one bug in other note.
2280 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2281 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2283 (tree_unroll_loops_completely): Make static.
2284 (pass_complete_unroll::execute): Use correct type for variable.
2285 (pass_complete_unrolli::execute): Fix formatting.
2286 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2288 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2290 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2291 enable when TARGET_PREFER_AVX128 is set.
2293 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2295 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2297 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2298 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2300 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2302 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2303 frequencies. Fix estimation of aggregate parameters.
2305 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2307 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2310 2017-11-23 Tom de Vries <tom@codesourcery.com>
2312 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2314 2017-11-23 Tom de Vries <tom@codesourcery.com>
2316 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2319 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2322 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2323 sibcall_value_pcrel_fdpic): Use local variable instead of
2325 (calli_tbr_rel): Add missing operand 2.
2326 (call_valuei_tbr_rel): Add missing operand 3.
2328 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2331 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2332 bitpos/bitsize covering the whole destination, use store_expr only if
2333 the complex mode is the same. Otherwise, use expand_normal and if
2334 it returns CONCAT, subreg each part separately instead of trying to
2335 subreg the whole result.
2337 2017-11-23 Richard Biener <rguenther@suse.de>
2339 PR tree-optimization/23094
2340 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2341 come here from walking over backedges in the first iteration.
2342 (vn_reference_lookup_3): Skip clobbers that store the same value.
2344 2017-11-23 Richard Biener <rguenther@suse.de>
2346 PR tree-optimization/81403
2347 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2348 a block we need a leader relative to.
2349 (phi_translate_1): For nary processing require a leader from
2350 get_representative_for given we run expression simplification
2351 using match-and-simplify. Remove previous fix.
2353 2017-11-22 Jeff Law <law@redhat.com>
2355 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2356 Use new method allocate_value_range rather than accessing the
2357 vrp_value_range_pool data member directly.
2358 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2359 to use extract_range_from_stmt method to avoid need for
2360 extract_range_from_assignment method.
2361 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2362 method rather than setting values_propgated data member directly.
2363 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2364 and values propagated data members and extract_range_from_assignment
2365 method. Reorder private data members to conform to standards.
2366 Add new methods set_lattice_propagation_complete and
2367 allocate_value_range.
2369 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2371 PR rtl-optimization/83030
2372 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2373 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2374 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2375 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2376 CROSSING_JUMP_P flag is already set before setting it.
2377 * cfgrtl.c (fixup_partition_crossing): Likewise.
2378 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2381 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2383 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2384 Handle the case where both arguments are using gen_const_vec_series.
2386 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2389 * pretty-print.c (pp_format): Move quoting implementation to
2390 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2391 to pass address of "quote" local.
2392 (pp_begin_quote): New function.
2393 (pp_end_quote): New function.
2394 * pretty-print.h (printer_fn): Convert penultimate param from bool
2396 (pp_begin_quote): New decl.
2397 (pp_end_quote): New decl.
2398 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2399 param from bool to bool *.
2400 * tree-diagnostic.h (default_tree_printer): Likewise.
2402 2017-11-22 Jeff Law <law@redhat.com>
2404 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2405 Initialize vr_values.
2406 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2407 vr_values via vr_values class instance rather than delegators.
2408 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2409 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2410 (evrp_range_analyzer::push_value_range): Likewise.
2411 (evrp_range_analyzer::pop_value_range): Likewise.
2412 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2413 most delegators. Those remaining are exposed as public interfaces.
2414 Make vr_values a pointer and private.
2415 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2417 (evrp_range_analyzer::get_vr_value): New method.
2418 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2419 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2420 (class evrp_dom_walker): Attach evrp_folder class, initialize
2421 it in the ctor. Remove temporary delegators.
2422 (evrp_dom_walker::before_dom_children): Call methods in attached
2423 evrp_range_analyzer class via class instance pointer. Use
2424 free value_range_constant_singleton to remove need for
2425 op_with_constant_singleton_value delegator method. Do not
2426 create a vrp_prop class instance for every call! Narrow
2427 scope of a couple variables.
2428 (evrp_dom_walker::cleanup): Call methods in attached
2429 evrp_range_analyzer class via class instance pointer.
2430 * vr-values.h (class vr_values): Privatize many methods and
2433 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2435 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2436 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2437 * expr.c (expand_expr_real_2): Likewise.
2438 * fold-const.c (const_unop): Likewise.
2439 * optabs-tree.c (optab_for_tree_code): Likewise.
2440 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2441 * tree-inline.c (estimate_operator_cost): Likewise.
2442 * tree-pretty-print.c (dump_generic_node): Likewise.
2443 (op_code_prio): Likewise.
2444 (op_symbol_code): Likewise.
2445 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2446 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2447 * internal-fn.c (direct_internal_fn_optab): New function.
2448 (direct_internal_fn_array, direct_internal_fn_supported_p
2449 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2450 * fold-const-call.c (fold_const_reduction): New function.
2451 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2453 * tree-vect-loop.c: Include internal-fn.h.
2454 (reduction_code_for_scalar_code): Rename to...
2455 (reduction_fn_for_scalar_code): ...this and return an internal
2457 (vect_model_reduction_cost): Take an internal_fn rather than
2459 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2461 (vectorizable_reduction): Use internal functions rather than tree
2462 codes for the reduction operation. Update calls to the functions
2464 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2465 Use calls to internal functions rather than REDUC tree codes.
2466 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2468 2017-11-22 Olivier Hainque <hainque@adacore.com>
2470 * config/vxworks.c (vxworks_override_options): Pick default
2471 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2472 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2473 DWARF_GNAT_ENCODINGS_DEFAULT.
2474 * config/vxworksae.h: Likewise.
2476 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2478 PR tree-optimization/83104
2479 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2482 2017-11-22 Marek Polacek <polacek@redhat.com>
2483 H.J. Lu <hongjiu.lu@intel.com>
2484 Jason Merrill <jason@redhat.com>
2489 * calls.c (initialize_argument_information): Call
2490 warn_parameter_passing_abi target hook.
2491 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2492 argument onto stack.
2493 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2494 * common.opt: Update -fabi-version description.
2495 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2496 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2498 (ix86_is_empty_record): New function.
2499 (ix86_warn_parameter_passing_abi): New function.
2500 (TARGET_EMPTY_RECORD_P): Redefine.
2501 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2502 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2503 * doc/tm.texi: Regenerated.
2504 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2505 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2506 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2507 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2508 instead of int_size_in_bytes.
2509 * expr.c (copy_blkmode_to_reg): Likewise.
2510 * function.c (aggregate_value_p): Return 0 for empty types.
2511 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2513 (locate_and_pad_parm): Call arg size_in_bytes instead
2515 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2516 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2517 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2519 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2520 (std_gimplify_va_arg_expr): Skip empty records. Call
2521 arg_size_in_bytes instead size_in_bytes.
2522 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2523 * tree-core.h (tree_type_common): Add empty_flag.
2524 (tree_decl_common): Update comments.
2525 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2527 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2528 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2530 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2531 * tree.c (default_is_empty_type): New function.
2532 (default_is_empty_record): New function.
2533 (arg_int_size_in_bytes): New function.
2534 (arg_size_in_bytes): New function.
2535 (get_ultimate_context): New function.
2536 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2537 TRANSLATION_UNIT_WARN_EMPTY_P.
2538 (default_is_empty_record, arg_int_size_in_bytes,
2539 arg_size_in_bytes, get_ultimate_context): Declare.
2541 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2543 * config/arm/arm.c (cmse_clear_registers): New function.
2544 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2545 code by call to cmse_clear_registers.
2546 (cmse_nonsecure_entry_clear_before_return): Likewise.
2548 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2550 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2551 (vdot_s32, vdotq_s32): New.
2552 (vdot_lane_u32, vdotq_lane_u32): New.
2553 (vdot_lane_s32, vdotq_lane_s32): New.
2556 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2559 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2560 for unsigned values with fewer HWIs than the precision.
2561 (test_overflow): New function.
2562 (wide_int_cc_tests): Call it.
2564 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2566 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2567 has the same mode class as Pmode.
2569 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2571 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2572 padding_bits_to_clear_ptr.
2573 (cmse_nonsecure_entry_clear_before_return): Likewise.
2575 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2577 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2578 auto_sbitap instead of integer bitfield to control register needing
2581 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2583 PR tree-optimization/83044
2584 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2585 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2586 that need the upper bound. Subtract offset from
2587 get_addr_base_and_unit_offset only if positive and subtract it
2588 before division by eltsize rather than after it.
2591 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2592 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2593 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2597 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2598 expand_binop with *mul_widen_optab, make sure at least one of the
2599 operands doesn't have VOIDmode.
2602 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2604 PR rtl-optimization/82044
2605 PR tree-optimization/82042
2606 * dse.c (record_store): Check for overflow.
2607 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2608 add_wild_read instead of clear_rhs_from_active_local_stores on
2609 overflow and log it into dump_file.
2611 2017-11-22 Richard Biener <rguenther@suse.de>
2613 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2614 fake edges to exit when looking for a place to insert.
2615 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2617 (insert_into_preds_of_block): Commit edge insertion immediately,
2618 assert that doesn't require new BBs.
2619 (fini_pre): Release expressions.
2620 (pass_pre::execute): Shuffle things around a bit, if the fn
2621 is too large do not compute AVAIL either as this is really the
2624 2017-11-22 Richard Biener <rguenther@suse.de>
2626 PR tree-optimization/83089
2627 * tree-if-conv.c (pass_if_conversion::execute): If anything
2628 changed reset SCEV and free the number of iteration estimates.
2630 2017-11-21 Martin Sebor <msebor@redhat.com>
2632 PR tree-optimization/82945
2633 * calls.h (warn_nonstring_bound): Remove unused function.
2635 2017-11-21 Martin Sebor <msebor@redhat.com>
2637 PR tree-optimization/82945
2638 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2639 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2640 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2642 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2643 * calls.h (get_attr_nonstring_decl): Declare new function.
2644 * doc/extend.texi (attribute nonstring): Update.
2645 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2646 get_attr_nonstring_decl and handle it.
2647 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2648 detection of nul-termination.
2649 (strlen_to_stridx): Change to a pointer.
2650 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2651 (pass_strlen::execute): Same.
2653 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2655 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2656 for the new option -mprefer-vector-width=[none|128|256|512].
2657 * config/i386/i386.c (ix86_target_string): remove old style options
2658 -mprefer-avx256 and make -mprefer-avx128 as alias.
2659 (ix86_option_override_internal): Apply defaults for the
2660 -mprefer-vector-width=[128|256] option.
2661 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2662 Implement macros to work with -mprefer-vector-width=.
2663 * config/i386/i386.opt: Implemented option
2664 -mprefer-vector-width=[none|128|256|512].
2665 * doc/invoke.texi: Documentation for
2666 -mprefer-vector-width=[none|128|256|512].
2668 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2670 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2672 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2674 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2676 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2678 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2680 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2681 * expr.c (expand_expr_real_2): Likewise.
2682 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2683 fold_binary_loc): Likewise.
2684 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2685 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2686 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2687 MINUS_EXPR transformations.
2688 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2689 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2690 * tree-inline.c (estimate_operator_cost): Likewise.
2691 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2692 op_symbol_code): Likewise.
2693 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2694 * vr-values.c (extract_range_from_binary_expr): Likewise.
2695 * varasm.c (initializer_constant_valid_p_1): Likewise.
2696 * tree.def: New tree code POINTER_DIFF_EXPR.
2698 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2700 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2701 integer suffix to movbe mnemonic.
2702 (*bswaphi2_movbe): Ditto.
2703 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2705 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2708 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2709 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2710 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2712 2017-11-21 Martin Liska <mliska@suse.cz>
2714 * tree-inline.c (expand_call_inline): Remove not needed
2717 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2718 Jakub Jelinek <jakub@redhat.com>
2721 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2722 Remove static keyword from f variable.
2724 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2726 PR tree-optimization/83086
2727 * gimple-ssa-store-merging.c
2728 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2729 rather than n.base_addr.
2731 2017-11-21 Martin Liska <mliska@suse.cz>
2733 PR rtl-optimization/82044
2734 PR tree-optimization/82042
2735 * dse.c (check_mem_read_rtx): Check for overflow.
2737 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2739 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2742 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2744 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2745 and its corresponding call arg location note.
2747 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2748 Andrew Burgess <andrew.burgess@embecosm.com>
2750 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2752 (arc_return_slot_offset): Likewise.
2753 (arc_eh_return_address_location): New declaration.
2754 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2755 (MUST_SAVE_REGISTER): Add exception handler case.
2756 (MUST_SAVE_RETURN_ADDR): Likewise.
2757 (arc_frame_pointer_required): Likewise.
2758 (arc_frame_pointer_needed): New function.
2759 (arc_compute_frame_size): Changed.
2760 (arc_expand_prologue): Likewise.
2761 (arc_expand_epilogue): Likewise.
2762 (arc_initial_elimination_offset): Likewise.
2763 (arc_return_slot_offset): Delete.
2764 (arc_eh_return_address_location): New function.
2765 (arc_builtin_setjmp_frame_value): Likewise.
2766 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2767 (EH_RETURN_STACKADJ_RTX): Define.
2768 (EH_RETURN_HANDLER_RTX): Likewise.
2769 * config/arc/arc.md (eh_return): Delete.
2771 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2773 * print-tree.h (debug_vec_tree): Remove prototype.
2774 * gdbinit.in (pvt): Remove macro.
2776 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2778 PR tree-optimization/83047
2779 * gimple-ssa-store-merging.c
2780 (imm_store_chain_info::output_merged_store): If the loads with the
2781 same vuse are in different basic blocks, for load_gsi pick a load
2782 location that is dominated by the other loads.
2785 * config/i386/i386.c (ix86_memmodel_check): Start
2786 -Winvalid-memory-model diagnostics with lowercase letter.
2789 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2790 set in_cold_section_p to the partition containing loc_list->first.
2791 When seeing loc_list->last_before_switch node, update secname and
2792 perform range_across_switch second partition handling only after that.
2795 * run-rtl-passes.c: Include debug.h.
2796 (run_rtl_passes): Call debug_hooks->assembly_start.
2797 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2801 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2802 OPTAB_DIRECT in calls to expand_simple_binop.
2804 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2807 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2809 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2812 * spellcheck.h (best_match::blithely_get_best_candidate): New
2815 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2817 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2818 with lower case letter.
2820 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2822 * config/i386/i386.md (bswaphi2): New expander.
2823 (*bswaphi2_movbe): New insn pattern.
2824 (bswaphi -> rorhi pepehole2): New peephole pattern.
2826 2017-11-20 Jeff Law <law@redhat.com>
2828 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2829 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2830 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2831 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2832 gimple-ssa-evrp-analyze.h.
2834 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2835 set BB_VISITED here.
2836 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2838 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2840 * tree-predcom.c: Add general comment on Store-Store chains.
2841 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2843 (get_chain_last_ref_at): Rename into...
2844 (get_chain_last_write_at): ...this.
2845 (get_chain_last_write_before_load): New function.
2846 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2847 CT_STORE_STORE when write reference is added.
2848 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2849 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2850 (initialize_root_vars_store_elim_1): Ditto.
2851 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2852 definition is created.
2853 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2854 chain by replacing it with dominant stored value.
2856 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2858 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2860 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2862 * vr-values.c (extract_range_from_binary_expr): Use a full range
2865 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2867 * config/arm/arm.md (R4_REGNUM): Define constant.
2868 (nonsecure_call_internal): Remove r4 clobber.
2869 (nonsecure_call_value_internal): Likewise.
2870 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2871 clobber and resequence match_operands.
2872 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2873 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2874 (nonsecure_call_value_reg_thumb2): Likewise.
2876 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2878 PR tree-optimization/78821
2879 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2880 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2881 to the address of the base rather than the base itself.
2882 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2883 (find_bswap_or_nop_finalize): New function.
2884 (find_bswap_or_nop): Use it.
2885 (bswap_replace): Return a tree rather than bool, change first
2886 argument from gimple * to gimple_stmt_iterator, allow inserting
2887 into an empty sequence, allow ins_stmt to be NULL - then emit
2888 all stmts into gsi. Fix up MEM_REF address gimplification.
2889 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2890 (struct store_immediate_info): Add N and INS_STMT non-static
2892 (store_immediate_info::store_immediate_info): Initialize them
2893 from newly added ctor args.
2894 (merged_store_group::apply_stores): Formatting fixes. Sort by
2896 (stmts_may_clobber_ref_p): For stores call also
2897 refs_anti_dependent_p.
2898 (gather_bswap_load_refs): New function.
2899 (imm_store_chain_info::try_coalesce_bswap): New method.
2900 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2901 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2902 (imm_store_chain_info::output_merged_store): Fail if number of
2903 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2904 and NOP_EXPR rhs_code.
2905 (pass_store_merging::process_store): Compute n and ins_stmt, if
2906 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
2907 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
2909 (pass_store_merging::execute): Calculate dominators.
2911 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
2912 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
2913 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
2914 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2915 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
2916 class pass_optimize_bswap, bswap_replace,
2917 pass_optimize_bswap::execute): Moved to ...
2918 * gimple-ssa-store-merging.c: ... this file.
2919 Include optabs-tree.h.
2920 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
2921 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2922 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
2923 anonymous namespace, remove static keywords.
2924 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
2925 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
2927 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
2930 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
2932 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
2934 * vec.h (debug_helper): New function.
2935 (DEFINE_DEBUG_VEC): New macro.
2936 * hash-set.h (debug_helper): New function.
2937 (DEFINE_DEBUG_HASH_SET): New macro.
2938 * cfg.c (debug_slim (edge)): New function.
2939 Call DEFINE_DEBUG_VEC for edges.
2940 Call DEFINE_DEBUG_HASH_SET for edges.
2941 * cfghooks.c (debug_slim (basic_block)): New function.
2942 Call DEFINE_DEBUG_VEC for basic blocks.
2943 Call DEFINE_DEBUG_HASH_SET for basic blocks.
2944 * print-tree.c (debug_slim): New function to handle trees.
2945 Call DEFINE_DEBUG_VEC for trees.
2946 Call DEFINE_DEBUG_HASH_SET for trees.
2947 (debug (vec<tree, va_gc>) &): Remove.
2948 (debug (<vec<tree, va_gc>) *): Remove.
2949 * print-rtl.c (debug_slim): New function to handle const_rtx.
2950 Call DEFINE_DEBUG_VEC for rtx_def.
2951 Call DEFINE_DEBUG_VEC for rtx_insn.
2952 Call DEFINE_DEBUG_HASH_SET for rtx_def.
2953 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
2954 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
2955 (debug (vec<rtx_insn *> *ptr): Remove.
2956 (debug_insn_vector): Remove.
2957 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
2959 2017-11-20 Tom de Vries <tom@codesourcery.com>
2961 PR rtl-optimization/82020
2962 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
2963 IF_THEN_ELSE condition.
2965 2017-11-19 Jeff Law <law@redhat.com>
2967 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2968 of degenerates resulting from ignoring an edge.
2970 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2973 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
2975 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2978 * profile-count.c (profile_count::to_sreal_scale): Fix return value
2979 for uninitialied counts.
2981 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2984 * tree-inline.c (estimate_num_insns): Set to 1 at least.
2986 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2989 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
2991 2017-11-19 Tom de Vries <tom@codesourcery.com>
2993 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
2996 2017-11-19 Tom de Vries <tom@codesourcery.com>
2998 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3001 2017-11-19 Tom de Vries <tom@codesourcery.com>
3003 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3006 2017-11-19 Tom de Vries <tom@codesourcery.com>
3008 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3010 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3012 2017-11-19 Tom de Vries <tom@codesourcery.com>
3014 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3016 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3017 semicolon after MCORE_EXPORT_NAME call.
3019 2017-11-19 Tom de Vries <tom@codesourcery.com>
3022 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3023 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3025 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3027 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3028 (cgraph_edge::clone): Cleanup updating of profile.
3029 * ipa-cp.c (update_profiling_info): Likewise.
3030 * ipa-inline-transform.c (inline_transform): Likewise.
3031 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3032 * ipa-split.c (execute_split_functions): Do not split when function
3034 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3035 * profile-count.c (profile_count::dump): Add global0.
3036 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3038 (profile_count::to_sreal_scale): Likewise.
3039 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3040 (profile_count::combine_with_ipa_count): New function.
3041 * profile-count.h (profile_guessed_global0adjusted): New.
3042 (profile_count::adjusted_zero): New.
3043 (profile_count::global0adjusted): New.
3044 (profile_count::combine_with_ipa_count): New.
3045 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3046 correct profile of return block of split functions.
3047 (copy_cfg_body): Remove unused profile_count.
3048 (copy_body): Likewise.
3049 (expand_call_inline): Update.
3050 (tree_function_versioning): Update.
3052 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3054 * hash-set.h (hash_set::empty): New.
3055 * tree-ssa-threadbackward.h: Delete.
3056 * tree-ssa-threadbackward.c (class thread_jumps): New.
3057 Move max_threaded_paths into class.
3058 (fsm_find_thread_path): Remove arguments that are now in class.
3059 (profitable_jump_thread_path): Rename to...
3060 (thread_jumps::profitable_jump_thread_path): ...this.
3061 (convert_and_register_jump_thread_path): Rename to...
3062 (thread_jumps::convert_and_register_current_path): ...this.
3063 (check_subpath_and_update_thread_path): Rename to...
3064 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3065 (register_jump_thread_path_if_profitable): Rename to...
3066 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3067 (handle_phi): Rename to...
3068 (thread_jumps::handle_phi): ...this.
3069 (handle_assignment): Rename to...
3070 (thread_jumps::handle_assignment): ...this.
3071 (fsm_find_control_statement_thread_paths): Rename to...
3072 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3073 (find_jump_threads_backwards): Rename to...
3074 (thread_jumps::find_jump_threads_backwards): ...this.
3075 Initialize path local data.
3076 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3077 from within thread_jumps class.
3078 (pass_early_thread_jumps::execute): Same.
3080 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3082 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3084 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3086 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3087 to canonical location.
3089 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3092 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3094 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3096 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3098 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3099 rcrt1.o%s/grcrt1.o%s for -static-pie.
3101 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3103 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3104 ix86_shift_rotate_cost): Break out from ...
3105 (ix86_rtx_costs): ... here.
3106 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3109 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3111 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3112 when entry block was promoted unlikely.
3113 (estimate_bb_frequencies): Increase frequency scale.
3114 * profile-count.h (profile_count): Export precision info.
3116 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3118 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3119 disturbin profile of entry block.
3121 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3123 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3126 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3128 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3129 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3131 2017-11-17 Jeff Law <law@redhat.com>
3133 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3134 from evrp_dom_walker class. Various methods moved into new class.
3135 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3136 (evrp_range_analyzer::enter): New method.
3137 (evrp_range_analyzer::leave): New method.
3138 (evrp_dom_walker): Remove delegators no longer needed by this class.
3139 Replace vr_values data member with evrp_range_analyzer
3141 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3142 method extracted from evrp_dom_walker::before_dom_children.
3143 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3144 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3146 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3147 Add private copy constructor and move assignment operators.
3148 Privatize methods and class data where trivially possible.
3149 (evrp_dom_walker::cleanup): New function, extracted from
3150 execute_early_vrp. Simplify access to class data.
3152 * vr-values.h (get_output_for_vrp): Prototype.
3153 * vr-values.c (get_output_for_vrp): New function extracted from
3154 vrp_visit_assignment_or_call and extract_range_from_stmt.
3155 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3157 2017-11-17 Luis Machado <luis.machado@linaro.org>
3159 * config/aarch64/aarch64.c
3160 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3161 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3162 tune_params::AUTOPREFETCHER_WEAK.
3164 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3167 * config/arm/arm.c (arm_valid_target_attribute_rec):
3168 Parse "arch=" and "+<ext>".
3169 (arm_valid_target_attribute_tree): Re-init global options.
3170 (arm_option_override): Make non-static.
3171 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3172 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3173 (__ARM_FEATURE_CRC32): Support undef.
3174 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3175 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3177 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3179 * gdbinit.in (break-on-diagnostic): New command.
3181 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3183 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3184 return to indirect jump for EH return if control-flow
3185 protection is enabled. Change explicit 'false' argument in
3186 pro_epilogue_adjust_stack with a value of
3188 * config/i386/i386.md (simple_return_indirect_internal):
3189 Remove SImode restriction to support 64-bit.
3191 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3193 * combine.c (added_notes_insn): New.
3194 (try_combine): Handle added_notes_insn like added_links_insn.
3195 Rewrite return value code.
3196 (distribute_notes): Set added_notes_insn to the earliest insn we added
3199 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3201 PR rtl-optimization/82621
3202 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3203 dest of one of those SETs is unused.
3205 2017-11-17 Richard Biener <rguenther@suse.de>
3208 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3209 * tree-pretty-print.c (dump_generic_node): Handle
3210 annot_expr_parallel_kind.
3211 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3212 * gimplify.c (gimple_boolify): Likewise.
3214 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3216 * config.gcc (extra_headers): Add cet.h for x86 targets.
3217 * config/i386/cet.h: New file.
3218 * doc/install.texi: Add --enable-cet/--disable-cet.
3220 2017-11-17 Richard Biener <rguenther@suse.de>
3222 PR tree-optimization/83017
3223 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3224 (gen_parallel_loop): Properly count iterations.
3225 (parallelize_loops): Handle loop->can_be_parallel independent
3226 of flag_loop_parallelize_all. Make static profitability test match
3228 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3229 * invoke.texi (parloops-min-per-thread): Document.
3231 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3233 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3234 upstreaming review comments.
3236 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3238 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3239 with fast unaligned access.
3240 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3242 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3244 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3245 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3247 2017-11-17 Richard Biener <rguenther@suse.de>
3249 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3250 folding of references.
3252 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3255 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3256 of replacing call to strncmp with corresponding call to strcmp when
3259 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3261 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3262 option prefer-avx256 for skylake-avx512 configuration.
3263 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3264 (get_builtin_code_for_version): Ditto.
3266 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3267 Monk Chiang <sh.chiang04@gmail.com>
3269 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3270 (FIXED_REGISTERS): Reserve more register numbers.
3271 (CALL_USED_REGISTERS): Likewise.
3272 (REG_ALLOC_ORDER): Likewise.
3273 (REG_CLASS_CONTENTS): Likewise.
3274 (REGISTER_NAMES): Likewise.
3276 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3277 Kito Cheng <kito.cheng@gmail.com>
3279 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3281 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3283 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3285 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3286 parameter list for vec_splats.
3288 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3290 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3292 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3294 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3297 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3298 with the required version of automake 1.11.6.
3300 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3302 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3304 (DU_C2_3_power9): Correct reservation combinations.
3305 (FP_DIV_power9, VEC_DIV_power9): New.
3306 (power9-alu): Split out rotate/shift...
3307 (power9-rot): ...to here, correct dispatch resource.
3308 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3310 (power9-fp): Correct latency.
3311 (power9-sdiv): Add div/sqrt resource.
3312 (power9-ddiv): Correct latency, add div/sqrt resource.
3313 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3314 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3316 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3318 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3320 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3321 switch statement mapping KF built-ins to TF built-ins if we don't
3322 have the proper ISA 3.0 assembler support.
3324 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3326 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3327 (lower_emutls_function_body): Do not compute it.
3329 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3331 * ipa-split.c (split_bb_info): Turn time to sreal.
3332 (split_point): Likewise.
3333 (dump_split_point): Likewise.
3334 (fine_split_points): Likewise.
3335 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3337 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3339 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3342 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3344 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3346 (expensive_function_p): Remove useless assert.
3347 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3349 2017-11-16 Martin Sebor <msebor@redhat.com>
3351 PR tree-optimization/82588
3352 PR tree-optimization/82583
3353 * tree-vrp.c (check_array_ref): Handle flexible array members,
3354 string literals, and inner indices.
3355 (search_for_addr_array): Add detail to diagnostics.
3357 2017-11-16 Nathan Sidwell <nathan@acm.org>
3361 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3362 (SET_DECL_ASSEMBLER_NAME): Forward to
3363 overwrite_decl_assembler_name.
3364 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3365 (overwrite_decl_assembler_name): Declare.
3366 * tree.c (overwrite_decl_assembler_name): New.
3367 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3368 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3369 (LANG_HOOKS_INITIALIZER): Add it.
3370 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3371 * langhooks.c (lhd_set_decl_assembler_name): Use
3372 SET_DECL_ASSEMBLER_NAME.
3373 (lhd_overwrite_decl_assembler_name): Default implementation.
3375 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3376 Jackson Woodruff <jackson.woodruff@arm.com>
3378 PR tree-optimization/71026
3379 * match.pd: Canonicalize constant multiplies in division.
3381 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3383 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3384 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3385 Remove OPT_fomit_frame_pointer entry.
3386 * common/config/arc/arc-common.c: Likewise.
3387 * common/config/arm/arm-common.c: Likewise.
3388 * common/config/avr/avr-common.c: Likewise.
3389 * common/config/c6x/c6x-common.c: Likewise.
3390 * common/config/cr16/cr16-common.c: Likewise.
3391 * common/config/cris/cris-common.c: Likewise.
3392 * common/config/epiphany/epiphany-common.c: Likewise.
3393 * common/config/fr30/fr30-common.c: Likewise.
3394 * common/config/frv/frv-common.c: Likewise.
3395 * common/config/ia64/ia64-common.c: Likewise.
3396 * common/config/iq2000/iq2000-common.c: Likewise.
3397 * common/config/lm32/lm32-common.c: Likewise.
3398 * common/config/m32r/m32r-common.c: Likewise.
3399 * common/config/mcore/mcore-common.c: Likewise.
3400 * common/config/microblaze/microblaze-common.c: Likewise.
3401 * common/config/mips/mips-common.c: Likewise.
3402 * common/config/mmix/mmix-common.c: Likewise.
3403 * common/config/mn10300/mn10300-common.c: Likewise.
3404 * common/config/nios2/nios2-common.c: Likewise.
3405 * common/config/pa/pa-common.c: Likewise.
3406 * common/config/pdp11/pdp11-common.c: Likewise.
3407 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3408 * common/config/riscv/riscv-common.c: Likewise.
3409 * common/config/rs6000/rs6000-common.c: Likewise.
3410 * common/config/rx/rx-common.c: Likewise.
3411 * common/config/s390/s390-common.c: Likewise.
3412 * common/config/sh/sh-common.c: Likewise.
3413 * common/config/sparc/sparc-common.c: Likewise.
3414 * common/config/tilegx/tilegx-common.c: Likewise.
3415 * common/config/tilepro/tilepro-common.c: Likewise.
3416 * common/config/v850/v850-common.c: Likewise.
3417 * common/config/visium/visium-common.c: Likewise.
3418 * common/config/xstormy16/xstormy16-common.c: Likewise.
3419 * common/config/xtensa/xtensa-common.c: Likewise.
3420 * invoke.texi (-fomit-frame-pointer): Update documentation.
3422 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3424 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3426 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3428 * cfg.c (scale_bbs_frequencies_int,
3429 cale_bbs_frequencies_gcov_type): Remove.
3430 * cfg.h (scale_bbs_frequencies_int,
3431 cale_bbs_frequencies_gcov_type): Remove.
3433 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3435 * tree-ssa-loop-manip.c
3436 (scale_dominated_blocks_in_loop): Update to profile counts.
3437 (tree_transform_and_unroll_loop): Likewise.
3439 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3441 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3442 scale_bbs_frequencies_int.
3444 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3446 * final.c (compute_alignments): Use counts rather than frequencies.
3448 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3450 * cfgloopanal.c: Include sreal.h
3451 (average_num_loop_insns): Use counts and sreal for accounting.
3453 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3455 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3458 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3459 Kito Cheng <kito.cheng@gmail.com>
3461 * config/nds32/constraints.md: Provide more constraints.
3462 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3463 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3464 support constraints usage.
3466 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3468 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3470 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3471 Kito Cheng <kito.cheng@gmail.com>
3473 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3474 * config/nds32/nds32.opt: Refine the layout.
3475 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3476 TARGET_EXT_STRING): Support new options.
3477 * config/nds32/nds32.h: Likewise.
3478 * config/nds32/nds32.md: Likewise.
3479 * config/nds32/nds32-predicates.c: Likewise.
3480 * config/nds32/constraints.md: Likewise.
3481 * common/config/nds32/nds32-common.c: Likewise.
3483 2017-11-16 Julia Koval <julia.koval@intel.com>
3486 * config/i386/gfniintrin.h: Add sse check.
3487 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3489 2017-11-16 Julia Koval <julia.koval@intel.com>
3491 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3492 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3493 (ix86_handle_option): Handle -mavx512vbmi2.
3494 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3495 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3496 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3497 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3498 (ix86_valid_target_attribute_inner_p): Ditto.
3499 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3500 * config/i386/i386.opt (mavx512vbmi2): New option.
3501 * doc/invoke.texi: Add new option.
3503 2017-11-16 Julia Koval <julia.koval@intel.com>
3505 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3506 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3507 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3508 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3509 _mm512_gf2p8mul_epi8): New intrinsics.
3510 * config/i386/i386-builtin-types.def
3511 (V64QI_FTYPE_V64QI_V64QI): New type.
3512 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3513 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3514 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3515 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3516 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3517 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3519 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3521 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3522 explicitly as a stream of bytes.
3524 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3526 * config/rs6000/altivec.h (vec_xst_be): New #define.
3527 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3528 and externalize from *altivec_vperm_<mode>_internal.
3529 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3531 (XL_BE_V8HI): Likewise.
3532 (XL_BE_V4SI): Likewise.
3533 (XL_BE_V4SI): Likewise.
3534 (XL_BE_V2DI): Likewise.
3535 (XL_BE_V4SF): Likewise.
3536 (XL_BE_V2DF): Likewise.
3537 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3538 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3539 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3540 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3541 VSX_BUILTIN_VEC_XST_BE.
3542 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3543 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3545 (altivec_init_builtins): Replace conditional calls to def_builtin
3546 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3547 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3548 with unconditional calls. Remove calls to def_builtin for
3549 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3550 __builtin_vec_xst_be.
3551 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3552 to define_expand, and add alternate RTL generation for P8.
3553 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3554 vsx_ld_elemrev_v8hi.
3555 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3556 add alternate RTL generation for P8.
3557 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3558 vsx_ld_elemrev_v16qi.
3559 (vsx_st_elemrev_v8hi): Convert define_insn
3560 to define_expand, and add alternate RTL generation for P8.
3561 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3562 vsx_st_elemrev_v8hi.
3563 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3564 add alternate RTL generation for P8.
3565 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3566 vsx_st_elemrev_v16qi.
3568 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3571 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3572 TARGET_AVX512ER check.
3573 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3574 neither -mzeroupper nor -mno-zeroupper is used and
3575 TARGET_EMIT_VZEROUPPER is set.
3576 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3577 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3579 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3581 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3582 folding of vector compares.
3583 (fold_build_vec_cmp): New helper function.
3584 (fold_compare_helper): New helper function.
3585 (builtin_function_type): Add compare builtins to the list of functions
3586 having unsigned arguments. Cosmetic updates to comment indentation.
3587 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3588 the not+eq combination.
3590 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3592 PR tree-optimization/82726
3593 PR tree-optimization/70754
3594 * tree-predcom.c (order_drefs_by_pos): New function.
3595 (combine_chains): Move code setting has_max_use_after to...
3596 (try_combine_chains): ...here. New parameter. Sort combined chains
3597 according to position information.
3598 (tree_predictive_commoning_loop): Update call to above function.
3599 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3601 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3603 PR tree-optimization/82726
3605 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3607 PR tree-optimization/70754
3608 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3609 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3610 combined stmt before it if not NULL.
3611 (combine_chains): Process refs reversely and compute dominance point
3615 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3617 PR tree-optimization/79663
3618 * tree-predcom.c (combine_chains): Process refs in reverse order
3619 only for ZERO length chains, and add explaining comment.
3621 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3623 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3624 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3625 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3627 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3629 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3631 2017-11-15 Martin Liska <mliska@suse.cz>
3633 * tree-cfg.c (pass_warn_function_return::execute):
3634 Compare warn_return_type for greater than zero.
3636 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3640 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3641 to return true on Xeon and not on Xeon Phi.
3642 (ix86_check_avx256_register): Changed to ...
3643 (ix86_check_avx_upper_register): ... this. Add extra check for
3644 VALID_AVX512F_REG_OR_XI_MODE.
3645 (ix86_avx_u128_mode_needed): Changed
3646 ix86_check_avx256_register to ix86_check_avx_upper_register.
3647 (ix86_check_avx256_stores): Changed to ...
3648 (ix86_check_avx_upper_stores): ... this. Changed
3649 ix86_check_avx256_register to ix86_check_avx_upper_register.
3650 (ix86_avx_u128_mode_after): Changed
3651 avx_reg256_found to avx_upper_reg_found. Changed
3652 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3653 (ix86_avx_u128_mode_entry): Changed
3654 ix86_check_avx256_register to ix86_check_avx_upper_register.
3655 (ix86_avx_u128_mode_exit): Ditto.
3656 * config/i386/i386.h: (host_detect_local_cpu): New define.
3658 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3660 * config/arm/xgene1.md (xgene1): Split into automatons
3661 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3662 (xgene1_f_load): Adjust reservations and/or types.
3663 (xgene1_f_store): Likewise.
3664 (xgene1_load_pair): Likewise.
3665 (xgene1_store_pair): Likewise.
3666 (xgene1_fp_load1): Likewise.
3667 (xgene1_load1): Likewise.
3668 (xgene1_store1): Likewise.
3669 (xgene1_move): Likewise.
3670 (xgene1_alu): Likewise.
3671 (xgene1_simd): Likewise.
3672 (xgene1_bfm): Likewise.
3673 (xgene1_neon_load1): Likewise.
3674 (xgene1_neon_store1): Likewise.
3675 (xgene1_neon_logic): Likewise.
3676 (xgene1_neon_st1): Likewise.
3677 (xgene1_neon_ld1r): Likewise.
3678 (xgene1_alu_cond): Added.
3679 (xgene1_shift_reg): Likwise.
3680 (xgene1_bfx): Likewise.
3681 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3683 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3686 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3688 (can_widen_mult_without_libcall): New function.
3689 (expand_mul_overflow): If only checking unsigned mul overflow,
3690 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3691 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3692 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3693 (expand_DIVMOD): Formatting fix.
3694 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3695 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3696 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3698 PR tree-optimization/82977
3699 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3700 constructed temporary to strlen_to_stridx.put.
3702 2017-11-15 Martin Liska <mliska@suse.cz>
3704 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3705 * configure: Regenerate.
3707 2017-11-15 Martin Liska <mliska@suse.cz>
3710 * config/sh/sh-mem.cc: Use proper probability for
3713 2017-11-14 Jeff Law <law@redhat.com>
3715 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3716 the red zone for stack_clash_protection_final_dynamic_probe targets
3717 when the total dynamic stack size is zero bytes.
3719 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3720 blocks is post order.
3722 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3724 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3725 * final.c (rest_of_clean_state): Set it for the
3726 -fcompare-debug dump.
3727 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3728 class when TDF_COMPARE_DEBUG is set.
3730 * dwarf2out.c (gen_producer_string): Discard
3733 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3736 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3737 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3738 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3739 (__TGMATH_REAL_2_3): Remove macros.
3740 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3741 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3743 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3745 (remquo): Define using __TGMATH_REAL_3.
3747 2017-11-14 Jeff Law <law@redhat.com>
3749 * vr-values.c: New file with contents extracted from tree-vrp.c.
3750 * Makefile.in (OBJS): Add vr-values.o
3751 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3752 (set_value_range, set_and_canonicalize_value_range): Likewise.
3753 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3754 (value_range_constant_singleton, symbolic_range_p): Likewise.
3755 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3756 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3757 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3758 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3759 (find_case_label_range, find_case_label_index): Likewise.
3760 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3761 (range_int_cst_singleton_p, value_inside_range): Likewise.
3762 (get_single_symbol): Likewise.
3763 (switch_update): Move structure definition here.
3764 (to_remove_edges, to_update_switch_stmts): Provide externs.
3765 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3766 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3767 (vrp_val_is_min, set_value_range): Likewise.
3768 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3769 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3770 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3771 (range_is_nonnull, range_int_cst_p): Likewwise.
3772 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3773 (get_single_symbol, operand_less_p): Likewise
3774 (compare_values_warnv, compare_values): Likewise.
3775 (value_inside_range, value_range_constant_singleton): Likewise.
3776 (zero_nonzero_bitgs_from_vr): Likewise.
3777 (extract_range_from_binary_expr_1): Likewise.
3778 (overflow_comparison_p): Likewise.
3779 (to_remove_edges, to_update_switch_stmts): Likewise.
3780 (find_case_label-index, find_case_label_range): Likewise.
3781 (switch_update, set_value_range_to_nonnegative): Remove.
3782 (set_value_range_to_truthvalue): Likewise.
3783 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3784 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3785 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3786 (find_case_label_ranges, test_for_singularity): Likewise.
3787 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3788 (x_vr_values): Move to its remaining use site.
3790 2017-11-10 Jeff Law <law@redhat.com>
3792 * vr-values.h (VR_INITIALIZER): Move #define here.
3793 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3794 * Makefile.in (OBJS): Add tree-evrp.o
3795 * tree-vrp.h (assert_info): Move structure definition here.
3796 (set_value_range_to_varying): Prototype.
3797 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3798 (infer_value_range, register_edge_assert_for): Likewise.
3799 (stmt_interesting_for_vrp): Likewise.
3800 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3801 (set_value_range_to_varying): No longer static.
3802 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3803 (infer_value_range, register_edge_assert_for): Likewise.
3805 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3807 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3808 generate the XXBRD instruction.
3810 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3811 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3812 as being compatible if -mabi=ieeelongdouble.
3813 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3814 to setup float128 built-ins with hardware support.
3815 (BU_FLOAT128_HW_2): Likewise.
3816 (BU_FLOAT128_HW_3): Likewise.
3817 (BU_FLOAT128_HW_VSX_1): Likewise.
3818 (BU_FLOAT128_HW_VSX_2): Likewise.
3819 (scalar_extract_expq): Change float128 built-in functions to
3820 accommodate having both KFmode and TFmode functions. Use the
3821 KFmode variant as the default.
3822 (scalar_extract_sigq): Likewise.
3823 (scalar_test_neg_qp): Likewise.
3824 (scalar_insert_exp_q): Likewise.
3825 (scalar_insert_exp_qp): Likewise.
3826 (scalar_test_data_class_qp): Likewise.
3827 (sqrtf128_round_to_odd): Delete processing the round to odd
3828 built-in functions as special built-in functions, and define them
3829 as float128 built-ins. Use the KFmode variant as the default.
3830 (truncf128_round_to_odd): Likewise.
3831 (addf128_round_to_odd): Likewise.
3832 (subf128_round_to_odd): Likewise.
3833 (mulf128_round_to_odd): Likewise.
3834 (divf128_round_to_odd): Likewise.
3835 (fmaf128_round_to_odd): Likewise.
3836 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3837 support for KFmode and TFmode xststdcqp calls.
3838 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3839 point, switch the built-in handlers for the get/set float128
3840 exponent, get float128 mantissa, float128 test built-ins, and the
3841 float128 round to odd built-in functions. Eliminate creating the
3842 float128 round to odd built-in functions as special built-ins.
3843 (rs6000_init_builtins): Eliminate special creation of the float128
3844 round to odd built-in functions.
3845 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3846 function insns to support both TFmode and KFmode variants.
3847 (xsxsigqp_<mode>): Likewise.
3848 (xsiexpqpf_<mode>): Likewise.
3849 (xsiexpqp_<mode>): Likewise.
3850 (xststdcqp_<mode>): Likewise.
3851 (xststdcnegqp_<mode>): Likewise.
3852 (xststdcqp_<mode>): Likewise.
3854 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3856 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3857 unused path_in_freq_ptr parameter.
3858 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3860 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3862 * ipa-inline.c (edge_badness): Dump sreal frequency.
3863 (compute_inlined_call_time): Match natural implementaiton ...
3864 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3865 forgotten division by CGRAPH_FREQ_BASE.
3867 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3869 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3870 Solaris 11. Update comment.
3871 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3873 * config/sol2.h (STARTFILE_SPEC): Likewise.
3874 * configure: Regenerate.
3876 2017-11-14 Carl Love <cel@us.ibm.com>
3878 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3879 le_ and be_ prefixes to swap* variables. Remove
3880 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3883 2017-11-14 Jason Merrill <jason@redhat.com>
3885 Support GTY((cache)) on hash_map.
3886 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3887 (ggc_cache_remove): Override it instead of ggc_mx.
3888 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3889 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3890 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3891 (simple_cache_map_traits): Override maybe_mx.
3892 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3893 (hash_map): Friend gt_cleare_cache.
3894 (gt_cleare_cache): New.
3895 * tree.h (tree_cache_traits): New hash_map traits class.
3896 (tree_cache_map): New typedef.
3898 2017-11-14 Richard Biener <rguenther@suse.de>
3900 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3901 paramter and handling.
3902 (cleanup_control_flow_bb): Likewise.
3903 (cleanup_control_flow_pre): New helper performing a DFS walk
3904 to call cleanup_control_flow_bb in PRE order.
3905 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3906 via cleanup_control_flow_pre.
3908 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
3910 * config/aarch64/aarch64-simd.md
3911 (aarch64_simd_bsl<mode>_internal): Remove DImode.
3912 (*aarch64_simd_bsl<mode>_alt): Likewise.
3913 (aarch64_simd_bsldi_internal): New.
3914 (aarch64_simd_bsldi_alt): Likewise.
3916 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3918 * tracer.c (better_p): Do not compare frequencies.
3919 * reg-stack.c (better_edge): Likewise.
3920 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
3923 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3925 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
3926 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
3927 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
3928 * lto-streamer-in.c (input_function): Use update_max_bb_count.
3929 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
3930 * predict.c (maybe_hot_frequency_p): Inline to ...
3931 (maybe_hot_count_p): ... here; rewrite to counts.
3932 (counts_to_freqs): Rename to ...
3933 (update_max_bb_count): ... this one.
3934 (expensive_function_p): Use counts.
3935 (estimate_bb_frequencies): Update.
3936 (rebuild_frequencies): Update.
3937 * predict.h (counts_to_freqs): Rename to ...
3938 (update_max_bb_count): ... this one.
3939 * profile.c (compute_branch_probabilities): Add debug info
3940 * tree-inline.c (expand_call_inline): Update debug info.
3941 (optimize_inline_calls): Use update_max_bb_count..
3942 (tree_function_versioning): Use update_max_bb_count..
3943 * value-prof.c (gimple_value_profile_transformations):
3944 Do not use update_max_bb_count.
3946 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3948 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
3949 always use frequencies.
3951 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3953 * bb-reorder.c: Remove frequencies from comments.
3954 (better_edge_p): Use profile counts.
3955 (find_traces): Dump profile counts.
3956 (rotate_loop): Use profile counts.
3957 (find_traces_1_round): Likewise.
3958 (connect_better_edge_p): Use counts instead of probabilities for
3960 (copy_bb_p): Drop early check for non-0 frequency.
3961 (sanitize_hot_paths): Update comments.
3963 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3965 * ipa-split.c (struct split_point): Add count.
3966 (consider_split): Do not compute incoming frequency; compute incoming
3967 count and store it to split_point.
3968 (split_function): Set count of the call to split part correctly.
3970 2017-11-13 Carl Love <cel@us.ibm.com>
3972 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
3974 2017-11-13 Tom Tromey <tom@tromey.com>
3976 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
3978 2017-11-13 Carl Love <cel@us.ibm.com>
3980 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3981 Add support for builtins:
3982 unsigned int vec_first_{,miss}_match_{,or_eos}index,
3983 vector {un,}signed {char,int,short},
3984 vector {un,}signed {char,int,short}) arguments.
3985 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
3986 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
3987 Add BU_P9V_AV_2 expansions for the builtins.
3988 * config/rs6000/altivec.h (vec_first_match_index,
3989 vec_first_mismatch_index, vec_first_match_or_eos_index,
3990 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
3991 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
3992 new extern declaration.
3993 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
3994 * config/rs6000/vsx.md (first_match_index_<mode>,
3995 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
3996 first_mismatch_or_eos_index_<mode>): Add define expand.
3997 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
3998 * doc/extend.texi: Update the built-in documenation file for the new
4001 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4003 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4004 into the min/max operations for _Float<N> and _Float<N>X types.
4006 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4009 * dwarf2out.c (do_eh_frame): New static variable.
4010 (dwarf2out_begin_prologue): Set it.
4011 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4013 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4015 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4017 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4019 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4022 * debug.h (dwarf2out_do_eh_frame): Declare.
4023 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4024 (dwarf2out_do_frame): Use it.
4025 (dwarf2out_do_cfi_asm): Likewise.
4026 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4027 (dwarf2out_assembly_start): Likewise.
4028 (dwarf2out_begin_prologue): Fix comment.
4029 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4030 if the target needs either debug or unwind DWARF2 info.
4031 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4032 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4033 -ftrapping-math, -ftrapv and -fwrapv.
4035 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4037 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4038 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4039 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4040 (estimate_edge_size_and_time): Likewise.
4041 (ipa_merge_fn_summary_after_inlining): Likewise.
4042 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4043 (compute_uninlined_call_time): Use sreal_frequency.
4044 (compute_inlined_call_time): Likewise.
4045 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4046 * profile-count.c: Include sreal.h.
4047 (profile_count::to_sreal_scale): New.
4048 * profile-count.h: Forward declare sreal.
4049 (profile_count::to_sreal_scale): Declare.
4051 2017-11-13 Nathan Sidwell <nathan@acm.org>
4053 * diagnostic.c (maybe_line_and_column): New.
4054 (diagnostic_get_location_text): Use it.
4055 (diagnostic_report_current_module): Likewise.
4056 (test_diagnostic_get_location_text): Add tests.
4058 2017-11-13 Luis Machado <luis.machado@linaro.org>
4060 * doc/md.texi (Specifying processor pipeline description): Fix
4061 incorrect latency for the div instruction example.
4063 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4065 PR tree-optimization/78821
4066 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4067 that bit_not_p is the same.
4068 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4069 (split_group): Count precisely bit_not_p bits in each statement.
4070 (invert_op): New function.
4071 (imm_store_chain_info::output_merged_store): Use invert_op to
4072 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4073 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4075 2017-11-13 Martin Liska <mliska@suse.cz>
4077 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4078 (struct source_info): Likewise.
4079 (add_branch_counts): Likewise.
4080 (add_line_counts): Likewise.
4081 (function_summary): Likewise.
4082 (output_intermediate_line): Likewise.
4083 (generate_results): Likewise.
4085 2017-11-13 Martin Liska <mliska@suse.cz>
4087 * gcov.c (struct block_info): Remove typedef for block_t.
4088 (struct line_info): Likewise.
4089 (line_info::has_block): Likewise.
4090 (EXIT_BLOCK): Likewise.
4091 (unblock): Likewise.
4092 (circuit): Likewise.
4093 (get_cycles_count): Likewise.
4094 (process_file): Likewise.
4095 (read_graph_file): Likewise.
4096 (solve_flow_graph): Likewise.
4097 (find_exception_blocks): Likewise.
4098 (add_line_counts): Likewise.
4099 (accumulate_line_info): Likewise.
4100 (output_line_details): Likewise.
4102 2017-11-13 Martin Liska <mliska@suse.cz>
4104 * gcov.c (struct arc_info): Remove typedef for arc_t.
4105 (struct line_info): Likewise.
4106 (add_branch_counts): Likewise.
4107 (output_branch_count): Likewise.
4108 (function_info::~function_info): Likewise.
4109 (circuit): Likewise.
4110 (output_intermediate_line): Likewise.
4111 (read_graph_file): Likewise.
4112 (solve_flow_graph): Likewise.
4113 (find_exception_blocks): Likewise.
4114 (add_line_counts): Likewise.
4115 (accumulate_line_info): Likewise.
4116 (output_line_details): Likewise.
4117 (output_function_details): Likewise.
4119 2017-11-13 Martin Liska <mliska@suse.cz>
4121 * gcov.c (struct function_info): Remove typedef for function_t.
4122 (struct source_info): Likewise.
4123 (source_info::get_functions_at_location): Likewise.
4124 (solve_flow_graph): Likewise.
4125 (find_exception_blocks): Likewise.
4126 (add_line_counts): Likewise.
4127 (output_intermediate_file): Likewise.
4128 (process_file): Likewise.
4129 (generate_results): Likewise.
4130 (release_structures): Likewise.
4131 (read_graph_file): Likewise.
4132 (read_count_file): Likewise.
4133 (accumulate_line_counts): Likewise.
4134 (output_lines): Likewise.
4136 2017-11-13 Martin Liska <mliska@suse.cz>
4138 * gcov.c (function_info::function_info): Remove num_counts
4139 and add vector<gcov_type>.
4140 (function_info::~function_info): Use the vector.
4141 (process_file): Likewise.
4142 (read_graph_file): Likewise.
4143 (read_count_file): Likewise.
4144 (solve_flow_graph): Likewise.
4146 2017-11-13 Martin Liska <mliska@suse.cz>
4148 * gcov.c (function_info::is_artificial): New function.
4149 (process_file): Erase all artificial early.
4150 (generate_results): Skip as all artificial are already
4153 2017-11-13 Martin Liska <mliska@suse.cz>
4155 * gcov.c (read_graph_file): Store to global vector of functions.
4156 (read_count_file): Iterate the vector.
4157 (process_file): Likewise.
4158 (generate_results): Likewise.
4159 (release_structures): Likewise.
4161 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4163 PR tree-optimization/82954
4164 * gimple-ssa-store-merging.c
4165 (imm_store_chain_info::coalesce_immediate_stores): If
4166 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4168 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4170 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4171 Upddate call to ENDIAN_LANE_N.
4172 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4173 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4174 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4175 and use aarch64_endian_lane_rtx.
4176 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4178 2017-11-12 Tom de Vries <tom@codesourcery.com>
4180 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4182 2017-11-12 Tom de Vries <tom@codesourcery.com>
4184 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4186 2017-11-12 Tom de Vries <tom@codesourcery.com>
4188 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4190 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4191 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4192 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4194 2017-11-11 Martin Sebor <msebor@redhat.com>
4197 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4200 * prefic.c (translate_name): Replace strncpy with memcpy to
4201 avoid -Wstringop-truncation.
4203 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4205 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4207 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4209 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4211 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4213 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4216 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4218 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4219 enable generating XXBRH if the value is in a vector register.
4220 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4221 value is in a vector register.
4222 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4223 register to register bswap64's instead of doing the GPR sequence
4224 used on previous machines.
4225 (bswapdi2_xxbrd): New insn.
4226 (bswapdi2_reg): Disallow on ISA 3.0.
4227 (register to register bswap64 splitter): Do not split the insn on
4228 ISA 3.0 systems that use XXBRD.
4230 2017-11-10 Martin Sebor <msebor@redhat.com>
4233 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4235 (find_subframework_file): Same.
4237 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4239 * auto-profile.c (afdo_indirect_call): Drop frequency.
4240 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4241 (cgraph_node::create_edge): Drop frequency argument.
4242 (cgraph_node::create_indirect_edge): Drop frequency argument.
4243 (cgraph_edge::make_speculative): Drop frequency arguments.
4244 (cgraph_edge::resolve_speculation): Do not update frequencies
4245 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4246 (cgraph_node::dump): Check consistency in IPA mode.
4247 (cgraph_edge::maybe_hot_p): Use IPA counter.
4248 (cgraph_edge::verify_count_and_frequency): Rename to ...
4249 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4250 (cgraph_node::verify_node): Update.
4251 * cgraph.h (struct cgraph_edge): Drop frequency.
4252 (cgraph_edge::frequency): New function.
4253 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4255 (cgraph_edge::rebuild_edges): Likewise.
4256 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4257 (duplicate_thunk_for_node): Do not pass frequency.
4258 (cgraph_node::create_clone): Scale only counts.
4259 (cgraph_node::create_virtual_clone): Do not pass frequency.
4260 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4261 (cgraph_node::create_version_clone): Do not pass frequency.
4262 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4263 (cgraph_node::expand_thunk): Do not pass frequency.
4264 (cgraph_node::create_wrapper): Do not pass frequency.
4265 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4267 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4268 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4269 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4270 (gather_caller_stats): Use frequency function.
4271 (ipcp_cloning_candidate_p): Use frequency function.
4272 (ipcp_propagate_stage): Use frequency function.
4273 (get_info_about_necessary_edges): Use frequency function.
4274 (update_profiling_info): Update only IPA profile.
4275 (update_specialized_profile): Use frequency functoin.
4276 (perhaps_add_new_callers): Update only IPA profile.
4277 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4278 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4279 (dump_ipa_call_summary): Use frequency function.
4280 (estimate_edge_size_and_time): Use frequency function.
4281 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4282 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4283 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4284 (update_noncloned_counts): ... ths one; scale counts only.
4285 (clone_inlined_nodes): Do not scale frequency.
4286 (inline_call): Do not pass frequency.
4287 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4288 (compute_inlined_call_time): Use IPA profile.
4289 (want_inline_small_function_p): Use IPA profile.
4290 (want_inline_self_recursive_call_p): Use IPA profile.
4291 (edge_badness): Use IPA profile.
4292 (lookup_recursive_calls): Use IPA profile.
4293 (recursive_inlining): Do not pass frequency.
4294 (resolve_noninline_speculation): Do not update frequency.
4295 (inline_small_functions): Collect max of IPA profile.
4296 (dump_overall_stats): Dump IPA porfile.
4297 (dump_inline_stats): Dump IPA porfile.
4298 (ipa_inline): Collect IPA stats.
4299 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4300 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4301 (ipa_propagate_frequency): Use frequency function.
4302 (ipa_profile): Cleanup.
4303 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4304 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4305 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4306 (input_node): Do not stream frequency.
4307 (input_edge): Do not stream frequency.
4308 (merge_profile_summaries): Scale only IPA profiles.
4309 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4310 * predict.c (drop_profile): Do not recompute frequency.
4311 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4312 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4313 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4314 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4315 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4316 * tree-inline.c (copy_bb): Do not scale frequency.
4317 (expand_call_inline): Do not scale frequency.
4318 (tree_function_versioning): Do not scale frequency.
4319 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4321 2017-11-10 Julia Koval <julia.koval@intel.com>
4323 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4324 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4325 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4326 (_mm256_maskz_gf2p8affine_epi64_epi8)
4327 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4328 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4329 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4330 (__builtin_ia32_vgf2p8affineqb_v32qi)
4331 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4332 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4334 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4338 (arm_option_override): Refactor.
4339 (arm_option_reconfigure_globals): New.
4340 (arm_options_perform_arch_sanity_checks): New.
4341 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4343 (arm_options_perform_arch_sanity_checks): Likewise
4345 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4347 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4348 (power9-qpmul): New.
4349 * rs6000/rs6000.md ("type" attr): Add qmul.
4350 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4351 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4352 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4354 2017-11-10 Martin Sebor <msebor@redhat.com>
4357 * builtins.c (compute_objsize): Handle arrays that
4358 compute_builtin_object_size likes to fail for. Make extern.
4359 * builtins.h (compute_objsize): Declare.
4360 (check_strncpy_sizes): New function.
4361 (expand_builtin_strncpy): Call check_strncpy_sizes.
4362 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4363 -Wstringop-truncation.
4364 (gimple_fold_builtin_strncat): Same.
4365 * gimple.c (gimple_build_call_from_tree): Set call location.
4366 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4367 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4368 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4369 (handle_builtin_strlen): Use strlen_to_stridx.
4370 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4372 Use strlen_to_stridx.
4373 (pass_strlen::execute): Release strlen_to_stridx.
4374 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4375 (-Wstringop-truncation): Document new option.
4377 2017-11-10 Martin Liska <mliska@suse.cz>
4379 PR gcov-profile/82702
4380 * gcov.c (main): Handle intermediate files in a different
4382 (get_gcov_intermediate_filename): New function.
4383 (output_gcov_file): Remove support of intermediate files.
4384 (generate_results): Allocate intermediate file.
4385 (release_structures): Clean-up properly fn_end.
4386 (output_intermediate_file): Start iterating with line 1.
4388 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4390 PR tree-optimization/82929
4391 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4392 ops_swapped_p non-static data member.
4393 (store_immediate_info::store_immediate_info): Clear it.
4394 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4395 ops set ops_swapped_p.
4396 (count_multiple_uses): Handle ops_swapped_p.
4398 2017-11-10 Martin Liska <mliska@suse.cz>
4400 * coverage.c (coverage_init): Stream information about
4401 support of has_unexecuted_blocks.
4402 * doc/gcov.texi: Document that.
4403 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4404 * gcov.c (read_graph_file): Likewise.
4405 (output_line_beginning): Fix a small issue with
4408 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4410 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4411 reference of trivial component.
4413 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4416 * gimple-ssa-store-merging.c
4417 (pass_store_merging::terminate_all_aliasing_chains): For
4418 gimple_store_p stmts also call refs_output_dependent_p.
4420 PR rtl-optimization/82913
4421 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4424 2017-11-09 Jeff Law <law@redhat.com>
4426 * vr-values.h: New file with vr_values class.
4427 * tree-vrp.c: Include vr-values.h
4428 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4429 data objects into the vr_values class.
4430 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4431 (get_value_range): Make it a member function within vr_values class.
4432 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4433 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4434 (op_with_constant_singleton_value_range): Likewise.
4435 (extract_range_for_var_from_comparison_expr): Likewise.
4436 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4437 (extract_range_from_binary_expr): Likewise.
4438 (extract_range_from_unary_expr): Likewise.
4439 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4440 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4441 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4442 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4443 (compare_name_with_value, compare_names): Likewise.
4444 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4445 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4446 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4447 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4448 (extract_range_from_phi_node): Likewise.
4449 (simplify_truth_ops_using_ranges): Likewise.
4450 (simplify_div_or_mod_using_ranges): Likewise.
4451 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4452 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4453 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4454 (simplify_float_conversion_using_ranges): Likewise.
4455 (simplify_internal_call_using_ranges): Likewise.
4456 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4457 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4458 poitner via x_vr_values for calls into gimple folder.
4459 (vrp_initialize_lattice): Make this the vr_values ctor.
4460 (vrp_free_lattice): Make this the vr_values dtor.
4461 (set_vr_value): New function.
4462 (class vrp_prop): Add vr_values data member. Add various member
4463 functions as well as member functions that delegate to vr_values.
4464 (check_array_ref): Make a member function within vrp_prop class.
4465 (search_for_addr_array, vrp_initialize): Likewise.
4466 (vrp_finalize): Likewise. Revamp to avoid direct access to
4467 vr_value, values_propagated, etc.
4468 (check_array_bounds): Extract vrp_prop class instance pointer from
4469 walk info structure. Use it to call member functions.
4470 (check_all_array_refs): Make a member function within vrp_prop class.
4471 Smuggle class instance pointer via walk info structure.
4472 (x_vr_values): New local static.
4473 (vrp_valueize): Use x_vr_values to get class instance.
4474 (vr_valueize_1): Likewise.
4475 (class vrp_folder): Add vr_values data member. Add various member
4476 functions as well as member functions that delegate to vr_values.
4477 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4478 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4479 class instance from vr_values. Use it to call member functions.
4480 (vrp_dom_walker): Add vr_values data member.
4481 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4482 instance via x_vr_values.
4483 (identify_jump_threads): Accept vr_values as argument. Store
4484 it into the walker structure.
4485 (evrp_dom_walker): Add vr_values class data member. Add various
4487 (evrp_dom_walker::try_find_new_range): Use vr_values data
4488 member to access the memory allocator.
4489 (evrp_dom_walker::before_dom_children): Store vr_values class
4490 instance into the vrp_folder class.
4491 (evrp_dom_walker::push_value_range): Rework to avoid direct
4492 access to num_vr_values and vr_value.
4493 (evrp_dom_walker::pop_value_range): Likewise.
4494 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4495 Use vr_values to get to dump_all_value_ranges member function.
4496 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4497 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4498 Pass vr_values class instance down to identify_jump_threads.
4499 Remove call to vrp_free_lattice.
4500 (debug_all_value_ranges): Remove.
4502 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4503 (vrp_folder): Likewise.
4505 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4506 Get it from the existing bitmap instead.
4507 (vrp_intersect_ranges_1): Likewise.
4509 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4511 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4513 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4514 set bit_not_p to it.
4515 (imm_store_chain_info::coalesce_immediate_stores): Break group
4516 if bit_not_p is different.
4517 (count_multiple_uses, split_group,
4518 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4519 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4520 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4521 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4522 to store_immediate_info ctor.
4524 2017-11-09 Jim Wilson <jimw@sifive.com>
4526 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4527 (scan_prog_file): Likewise.
4529 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4531 * bb-reorder.c (max_entry_frequency): Remove.
4532 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4533 connect_traces, push_to_next_round_p): Remove prototypes.
4534 (find_traces_1_round): Use counts only.
4535 (push_to_next_round_p): Likewise.
4536 (find_traces): Likewise.
4537 (rotate_loop): Likewise.
4538 (find_traces_1_round): Likewise.
4539 (connect_traces): Likewise.
4540 (edge_order): Likewise.
4542 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4544 * config/arm/arm.c (output_return_instruction): Add comments to
4545 indicate requirement for cmse_nonsecure_entry return to account
4546 for the size of clearing instruction output here.
4547 (thumb_exit): Likewise.
4548 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4549 return in hardfloat mode.
4551 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4553 * config/rs6000/rs6000.c (machine_function): Add a bool,
4554 "toc_is_wrapped_separately".
4555 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4556 if it wasn't explicitly set or unset, we are optimizing for speed, and
4557 doing separate shrink-wrapping.
4558 (rs6000_get_separate_components): Enable the TOC component if
4559 saving the TOC register in the prologue.
4560 (rs6000_components_for_bb): Handle the TOC component.
4561 (rs6000_emit_prologue_components): Store the TOC register where needed.
4562 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4563 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4565 2017-11-09 Martin Jambor <mjambor@suse.cz>
4567 * ipa-param-manipulation.c: New file.
4568 * ipa-param-manipulation.h: Likewise.
4569 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4570 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4571 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4572 (ipa_parm_adjustment): Likewise.
4573 (ipa_parm_adjustment_vec): Likewise.
4574 (ipa_get_vector_of_formal_parms): Moved declaration to
4575 ipa-param-manipulation.h.
4576 (ipa_get_vector_of_formal_parm_types): Likewise.
4577 (ipa_modify_formal_parameters): Likewise.
4578 (ipa_modify_call_arguments): Likewise.
4579 (ipa_combine_adjustments): Likewise.
4580 (ipa_dump_param_adjustments): Likewise.
4581 (ipa_modify_expr): Likewise.
4582 (ipa_get_adjustment_candidate): Likewise.
4583 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4584 ipa-param-manipulation.c.
4585 (ipa_get_vector_of_formal_parm_types): Likewise.
4586 (ipa_modify_formal_parameters): Likewise.
4587 (ipa_modify_call_arguments): Likewise.
4588 (ipa_modify_expr): Likewise.
4589 (get_ssa_base_param): Likewise.
4590 (ipa_get_adjustment_candidate): Likewise.
4591 (index_in_adjustments_multiple_times_p): Likewise.
4592 (ipa_combine_adjustments): Likewise.
4593 (ipa_dump_param_adjustments): Likewise.
4594 * tree-sra.c: Also include ipa-param-manipulation.h
4595 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4598 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4599 Alan Hayward <alan.hayward@arm.com>
4600 David Sherwood <david.sherwood@arm.com>
4602 * doc/sourcebuild.texi (vect_masked_store): Document.
4604 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4605 Alan Hayward <alan.hayward@arm.com>
4606 David Sherwood <david.sherwood@arm.com>
4608 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4610 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4611 Alan Hayward <alan.hayward@arm.com>
4612 David Sherwood <david.sherwood@arm.com>
4614 * doc/sourcebuild.texi (vect_variable_length): Document.
4616 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4617 Alan Hayward <alan.hayward@arm.com>
4618 David Sherwood <david.sherwood@arm.com>
4620 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4622 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4623 Alan Hayward <alan.hayward@arm.com>
4624 David Sherwood <david.sherwood@arm.com>
4626 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4628 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4629 Alan Hayward <alan.hayward@arm.com>
4630 David Sherwood <david.sherwood@arm.com>
4632 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4633 previously undocumented selectors.
4634 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4636 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4638 * doc/rtl.texi (const_vector): Say that elements can be
4639 const_wide_ints too.
4640 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4641 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4642 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4643 * optabs.c (expand_vector_broadcast): Likewise.
4645 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4646 Alan Hayward <alan.hayward@arm.com>
4647 David Sherwood <david.sherwood@arm.com>
4649 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4650 scaled index even if the unscaled address was invalid.
4651 Don't increase the complexity of using a scale in that case.
4653 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4654 Alan Hayward <alan.hayward@arm.com>
4655 David Sherwood <david.sherwood@arm.com>
4657 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4658 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4659 UNITS_PER_WORD bytes.
4660 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4661 based on REGMODE_NATURAL_SIZE of the inner mode.
4662 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4663 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4664 * expmed.c (lowpart_bit_field_p): Divide the value up into
4665 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4666 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4667 whether something is likely to occupy more than one register.
4669 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4672 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4673 profile_count::adjust_for_ipa_scaling.
4674 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4675 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4677 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4679 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4681 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4682 (split_group): Add total_orig and total_new arguments, estimate the
4683 number of statements related to the store group without store merging
4684 and with store merging.
4685 (imm_store_chain_info::output_merged_store): Adjust split_group
4686 callers, punt if estimated number of statements with store merging
4687 is not smaller than estimated number of statements without it.
4689 (handled_load): Remove has_single_use checks.
4690 (pass_store_merging::process_store): Likewise.
4692 2017-11-09 Richard Biener <rguenther@suse.de>
4694 PR tree-optimization/82902
4695 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4697 2017-11-09 Martin Liska <mliska@suse.cz>
4700 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4703 2017-11-09 Martin Liska <mliska@suse.cz>
4705 PR tree-optimization/82669
4706 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4708 2017-11-09 Martin Liska <mliska@suse.cz>
4710 PR gcov-profile/48463
4711 * coverage.c (coverage_begin_function): Output also end locus
4712 of a function and information whether the function is
4714 * gcov-dump.c (tag_function): Parse and print the information.
4715 * gcov.c (INCLUDE_MAP): Add include.
4716 (INCLUDE_SET): Likewise.
4717 (struct line_info): Move earlier in the source file because
4718 of vector<line_info> in function_info structure.
4719 (line_info::line_info): Likewise.
4720 (line_info::has_block): Likewise.
4721 (struct source_info): Add new member index.
4722 (source_info::get_functions_at_location): New function.
4723 (function_info::group_line_p): New function.
4724 (output_intermediate_line): New function.
4725 (output_intermediate_file): Use the mentioned function.
4726 (struct function_start): New.
4727 (struct function_start_pair_hash): Likewise.
4728 (process_file): Add code that identifies group functions.
4729 Assign lines either to global or function scope.
4730 (generate_results): Skip artificial functions.
4731 (find_source): Assign index for each source file.
4732 (read_graph_file): Read new flag artificial and end_line.
4733 (add_line_counts): Assign it either to global of function scope.
4734 (accumulate_line_counts): Isolate core of the function to
4735 accumulate_line_info and call it for both function and global
4737 (accumulate_line_info): New function.
4738 (output_line_beginning): Fix GNU coding style.
4739 (print_source_line): New function.
4740 (output_line_details): Likewise.
4741 (output_function_details): Likewise.
4742 (output_lines): Iterate both source (global) scope and function
4744 (struct function_line_start_cmp): New class.
4745 * doc/gcov.texi: Reflect changes in documentation.
4747 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4750 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4751 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4752 and similarly for not instead of neg.
4754 2017-11-08 Andi Kleen <ak@linux.intel.com>
4756 * config/i386/i386.opt: Add -mforce-indirect-call.
4757 * config/i386/predicates.md: Check for flag_force_indirect_call.
4758 * doc/invoke.texi: Document -mforce-indirect-call
4760 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4762 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4764 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4765 config/riscv/riscv.c (predict.h): New include.
4766 (riscv_slow_unaligned_access_p): No longer static.
4767 (riscv_block_move_straight): Add require.
4768 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4770 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4773 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4774 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4775 nonimmediate_operand predicate for operand 1 instead of
4778 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4780 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4782 * config/aarch64/constraints.md (Uml): New constraint.
4783 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4786 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4788 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4789 of two vec_duplicates into a vec_concat.
4791 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4793 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4794 Simplify vec_merge of vec_duplicate and vec_concat.
4795 * config/aarch64/constraints.md (Utq): New constraint.
4796 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4799 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4801 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4802 Simplify vec_merge of vec_duplicate and const_vector.
4803 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4805 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4806 mode iterator. Update predicate on operand 1 to
4807 handle non-const_vec constants. Delete constraints.
4808 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4810 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4812 PR tree-optimization/78821
4813 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4815 (store_operand_info::store_operand_info): Initialize it to false.
4816 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4817 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4818 store in the group, and if chain_info is non-NULL, to ignore altogether
4820 (compatible_load_p): Fail if bit_not_p does not match.
4821 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4822 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4823 (pass_store_merging::process_store): Adjust
4824 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4825 call terminate_all_aliasing_chains newly when adding a store into
4826 a chain with non-NULL chain_info.
4828 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4830 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4832 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4834 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4836 (aarch64_layout_frame): Initialise emit_frame_chain.
4837 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4838 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4840 2017-11-08 Martin Liska <mliska@suse.cz>
4842 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4843 of gimple_call_internal_p.
4845 2017-11-07 Tom de Vries <tom@codesourcery.com>
4847 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4849 2017-11-07 Tom de Vries <tom@codesourcery.com>
4851 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4854 2017-11-08 Martin Liska <mliska@suse.cz>
4857 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4859 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4861 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4863 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4866 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4867 (*cmp<mode>_ccz_1): New insn with $k alternative.
4870 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4871 RTX_COMM_COMPARE as commutative as well.
4872 (ix86_binary_operator_ok): Formatting fix.
4873 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4874 *<code><mode>3<mask_name><round_saeonly_name>,
4875 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4876 *<s>mul<mode>3_highpart<mask_name>,
4877 *vec_widen_umult_even_v16si<mask_name>,
4878 *vec_widen_umult_even_v8si<mask_name>,
4879 *vec_widen_umult_even_v4si<mask_name>,
4880 *vec_widen_smult_even_v16si<mask_name>,
4881 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4882 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4883 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4884 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4885 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4886 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4887 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4888 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4889 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4890 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4891 ix86_binary_operator_ok. Formatting fixes.
4892 (*<plusminus_insn><mode>3<mask_name><round_name>,
4893 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4896 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4898 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4899 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4900 eq and ne if TARGET_ISEL.
4901 (cmp): New code_iterator.
4902 (UNS, UNSU_, UNSIK): New code_attrs.
4903 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4904 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4906 ("eq<mode>3"): ... this.
4907 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
4909 ("ne<mode>3"): ... this.
4911 2017-11-07 Julia Koval <julia.koval@intel.com>
4914 * common/config/i386/i386-common.c
4915 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
4916 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
4917 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
4918 * config/i386/i386.opt: Ditto.
4919 * config/i386/i386.c (ix86_target_string): Ditto.
4920 (ix86_option_override_internal): Ditto.
4921 (ix86_init_mpx_builtins): Move MPX to args2.
4922 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
4923 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4924 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4925 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4926 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4927 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4928 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
4930 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
4933 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
4934 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
4935 (zero-extendsidi peephole2): Remove peephole.
4937 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4940 * stor-layout.c: Include attribs.h.
4941 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
4942 explicit lookup of "aligned" attribute.
4944 2017-11-07 Andrew Waterman <andrew@sifive.com>
4946 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
4947 (riscv_expand_block_move): Likewise.
4948 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
4950 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
4951 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
4952 * config/riscv/riscv.c (riscv_block_move_straight): New function.
4953 (riscv_adjust_block_mem): Likewise.
4954 (riscv_block_move_loop): Likewise.
4955 (riscv_expand_block_move): Likewise.
4956 * config/riscv/riscv.md (movmemsi): New pattern.
4958 2017-11-07 Michael Clark <michaeljclark@mac.com>
4960 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
4961 (MUSL_DYNAMIC_LINKER): Likewise.
4963 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
4965 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
4966 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
4969 2017-11-07 Richard Biener <rguenther@suse.de>
4971 * match.pd: Fix build.
4973 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
4974 Jackson Woodruff <jackson.woodruff@arm.com>
4976 PR tree-optimization/71026
4977 * match.pd: Canonicalize negate in division.
4979 2017-11-07 Sudakshina Das <sudi.das@arm.com>
4982 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
4984 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4986 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
4987 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
4989 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4991 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
4992 non-scalar integral types.
4993 * match.pd (negate_expr_p): Handle MINUS_EXPR.
4994 (-(A-B), -(~A)): New transformations.
4996 2017-11-07 Tom de Vries <tom@codesourcery.com>
4998 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
4999 semicolon after "do {} while (0)".
5000 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5001 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5002 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5003 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5004 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5006 2017-11-07 Tom de Vries <tom@codesourcery.com>
5008 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5009 after "do {} while (0)".
5010 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5011 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5012 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5013 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5014 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5016 2017-11-07 Tom de Vries <tom@codesourcery.com>
5019 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5021 (arm_rtx_costs_internal): Add missing semicolon after
5022 HANDLE_NARROW_SHIFT_ARITH call.
5024 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5026 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5027 disable isel if it was not set explicitly.
5029 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5031 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5032 (add_type_attribute) likewise.
5034 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5036 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5037 of struct ix86_frame.
5038 (ix86_initial_elimination_offset): Likewise.
5039 (ix86_expand_split_stack_prologue): Likewise.
5041 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5043 * tree-vrp.h (enum value_range_type): Update stale comment.
5045 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5046 Alan Hayward <alan.hayward@arm.com>
5047 David Sherwood <david.sherwood@arm.com>
5049 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5050 (aarch64_expand_vec_perm_const): Take the number of units too.
5051 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5052 (aarch64_expand_vec_perm_const): Likewise.
5053 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5054 (vec_perm<mode>): Update accordingly.
5056 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5057 Alan Hayward <alan.hayward@arm.com>
5058 David Sherwood <david.sherwood@arm.com>
5060 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5061 Take the number of units too.
5062 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5063 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5064 but check for a vector mode before rather than after the call.
5065 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5066 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5067 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5068 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5069 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5070 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5071 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5072 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5073 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5074 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5075 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5076 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5077 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5078 (aarch64_sqdmull2_n<mode>): Update accordingly.
5080 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5081 Alan Hayward <alan.hayward@arm.com>
5082 David Sherwood <david.sherwood@arm.com>
5084 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5085 the number of units too.
5086 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5087 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5088 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5089 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5090 (vec_store_lanesxi<mode>): Update accordingly.
5092 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5093 Alan Hayward <alan.hayward@arm.com>
5094 David Sherwood <david.sherwood@arm.com>
5096 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5097 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5098 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5099 of units rather than the mode.
5100 * config/aarch64/iterators.md (nunits): New mode attribute.
5101 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5102 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5103 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5104 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5105 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5106 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5107 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5108 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5109 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5110 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5111 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5112 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5113 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5114 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5115 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5116 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5117 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5118 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5119 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5120 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5121 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5122 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5123 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5124 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5125 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5126 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5127 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5128 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5129 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5130 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5131 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5132 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5133 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5134 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5135 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5136 (aarch64_simd_vec_setv2di): Likewise.
5138 2017-11-06 Carl Love <cel@us.ibm.com>
5140 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5142 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5143 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5145 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5147 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5148 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5149 Add power 8 macro expansions.
5150 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5151 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5152 power 8 instructions. (VSX_XXBR): Add iterator.
5154 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5156 * config/arm/arm.md (predicable_short_it): Change default to "no",
5157 improve documentation, remove uses that are identical to the default.
5158 (enabled_for_depr_it): Rename to enabled_for_short_it.
5159 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5160 Remove default uses.
5161 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5162 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5163 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5164 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5166 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5169 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5170 float128 helper macros, which are no longer used after deleting
5171 the old 'q' built-in functions, and moving the round to odd
5172 built-in functions to being special built-in functions.
5173 (BU_FLOAT128_2): Likewise.
5174 (BU_FLOAT128_1_HW): Likewise.
5175 (BU_FLOAT128_2_HW): Likewise.
5176 (BU_FLOAT128_3_HW): Likewise.
5177 (FABSQ): Delete old 'q' built-in functions.
5178 (COPYSIGNQ): Likewise.
5179 (SQRTF128_ODD): Move round to odd built-in functions to be
5180 special built-in functions, so that we can handle
5181 -mabi=ieeelongdouble.
5182 (TRUNCF128_ODD): Likewise.
5183 (ADDF128_ODD): Likewise.
5184 (SUBF128_ODD): Likewise.
5185 (MULF128_ODD): Likewise.
5186 (DIVF128_ODD): Likewise.
5187 (FMAF128_ODD): Likewise.
5188 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5189 built-in names to 'f128'.
5190 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5191 old 'q' built-in functions, as the machine independent code for
5192 'f128' built-in functions handles this.
5193 (rs6000_expand_builtin): Add expansion for float128 round to odd
5194 functions, keying off on -mabi=ieeelongdouble of whether to use
5195 the KFmode or TFmode variant.
5196 (rs6000_init_builtins): Initialize the _Float128 round to odd
5198 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5199 _Float128 'q' built-in functions are now mapped into the new
5200 'f128' built-in functions.
5202 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5204 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5205 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5207 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5210 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5211 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5212 * toplev.c: Include "ipa-fnsummary.h".
5213 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5215 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5217 PR tree-optimization/82838
5218 * gimple-ssa-store-merging.c
5219 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5220 on a separate gimple_seq which is then appended to seq.
5222 2017-11-06 Jeff Law <law@redhat.com>
5225 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5226 (get_probe_interval): New functions.
5227 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5228 (ix86_adjust_stack_and_probe): Likewise.
5229 (output_adjust_stack_and_probe): Likewise.
5230 (ix86_emit_probe_stack_range): Likewise.
5231 (ix86_expand_prologue): Likewise.
5233 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5235 PR tree-optimization/82816
5236 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5237 if the modes of the two types are the same.
5238 (convert_plusminus_to_widen): Likewise.
5240 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5242 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5244 (usadv16qi): New define_expand.
5245 (usadv8hi): New define_expand.
5247 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5250 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5252 (inline_transform): Likewise.
5253 * predict.c (counts_to_freqs): Remove useless conditional.
5254 * profile-count.h (profile_count::apply_scale): Move sanity check.
5255 * tree-inline.c (copy_bb): Always scale.
5256 (copy_cfg_body): Likewise.
5258 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5261 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5263 (*cmp_ite0): Add enabled_for_depr_it attribute.
5264 (*cmp_ite1): Likewise.
5266 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5268 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5271 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5273 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5274 return the value by pointer.
5275 (extract_range_from_multiplicative_op_1): Update accordingly.
5276 Return as soon as an operation fails.
5278 2017-11-05 Tom de Vries <tom@codesourcery.com>
5281 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5282 (DEF_SANITIZER_BUILTIN): ... here.
5283 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5284 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5286 2017-11-05 Tom de Vries <tom@codesourcery.com>
5289 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5291 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5292 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5293 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5295 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5296 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5297 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5299 2017-11-05 Tom de Vries <tom@codesourcery.com>
5302 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5305 2017-11-04 Michael Clark <michaeljclark@mac.com>
5307 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5308 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5310 (*addsi3_extended): Likewise.
5311 (*addsi3_extended2): Likewise.
5312 (<optab>si3): Likewise.
5313 (<optab>di3): Likewise.
5314 (<optab><mode>3): Likewise.
5315 (<*optabe>si3_internal): Likewise.
5316 (zero_extendqi<SUPERQI:mode>2): Likewise.
5317 (*add<mode>hi3): Likewise.
5318 (*xor<mode>hi3): Likewise.
5319 (<optab>di3): Likewise.
5320 (*<optab>si3_extend): Likewise.
5321 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5322 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5323 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5325 2017-11-04 Andrew Waterman <andrew@sifive.com>
5327 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5328 TARGET_STRICT_ALIGN based upon -mtune argument.
5330 2017-11-04 Andrew Waterman <andrew@sifive.com>
5332 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5334 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5336 * config/i386/i386.c (choose_basereg): Use optional scratch
5337 register and add assertion.
5338 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5339 needed, and don't allocate stack.
5340 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5341 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5342 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5344 2017-11-03 Jeff Law <law@redhat.com>
5346 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5347 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5348 to probe at the start of a noreturn function.
5350 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5352 PR tree-optimization/78821
5353 * gimple-ssa-store-merging.c: Update the file comment.
5354 (MAX_STORE_ALIAS_CHECKS): Define.
5355 (struct store_operand_info): New type.
5356 (store_operand_info::store_operand_info): New constructor.
5357 (struct store_immediate_info): Add rhs_code and ops data members.
5358 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5359 and op1r arguments to the ctor, initialize corresponding data members.
5360 (struct merged_store_group): Add load_align_base and load_align
5362 (merged_store_group::merged_store_group): Initialize them.
5363 (merged_store_group::do_merge): Update them.
5364 (merged_store_group::apply_stores): Pick the constant for
5365 encode_tree_to_bitpos from one of the two operands, or skip
5366 encode_tree_to_bitpos if neither operand is a constant.
5367 (class pass_store_merging): Add process_store method decl. Remove
5368 bool argument from terminate_all_aliasing_chains method decl.
5369 (pass_store_merging::terminate_all_aliasing_chains): Remove
5370 var_offset_p argument and corresponding handling.
5371 (stmts_may_clobber_ref_p): New function.
5372 (compatible_load_p): New function.
5373 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5374 if there is overlap and rhs_code is not INTEGER_CST. For
5375 non-overlapping stores terminate group if rhs is not mergeable.
5376 (get_alias_type_for_stmts): Change first argument from
5377 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5378 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5379 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5381 (get_location_for_stmts): Change first argument from
5382 auto_vec<gimple *> & to vec<gimple *> &.
5383 (struct split_store): Remove orig_stmts data member, add orig_stores.
5384 (split_store::split_store): Create orig_stores rather than orig_stmts.
5385 (find_constituent_stmts): Renamed to ...
5386 (find_constituent_stores): ... this. Change second argument from
5387 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5388 to info structures rather than the statements.
5389 (split_group): Rename ALLOW_UNALIGNED argument to
5390 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5391 it. Adjust find_constituent_stores caller.
5392 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5393 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5394 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5395 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5396 (mem_valid_for_store_merging): New function.
5397 (handled_load): New function.
5398 (pass_store_merging::process_store): New method.
5399 (pass_store_merging::execute): Use process_store method. Adjust
5400 terminate_all_aliasing_chains caller.
5402 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5404 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5405 Return true for more constants, symbols and label references.
5406 (aarch64_valid_floating_const): Remove unused function.
5408 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5411 * config/aarch64/aarch64.c (aarch64_layout_frame):
5412 Undo forcing of LR at bottom of frame.
5414 2017-11-03 Jeff Law <law@redhat.com>
5417 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5418 for int_registers_saved.
5420 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5421 extracted from tree-ssa-dom.c.
5422 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5423 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5424 (record_equivalences_from_incoming_edge): Add additional argument
5425 to single_pred_edge_ignoring_loop_edges call.
5426 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5427 (uncprop_dom_walker::before_dom_children): Add additional argument
5428 to single_pred_edge_ignoring_loop_edges call.
5429 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5430 single_pred_edge_ignoring_loop_edges rather than open coding.
5431 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5433 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5435 * match.pd (-(-A)): Rewrite.
5437 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5439 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5440 (rs6000_emit_int_cmove): New declaration.
5441 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5442 (rs6000_emit_sISEL): Delete.
5443 (rs6000_emit_int_cmove): Make non-static.
5444 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5445 instead of rs6000_emit_sISEL.
5447 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5449 * asan.c (create_cond_insert_point): Maintain profile.
5450 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5452 * basic-block.h (struct basic_block_def): Remove frequency.
5453 (EDGE_FREQUENCY): Use to_frequency
5454 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5456 (find_traces): Update to use to_frequency.
5457 (find_traces_1_round): Likewise; use only IPA counts.
5458 (bb_to_key): Likewise.
5459 (connect_traces): Use IPA counts only.
5460 (copy_bb_p): Update to use to_frequency.
5461 (fix_up_crossing_landing_pad): Likewise.
5462 (sanitize_hot_paths): Likewise.
5463 * bt-load.c (basic_block_freq): Likewise.
5464 * cfg.c (init_flow): Set count_max to uninitialized.
5465 (check_bb_profile): Remove frequencies; check counts.
5466 (dump_bb_info): Do not dump frequencies.
5467 (update_bb_profile_for_threading): Update counts only.
5468 (scale_bbs_frequencies_int): Likewise.
5469 (MAX_SAFE_MULTIPLIER): Remove.
5470 (scale_bbs_frequencies_gcov_type): Update counts only.
5471 (scale_bbs_frequencies_profile_count): Update counts only.
5472 (scale_bbs_frequencies): Update counts only.
5473 * cfg.h (struct control_flow_graph): Add count-max.
5474 (update_bb_profile_for_threading): Update prototype.
5475 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5476 (find_many_sub_basic_blocks): Likewise.
5477 * cfgcleanup.c (try_forward_edges): Likewise.
5478 (try_crossjump_to_edge): Likewise.
5479 * cfgexpand.c (expand_gimple_cond): Likewise.
5480 (expand_gimple_tailcall): Likewise.
5481 (construct_init_block): Likewise.
5482 (construct_exit_block): Likewise.
5483 * cfghooks.c (verify_flow_info): Check consistency of counts.
5484 (dump_bb_for_graph): Do not dump frequencies.
5485 (split_block_1): Do not update frequencies.
5486 (split_edge): Do not update frequencies.
5487 (make_forwarder_block): Do not update frequencies.
5488 (duplicate_block): Do not update frequencies.
5489 (account_profile_record): Do not update frequencies.
5490 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5491 for global heuristics.
5492 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5493 (expected_loop_iterations_unbounded): Use counts only.
5494 * cfgloopmanip.c (scale_loop_profile): Simplify.
5495 (create_empty_loop_on_edge): Simplify
5497 (duplicate_loop_to_header_edge): Simplify
5498 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5499 (update_br_prob_note): Take care of removing note when profile
5501 (relink_block_chain): Do not dump frequency.
5502 (rtl_account_profile_record): Use to_frequency.
5503 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5504 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5505 (cgraph_update_edges_for_call_stmt_node): Likewise.
5506 (cgraph_edge::verify_count_and_frequency): Update.
5507 (cgraph_node::verify_node): Temporarily disable frequency verification.
5508 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5509 to_cgraph_frequency.
5510 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5511 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5513 (cgraph_node::expand_thunk): Update profile.
5514 * except.c (dw2_build_landing_pads): Do not update frequency.
5515 * final.c (compute_alignments): Use to_frequency.
5516 (dump_basic_block_info): Do not dump frequency.
5517 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5518 (dump_gimple_bb_header): Do not dump frequency.
5519 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5521 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5522 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5523 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5524 (init_before_recovery): Do not update frequency.
5525 (sched_create_recovery_edges): Do not update frequency.
5526 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5527 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5528 (ipa_cp_c_finalize): Set max_count to uninitialized.
5529 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5530 (param_change_prob): Use counts.
5531 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5533 * ipa-split.c (consider_split): Use to_frequency.
5534 (split_function): Use to_frequency.
5535 * ira-build.c (loop_compare_func): Likewise.
5536 (mark_loops_for_removal): Likewise.
5537 (mark_all_loops_for_removal): Likewise.
5538 * loop-doloop.c (doloop_modify): Do not update frequency.
5539 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5541 * lto-streamer-in.c (input_function): Update count_max.
5542 * omp-expand.c (expand_omp_taskreg): Update count_max.
5543 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5544 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5545 (maybe_hot_count_p): Use ipa counts only.
5546 (maybe_hot_bb_p): Simplify.
5547 (maybe_hot_edge_p): Simplify.
5548 (probably_never_executed): Do not take frequency argument.
5549 (probably_never_executed_bb_p): Do not pass frequency.
5550 (probably_never_executed_edge_p): Likewise.
5551 (combine_predictions_for_bb): Check that profile is nonzero.
5552 (propagate_freq): Do not set frequency.
5553 (drop_profile): Simplify.
5554 (counts_to_freqs): Simplify.
5555 (expensive_function_p): Use to_frequency.
5556 (propagate_unlikely_bbs_forward): Simplify.
5557 (determine_unlikely_bbs): Simplify.
5558 (estimate_bb_frequencies): Add hack to silence graphite issues.
5559 (compute_function_frequency): Use ipa counts.
5560 (pass_profile::execute): Update.
5561 (rebuild_frequencies): Use counts only.
5562 (force_edge_cold): Use counts only.
5563 * profile-count.c (profile_count::dump): Dump new count types.
5564 (profile_count::differs_from_p): Check compatiblity.
5565 (profile_count::to_frequency): New function.
5566 (profile_count::to_cgraph_frequency): New function.
5567 * profile-count.h (struct function): Declare.
5568 (enum profile_quality): Add profile_guessed_local and
5569 profile_guessed_global0.
5570 (class profile_proability): Decrease number of bits to 29;
5571 update from_reg_br_prob_note and to_reg_br_prob_note.
5572 (class profile_count: Update comment; decrease number of bits
5573 to 61. Check compatibility.
5574 (profile_count::compatible_p): New private member function.
5575 (profile_count::ipa_p): New member function.
5576 (profile_count::operator<): Handle global zero correctly.
5577 (profile_count::operator>): Handle global zero correctly.
5578 (profile_count::operator<=): Handle global zero correctly.
5579 (profile_count::operator>=): Handle global zero correctly.
5580 (profile_count::nonzero_p): New member function.
5581 (profile_count::force_nonzero): New member function.
5582 (profile_count::max): New member function.
5583 (profile_count::apply_scale): Handle IPA scalling.
5584 (profile_count::guessed_local): New member function.
5585 (profile_count::global0): New member function.
5586 (profile_count::ipa): New member function.
5587 (profile_count::to_frequency): Declare.
5588 (profile_count::to_cgraph_frequency): Declare.
5589 * profile.c (OVERLAP_BASE): Delete.
5590 (compute_frequency_overlap): Delete.
5591 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5592 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5593 * sched-ebb.c (rank): Use counts only.
5594 * shrink-wrap.c (handle_simple_exit): Use counts only.
5595 (try_shrink_wrapping): Use counts only.
5596 (place_prologue_for_one_component): Use counts only.
5597 * tracer.c (find_best_predecessor): Use to_frequency.
5598 (find_trace): Use to_frequency.
5599 (tail_duplicate): Use to_frequency.
5600 * trans-mem.c (expand_transaction): Do not update frequency.
5601 * tree-call-cdce.c: Do not update frequency.
5602 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5603 (gimple_merge_blocks): Likewise.
5604 (gimple_split_edge): Likewise.
5605 (gimple_duplicate_sese_region): Likewise.
5606 (gimple_duplicate_sese_tail): Likewise.
5607 (move_sese_region_to_fn): Likewise.
5608 (gimple_account_profile_record): Likewise.
5609 (insert_cond_bb): Likewise.
5610 * tree-complex.c (expand_complex_div_wide): Likewise.
5611 * tree-eh.c (lower_resx): Update profile.
5612 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5614 (initialize_cfun): Do not initialize frequencies
5615 (freqs_to_counts): Delete.
5616 (copy_cfg_body): Ignore count parameter.
5617 (copy_body): Update.
5618 (expand_call_inline): Update count_max.
5619 (optimize_inline_calls): Update count_max.
5620 (tree_function_versioning): Update count_max.
5621 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5622 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5624 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5625 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5626 (try_peel_loop): Likewise.
5627 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5629 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5630 (tree_transform_and_unroll_loop): Do not use frequencies
5631 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5632 Use reliable prediction only.
5633 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5634 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5635 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5636 probability scaling.
5637 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5638 not update frequency
5639 (any_remaining_duplicated_blocks): Likewise.
5640 (update_profile): Likewise.
5641 (estimated_freqs_path): Delete.
5642 (freqs_to_counts_path): Delete.
5643 (clear_counts_path): Delete.
5644 (ssa_fix_duplicate_block_edges): Likewise.
5645 (duplicate_thread_path): Likewise.
5646 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5647 * tree-tailcall.c (decrease_profile): Do not update frequency.
5648 (eliminate_tail_call): Likewise.
5649 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5650 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5651 (optimize_mask_stores): Likewise.
5652 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5653 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5654 (ubsan_expand_ptr_ifn): Update profile.
5655 * value-prof.c (gimple_ic): Simplify.
5656 * value-prof.h (gimple_ic): Update prototype.
5657 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5658 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5660 (want_inline_self_recursive_call_p): Likewise.
5661 (resolve_noninline_speculation): Only cummulate defined counts.
5662 (inline_small_functions): Use nonzero_p.
5663 (ipa_inline): Do not access freed node.
5665 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5667 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5668 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5670 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5672 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5673 non-legitimate address.
5675 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5677 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5678 (*lt0_<mode>di, *lt0_<mode>si): New.
5680 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5682 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5683 TARGET_PAIRED_FLOAT.
5685 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5686 Jim Wilson <jim.wilson@linaro.org>
5688 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5689 * config/aarch64/aarch64-tune.md: Regenerated.
5690 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5691 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5693 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5695 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5696 (arc_expand_prologue): Restore blink for millicode.
5697 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5699 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5702 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5703 gen_vec_duplicate after forcing the scalar into a register.
5705 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5707 * combine (try_combine): Print the insns input to try_combine to the
5710 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5713 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5714 Remove second argument from aarch64_process_target_attr call.
5715 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5717 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5719 (aarch64_handle_attr_arch): Remove second argument.
5720 (aarch64_handle_attr_cpu): Ditto.
5721 (aarch64_handle_attr_tune): Ditto.
5722 (aarch64_handle_attr_isa_flags): Ditto.
5723 (aarch64_process_one_target_attr): Ditto.
5724 (aarch64_process_target_attr): Ditto.
5725 (aarch64_option_valid_attribute_p): Remove second argument.
5726 on aarch64_process_target_attr call.
5728 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5730 * diagnostic.c: Include "selftest-diagnostic.h".
5731 (selftest::assert_location_text): New function.
5732 (selftest::test_diagnostic_get_location_text): New function.
5733 (selftest::diagnostic_c_tests): Call it.
5735 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5737 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5738 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5739 (class selftest::test_diagnostic_context): Move to...
5740 * selftest-diagnostic.c: New file.
5741 * selftest-diagnostic.h: New file.
5743 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5745 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5746 offset range for FT32B.
5747 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5748 * config/ft32/ft32.md: Add TARGET_NOPM.
5749 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5750 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5752 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5754 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5756 2017-11-02 Jeff Law <law@redhat.com>
5758 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5759 virtual keyword on FINAL OVERRIDE members.
5761 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5762 virtuals together. Add virtual destructor.
5763 (substitute_and_fold_engine): Similarly.
5765 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5767 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5769 2017-11-02 Richard Biener <rguenther@suse.de>
5771 PR tree-optimization/82795
5772 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5774 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5776 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5778 (gcc_GAS_CHECK_FEATURE): Remove.
5779 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5780 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5781 * configure: Regenerate.
5783 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5785 * config/arc/arc.c (hwloop_optimize): Account for empty
5788 2017-11-02 Richard Biener <rguenther@suse.de>
5791 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5792 Truncate ARRAY_REF index and element size.
5794 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5796 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5798 2017-11-01 Jeff Law <law@redhat.com>
5800 * tree-ssa-ccp.c (ccp_folder): New class derived from
5801 substitute_and_fold_engine.
5802 (ccp_folder::get_value): New member function.
5803 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5804 (ccp_fold_stmt): Remove prototype.
5805 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5806 * tree-ssa-copy.c (copy_folder): New class derived from
5807 substitute_and_fold_engine.
5808 (copy_folder::get_value): Renamed from get_value.
5809 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5810 * tree-vrp.c (vrp_folder): New class derived from
5811 substitute_and_fold_engine.
5812 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5813 (vrp_folder::get_value): New member function.
5814 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5815 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5816 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5817 provide a class interface to folder/substitute routines.
5818 (ssa_prop_fold_stmt_fn): Remove typedef.
5819 (ssa_prop_get_value_fn): Likewise.
5820 (subsitute_and_fold): Remove prototype.
5821 (replace_uses_in): Likewise.
5822 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5823 Renamed from replace_uses_in. Call the virtual member function
5824 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5825 (substitute_and_fold_dom_walker): Remove initialization of
5826 data member entries for calbacks. Add substitute_and_fold_engine
5827 member and initialize it.
5828 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5829 member functions for get_value, replace_phi_args_in c
5830 replace_uses_in, and fold_stmt calls.
5831 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5832 substitute_and_fold. Remove assert. Update ctor call.
5834 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5835 (ssa_prop_visit_phi_fn): Likewise.
5836 (class ssa_propagation_engine): New class to provide an interface
5838 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5840 (ssa_prop_visit_phi): Likewise.
5841 (ssa_propagation_engine::simulate_stmt): Moved into class.
5842 Call visit_phi/visit_stmt from the class rather than via
5843 file scoped static variables.
5844 (ssa_propagation_engine::simulate_block): Moved into class.
5845 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5846 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5847 set file scoped statics for the visit_stmt/visit_phi callbacks.
5848 * tree-complex.c (complex_propagate): New class derived from
5849 ssa_propagation_engine.
5850 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5851 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5852 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5854 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5855 ssa_propagation_engine.
5856 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5857 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5858 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5859 * tree-ssa-copy.c (copy_prop): New class derived from
5860 ssa_propagation_engine.
5861 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5862 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5863 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5864 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5865 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5866 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5867 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5869 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5871 PR rtl-optimization/82778
5872 PR rtl-optimization/82597
5873 * compare-elim.c (struct comparison): Add in_a_setter field.
5874 (find_comparison_dom_walker::before_dom_children): Remove killed
5875 bitmap and df_simulate_find_defs call, instead walk the defs.
5876 Compute last_setter and initialize in_a_setter. Merge definitions
5877 with first initialization for a few variables.
5878 (try_validate_parallel): Use insn_invalid_p instead of
5879 recog_memoized. Return insn rather than just the pattern.
5880 (try_merge_compare): Fix up comment. Don't uselessly test if
5881 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5883 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5886 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5887 Alan Hayward <alan.hayward@arm.com>
5888 David Sherwood <david.sherwood@arm.com>
5890 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5891 aarch64_hard_regno_nregs to get the number of registers
5894 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5895 Alan Hayward <alan.hayward@arm.com>
5896 David Sherwood <david.sherwood@arm.com>
5898 * config/aarch64/constraints.md (Upl): Rename to...
5900 * config/aarch64/aarch64.md
5901 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5904 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5905 Alan Hayward <alan.hayward@arm.com>
5906 David Sherwood <david.sherwood@arm.com>
5908 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
5909 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
5912 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5913 Alan Hayward <alan.hayward@arm.com>
5914 David Sherwood <david.sherwood@arm.com>
5916 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
5917 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
5918 (aarch64_evpc_dup): Generate rtl direcly, rather than using
5920 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
5921 of a single element.
5922 * config/aarch64/iterators.md: Add a comment above the permute
5923 unspecs to say that they are generated directly by
5924 aarch64_expand_vec_perm_const.
5925 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
5927 2017-11-01 Nathan Sidwell <nathan@acm.org>
5929 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
5931 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5933 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
5934 and medany code models, and describe what they do.
5936 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5938 Revert accidental duplicate:
5940 * combine.c (can_change_dest_mode): Reject changes in
5941 REGMODE_NATURAL_SIZE.
5943 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
5945 PR rtl-optimization/64682
5946 PR rtl-optimization/69567
5947 PR rtl-optimization/69737
5948 PR rtl-optimization/82683
5949 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
5950 register mentioned in the note, drop the note, unless it came from I3,
5951 in which case it should go to I3 again.
5953 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5955 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
5956 and return false if not.
5957 (clear_bytes_written_by, live_bytes_read): Update accordingly.
5959 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5961 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
5962 range is known to be empty.
5964 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5965 Alan Hayward <alan.hayward@arm.com>
5966 David Sherwood <david.sherwood@arm.com>
5968 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
5969 and CONST_VECTOR_NUNITS instead of computing the number of units from
5970 the byte sizes of the vector and element.
5971 (simplify_binary_operation_1): Likewise.
5972 (simplify_const_binary_operation): Likewise.
5973 (simplify_ternary_operation): Likewise.
5975 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5976 Alan Hayward <alan.hayward@arm.com>
5977 David Sherwood <david.sherwood@arm.com>
5979 * var-tracking.c (INT_MEM_OFFSET): Replace with...
5980 (int_mem_offset): ...this new function.
5981 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
5982 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
5983 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
5986 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5987 Alan Hayward <alan.hayward@arm.com>
5988 David Sherwood <david.sherwood@arm.com>
5990 * lower-subreg.c (interesting_mode_p): New function.
5991 (compute_costs, find_decomposable_subregs, decompose_register)
5992 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
5993 (resolve_clobber, dump_choices): Use it.
5995 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5996 Alan Hayward <alan.hayward@arm.com>
5997 David Sherwood <david.sherwood@arm.com>
5999 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6001 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6002 Alan Hayward <alan.hayward@arm.com>
6003 David Sherwood <david.sherwood@arm.com>
6005 * alias.c (find_base_value, find_base_term): Only process integer
6006 truncations. Check the precision rather than the size.
6008 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6009 Alan Hayward <alan.hayward@arm.com>
6010 David Sherwood <david.sherwood@arm.com>
6012 * machmode.h (is_narrower_int_mode): New function
6013 * optabs.c (expand_float, expand_fix): Use it.
6014 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6016 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6017 Alan Hayward <alan.hayward@arm.com>
6018 David Sherwood <david.sherwood@arm.com>
6020 * rtl.h (narrower_subreg_mode): New function.
6021 * ira-color.c (update_costs_from_allocno): Use it.
6023 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6024 Alan Hayward <alan.hayward@arm.com>
6025 David Sherwood <david.sherwood@arm.com>
6027 * optabs-query.h (convert_optab_p): New function, split out from...
6028 (convert_optab_handler): ...here.
6029 (widening_optab_handler): Delete.
6030 (find_widening_optab_handler): Remove permit_non_widening parameter.
6031 (find_widening_optab_handler_and_mode): Likewise. Provide an
6032 override that operates on mode class wrappers.
6033 * optabs-query.c (widening_optab_handler): Delete.
6034 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6035 parameter. Assert that the two modes are the same class and that
6036 the "from" mode is narrower than the "to" mode. Use
6037 convert_optab_handler instead of widening_optab_handler.
6038 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6039 instead of widening_optab_handler.
6040 * expr.c (expand_expr_real_2): Update calls to
6041 find_widening_optab_handler.
6042 * optabs.c (expand_widen_pattern_expr): Likewise.
6043 (expand_binop_directly): Take the insn_code as a parameter.
6044 (expand_binop): Only call find_widening_optab_handler for
6045 conversion optabs; use optab_handler otherwise. Update calls
6046 to find_widening_optab_handler and expand_binop_directly.
6047 Use convert_optab_handler instead of widening_optab_handler.
6048 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6049 find_widening_optab_handler and use scalar_mode rather than
6051 (convert_plusminus_to_widen): Likewise.
6053 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6054 Alan Hayward <alan.hayward@arm.com>
6055 David Sherwood <david.sherwood@arm.com>
6057 * machmode.h (fixed_size_mode): New class.
6058 * rtl.h (get_pool_mode): Return fixed_size_mode.
6059 * gengtype.c (main): Add fixed_size_mode.
6060 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6061 (get_raw_arg_mode): Likewise.
6062 * doc/tm.texi: Regenerate.
6063 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6064 * targhooks.c (default_get_reg_raw_mode): Likewise.
6065 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6066 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6067 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6068 (msp430_get_raw_result_mode): Likewise.
6069 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6070 * dbxout.c (dbxout_parms): Require fixed-size modes.
6071 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6072 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6073 * omp-low.c (lower_oacc_reductions): Likewise.
6074 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6075 (simplify_subreg): Update accordingly.
6076 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6077 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6078 that aren't fixed-size.
6079 (get_pool_mode): Return a fixed_size_mode.
6080 (output_constant_pool_2): Take a fixed_size_mode.
6082 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6083 Alan Hayward <alan.hayward@arm.com>
6084 David Sherwood <david.sherwood@arm.com>
6086 * doc/rtl.texi (vec_series): Document.
6087 (const): Say that the operand can be a vec_series.
6088 * rtl.def (VEC_SERIES): New rtx code.
6089 * rtl.h (const_vec_series_p_1): Declare.
6090 (const_vec_series_p): New function.
6091 * emit-rtl.h (gen_const_vec_series): Declare.
6092 (gen_vec_series): Likewise.
6093 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6094 (gen_vec_series): Likewise.
6095 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6096 * simplify-rtx.c (simplify_unary_operation): Handle negations
6098 (simplify_binary_operation_series): New function.
6099 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6100 (test_vector_ops_series): New function.
6101 (test_vector_ops): Call it.
6102 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6103 gen_const_vec_series.
6104 (altivec_lvsr): Likewise.
6105 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6107 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6108 Alan Hayward <alan.hayward@arm.com>
6109 David Sherwood <david.sherwood@arm.com>
6111 * doc/rtl.texi (const): Update description of address constants.
6112 Say that vector constants are allowed too.
6113 * common.md (E, F): Use CONSTANT_P instead of checking for
6115 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6116 checking for CONST_VECTOR.
6117 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6119 * expr.c (expand_expr_real_2): Check for vector modes instead
6120 of checking for CONST_VECTOR.
6121 * rtl.h (const_vec_p): New function.
6122 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6123 (unwrap_const_vec_duplicate): Handle them here too.
6125 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6126 David Malcolm <dmalcolm@redhat.com>
6127 Alan Hayward <alan.hayward@arm.com>
6128 David Sherwood <david.sherwood@arm.com>
6130 * rtl.h (vec_duplicate_p): New function.
6131 * selftest-rtl.c (assert_rtx_eq_at): New function.
6132 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6133 (assert_rtx_eq_at): Declare.
6134 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6135 * selftest-run-tests.c (selftest::run_tests): Call it.
6136 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6137 (simplify_unary_operation_1): Recursively handle vector duplicates.
6138 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6140 (simplify_subreg): Handle subregs of vector duplicates.
6141 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6142 (selftest::simplify_rtx_c_tests): New functions.
6144 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6145 Alan Hayward <alan.hayward@arm.com>
6146 David Sherwood <david.sherwood@arm.com>
6148 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6149 (gen_vec_duplicate): Likewise.
6150 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6152 (gen_const_vector): ...here.
6153 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6154 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6155 whose elements are all equal.
6156 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6157 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6158 (simplify_relational_operation): Likewise.
6159 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6161 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6162 (aarch64_expand_vector_init): Likewise.
6163 * config/arm/arm.c (neon_vdup_constant): Likewise.
6164 (neon_expand_vector_init): Likewise.
6165 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6166 (arm_block_set_unaligned_vect): Likewise.
6167 (arm_block_set_aligned_vect): Likewise.
6168 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6169 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6170 (expand_vec_perm_even_odd_pack): Likewise.
6171 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6172 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6173 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6174 gen_const_vec_duplicate.
6175 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6176 * config/mips/mips.c (mips_gen_const_int_vector): Use
6177 gen_const_vec_duplicate.
6178 (mips_expand_vector_init): Use CONST0_RTX.
6179 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6180 (define_split): Use gen_const_vec_duplicate.
6181 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6182 (define_split): Use gen_const_vec_duplicate.
6183 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6184 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6185 * config/spu/spu.c (spu_const): Likewise.
6187 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6188 Alan Hayward <alan.hayward@arm.com>
6189 David Sherwood <david.sherwood@arm.com>
6191 * combine.c (can_change_dest_mode): Reject changes in
6192 REGMODE_NATURAL_SIZE.
6194 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6196 * configure.ac (--enable-libssp): New.
6197 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6198 trying to determine target-specific default. Adjust indentation.
6199 * configure: Regenerated.
6200 * doc/install.texi (Configuration): Expand --disable-libssp
6203 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6205 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6208 2017-10-31 Martin Jambor <mjambor@suse.cz>
6211 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6213 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6215 * auto-profile.c (autofdo_source_profile::read): Use
6216 UNKNOWN_LOCATION rather than 0.
6217 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6218 (warning_at): ...this overload.
6219 (warning_at_rich_loc_n): Rename to...
6220 (warning_n): ...this overload.
6221 (error_at_rich_loc): Rename to...
6222 (error_at): ...this overload.
6223 (pedwarn_at_rich_loc): Rename to...
6224 (pedwarn): ...this overload.
6225 (permerror_at_rich_loc): Rename to...
6226 (permerror): ...this overload.
6227 (inform_at_rich_loc): Rename to...
6228 (inform): ...this overload.
6229 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6230 (diagnostic_n_impl_richloc): Rename to...
6231 (diagnostic_n_impl): ...this rich_location *-based decl.
6232 (inform_at_rich_loc): Rename to...
6233 (inform): ...this, and add an assertion.
6234 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6235 (warning_at_rich_loc): Rename to...
6236 (warning_at): ...this, and add an assertion.
6237 (warning_at_rich_loc_n): Rename to...
6238 (warning_n): ...this, and add an assertion.
6239 (warning_n): Update location_t-based implementation for removal of
6240 location_t-based diagnostic_n_impl.
6241 (pedwarn_at_rich_loc): Rename to...
6242 (pedwarn): ...this, and add an assertion.
6243 (permerror_at_rich_loc): Rename to...
6244 (permerror): ...this, and add an assertion.
6245 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6246 (error_at_rich_loc): Rename to...
6247 (error_at): ...this, and add an assertion.
6248 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6249 (driver::do_spec_on_infiles): Likewise.
6250 * substring-locations.c (format_warning_va): Update for renaming
6251 of inform_at_rich_loc.
6253 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6255 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6256 _Float<N>X built-in functions so that the variant without the
6257 "__builtin_" prefix is only enabled for the GNU C and Objective C
6258 languages when they are in non-strict ANSI/ISO mode.
6259 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6260 * target.def (floatn_builtin_p): Add a target hook to control
6261 whether _Float<N> and _Float<N>X built-in functions without the
6262 "__builtin_" prefix are enabled, and return true for C and
6263 Objective C in the default hook. Include langhooks.h in
6265 * targhooks.h (default_floatn_builtin_p): Likewise.
6266 * targhooks.c (default_floatn_builtin_p): Likewise.
6267 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6268 floatn_builtin_p target hook.
6269 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6271 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6272 Eric Botcazou <ebotcazou@adacore.com>
6274 PR rtl-optimization/81803
6275 * lra-constraints.c (curr_insn_transform): Also reload the whole
6276 register for a strict subreg no wider than a word if this is for
6277 a WORD_REGISTER_OPERATIONS target.
6279 2017-10-31 Jason Merrill <jason@redhat.com>
6281 * gdbinit.in: Skip over inlines from timevar.h.
6283 2017-10-31 Martin Liska <mliska@suse.cz>
6285 * doc/gcov.texi: Document new option.
6286 * gcov.c (print_usage): Likewise print it.
6287 (process_args): Support the argument.
6288 (format_count): New function.
6289 (format_gcov): Use the function.
6291 2017-10-31 Martin Liska <mliska@suse.cz>
6293 * gcov.c (struct name_map): do not use typedef.
6294 Define operator== and operator<.
6295 (name_search): Remove.
6296 (name_sort): Remove.
6297 (main): Do not allocate names.
6298 (process_file): Add vertical space.
6299 (generate_results): Use std::find.
6300 (release_structures): Do not release memory.
6301 (find_source): Use std::find.
6303 2017-10-31 Martin Liska <mliska@suse.cz>
6305 * gcov.c (struct line_info): Remove it's typedef.
6306 (line_info::line_info): Add proper ctor.
6307 (line_info::has_block): Do not use a typedef.
6308 (struct source_info): Do not use typedef.
6309 (circuit): Likewise.
6310 (get_cycles_count): Likewise.
6311 (output_intermediate_file): Iterate via vector iterator.
6312 (add_line_counts): Use std::vector methods.
6313 (accumulate_line_counts): Likewise.
6314 (output_lines): Likewise.
6316 2017-10-31 Martin Liska <mliska@suse.cz>
6318 * gcov.c (struct source_info): Remove typedef.
6319 (source_info::source_info): Add proper ctor.
6320 (accumulate_line_counts): Use struct, not it's typedef.
6321 (output_gcov_file): Likewise.
6322 (output_lines): Likewise.
6323 (main): Do not allocate an array.
6324 (output_intermediate_file): Use size of vector container.
6325 (process_file): Resize the vector.
6326 (generate_results): Do not preallocate, use newly added vector
6328 (release_structures): Do not release sources.
6329 (find_source): Use vector methods.
6330 (add_line_counts): Do not use typedef.
6332 2017-10-31 Martin Liska <mliska@suse.cz>
6334 * doc/gcov.texi: Document that.
6335 * gcov.c (add_line_counts): Mark lines with a non-executed
6337 (output_line_beginning): Handle such lines.
6338 (output_lines): Pass new argument.
6339 (output_intermediate_file): Print it in intermediate format.
6341 2017-10-31 Martin Liska <mliska@suse.cz>
6343 * color-macros.h: New file.
6344 * diagnostic-color.c: Factor out color related to macros to
6346 * doc/gcov.texi: Document -k option.
6347 * gcov.c (INCLUDE_STRING): Include string.h.
6348 (print_usage): Add -k option.
6349 (process_args): Parse it.
6350 (pad_count_string): New function.
6351 (output_line_beginning): Likewise.
6352 (DEFAULT_LINE_START): New macro.
6353 (output_lines): Support color output.
6355 2017-10-31 Martin Liska <mliska@suse.cz>
6357 PR gcov-profile/82633
6358 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6359 their interaction with GCOV infrastructure.
6360 * configure.ac: Add -fkeep-{inline,static}-functions to
6362 * configure: Regenerate.
6364 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6367 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6369 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6372 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6373 into a register if it does not fit into an immediate offset field.
6375 2017-10-31 Olivier Hainque <hainque@adacore.com>
6377 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6379 2017-10-31 Julia Koval <julia.koval@intel.com>
6381 * config.gcc: Add gfniintrin.h.
6382 * config/i386/gfniintrin.h: New.
6383 * config/i386/i386-builtin-types.def
6384 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6385 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6386 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6387 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6388 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6389 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6390 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6391 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6392 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6393 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6394 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6395 * config/i386/immintrin.h: Include gfniintrin.h.
6396 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6398 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6400 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6402 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6404 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6405 (arm_ashldi3_1bit): Remove pattern.
6406 (ashrdi3): Remove shift by 1 expansion.
6407 (arm_ashrdi3_1bit): Remove pattern.
6408 (lshrdi3): Remove shift by 1 expansion.
6409 (arm_lshrdi3_1bit): Remove pattern.
6410 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6411 cost of ashldi3 by 1.
6412 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6413 (<shift>di3_neon): Likewise.
6415 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6417 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6418 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6419 and (*aarch64_simd_mov<VQ:mode>).
6420 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6421 pattern alternative.
6422 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6423 attributes to match pattern alternative.
6425 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6427 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6428 * config/rs6000/emmintrin.h: New file.
6429 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6431 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6433 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6434 * (movdi_vfp_cortexa8): Remove pattern.
6436 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6438 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6440 (Specific, alpha*-dec-osf5.1): Remove.
6441 (Specific, mips-sgi-irix5): Remove.
6442 (Specific, mips-sgi-irix6): Remove.
6444 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6447 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6448 arguments to clear_bit_region_be.
6450 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6452 * gimplify.c: Include memmodel.h.
6454 2017-10-30 Martin Jambor <mjambor@suse.cz>
6456 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6457 condition whether loop should be executed at all.
6459 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6461 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6462 gimple folding of vec_madd() intrinsics.
6463 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6464 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6465 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6467 2017-10-30 Richard Biener <rguenther@suse.de>
6469 PR tree-optimization/82762
6471 2017-10-23 Richard Biener <rguenther@suse.de>
6473 PR tree-optimization/82129
6475 2017-08-01 Richard Biener <rguenther@suse.de>
6477 PR tree-optimization/81181
6478 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6479 (compute_antic): ... end of iteration here.
6481 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6483 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6484 -std=iso9899:2017 and -std=gnu17.
6485 * doc/standards.texi (C Language): Document C17 support.
6486 * doc/cpp.texi (Overview): Mention -std=c17.
6487 (Standard Predefined Macros): Document C11 and C17 values of
6488 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6489 * doc/extend.texi (Inline): Do not list individual options for
6490 standards newer than C99.
6491 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6493 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6496 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6498 * asan.c (asan_finish_file): Align asan globals array by shadow
6501 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6504 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6505 (struct store_immediate_info): Add bitregion_start and bitregion_end
6507 (store_immediate_info::store_immediate_info): Add brs and bre
6508 arguments and initialize bitregion_{start,end} from those.
6509 (struct merged_store_group): Add bitregion_start, bitregion_end,
6510 align_base and mask fields. Drop unnecessary struct keyword from
6511 struct store_immediate_info. Add do_merge method.
6512 (clear_bit_region_be): Use memset instead of loop storing zeros.
6513 (merged_store_group::do_merge): New method.
6514 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6515 stores as long as the surrounding bitregions have no gaps.
6516 (merged_store_group::merge_overlapping): Use do_merge.
6517 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6518 is byte aligned, rather than requiring that start and width are
6519 byte aligned. Drop unnecessary struct keyword from
6520 struct store_immediate_info. Allocate and populate also mask array.
6521 Make start of the arrays relative to bitregion_start rather than
6522 start and size them according to bitregion_{end,start} difference.
6523 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6524 struct store_immediate_info.
6525 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6526 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6527 struct keyword from struct store_immediate_info.
6528 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6529 between stores as long as the surrounding bitregions have no gaps.
6531 (struct split_store): Add orig non-static data member.
6532 (split_store::split_store): Initialize orig to false.
6533 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6534 if there is exactly a single original stmt. Change stmts argument
6535 to pointer from reference, if NULL, don't push anything to it. Add
6536 first argument, use it to optimize skipping over orig stmts that
6537 are known to be before bitpos already. Simplify.
6538 (split_group): Return unsigned int count how many stores are or
6539 would be needed rather than a bool. Add allow_unaligned argument.
6540 Change split_stores argument from reference to pointer, if NULL,
6541 only do a dry run computing how many stores would be produced.
6542 Rewritten algorithm to use both alignment and misalign if
6543 !allow_unaligned and handle bitfield stores with gaps.
6544 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6545 from bitregion_start instead of start. Compute allow_unaligned
6546 here, if true, do 2 split_group dry runs to compute which one
6547 produces fewer stores and prefer aligned if equal. Punt if
6548 new count is bigger or equal than original before emitting any
6549 statements, rather than during that. Remove no longer needed
6550 new_ssa_names tracking. Replace num_stmts with
6551 split_stores.length (). Use 32-bit stack allocated entries
6552 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6553 if possible. Handle bitfields with gaps.
6554 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6555 Compute bitregion_{start,end} for the stores and construct
6556 store_immediate_info with that. Formatting fixes.
6558 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6561 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6562 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6564 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6566 * gimplify.c: Include tm_p.h.
6568 * common.opt (gcoff): Re-add as ignored option.
6569 (gcoff1, gcoff2, gcoff3): Likewise.
6571 * Makefile.in (OBJS): Delete sdbout.o.
6572 (GTFILES): Delete $(srcdir)/sdbout.c.
6573 * debug.h: Delete sdb_debug_hooks.
6574 * final.c: Delete sdbout.h include.
6575 (final_scan_insn): Delete SDB_DEBUG check.
6576 (rest_of_clean_state): Likewise.
6577 * output.h: Delete sdb_begin_function_line.
6580 * toplev.c: Delete sdbout.h include.
6581 (process_options): Delete SDB_DEBUG check.
6582 * tree-core.h (tree_type_common): Delete pointer field of
6584 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6585 TYPE_SYMTAB_POINTER.
6586 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6587 (TYPE_SYMTAB_IS_POINTER): Delete.
6588 (TYPE_SYMTAB_IS_DIE): Renumber.
6589 * xcoffout.c: Refer to former sdbout.c file.
6590 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6592 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6593 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6594 (Debugging Options): Delete -gcoff.
6595 (-gstabs): Delete SDB reference.
6597 (-gcoff@var{level}): Delete.
6598 * doc/passes.texi (Debugging information output): Delete SDB and
6599 sdbout.c references.
6600 * doc/tm.texi: Regenerate.
6601 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6602 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6604 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6605 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6606 (SDB_DEBUGGING_INFO): Delete.
6607 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6608 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6609 * target.def (output_source_filename): Delete COFF reference.
6611 * common.opt (gcoff): Delete.
6612 (gxcoff+): Update Negative chain.
6613 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6615 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6616 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6617 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6619 (expand_function_start): Change sdb reference to past tense.
6620 (expand_function_end): Change sdb reference to past tense.
6621 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6622 * opts.c (debug_type_names): Delete coff entry.
6623 (common_handle_option): Delete OPT_gcoff case.
6624 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6626 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6627 * config/cris/cris.h: Delete SDB reference in comment.
6628 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6629 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6630 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6631 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6633 (ix86_expand_prologue): Likewise.
6634 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6635 * config/ia64/ia64.h: Likewise.
6636 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6637 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6639 * config/mmix/mmix.h: Likewise.
6640 * config/nds32/nds32.c: Likewise.
6641 * config/stormy/storym16.h: Likewise.
6642 * config/visium/visium.h: Likewise.
6643 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6645 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6647 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6648 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6649 FRAME_POINTER_REGNUM point at high end of local var area.
6651 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6653 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6654 Move comment around. Do not reset best_edge for a copiable
6655 destination if the copy would cause a partition change.
6656 (better_edge_p): Remove redundant check.
6658 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6660 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6662 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6664 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6665 for math functions that have _Float<N> and _Float<N>X variants.
6666 (mathfn_built_in_2): Add support for math functions that have
6667 _Float<N> and _Float<N>X variants.
6668 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6669 (expand_builtin_mathfn_ternary): Add support for fma with
6670 _Float<N> and _Float<N>X variants.
6671 (expand_builtin): Likewise.
6672 (fold_builtin_3): Likewise.
6673 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6674 create math function _Float<N> and _Float<N>X variants as external
6676 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6677 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6678 the __builtin_ prefix and if not strict ansi, without the prefix.
6679 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6680 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6681 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6682 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6683 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6684 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6685 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6686 function signatures for fma _Float<N> and _Float<N>X variants.
6687 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6688 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6689 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6690 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6691 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6692 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6693 * gencfn-macros.c (print_case_cfn): Add support for math functions
6694 that have _Float<N> and _Float<N>X variants.
6695 (print_define_operator_list): Likewise.
6696 (fltfn_suffixes): Likewise.
6698 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6699 for math functions that have _Float<N> and _Float<N>X variants.
6700 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6701 and _Float<N>X variants.
6702 (COPYSIGN): Likewise.
6705 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6706 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6708 (integer_valued_read_call_p): Likewise.
6709 * fold-const-call.c (fold_const_call_ss): Likewise.
6710 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6711 _Float<N> and _Float<N>X variants.
6712 (fold_const_call_ssss): Add support for fma _Float<N> and
6713 _Float<N>X variants.
6714 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6715 support for copysign and fma _Float<N> and _Float<N>X variants.
6716 (backprop::process_builtin_call_use): Likewise.
6717 * tree-call-cdce.c (can_test_argument_range); Add support for
6718 sqrt _Float<N> and _Float<N>X variants.
6719 (edom_only_function): Likewise.
6720 (get_no_error_domain): Likewise.
6721 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6722 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6723 copysign _Float<N> and _Float<N>X variants.
6724 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6725 handled by machine independent code.
6726 (FMAF128): Likewise.
6727 * doc/cpp.texi (Common Predefined Macros): Document defining
6728 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6729 fma _Float<N> and _Float<N>X variants.
6731 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6734 * config/i386/i386-modes.def (CCFPU): Remove definition.
6735 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6736 (ix86_cc_modes_compatible): Ditto.
6737 (ix86_expand_carry_flag_compare): Ditto.
6738 (ix86_expand_int_movcc): Ditto.
6739 (ix86_expand_int_addcc): Ditto.
6740 (ix86_reverse_condition): Ditto.
6741 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6742 Return true/false for unordered/ordered fp comparisons.
6743 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6744 (ix86_prepare_fp_compare_args): Update for rename.
6745 (ix86_expand_fp_compare): Update for rename. Generate unordered
6746 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6747 (ix86_expand_sse_compare_and_jump): Ditto.
6748 * config/i386/predicates.md (fcmov_comparison_operator):
6749 Remove CCFPU mode handling.
6750 (ix86_comparison_operator): Ditto.
6751 (ix86_carry_flag_operator): Ditto.
6752 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6753 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6754 (*cmpu<mode>_cc_i387): Ditto.
6755 (FPCMP): Remove mode iterator.
6756 (unord): Remove mode attribute.
6757 (unord_subst): New define_subst transformation
6758 (unord): New define_subst attribute.
6760 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6761 (*cmpi<unord>xf_i387): Ditto.
6762 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6763 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6764 using unord_subst transformation.
6765 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6766 (round_saeonly): Also handle CCFP mode.
6767 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6768 Remove UNSPEC_SAHF unspec handling.
6770 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6772 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6774 2017-10-27 Jeff Law <law@redhat.com>
6776 * gimple-ssa-sprintf.c: Include domwalk.h.
6777 (class sprintf_dom_walker): New class, derived from dom_walker.
6778 (sprintf_dom_walker::before_dom_children): New function.
6779 (struct call_info): Moved into sprintf_dom_walker class
6780 (compute_formath_length, handle_gimple_call): Likewise.
6781 (sprintf_length::execute): Call the dominator walker rather
6782 than walking the statements.
6784 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6785 gimple statement locations.
6786 (check_array_bounds): Corresponding changes. Get the statement's
6787 location directly from wi->stmt.
6789 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6792 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6794 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6796 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6797 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6799 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6802 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6803 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6804 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6805 maybe_get_pool_constant.
6806 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6809 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6811 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6812 2.26 caveat. Update gas and gld versions.
6813 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6816 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6818 * cgraph.h (set_malloc_flag): Declare.
6819 * cgraph.c (set_malloc_flag_1): New function.
6820 (set_malloc_flag): Likewise.
6821 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6822 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6824 (read_ipa_call_summary): Add support for reading is_return_callee.
6825 (write_ipa_call_summary): Stream is_return_callee.
6826 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6827 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6828 ipa-prop.h, ipa-fnsummary.h.
6829 (pure_const_names): Change to static.
6830 (malloc_state_e): Define.
6831 (malloc_state_names): Define.
6832 (funct_state_d): Add field malloc_state.
6833 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6834 (check_retval_uses): New function.
6835 (malloc_candidate_p): Likewise.
6836 (analyze_function): Add support for malloc attribute.
6837 (pure_const_write_summary): Stream malloc_state.
6838 (pure_const_read_summary): Add support for reading malloc_state.
6839 (dump_malloc_lattice): New function.
6840 (propagate_malloc): New function.
6841 (warn_function_malloc): New function.
6842 (ipa_pure_const::execute): Call propagate_malloc and
6843 ipa_free_fn_summary.
6844 (pass_local_pure_const::execute): Add support for malloc attribute.
6845 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6846 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6848 2017-10-27 Martin Liska <mliska@suse.cz>
6850 PR gcov-profile/82457
6851 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6852 for fork-like functions to be properly instrumented.
6854 2017-10-27 Richard Biener <rguenther@suse.de>
6857 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6858 info when we redirected EH.
6860 2017-10-26 Michael Collison <michael.collison@arm.com>
6862 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6864 (<optab>_trunchf<GPI:mode>2: New pattern.
6865 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6866 * config/aarch64/iterators.md (wv): New mode attribute.
6867 (vf, VF): New mode attributes.
6868 (vgp, VGP): New mode attributes.
6869 (s): Update attribute with SImode and DImode prefixes.
6871 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6873 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6874 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6875 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6876 (nios2_option_overide): Initialize it. Don't allow R0-relative
6877 addressing with PIC.
6878 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6879 (nios2_symbolic_constant_p): Likewise.
6880 (nios2_legitimate_address_p): Likewise.
6881 (nios2_r0rel_section_name_p): New.
6882 (nios2_symbol_ref_in_r0rel_data_p): New.
6883 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6884 (r0rel_constant_p): New.
6885 (nios2_print_operand_address): Handle r0rel_constant_p.
6886 (nios2_cdx_narrow_form_p): Likewise.
6887 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6888 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6889 (Nios II Options): Document -mr0rel-sec.
6891 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6893 * config/nios2/nios2.c: Include xregex.h.
6894 (nios2_gprel_sec_regex): New.
6895 (nios2_option_overide): Initialize it. Don't allow GP-relative
6896 addressing with PIC.
6897 (nios2_small_section_name_p): Check for regex match.
6898 * config/nios2/nios2.opt (mgprel-sec=): New option.
6899 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6900 (Nios II Options): Document -mgprel-sec.
6902 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6904 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6906 2017-10-26 Tom de Vries <tom@codesourcery.com>
6908 PR tree-optimization/82707
6909 * gimple.c (gimple_copy): Fix unsharing of
6910 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
6912 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
6914 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
6915 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
6916 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
6917 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
6918 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
6919 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
6920 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
6921 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
6922 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
6923 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
6924 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
6925 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
6926 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
6927 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
6928 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
6929 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
6930 _mm512_mask_cmpunord_ps_mask): New intrinsics.
6932 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6934 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
6936 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
6937 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
6938 warning to rs6000.c. Remove the Undocumented flag, since it has
6940 (-mabi=ibmlongdouble): Likewise.
6941 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
6942 already set, set the default format for long double.
6943 (rs6000_debug_reg_global): Print whether long double is IBM or
6945 (rs6000_option_override_internal): Rework setting long double
6946 format. Only warn if the user is changing the long double default
6947 and they did not use -Wno-psabi.
6948 * doc/invoke.texi (PowerPC options): Update the documentation for
6949 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
6951 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6952 Alan Hayward <alan.hayward@arm.com>
6953 David Sherwood <david.sherwood@arm.com>
6955 * rtl.h (wider_subreg_mode): New function.
6956 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
6957 rather than an unsigned int *.
6958 * ira-color.c (regno_max_ref_width): Replace with...
6959 (regno_max_ref_mode): ...this new variable.
6960 (coalesced_pseudo_reg_slot_compare): Update accordingly.
6961 Use wider_subreg_mode.
6962 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
6963 rather than an unsigned int *.
6964 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
6965 (process_alt_operands): Likewise.
6966 (invariant_p): Likewise.
6967 * lra-spills.c (assign_mem_slot): Likewise.
6968 (add_pseudo_to_slot): Likewise.
6969 * lra.c (collect_non_operand_hard_regs): Likewise.
6970 (add_regs_to_insn_regno_info): Likewise.
6971 * reload1.c (regno_max_ref_width): Replace with...
6972 (regno_max_ref_mode): ...this new variable.
6973 (reload): Update accordingly. Update call to
6974 ira_sort_regnos_for_alter_reg.
6975 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
6976 (init_eliminable_invariants): Update to use regno_max_ref_mode.
6977 (scan_paradoxical_subregs): Likewise.
6979 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6981 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
6982 (aarch64_frame): Add emit_frame_chain boolean.
6983 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6984 Move eh_return case to aarch64_layout_frame.
6985 (aarch64_layout_frame): Initialize emit_frame_chain.
6986 (aarch64_expand_prologue): Use emit_frame_chain.
6988 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6990 * config/aarch64/aarch64.c (aarch64_layout_frame):
6991 Ensure LR is always stored at the bottom of the callee-saves.
6992 Remove rarely used frame layout which saves callee-saves at top of
6993 frame, so the store of LR can be used as a valid probe in all cases.
6995 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6997 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
6998 Improve unaligned TImode/TFmode base/offset split.
7000 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7001 Alan Hayward <alan.hayward@arm.com>
7002 David Sherwood <david.sherwood@arm.com>
7004 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7005 * combine.c (find_single_use_1): Likewise.
7006 (expand_field_assignment): Likewise.
7007 (move_deaths): Likewise.
7008 * lra-constraints.c (simplify_operand_subreg): Likewise.
7009 (curr_insn_transform): Likewise.
7010 * lra.c (collect_non_operand_hard_regs): Likewise.
7011 (add_regs_to_insn_regno_info): Likewise.
7012 * rtlanal.c (reg_referenced_p): Likewise.
7013 (covers_regno_no_parallel_p): Likewise.
7015 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7017 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7018 Don't print any bits outside the precision of the value.
7019 * wide-int.cc (test_printing): Add some new tests.
7021 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7023 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7024 supports -xbrace_comment option.
7025 * configure: Regenerate.
7026 * config.in: Regenerate.
7027 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7028 (ASM_CPU_SPEC): Use it.
7030 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7032 * target.def (static_rtx_alignment): New hook.
7033 * targhooks.h (default_static_rtx_alignment): Declare.
7034 * targhooks.c (default_static_rtx_alignment): New function.
7035 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7036 * doc/tm.texi: Regenerate.
7037 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7038 instead of targetm.constant_alignment. Remove call to
7040 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7041 (cris_preferred_mininum_alignment): New function, split out from...
7042 (cris_constant_alignment): ...here.
7043 (cris_static_rtx_alignment): New function.
7044 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7046 (ix86_constant_alignment): ...here.
7047 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7048 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7049 (mmix_static_rtx_alignment): New function.
7050 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7051 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7053 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7056 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7057 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7059 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7061 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7062 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7063 mark as a sign-extending load.
7064 (local_pic_load_u): Define.
7066 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7069 * fold-const.c (operand_equal_for_comparison_p): Also return true
7070 if ARG0 is a simple variant of ARG1 with narrower precision.
7071 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7073 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7075 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7077 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7078 scatter_static, scatter_per_elt.
7079 * x86-tune-costs.h: Add new cost entries.
7081 2017-10-25 Richard Biener <rguenther@suse.de>
7083 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7084 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7085 class pass_fre): Move to ...
7086 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7087 class pass_fre): ... here and adjust for statistics.
7089 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7092 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7093 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7094 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7095 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7098 2017-10-25 Richard Biener <rguenther@suse.de>
7100 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7101 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7102 eliminate_push_avail, eliminate_insert): Move inside...
7103 (class eliminate_dom_walker): ... this class in preparation
7105 (fini_eliminate): Remove by merging with ...
7106 (eliminate): ... this function. Adjust for class changes.
7107 (pass_pre::execute): Remove fini_eliminate call.
7108 (pass_fre::execute): Likewise.
7110 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7113 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7114 (VPERMI2, VPERMI2I): New mode iterators.
7115 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7116 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7118 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7119 mode iterator. Remove 3 old define_insn patterns.
7120 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7121 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7122 VPERMI2 mode iterator, remove the other two expanders.
7123 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7124 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7125 instructions, remove the other two patterns.
7126 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7127 mode iterator, remove the other two patterns.
7128 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7129 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7130 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7131 and adjust argument order accordingly.
7132 (ix86_expand_vec_perm): Adjust caller.
7133 (expand_vec_perm_1): Likewise.
7134 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7135 (expand_vec_perm_vpermt2_vpshub2): ... this.
7136 (ix86_expand_vec_perm_const_1): Adjust caller.
7137 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7140 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7141 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7142 (vec_shl_<mode>): Remove unused expander.
7143 (avx512bw_<shift_insn><mode>3): New define_insn.
7144 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7145 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7147 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7150 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7153 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7155 PR rtl-optimization/82396
7156 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7157 (autopref_multipass_init): Simplify initialization.
7158 (autopref_rank_data): Simplify sort order.
7159 * gcc/sched-int.h (autopref_multipass_data_): Remove
7160 multi_mem_insn_p, min_offset and max_offset.
7162 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7165 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7166 Check special value of flag_omit_frame_pointer.
7167 (aarch64_can_eliminate): Likewise.
7168 (aarch64_override_options_after_change_1): Simplify handling of
7169 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7171 2017-10-24 Richard Biener <rguenther@suse.de>
7173 PR tree-optimization/82697
7174 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7175 zero for conditional load and unconditional store.
7177 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7179 * doc/install.texi: Document bootstrap-cet.
7181 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7184 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7185 ENDBR instruction at function entrance if function is only
7188 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7191 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7192 patterns to better describe from which operation the CF is computed.
7193 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7194 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7195 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7196 is 0, use _0 suffixed expanders instead of emitting a comparison
7199 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7201 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7202 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7204 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7207 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7208 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7209 * loop-iv.c (iv_get_reaching_def): Likewise.
7210 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7211 variable is promoted and the partition contains undefined values.
7213 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7215 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7217 (nios2_address_cost): Define.
7218 (nios2_legitimize_address): Recognize (exp + constant) directly.
7219 (TARGET_ADDRESS_COST): Define.
7221 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7223 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7224 (nios2_symbolic_memory_operand_p): Declare.
7225 (nios2_split_large_constant): Declare.
7226 (nios2_split_symbolic_memory_operand): Declare.
7227 * config/nios2/nios2.c: Adjust includes.
7228 (nios2_symbolic_constant_allowed): New.
7229 (nios2_symbolic_constant_p): New.
7230 (nios2_plus_symbolic_constant_p): New.
7231 (nios2_valid_addr_expr_p): Recognize addresses involving
7233 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7234 (nios2_symbolic_memory_operand_p): New.
7235 (nios2_large_constant_p): New.
7236 (nios2_split_large_constant): New.
7237 (nios2_split_plus_large_constant): New.
7238 (nios2_split_symbolic_memory_operand): New.
7239 (nios2_legitimize_address): Code refactoring. Handle addresses
7240 involving symbolic constants.
7241 (nios2_emit_move_sequence): Likewise.
7242 (nios2_print_operand): Improve error output.
7243 (nios2_print_operand_address): Handle LO_SUM.
7244 (nios2_cdx_narrow_form_p): Likewise.
7245 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7246 operands involving symbolic constants.
7247 (movhi_internal, movsi_internal): Likewise.
7248 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7249 (extendhisi2, extendqi<mode>2): Likewise.
7251 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7253 * tree-pass.h (PROP_rtl_split_insns): Define.
7254 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7256 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7258 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7260 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7263 * target.def (const_not_ok_for_debug_p): Default to
7264 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7265 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7266 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7267 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7268 which targetm.const_not_ok_for_debug_p returned true.
7269 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7271 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7273 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7274 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7275 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7277 (ix86_const_not_ok_for_debug_p): New function.
7278 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7279 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7281 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7284 * system.h: Conditionally include "unique-ptr.h" if
7285 INCLUDE_UNIQUE_PTR is defined.
7286 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7287 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7289 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7291 * config/rl78/rl78.md: New define_expand "subdi3".
7293 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7296 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7297 DF_REF_INSN if DF_REF_INSN_INFO is false.
7299 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7301 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7302 xmm_move instead of sse_move.
7303 (sse_store_index): New function.
7304 (ix86_register_move_cost): Be more sensible about mismatch stall;
7305 model AVX moves correctly; make difference between sse->integer and
7307 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7308 moves; make difference between SSE and AVX.
7309 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7310 and zmm_move. Increase size of sse load and store tables;
7311 add unaligned load and store tables; add ssemmx_to_integer.
7312 * x86-tune-costs.h: Update all entries according to real
7313 move latencies from Agner Fog's manual and chip documentation.
7315 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7318 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7319 * config/i386/constraints.md (Wf): New constraint.
7320 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7321 (cmp<dwi>_doubleword): Removed.
7322 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7323 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7324 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7325 expand with cmp<dwi>_doubleword. For LTU and GEU use
7326 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7328 * common.opt (gcolumn-info): Enable by default.
7329 * doc/invoke.texi (gcolumn-info): Document new default.
7331 2017-10-23 Richard Biener <rguenther@suse.de>
7333 PR tree-optimization/82672
7334 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7335 Fold the stmt if we propagated into it.
7337 2017-10-23 Richard Biener <rguenther@suse.de>
7339 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7340 (bitmap_remove_expr_from_set): ... this. All callers call this
7341 for non-constant values.
7342 (bitmap_set_subtract): Rename to...
7343 (bitmap_set_subtract_expressions): ... this. Adjust and
7345 (bitmap_set_contains_value): Remove superfluous check.
7346 (bitmap_set_replace_value): Inline into single caller ...
7347 (bitmap_value_replace_in_set): ... here and simplify.
7348 (dependent_clean): Merge into ...
7349 (clean): ... this using an overload. Adjust.
7350 (prune_clobbered_mems): Adjust.
7351 (compute_antic_aux): Likewise.
7352 (compute_partial_antic_aux): Likewise.
7354 2017-10-23 Richard Biener <rguenther@suse.de>
7356 PR tree-optimization/82129
7358 2017-08-01 Richard Biener <rguenther@suse.de>
7360 PR tree-optimization/81181
7361 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7362 (compute_antic): ... end of iteration here.
7364 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7366 * target.def (starting_frame_offset): New hook.
7367 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7368 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7369 * doc/tm.texi.in: Regenerate.
7370 * hooks.h (hook_hwi_void_0): Declare.
7371 * hooks.c (hook_hwi_void_0): New function.
7372 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7373 STARTING_FRAME_OFFSET.
7374 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7375 * reload1.c (reload): Likewise.
7376 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7377 instead of STARTING_FRAME_OFFSET.
7378 * function.c (try_fit_stack_local): Likewise.
7379 (assign_stack_local_1): Likewise
7380 (instantiate_virtual_regs): Likewise.
7381 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7382 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7383 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7384 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7385 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7386 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7387 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7388 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7389 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7390 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7391 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7392 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7393 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7394 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7395 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7396 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7397 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7398 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7399 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7400 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7401 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7402 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7403 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7404 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7405 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7406 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7407 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7408 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7409 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7410 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7411 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7412 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7413 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7414 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7415 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7416 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7417 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7418 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7419 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7420 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7421 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7422 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7423 return a HOST_WIDE_INT.
7424 (avr_builtin_setjmp_frame_value): Use it instead of
7425 STARTING_FRAME_OFFSET.
7426 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7427 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7428 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7430 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7431 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7432 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7433 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7434 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7435 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7436 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7437 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7438 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7439 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7440 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7441 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7443 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7444 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7445 * config/mips/mips.c (mips_compute_frame_info): Refer to
7446 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7447 (mips_starting_frame_offset): New function.
7448 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7449 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7450 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7451 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7452 and return a HOST_WIDE_INT.
7453 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7454 (mmix_initial_elimination_offset): Refer to
7455 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7456 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7457 * config/pa/pa.c (pa_starting_frame_offset): New function.
7458 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7459 (pa_expand_prologue): Likewise.
7460 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7461 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7462 !FRAME_GROWS_DOWNWARD handling to...
7463 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7464 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7465 !FRAME_GROWS_DOWNWARD handling to...
7466 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7467 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7468 !FRAME_GROWS_DOWNWARD handling to...
7469 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7470 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7472 (rs6000_starting_frame_offset): New function.
7473 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7474 !FRAME_GROWS_DOWNWARD handling to...
7475 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7476 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7477 !FRAME_GROWS_DOWNWARD handling to...
7478 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7479 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7480 !FRAME_GROWS_DOWNWARD handling to...
7481 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7482 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7483 (rs6000_starting_frame_offset): New function.
7484 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7485 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7486 * config/vax/vax.c (vax_starting_frame_offset): New function.
7487 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7488 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7489 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7490 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7491 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7492 * system.h (STARTING_FRAME_OFFSET): Poison.
7494 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7496 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7497 SCALAR_TYPE_MODE instead of TYPE_MODE.
7499 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7500 Alan Hayward <alan.hayward@arm.com>
7501 David Sherwood <david.sherwood@arm.com>
7503 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7505 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7506 Alan Hayward <alan.hayward@arm.com>
7507 David Sherwood <david.sherwood@arm.com>
7509 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7511 2017-10-23 Richard Biener <rguenther@suse.de>
7513 PR tree-optimization/82129
7514 * tree-ssa-pre.c (bitmap_set_and): Remove.
7515 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7516 canonicalizing expressions in the set to those with lowest
7517 ID rather than taking that from the first edge.
7519 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7521 * combine.c (rtx_equal_for_field_assignment_p): Use
7522 byte_lowpart_offset.
7524 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7525 Alan Hayward <alan.hayward@arm.com>
7526 David Sherwood <david.sherwood@arm.com>
7528 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7529 to a SUBREG_PROMOTED_VAR.
7531 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7532 Alan Hayward <alan.hayward@arm.com>
7533 David Sherwood <david.sherwood@arm.com>
7535 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7536 (expand_debug_source_expr): Likewise.
7537 * combine.c (combine_simplify_rtx): Likewise.
7538 * cse.c (fold_rtx): Likewise.
7539 * optabs.c (expand_float): Likewise.
7540 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7541 (simplify_binary_operation_1): Likewise.
7543 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7544 Alan Hayward <alan.hayward@arm.com>
7545 David Sherwood <david.sherwood@arm.com>
7547 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7548 (record_promoted_value): Likewise.
7549 * expr.c (expand_expr_real_2): Likewise.
7550 * ree.c (update_reg_equal_equiv_notes): Likewise.
7551 (combine_set_extension): Likewise.
7552 * rtlanal.c (low_bitmask_len): Likewise.
7553 * simplify-rtx.c (neg_const_int): Likewise.
7554 (simplify_binary_operation_1): Likewise.
7556 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7557 Alan Hayward <alan.hayward@arm.com>
7558 David Sherwood <david.sherwood@arm.com>
7560 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7561 * regcprop.c (maybe_mode_change): Likewise.
7562 * reload1.c (alter_reg): Likewise.
7564 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7566 * inchash.h (inchash::hash::add_wide_int): New function.
7567 * lto-streamer-out.c (hash_tree): Use it.
7569 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7571 * inchash.h (inchash::hash::add_wide_int): Rename to...
7572 (inchash::hash::add_hwi): ...this.
7573 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7574 (polymorphic_call_target_hasher::hash): Likewise.
7575 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7576 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7577 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7578 * lto-streamer-out.c (hash_tree): Likewise.
7579 * optc-save-gen.awk: Likewise.
7580 * tree.c (add_expr): Likewise.
7582 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7585 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7586 for ordered inequality comparisons even with TARGET_IEEE_FP.
7588 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7591 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7592 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7593 Expand with cmp<dwi>_doubleword.
7595 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7597 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7598 List CET intrinsics.
7599 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7600 specific to -fcf-protection option.
7602 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7604 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7605 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7606 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7607 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7608 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7609 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7610 (extra_objs): Add cet.o for Linux/x86 targets.
7611 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7612 * config/i386/cet.c: New file.
7613 * config/i386/cetintrin.h: Likewise.
7614 * config/i386/t-cet: Likewise.
7615 * config/i386/cpuid.h (bit_SHSTK): New.
7616 (bit_IBT): Likewise.
7617 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7618 pass IBT and SHSTK bits.
7619 * config/i386/i386-builtin-types.def
7620 (VOID_FTYPE_UNSIGNED_PVOID): New.
7621 (VOID_FTYPE_UINT64_PVOID): Likewise.
7622 * config/i386/i386-builtin.def: Add CET intrinsics.
7623 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7624 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7625 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7626 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7628 * config/i386/i386.c (rest_of_insert_endbranch): New.
7629 (pass_data_insert_endbranch): Likewise.
7630 (pass_insert_endbranch): Likewise.
7631 (make_pass_insert_endbranch): Likewise.
7632 (ix86_notrack_prefixed_insn_p): Likewise.
7633 (ix86_target_string): Add -mibt, -mshstk flags.
7634 (ix86_option_override_internal): Add flag_cf_protection
7636 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7637 (ix86_print_operand): Add 'notrack' prefix output.
7638 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7639 (ix86_expand_builtin): Expand CET intrinsics.
7640 (x86_output_mi_thunk): Add 'endbranch' instruction.
7641 * config/i386/i386.h (TARGET_IBT): New.
7642 (TARGET_IBT_P): Likewise.
7643 (TARGET_SHSTK): Likewise.
7644 (TARGET_SHSTK_P): Likewise.
7645 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7646 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7647 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7648 (builtin_setjmp_setup): New pattern.
7649 (builtin_longjmp): Likewise.
7650 (rdssp<mode>): Likewise.
7651 (incssp<mode>): Likewise.
7652 (saveprevssp): Likewise.
7653 (rstorssp): Likewise.
7654 (wrss<mode>): Likewise.
7655 (wruss<mode>): Likewise.
7656 (setssbsy): Likewise.
7657 (clrssbsy): Likewise.
7658 (nop_endbr): Likewise.
7659 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7661 * config/i386/immintrin.h: Include <cetintrin.h>.
7662 * config/i386/linux-common.h
7663 (file_end_indicate_exec_stack_and_cet): New prototype.
7664 (TARGET_ASM_FILE_END): New.
7666 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7668 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7669 latencies instead of having separate table; make difference between
7670 integer and float costs.
7671 * i386.h (processor_costs): Remove scalar_stmt_cost,
7672 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7673 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7675 * x86-tune-costs.h: Remove entries which has been removed in
7676 procesor_costs from all tables; make cond_taken_branch_cost
7677 and cond_not_taken_branch_cost COST_N_INSNS based.
7679 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7681 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7683 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7685 * config/i386/i386.md (isa): Remove fma_avx512f.
7686 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7687 <avx512>_fmadd_<mode>_mask3<round_name>,
7688 <avx512>_fmsub_<mode>_mask<round_name>,
7689 <avx512>_fmsub_<mode>_mask3<round_name>,
7690 <avx512>_fnmadd_<mode>_mask<round_name>,
7691 <avx512>_fnmadd_<mode>_mask3<round_name>,
7692 <avx512>_fnmsub_<mode>_mask<round_name>,
7693 <avx512>_fnmsub_<mode>_mask3<round_name>,
7694 <avx512>_fmaddsub_<mode>_mask<round_name>,
7695 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7696 <avx512>_fmsubadd_<mode>_mask<round_name>,
7697 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7698 (*vec_widen_umult_even_v16si<mask_name>,
7699 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7700 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7702 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7704 * extend.texi: Add 'nocf_check' documentation.
7705 * gimple.texi: Add second parameter to
7706 gimple_build_call_from_tree.
7707 * invoke.texi: Add -fcf-protection documentation.
7708 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7710 2017-10-20 Richard Biener <rguenther@suse.de>
7712 PR tree-optimization/82473
7713 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7714 the largest input type.
7716 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7718 * c-attribs.c (handle_nocf_check_attribute): New function.
7719 (c_common_attribute_table): Add 'nocf_check' handling.
7720 * gimple-parser.c: Add second argument NULL to
7721 gimple_build_call_from_tree.
7722 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7723 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7725 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7726 * common.opt: Add fcf-protection flag.
7727 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7728 * flag-types.h: Add enum cf_protection_level.
7729 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7730 Add 'nocf_check' attribute propagation to gimple call.
7731 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7732 (gimple_build_call_from_tree): Update prototype.
7733 (gimple_call_nocf_check_p): New function.
7734 (gimple_call_set_nocf_check): Likewise.
7735 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7736 * ipa-icf.c: Add nocf_check attribute in statement hash.
7737 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7738 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7739 * toplev.c (process_options): Add flag_cf_protection handling.
7741 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7743 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7745 2017-10-20 Richard Biener <rguenther@suse.de>
7747 PR tree-optimization/82603
7748 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7749 remove false predicated stores.
7751 2017-10-20 Richard Biener <rguenther@suse.de>
7753 * graphite-isl-ast-to-gimple.c
7754 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7755 Remove return value and simplify, dump copied stmt after lhs
7757 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7758 Reduce dump verbosity.
7759 (gsi_insert_earliest): Likewise.
7760 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7761 * graphite.c (print_global_statistics): Adjust dumping.
7762 (print_graphite_scop_statistics): Likewise.
7763 (print_graphite_statistics): Do not dump loops here.
7764 (graphite_transform_loops): But here.
7766 2017-10-20 Nicolas Roche <roche@adacore.com>
7768 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7769 * configure: Regenerate.
7771 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7774 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7775 functions when optimizing replace GIMPLE_RETURN stmts with
7776 calls to __builtin_unreachable ().
7779 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7780 for -fsanitize=thread link of executables.
7781 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7782 link of executables.
7785 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7787 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7788 define_insns for logical vector shifts to use VI248_AVX512BW
7789 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7790 condition, useless isa and prefix attributes. Change the first
7791 2 of these define_insns to ...
7792 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7793 define_insn for avx512vl.
7794 (<shift_insn><mode>3): ... and this, new define_insn without
7795 masking for non-avx512vl.
7798 * config/i386/sse.md (*andnot<mode>3,
7799 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7800 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7801 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7802 not applied use empty suffix even for TARGET_AVX512VL.
7803 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7804 is applied, supply evex,evex or evex,evex,evex instead of just
7807 2017-10-20 Julia Koval <julia.koval@intel.com>
7809 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7810 (OPTION_MASK_ISA_GFNI_UNSET): New.
7811 (ix86_handle_option): Handle OPT_mgfni.
7812 * config/i386/cpuid.h (bit_GFNI): New.
7813 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7814 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7815 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7816 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7817 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7818 * config/i386/i386.opt: Add mgfni.
7820 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7822 * config/msp430/msp430.c (msp430_option_override): Disable
7823 -fdelete-null-pointer-checks.
7824 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7826 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7828 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7829 of x87 and SSE instructions.
7831 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7833 * asan.c (create_cond_insert_point): Do not update edge count.
7834 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7835 (afdo_propagate_circuit): Likewise.
7836 (afdo_calculate_branch_prob): Likewise.
7837 (afdo_annotate_cfg): Likewise.
7838 * basic-block.h (struct edge_def): Remove count.
7839 (edge_def::count): New accessor.
7840 * bb-reorder.c (rotate_loop): Update.
7841 (find_traces_1_round): Update.
7842 (connect_traces): Update.
7843 (sanitize_hot_paths): Update.
7844 * cfg.c (unchecked_make_edge): Update.
7845 (make_single_succ_edge): Update.
7846 (check_bb_profile): Update.
7847 (dump_edge_info): Update.
7848 (update_bb_profile_for_threading): Update.
7849 (scale_bbs_frequencies_int): Update.
7850 (scale_bbs_frequencies_gcov_type): Update.
7851 (scale_bbs_frequencies_profile_count): Update.
7852 (scale_bbs_frequencies): Update.
7853 * cfganal.c (connect_infinite_loops_to_exit): Update.
7854 * cfgbuild.c (compute_outgoing_frequencies): Update.
7855 (find_many_sub_basic_blocks): Update.
7856 * cfgcleanup.c (try_forward_edges): Update.
7857 (try_crossjump_to_edge): Update
7858 * cfgexpand.c (expand_gimple_cond): Update
7859 (expand_gimple_tailcall): Update
7860 (construct_exit_block): Update
7861 * cfghooks.c (verify_flow_info): Update
7862 (redirect_edge_succ_nodup): Update
7863 (split_edge): Update
7864 (make_forwarder_block): Update
7865 (duplicate_block): Update
7866 (account_profile_record): Update
7867 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7868 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7869 * cfgloopmanip.c (scale_loop_profile): Update.
7871 (lv_adjust_loop_entry_edge): Update.
7872 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7873 (force_nonfallthru_and_redirect): Update.
7874 (purge_dead_edges): Update.
7875 (rtl_flow_call_edges_add): Update.
7876 * cgraphunit.c (init_lowered_empty_function): Update.
7877 (cgraph_node::expand_thunk): Update.
7878 * gimple-pretty-print.c (dump_probability): Update.
7879 (dump_edge_probability): Update.
7880 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7881 * haifa-sched.c (sched_create_recovery_edges): Update.
7882 * hsa-gen.c (convert_switch_statements): Update.
7883 * ifcvt.c (dead_or_predicable): Update.
7884 * ipa-inline-transform.c (inline_transform): Update.
7885 * ipa-split.c (split_function): Update.
7886 * ipa-utils.c (ipa_merge_profiles): Update.
7887 * loop-doloop.c (add_test): Update.
7888 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7889 * lto-streamer-in.c (input_cfg): Update.
7890 (input_function): Update.
7891 * lto-streamer-out.c (output_cfg): Update.
7892 * modulo-sched.c (sms_schedule): Update.
7893 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7894 * predict.c (maybe_hot_edge_p): Update.
7895 (unlikely_executed_edge_p): Update.
7896 (probably_never_executed_edge_p): Update.
7897 (dump_prediction): Update.
7898 (drop_profile): Update.
7899 (propagate_unlikely_bbs_forward): Update.
7900 (determine_unlikely_bbs): Update.
7901 (force_edge_cold): Update.
7902 * profile.c (compute_branch_probabilities): Update.
7903 * reg-stack.c (better_edge): Update.
7904 * shrink-wrap.c (handle_simple_exit): Update.
7905 * tracer.c (better_p): Update.
7906 * trans-mem.c (expand_transaction): Update.
7907 (split_bb_make_tm_edge): Update.
7908 * tree-call-cdce.c: Update.
7909 * tree-cfg.c (gimple_find_sub_bbs): Update.
7910 (gimple_split_edge): Update.
7911 (gimple_duplicate_sese_region): Update.
7912 (gimple_duplicate_sese_tail): Update.
7913 (gimple_flow_call_edges_add): Update.
7914 (insert_cond_bb): Update.
7915 (execute_fixup_cfg): Update.
7916 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
7917 * tree-complex.c (expand_complex_div_wide): Update.
7918 * tree-eh.c (lower_resx): Update.
7919 (unsplit_eh): Update.
7920 (cleanup_empty_eh_move_lp): Update.
7921 * tree-inline.c (copy_edges_for_bb): Update.
7922 (freqs_to_counts): Update.
7923 (copy_cfg_body): Update.
7924 * tree-ssa-dce.c (remove_dead_stmt): Update.
7925 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7926 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7927 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7928 (unloop_loops): Update.
7929 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7930 * tree-ssa-loop-split.c (connect_loops): Update.
7931 (split_loop): Update.
7932 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7933 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7934 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7935 * tree-ssa-reassoc.c (branch_fixup): Update.
7936 * tree-ssa-tail-merge.c (replace_block_by): Update.
7937 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7938 (compute_path_counts): Update.
7939 (update_profile): Update.
7940 (recompute_probabilities): Update.
7941 (update_joiner_offpath_counts): Update.
7942 (estimated_freqs_path): Update.
7943 (freqs_to_counts_path): Update.
7944 (clear_counts_path): Update.
7945 (ssa_fix_duplicate_block_edges): Update.
7946 (duplicate_thread_path): Update.
7947 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7948 (case_bit_test_cmp): Update.
7949 (collect_switch_conv_info): Update.
7950 (gen_inbound_check): Update.
7951 (do_jump_if_equal): Update.
7952 (emit_cmp_and_jump_insns): Update.
7953 * tree-tailcall.c (decrease_profile): Update.
7954 (eliminate_tail_call): Update.
7955 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7956 (vect_do_peeling): Update.
7957 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7958 * ubsan.c (ubsan_expand_null_ifn): Update.
7959 (ubsan_expand_ptr_ifn): Update.
7960 * value-prof.c (gimple_divmod_fixed_value): Update.
7961 (gimple_mod_pow2): Update.
7962 (gimple_mod_subtract): Update.
7963 (gimple_ic): Update.
7964 (gimple_stringop_fixed_value): Update.
7966 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7969 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
7971 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
7973 PR rtl-optimization/82395
7974 * ira-color.c (allocno_priority_compare_func): Fix comparison step
7975 based on non_spilled_static_chain_regno_p.
7977 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7979 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
7980 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
7981 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
7983 2017-10-19 Martin Sebor <msebor@redhat.com>
7985 PR tree-optimization/82596
7986 * tree.c (array_at_struct_end_p): Handle STRING_CST.
7988 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7990 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
7991 (get_mem_refs_of_builtin_call): Likewise.
7992 * builtins.c (expand_builtin_apply): Adjust call to
7993 allocate_dynamic_stack_space.
7994 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
7995 the third argument to allocate_dynamic_stack_space, otherwise -1.
7996 (expand_builtin): Deal with all alloca variants.
7997 (is_inexpensive_builtin): Likewise.
7998 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
7999 * calls.c (special_function_p): Deal with all alloca variants.
8000 (initialize_argument_information): Adjust call to
8001 allocate_dynamic_stack_space.
8002 (expand_call): Likewise.
8003 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8004 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8005 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8006 use it for the stack usage computation.
8007 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8008 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8009 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8010 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8011 (in_loop_p): Remove first argument and useless check.
8012 (pass_walloca::execute): Remove useless test and adjust call to above.
8013 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8014 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8015 (gimplify_call_expr): Deal with all alloca variants.
8016 * hsa-gen.c (gen_hsa_alloca): Likewise.
8017 (gen_hsa_insns_for_call): Likewise.
8018 * ipa-pure-const.c (special_builtin_state): Likewise.
8019 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8020 * tree-object-size.c (alloc_object_size): Likewise.
8021 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8022 (call_may_clobber_ref_p_1): Likewise.
8023 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8024 (ccp_fold_stmt): Likewise.
8025 (optimize_stack_restore): Likewise.
8026 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8027 (mark_all_reaching_defs_necessary_1): Likewise.
8028 (propagate_necessity): Likewise.
8029 (eliminate_unnecessary_stmts): Likewise.
8030 * tree.c (build_common_builtin_nodes): Build
8031 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8032 (build_alloca_call_expr): New function.
8033 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8034 (CASE_BUILT_IN_ALLOCA): Likewise.
8035 (build_alloca_call_expr): Declare.
8036 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8038 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8041 * dwarf2out.c (new_die_raw): New static inline function.
8042 (new_die): Use it to create the DIE.
8043 (add_AT_external_die_ref): Likewise.
8044 (clone_die): Likewise.
8045 (clone_as_declaration): Likewise.
8046 (dwarf2out_vms_debug_main_pointer): Likewise.
8047 (base_type_die): Likewise. Remove early return for corner cases.
8048 Do not call add_pubtype on the DIE here.
8049 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8050 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8051 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8052 native order exists for base types, attach the DIE manually and call
8053 add_pubtype on it. Do not equate a reverse order DIE to the type.
8055 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8057 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8058 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8060 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8062 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8064 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8065 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8066 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8067 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8068 * opts.c (sanitizer_opts): Add builtin.
8069 * ubsan.c (instrument_builtin): New function.
8070 (pass_ubsan::execute): Call it.
8071 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8072 * doc/invoke.texi: Document -fsanitize=builtin.
8074 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8075 builtins, store max (log2 (align), 0) into uchar field instead of
8076 align into uptr field.
8077 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8078 store uchar 0 field instead of uptr 0 field.
8079 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8080 instead of passing one address of struct with 2 locations pass
8081 two addresses of structs with 1 location each.
8082 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8083 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8084 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8085 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8086 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8087 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8088 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8089 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8091 2017-10-19 Martin Liska <mliska@suse.cz>
8094 * file-find.c (remove_prefix): Remove.
8095 * file-find.h (remove_prefix): Likewise.
8096 * gcc-ar.c: Remove smartness of lookup.
8098 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8100 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8101 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8102 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8104 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8107 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8108 (setcc + and to xor + setcc): New peephole2.
8110 2017-10-19 Tom de Vries <tom@codesourcery.com>
8112 * doc/sourcebuild.texi (Test Directives, Variants of
8113 dg-require-support): Add dg-require-stack-size.
8115 2017-10-19 Martin Liska <mliska@suse.cz>
8118 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8119 that have a large alignment.
8120 (gimplify_target_expr): Likewise.
8122 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8124 PR rtl-optimization/82602
8125 * ira.c (rtx_moveable_p): Return false for volatile asm.
8127 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8130 * config/i386/i386-modes.def (CCGZ): New CC mode.
8131 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8132 * config/i386/predicates.md (ix86_comparison_operator):
8134 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8135 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8136 with double-word subtraction.
8137 (put_condition_code): Handle CCGZmode.
8139 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8141 * wide-int.cc (debug (const wide_int &)): New.
8142 (debug (const wide_int *)): New.
8143 (debug (const widest_int &)): New.
8144 (debug (const widest_int *)): New.
8146 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8149 * lra-constraints.c (curr_insn_transform): Use non-input operand
8150 instead of output one for matched reload.
8152 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8154 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8155 (tree-ssa-loop-ivopts.h): New header file.
8156 (struct builtin_info): New fields.
8157 (classify_builtin_1): Compute and record base and offset parts for
8158 memset builtin partition by calling strip_offset.
8159 (offset_cmp, fuse_memset_builtins): New functions.
8160 (finalize_partitions): Fuse adjacent memset partitions by calling
8162 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8163 Expose the interface.
8164 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8166 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8168 PR tree-optimization/82574
8169 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8170 that data reference must be executed exactly once per iteration
8171 against the outermost loop in nest.
8172 (classify_partition): Update call to above function.
8174 2017-10-18 Richard Biener <rguenther@suse.de>
8176 PR tree-optimization/82591
8177 * graphite.c (graphite_transform_loops): Move code gen message
8179 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8180 Here. Handle scop_to_isl_ast failing.
8181 (scop_to_isl_ast): Limit the number of ISL operations.
8183 2017-10-18 Richard Biener <rguenther@suse.de>
8185 * graphite-isl-ast-to-gimple.c
8186 (translate_isl_ast_to_gimple::set_rename): Simplify.
8187 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8188 (graphite_copy_stmts_from_block): ... here.
8189 (copy_bb_and_scalar_dependences): Simplify.
8190 (add_parameters_to_ivs_params): Canonicalize.
8191 (generate_entry_out_of_ssa_copies): Simplify.
8192 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8193 by passing in ISL dimension.
8194 (parameter_index_in_region_1): Rename to ...
8195 (parameter_index_in_region): ... this.
8196 (extract_affine): Adjust assert, pass down parameter index.
8197 (add_param_constraints): Use range-info when available.
8198 (build_scop_context): Adjust.
8199 * sese.c (new_sese_info): Adjust.
8200 (free_sese_info): Likewise.
8201 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8202 Remove unused typedefs.
8203 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8205 2017-10-18 Martin Liska <mliska@suse.cz>
8207 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8209 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8211 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8212 (s390_sched_init): Do not reset s390_sched_state if we entered the
8213 current basic block via a fallthru edge and all others are unlikely.
8215 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8217 * config/s390/s390.c (NUM_SIDES): New variable.
8218 (LONGRUNNING_THRESHOLD): New variable.
8219 (LATENCY_FACTOR): New variable.
8220 (s390_sched_score): Decrease score for long-running instructions on
8222 (s390_sched_variable_issue): Perform bookkeeping for long-running
8225 2017-10-18 Richard Biener <rguenther@suse.de>
8227 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8228 Simplify with removal of the parameter rename map.
8229 (set_rename): Likewise.
8230 (should_copy_to_new_region): Likewise.
8231 (graphite_copy_stmts_from_block): Likewise.
8232 (copy_bb_and_scalar_dependences): Remove initialization of
8233 unused copied_bb_map.
8235 (copy_internal_parameters): Likewise.
8236 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8237 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8238 Use INTEGRAL_TYPE_P.
8239 (parameter_index_in_region_1): Rename to ...
8240 (assign_parameter_index_in_region): ... this. Assert we have
8241 a parameter we handle.
8242 (scan_tree_for_params): Adjust.
8243 * sese.h (parameter_rename_map_t): Remove.
8244 (struct sese_info_t): Remove unused parameter_rename_map and
8245 copied_bb_map members.
8246 * sese.c (new_sese_info): Adjust.
8247 (free_sese_info): Likewise.
8249 2017-10-18 Martin Liska <mliska@suse.cz>
8252 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8253 on an abnormal edge.
8255 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8257 * doc/invoke.texi (ffunction-sections and fdata-sections):
8260 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8262 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8263 the use statement can throw internally.
8265 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8267 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8268 any RTX present on the RHS of a SET.
8269 * compare-elim.c (try_eliminate_compare): Restore comment.
8271 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8273 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8274 may be also called on tcc_type nodes.
8275 * langhooks.c (lhd_tree_size): Likewise.
8277 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8279 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8280 format_warning_at_substring.
8281 (maybe_warn): Convert source_range * param to a location_t. Pass
8282 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8283 (format_directive): Remove code to extract source_ranges and
8284 source_range * in favor of just a location_t.
8285 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8287 * substring-locations.c (format_warning_va): Convert
8288 source_range * param to a location_t.
8289 (format_warning_at_substring): Likewise.
8290 * substring-locations.h (format_warning_va): Likewise.
8291 (format_warning_at_substring): Likewise.
8293 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8295 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8297 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8298 and scatter/gather ops.
8300 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8301 vec_gather_load and vec_scatter_store.
8302 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8303 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8304 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8305 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8306 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8307 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8309 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8311 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8312 Detect FTST instruction and handle its register pops. Only pop
8313 second operand if can_pop_second_op is true.
8314 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8315 set can_pop_second_op to false in the compare_for_stack_reg call.
8317 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8318 output_fp_compare for stack register operands.
8319 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8320 instructions here. Do not emit stack register pops here. Assert
8321 that FCOMPP pops next to top stack register. Rewrite function.
8323 2017-10-17 Nathan Sidwell <nathan@acm.org>
8326 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8327 use DECL_ASSEMBLER_NAME_RAW.
8330 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8333 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8334 Wilco Dijkstra <wilco.dijkstra@arm.com>
8336 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8337 converstion to Pmode from the buf_addr.
8339 2017-10-17 Richard Biener <rguenther@suse.de>
8341 * graphite-dependences.c (scop_get_reads_and_writes): Change
8342 output parameters to references.
8344 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8346 PR 71026/tree-optimization
8347 * fold-const.c (distribute_real_division): Removed.
8348 (fold_binary_loc): Remove calls to distribute_real_divison.
8350 2017-10-17 Richard Biener <rguenther@suse.de>
8352 * graphite-scop-detection.c
8353 (scop_detection::stmt_has_simple_data_refs_p): Always use
8354 the full nest as region.
8355 (try_generate_gimple_bb): Likewise.
8356 * sese.c (scalar_evolution_in_region): Simplify now that
8357 SCEV can handle instantiation in regions.
8358 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8359 in the non-loop part of a function if requested.
8361 2017-10-17 Richard Biener <rguenther@suse.de>
8363 PR tree-optimization/82563
8364 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8366 (graphite_regenerate_ast_isl): Call it.
8367 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8369 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8371 PR tree-optimization/82549
8372 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8373 Formatting fixes. Instead of calling make_bit_field_ref with negative
8376 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8378 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8379 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8380 _mm_maskz_reduce_ss): New.
8381 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8382 __builtin_ia32_reducess_mask): Ditto..
8383 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8384 * config/i386/sse.md (reduces<mode>): Renamed to ...
8385 (reduces<mode><mask_scalar_name>): ... this.
8386 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8388 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8389 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8391 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8393 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8394 * selftest-run-tests.c (selftest::run_tests): Call
8395 selftest::unique_ptr_tests_cc_tests.
8396 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8397 * unique-ptr-tests.cc: New file.
8399 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8402 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8404 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8405 (make_hard_regno_born, make_hard_regno_dead): Update
8406 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8408 2017-10-16 Jeff Law <law@redhat.com>
8410 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8412 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8414 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8416 2017-10-16 Olivier Hainque <hainque@adacore.com>
8418 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8419 with_cpu if we were configured for an e500v2 target cpu name.
8421 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8423 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8424 * doc/invoke.texi: Document +nodsp as a valid extension for
8427 2017-10-16 Martin Liska <mliska@suse.cz>
8429 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8430 (test_set_range): Likewise.
8431 (test_range_functions): Rename to ...
8432 (test_bit_in_range): ... this.
8433 (sbitmap_c_tests): Add new test.
8435 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8437 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8439 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8440 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8442 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8444 * config/aarch64/aarch64-builtins.c
8445 (aarch64_types_quadopu_lane_qualifiers): New.
8446 (TYPES_QUADOPU_LANE): New.
8447 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8448 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8449 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8450 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8451 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8452 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8453 (Vdottype, DOTPROD): New.
8454 (sur): Add SDOT and UDOT.
8456 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8458 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8459 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8460 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8462 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8463 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8464 Enable TARGET_DOTPROD.
8465 (cortex-a75.cortex-a55): Likewise.
8466 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8468 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8470 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8471 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8473 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8475 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8476 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8477 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8478 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8479 * config/arm/types.md (neon_dot, neon_dot_q): New.
8480 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8482 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8484 * config/arm/arm.h (TARGET_DOTPROD): New.
8485 * config/arm/arm.c (arm_arch_dotprod): New.
8486 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8487 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8488 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8489 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8490 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8491 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8492 * doc/invoke.texi (armv8.2-a): Document dotprod
8494 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8496 * i386.c (ix86_vec_cost): New function.
8497 (ix86_rtx_costs): Handle vector operations better.
8498 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8499 * x86-tune-costs.h: Add new costs to all tables.
8501 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8503 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8505 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8506 divsd, sqrtss and sqrtsd
8507 * x86-tune-costs.h: Add new entries to all costs.
8508 (znver1_cost): Fix to match real instruction latencies.
8510 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8511 Michael Collison <michael.collison@arm.com>
8513 * compare-elim.c: Include emit-rtl.h.
8514 (can_merge_compare_into_arith): New function.
8515 (try_validate_parallel): Likewise.
8516 (try_merge_compare): Likewise.
8517 (try_eliminate_compare): Call the above when no previous clobber
8519 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8522 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8526 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8527 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8531 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8532 to be any operand_equal_p operands. For & (B - 1) require
8533 B to be power of 2. Recognize
8534 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8536 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8539 * optabs.c (expand_memory_blockage): Fix call of
8540 targetm.have_memory_blockage.
8542 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8545 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8546 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8547 objects to extra_objs instead of overwriting it.
8549 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8551 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8552 Use any_fp_register_operand as operand[3] predicate. Simplify
8553 equality test for operands[2] and operands[4] memory location.
8554 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8555 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8556 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8557 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8558 any_fp_register_operand as operand[1] predicate. Simplify
8559 equality test for operands[0] and operands[3] memory location.
8560 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8561 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8562 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8564 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8566 * target-insns.def: Add memory_blockage.
8567 * optabs.c (expand_memory_blockage): New function.
8568 (expand_asm_memory_barrier): Rename ...
8569 (expand_asm_memory_blockage): ... to this.
8570 (expand_mem_thread_fence): Call expand_memory_blockage
8571 instead of expand_asm_memory_barrier.
8572 (expand_mem_singnal_fence): Ditto.
8573 (expand_atomic_load): Ditto.
8574 (expand_atomic_store): Ditto.
8575 * doc/md.texi (Standard Pattern Names For Generation):
8576 Document memory_blockage instruction pattern.
8578 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8580 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8581 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8582 * config/rl78/rl78.md: New define_expand "adddi3".
8584 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8586 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8589 2017-10-13 Jeff Law <law@redhat.com>
8591 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8593 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8596 * internal-fn.c (expand_mul_overflow): If both operands have
8597 the same highpart of -1 or 0 and the topmost bit of lowpart
8598 is different, overflow is if res <= 0 rather than res < 0.
8600 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8602 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8603 TARGET_P9_VECTOR code for unaligned_load case.
8605 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8607 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8609 2017-10-13 Nathan Sidwell <nathan@acm.org>
8611 * tree-core.h (tree_contains_struct): Make bool.
8612 * tree.c (tree_contains_struct): Likewise.
8613 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8614 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8615 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8616 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8617 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8619 2017-10-13 Richard Biener <rguenther@suse.de>
8621 * graphite-isl-ast-to-gimple.c
8622 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8623 parameters and dominance check.
8624 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8625 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8626 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8627 Do not update SSA form here or do intermediate IL verification.
8628 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8629 (graphite_initialize): Remove check on the number of loops in
8630 the function and inline into graphite_transform_loops.
8631 (graphite_finalize): Inline into graphite_transform_loops.
8632 (graphite_transform_loops): Perform SSA update and IL verification
8634 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8636 2017-10-13 Richard Biener <rguenther@suse.de>
8638 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8639 graphite_expression_type_precision): Avoid global constructor
8641 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8642 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8643 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8644 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8645 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8646 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8648 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8651 * config/i386/i386.h (ix86_red_zone_size): New.
8652 * config/i386/i386.md (push peephole2s): Replace
8653 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8655 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8656 Alan Hayward <alan.hayward@arm.com>
8657 David Sherwood <david.sherwood@arm.com>
8659 * combine.c (can_change_dest_mode): Reject changes in
8660 REGMODE_NATURAL_SIZE.
8662 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8663 Alan Hayward <alan.hayward@arm.com>
8664 David Sherwood <david.sherwood@arm.com>
8666 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8667 (expand_debug_source_expr): Likewise.
8668 * combine.c (combine_simplify_rtx): Likewise.
8669 * cse.c (fold_rtx): Likewise.
8670 * fwprop.c (canonicalize_address): Likewise.
8671 * targhooks.c (default_shift_truncation_mask): Likewise.
8673 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8674 Alan Hayward <alan.hayward@arm.com>
8675 David Sherwood <david.sherwood@arm.com>
8677 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8678 (widened_mode): Likewise.
8679 (expand_unop): Likewise.
8680 * ree.c (transform_ifelse): Likewise.
8681 (merge_def_and_ext): Likewise.
8682 (combine_reaching_defs): Likewise.
8683 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8685 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8686 Alan Hayward <alan.hayward@arm.com>
8687 David Sherwood <david.sherwood@arm.com>
8689 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8690 * combine.c (gen_lowpart_for_combine): Likewise.
8691 * dwarf2out.c (rtl_for_decl_location): Likewise.
8692 * final.c (alter_subreg): Likewise.
8693 * rtlhooks.c (gen_lowpart_general): Likewise.
8694 (gen_lowpart_if_possible): Likewise.
8696 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8697 Alan Hayward <alan.hayward@arm.com>
8698 David Sherwood <david.sherwood@arm.com>
8700 * calls.c (expand_call): Use subreg_lowpart_offset.
8701 * cse.c (cse_insn): Likewise.
8702 * regcprop.c (copy_value): Likewise.
8703 (copyprop_hardreg_forward_1): Likewise.
8705 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8708 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8709 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8710 =Q constraints to +Q and into insn condition add check
8711 that operands[0] and operands[1] are equal.
8712 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8713 =Q constraints to +Q and into insn condition add check
8714 that operands[0] is equal to either operands[1] or operands[2].
8717 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8718 instead of handling MINUS_EXPR twice (once for each argument),
8719 canonicalize operand order and handle just once, use rtype where
8720 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8723 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8724 any values of __C while still being pattern recognizable as a simple
8727 2017-10-13 Richard Biener <rguenther@suse.de>
8729 PR tree-optimization/82451
8731 2017-10-02 Richard Biener <rguenther@suse.de>
8733 PR tree-optimization/82355
8734 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8735 a mapping for the enclosing loop but avoid generating one for
8737 (copy_bb_and_scalar_dependences): Remove premature codegen
8738 error on PHIs in blocks duplicated into multiple places.
8739 * graphite-scop-detection.c
8740 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8741 in the region use it as loop and nest to analyze the DR in.
8742 (try_generate_gimple_bb): Likewise.
8743 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8744 (add_loop_constraints): For blocks in a loop not in the region
8745 create a dimension with a single iteration.
8746 * sese.h (gbb_loop_at_index): Remove assert.
8748 * cfgloop.c (loop_preheader_edge): For the loop tree root
8749 return the single successor of the entry block.
8750 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8751 Reset the SCEV hashtable and niters.
8752 * graphite-scop-detection.c
8753 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8754 assert that we only have POLYNOMIAL_CHREC that vary in loops
8755 contained in the region.
8756 (scop_detection::graphite_can_represent_expr): Adjust.
8757 (scop_detection::stmt_has_simple_data_refs_p): For loops
8758 not in the region set loop to NULL. The nest is now the
8759 entry edge to the region.
8760 (try_generate_gimple_bb): Likewise.
8761 * sese.c (scalar_evolution_in_region): Adjust for
8762 instantiate_scev change.
8763 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8764 Make nest parameter the edge into the region.
8765 (create_data_ref): Likewise.
8766 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8767 entry edge into a region and adjust instantiate_scev calls.
8768 (create_data_ref): Likewise.
8769 (graphite_find_data_references_in_stmt): Likewise.
8770 (find_data_references_in_stmt): Pass the loop preheader edge
8771 from the nest argument.
8772 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8773 parameter the edge into the region.
8774 (instantiate_parameters): Use the loop preheader edge as entry.
8775 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8777 (get_instantiated_value_entry): Make instantiate_below parameter
8778 the edge into the region.
8779 (instantiate_scev_name): Likewise. Adjust dominance checks,
8780 when we cannot use loop-based instantiation instantiate by
8781 walking use-def chains.
8782 (instantiate_scev_poly): Adjust.
8783 (instantiate_scev_binary): Likewise.
8784 (instantiate_scev_convert): Likewise.
8785 (instantiate_scev_not): Likewise.
8786 (instantiate_array_ref): Remove.
8787 (instantiate_scev_3): Likewise.
8788 (instantiate_scev_2): Likewise.
8789 (instantiate_scev_1): Likewise.
8790 (instantiate_scev_r): Do not blindly handle N-operand trees.
8791 Do not instantiate array-refs. Handle all constants and invariants.
8792 (instantiate_scev): Make instantiate_below parameter
8793 the edge into the region.
8794 (resolve_mixers): Use the loop preheader edge for the region
8795 parameter to instantiate_scev_r.
8796 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8798 2017-10-13 Richard Biener <rguenther@suse.de>
8800 PR tree-optimization/82525
8801 * graphite-isl-ast-to-gimple.c
8802 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8804 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8805 Fail code generation when we cannot represent the isl integer.
8806 (binary_op_to_tree): Elide modulo operations that are no-ops
8807 in the type we code generate. Remove now superfluous code
8810 2017-10-13 Richard Biener <rguenther@suse.de>
8812 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8813 (scop_detection::harmful_loop_in_region): Remove premature
8814 IV type restriction.
8815 (scop_detection::graphite_can_represent_scev): We can handle
8816 pointer IVs just fine.
8818 2017-10-13 Alan Modra <amodra@gmail.com>
8820 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8821 "Clobbers and Scratch Registers". Add paragraph on
8822 alternative to clobbers for scratch registers and OpenBLAS
8825 2017-10-13 Alan Modra <amodra@gmail.com>
8827 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8828 example of a memory input for a string of known length. Move
8829 commentary out of table. Add a number of new examples
8830 covering array memory inputs.
8832 2017-10-12 Martin Liska <mliska@suse.cz>
8834 PR tree-optimization/82493
8835 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8836 (test_range_functions): New function.
8837 (sbitmap_c_tests): Likewise.
8838 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8839 * selftest.h (sbitmap_c_tests): New function.
8841 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8843 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8845 * config/rs6000/amo.h: Fix spacing issue.
8847 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8850 * config/i386/i386.md (*ashl<mode>3_mask_1,
8851 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8852 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8855 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8857 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8858 (profile_probability): Set max_probability
8859 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8861 * profile-count.c (profile_probability::differs_from_p): Do not
8862 rely on max_probaiblity == 10000
8864 2017-10-12 Jeff Law <law@redhat.com>
8866 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8869 2017-10-12 Martin Sebor <msebor@redhat.com>
8873 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8874 (handle_alias_pairs): Call it.
8875 * common.opt (-Wattribute-alias): New option.
8876 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8877 * doc/invoke.texi (-Wattribute-alias): Document.
8879 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8882 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8884 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8886 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8887 (make_hard_regno_born, make_hard_regno_dead): Update
8888 bb_killed_pseudos and bb_gen_pseudos.
8890 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8892 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8894 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8896 * config/alpha/alpha.c (alpha_split_conditional_move):
8897 Use std::swap instead of manually swapping.
8898 (alpha_stdarg_optimize_hook): Ditto.
8899 (alpha_canonicalize_comparison): Ditto.
8901 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8903 * tree-loop-distribution.c (struct builtin_info): New struct.
8904 (struct partition): Refactor fields into struct builtin_info.
8905 (partition_free): Free struct builtin_info.
8906 (build_size_arg_loc, build_addr_arg_loc): Delete.
8907 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8908 information from struct builtin_info.
8909 (find_single_drs): New function refactored from classify_partition.
8910 Also moved builtin validity checks to this function.
8911 (compute_access_range, alloc_builtin): New functions.
8912 (classify_builtin_st, classify_builtin_ldst): New functions.
8913 (classify_partition): Refactor code into functions find_single_drs,
8914 classify_builtin_st and classify_builtin_ldst.
8915 (distribute_loop): Don't do runtime alias check when distributing
8917 (find_seed_stmts_for_distribution): New function.
8918 (pass_loop_distribution::execute): Refactor code finding seed
8919 stmts into above function. Support distribution for the innermost
8920 two-level loop nest. Adjust dump information.
8922 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8924 * tree-loop-distribution.c: Adjust the general comment.
8925 (NUM_PARTITION_THRESHOLD): New macro.
8926 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
8927 (classify_partition): Skip builtin pattern of loop nest's inner loop.
8928 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
8929 in call to build_partition_graph.
8930 (finalize_partitions): New parameter. Make loop distribution more
8931 conservative by fusing more partitions.
8932 (distribute_loop): Don't do runtime alias check in case of loop nest
8934 (find_seed_stmts_for_distribution): New function.
8935 (prepare_perfect_loop_nest): New function.
8936 (pass_loop_distribution::execute): Refactor code finding seed stmts
8937 and loop nest into above functions. Support loop nest distribution.
8938 Adjust dump information accordingly.
8940 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8942 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
8943 and set PTYPE_SEQUENTIAL for merged partition.
8945 2017-10-12 Richard Biener <rguenther@suse.de>
8947 PR tree-optimization/69728
8949 2017-09-19 Richard Biener <rguenther@suse.de>
8951 PR tree-optimization/69728
8952 * graphite-sese-to-poly.c (schedule_error): New global.
8953 (add_loop_schedule): Handle empty domain by failing the
8955 (build_original_schedule): Handle schedule_error.
8957 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
8958 domain by returning an unchanged schedule.
8960 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8962 * genrecog.c (validate_pattern): For VEC_SELECT verify that
8963 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
8965 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
8967 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
8970 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8973 * expr.c (store_field): Don't optimize away bitsize == 0 store
8974 from CALL_EXPR with addressable return type.
8976 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8978 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
8979 * config/rs6000/rs6000.md (sel): Delete mode attribute.
8980 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
8981 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
8982 TARGET_ISEL instead of TARGET_ISEL<sel>.
8984 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
8986 * config/rs6000/rs6000.c
8987 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
8989 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8991 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
8993 (reg_or_cint_operand): Fix comment.
8994 (reg_or_zero_operand): New predicate.
8995 * config/rs6000/rs6000-protos.h (output_isel): Delete.
8996 * config/rs6000/rs6000.c (output_isel): Delete.
8997 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
8998 instead of reg_or_cint_operand. Output instruction directly (not via
9000 (isel_unsigned_<mode>): Ditto.
9001 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9002 gpc_reg_operand. Add an instruction alternative for this. Output
9003 instruction directly.
9004 (*isel_reversed_unsigned_<mode>): Ditto.
9006 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9008 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9009 (TARGET_CANONICALIZE_COMPARISON): Define.
9011 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9014 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9015 Check whether the dest is REG before adding REG_EQUIV note.
9017 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9020 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9022 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9023 (make_hard_regno_born, make_hard_regno_dead): Update
9024 bb_killed_pseudos and bb_gen_pseudos.
9026 2017-10-11 Nathan Sidwell <nathan@acm.org>
9028 * incpath.h (enum incpath_kind): Name enum, prefix values.
9029 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9030 * incpath.c (heads, tails): Use INC_MAX.
9031 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9032 (merge_include_chains, split_quote_chain,
9033 register_include_chains): Update incpath_kind names.
9034 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9035 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9037 (add_framework_path, darwin_register_objc_includes): Likewise.
9038 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9040 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9042 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9043 Do not use float_operator operator predicate.
9044 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9045 * config/i386/predicates.md (float_operator): Remove predicate.
9047 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9049 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9050 (*jccxf_i387): Ditto.
9051 (*jcc<mode>_i387): Ditto.
9052 (*jccu<mode>_i387): Ditto.
9053 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9054 (*jcc_*_i387 splitters): Remove.
9055 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9056 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9057 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9060 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9062 * profile-count.h (slow_safe_scale_64bit): New function.
9063 (safe_scale_64bit): New inline.
9064 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9065 * profile-count.c: Include wide-int.h
9066 (slow_safe_scale_64bit): New.
9068 2017-10-11 Nathan Sidwell <nathan@acm.org>
9070 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9071 HAS_DECL_ASSEMBLER_NAME_P.
9072 * gimple-expr.c (gimple_decl_printable_name: Check
9073 HAS_DECL_ASSEMBLER_NAME_P too.
9074 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9075 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9076 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9077 * passes.c (rest_of_decl_compilation): Check
9078 HAS_DECL_ASSEMBLER_NAME_P too.
9079 * recog.c (verify_changes): Likewise.
9080 * tree-pretty-print.c (dump_decl_name): Likewise.
9081 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9083 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9084 (SET_DECL_ASSEMBLER_NAME): Use it.
9085 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9086 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9087 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9089 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9091 * config.gcc (i386, x86_64): Add extra objects.
9092 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9093 (ix86_min_insn_size): Declare.
9094 (ix86_issue_rate): Declare.
9095 (ix86_adjust_cost): Declare.
9096 (ia32_multipass_dfa_lookahead): Declare.
9097 (ix86_macro_fusion_p): Declare.
9098 (ix86_macro_fusion_pair_p): Declare.
9099 (ix86_bd_has_dispatch): Declare.
9100 (ix86_bd_do_dispatch): Declare.
9101 (ix86_core2i7_init_hooks): Declare.
9102 (ix86_atom_sched_reorder): Declare.
9103 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9104 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9105 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9106 (rip_relative_addr_p): Rename to ...
9107 (ix86_rip_relative_addr_p): ... this one; export.
9108 (memory_address_length): Update.
9109 (ix86_issue_rate): Move to x86-tune-sched.c.
9110 (ix86_flags_dependent): Move to x86-tune-sched.c.
9111 (ix86_agi_dependent): Move to x86-tune-sched.c.
9112 (exact_dependency_1): Move to x86-tune-sched.c.
9113 (exact_store_load_dependency): Move to x86-tune-sched.c.
9114 (ix86_adjust_cost): Move to x86-tune-sched.c.
9115 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9116 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9117 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9118 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9119 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9120 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9121 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9122 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9123 (min_insn_size): Rename to ...
9124 (ix86_min_insn_size): ... this one; export.
9125 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9126 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9127 (core2i7_first_cycle_multipass_backtrack): Move to
9128 x86-tune-sched-core.c.
9129 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9130 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9131 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9132 (ix86_avoid_jump_mispredicts): Update.
9133 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9134 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9135 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9136 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9137 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9138 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9139 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9140 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9141 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9142 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9143 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9144 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9145 (BIG): Move to ix86-tune-sched-bd.c.
9146 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9147 (enum insn_path): Move to ix86-tune-sched-bd.c.
9148 (get_mem_group): Move to ix86-tune-sched-bd.c.
9149 (is_cmp): Move to ix86-tune-sched-bd.c.
9150 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9151 (is_branch): Move to ix86-tune-sched-bd.c.
9152 (is_prefetch): Move to ix86-tune-sched-bd.c.
9153 (init_window): Move to ix86-tune-sched-bd.c.
9154 (allocate_window): Move to ix86-tune-sched-bd.c.
9155 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9156 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9157 (process_end_window): Move to ix86-tune-sched-bd.c.
9158 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9159 (find_constant): Move to ix86-tune-sched-bd.c.
9160 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9161 (has_immediate): Move to ix86-tune-sched-bd.c.
9162 (get_insn_path): Move to ix86-tune-sched-bd.c.
9163 (get_insn_group): Move to ix86-tune-sched-bd.c.
9164 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9165 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9166 (add_insn_window): Move to ix86-tune-sched-bd.c.
9167 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9168 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9169 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9170 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9171 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9172 (do_dispatch): Move to ix86-tune-sched-bd.c.
9173 (has_dispatch): Move to ix86-tune-sched-bd.c.
9174 * config/i386/t-i386: Add new object files.
9175 * config/i386/x86-tune-costs.h: New file.
9176 * config/i386/x86-tune-sched-atom.c: New file.
9177 * config/i386/x86-tune-sched-bd.c: New file.
9178 * config/i386/x86-tune-sched-core.c: New file.
9179 * config/i386/x86-tune-sched.c: New file.
9181 2017-10-11 Liu Hao <lh_mouse@126.com>
9183 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9184 the generic version below instead.
9185 (should_colorize): Recognize Windows consoles as terminals
9187 * pretty-print.c [__MINGW32__] (write_all): New function.
9188 [__MINGW32__] (find_esc_head): Likewise.
9189 [__MINGW32__] (find_esc_terminator): Likewise.
9190 [__MINGW32__] (eat_esc_sequence): Likewise.
9191 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9193 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9196 2017-10-11 Richard Biener <rguenther@suse.de>
9198 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9199 Properly call analyze_scalar_evolution with the loop of the stmt.
9201 2017-10-11 Richard Biener <rguenther@suse.de>
9203 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9204 * tree-core.h (tree_base): Add chrec_var union member.
9205 * tree.h (CHREC_VAR): Remove.
9206 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9207 * tree-chrec.h (build_polynomial_chrec): Adjust.
9208 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9209 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9211 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9213 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9214 * match.pd: ... here.
9215 ((T) X == (T) Y): Relax condition.
9217 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9219 PR tree-optimization/82472
9220 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9222 (break_alias_scc_partitions): Update postorder number.
9224 2017-10-11 Martin Liska <mliska@suse.cz>
9227 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9229 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9231 2017-10-11 Martin Liska <mliska@suse.cz>
9233 * print-rtl.c (print_insn): Move declaration of idbuf
9234 to same scope as name.
9236 2017-10-11 Martin Liska <mliska@suse.cz>
9241 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9244 2017-10-11 Martin Liska <mliska@suse.cz>
9247 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9250 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9252 * config.gcc (default_gnu_indirect_function): Default to yes for
9253 arm*-*-linux* with glibc.
9255 2017-10-11 Richard Biener <rguenther@suse.de>
9257 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9258 default-defs and types we do not want to analyze.
9259 (interpret_loop_phi): Replace unreachable code with an assert.
9260 (compute_scalar_evolution_in_loop): Remove and inline ...
9261 (analyze_scalar_evolution_1): ... here, replacing condition with
9262 what makes the intent clearer. Remove handling of cases
9263 get_scalar_evolution now handles.
9265 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9267 PR rtl-optimization/81434
9268 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9269 comment for main loop. In sched_group_found if, also add checks for
9270 pass and min_cost_group.
9272 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9274 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9275 (rs6000_insn_cost): New function.
9276 * config/rs6000/rs6000.md (cost): New attribute.
9278 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9279 H.J. Lu <hongjiu.lu@intel.com>
9283 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9284 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9285 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9286 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9287 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9288 OPTION_MASK_ISA_MMX and some other bit set require both
9289 mmx and the ISAs without the mmx bit.
9290 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9291 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9292 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9293 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9294 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9295 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9296 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9297 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9298 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9299 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9300 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9301 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9302 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9303 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9304 Add OPTION_MASK_ISA_MMX.
9306 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9308 * config.gcc (armv7*-*-freebsd*): New target.
9309 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9311 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9313 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9314 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9317 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9319 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9320 a template parameter.
9321 (WIDE_INT_REF_FOR): Update accordingly.
9322 * tree.h (wi::int_traits <const_tree>): Delete.
9323 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9324 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9325 (wi::tree_to_wide_ref): New typedef.
9326 (wi::to_wide): New function.
9327 * calls.c (get_size_range): Use wi::to_wide when operating on
9329 * cgraph.c (cgraph_node::create_thunk): Likewise.
9330 * config/i386/i386.c (ix86_data_alignment): Likewise.
9331 (ix86_local_alignment): Likewise.
9332 * dbxout.c (stabstr_O): Likewise.
9333 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9334 * expr.c (const_vector_from_tree): Likewise.
9335 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9336 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9337 (fold_negate_expr_1, int_const_binop_1, const_binop)
9338 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9339 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9340 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9341 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9342 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9343 (fold_not_const, round_up_loc): Likewise.
9344 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9345 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9346 (alloca_call_type): Likewise.
9347 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9348 * godump.c (go_output_typedef): Likewise.
9349 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9350 * internal-fn.c (get_min_precision): Likewise.
9351 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9352 * ipa-polymorphic-call.c
9353 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9354 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9355 (ipa_modify_call_arguments): Likewise.
9356 * match.pd: Likewise.
9357 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9358 * print-tree.c (print_node_brief, print_node): Likewise.
9359 * stmt.c (expand_case): Likewise.
9360 * stor-layout.c (layout_type): Likewise.
9361 * tree-affine.c (tree_to_aff_combination): Likewise.
9362 * tree-cfg.c (group_case_labels_stmt): Likewise.
9363 * tree-data-ref.c (dr_analyze_indices): Likewise.
9364 (prune_runtime_alias_test_list): Likewise.
9365 * tree-dump.c (dequeue_and_dump): Likewise.
9366 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9367 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9368 * tree-pretty-print.c (dump_generic_node): Likewise.
9369 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9370 (simple_iv_with_niters): Likewise.
9371 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9372 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9373 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9374 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9375 (refine_value_range_using_guard, number_of_iterations_ne_max)
9376 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9377 (get_cst_init_from_scev, record_nonwrapping_iv)
9378 (scev_var_range_cant_overflow): Likewise.
9379 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9380 * tree-ssa-pre.c (compute_avail): Likewise.
9381 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9382 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9383 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9384 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9385 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9386 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9387 (dump_case_nodes, try_switch_expansion): Likewise.
9388 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9389 (vect_do_peeling): Likewise.
9390 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9391 * tree-vect-stmts.c (vectorizable_load): Likewise.
9392 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9393 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9394 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9395 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9396 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9397 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9398 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9399 (evrp_dom_walker::before_dom_children): Likewise.
9400 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9401 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9402 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9403 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9404 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9405 (get_range_pos_neg): Likewise.
9406 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9407 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9408 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9409 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9410 * config/avr/avr.c (avr_fold_builtin): Likewise.
9411 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9412 * config/msp430/msp430.c (msp430_attr): Likewise.
9413 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9414 * config/powerpcspe/powerpcspe-c.c
9415 (altivec_resolve_overloaded_builtin): Likewise.
9416 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9417 (rs6000_expand_ternop_builtin): Likewise.
9418 * config/rs6000/rs6000-c.c
9419 (altivec_resolve_overloaded_builtin): Likewise.
9420 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9421 (rs6000_expand_ternop_builtin): Likewise.
9422 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9424 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9426 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9427 when copying loop nest with only one inner loop.
9429 2017-10-10 Richard Biener <rguenther@suse.de>
9431 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9432 blocks if SCEV is active.
9433 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9435 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9436 (scev_initialize): Assert we are not yet initialized.
9438 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9440 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9441 inner loop's exit stmt by making it always exit the loop, otherwise
9442 we would generate an infinite empty loop.
9444 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9446 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9447 renaming variables in new preheader if it's deleted.
9449 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9451 * tree-loop-distribution.c (struct partition): Remove unused field
9452 loops of the structure.
9453 (partition_alloc, partition_free): Ditto.
9454 (build_rdg_partition_for_vertex): Ditto.
9456 2017-10-09 Jeff Law <law@redhat.com>
9458 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9459 return type to match prototype and documentation.
9461 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9463 * config/rs6000/rs6000.c (processor_costs): Move to ...
9464 * config/rs6000/rs6000.h: ... here.
9465 (rs6000_cost): Declare.
9467 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9469 * except.c (setjmp_fn): New global variable.
9470 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9471 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9472 if DONT_USE_BUILTIN_SETJMP is defined.
9474 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9476 * target.def (insn_cost): New hook.
9477 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9478 * doc/tm.texi: Regenerate.
9479 * rtlanal.c (insn_cost): Use the new hook.
9481 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9483 * combine.c (combine_validate_cost): Compute the new insn_cost,
9484 not just pattern_cost.
9485 (try_combine): Adjust comment.
9487 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9489 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9491 * combine.c (uid_insn_cost): Adjust comment.
9492 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9494 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9495 * dse.c (find_shift_sequence): Ditto.
9496 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9497 (bb_valid_for_noce_process_p): Use pattern_cost.
9498 * rtl.h (insn_rtx_cost): Delete.
9499 (pattern_cost): New prototype.
9500 (insn_cost): New prototype.
9501 * rtlanal.c (insn_rtx_cost): Rename to...
9502 (pattern_cost): ... this.
9505 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9507 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9508 (*jcc<mode>_0_r_i387): Ditto.
9509 (*jccxf_r_i387): Ditto.
9510 (*jcc<mode>_r_i387): Ditto.
9511 (*jccu<mode>_r_i387): Ditto.
9512 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9513 (*jcc): Rename from *jcc_1.
9515 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9517 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9518 deferred rescans after the lvx/stvx recombination pre-pass.
9520 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9522 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9523 memory operation instruction support.
9524 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9525 (rs6000-ibm-aix[789]*): Likewise.
9526 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9527 Document new functions.
9529 2017-10-09 Richard Biener <rguenther@suse.de>
9531 PR tree-optimization/82397
9532 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9533 equality only for semantically equal trees.
9535 2017-10-09 Richard Biener <rguenther@suse.de>
9537 PR tree-optimization/82449
9538 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9539 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9540 allow constant addresses.
9541 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9544 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9546 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9549 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9552 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9555 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9558 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9560 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9563 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9564 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9566 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9568 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9569 (WI_BINARY_PREDICATE_RESULT): Likewise.
9570 (wi::binary_traits::operator_result): New type.
9571 (wi::binary_traits::predicate_result): Likewise.
9572 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9573 (generic_wide_int::operator==, generic_wide_int::operator!=)
9574 (generic_wide_int::operator&, generic_wide_int::and_not)
9575 (generic_wide_int::operator|, generic_wide_int::or_not)
9576 (generic_wide_int::operator^, generic_wide_int::operator+
9577 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9579 (operator~, unary operator-, operator==, operator!=, operator&)
9580 (operator|, operator^, operator+, binary operator-, operator*): New
9582 * expr.c (get_inner_reference): Use wi::bit_and_not.
9583 * fold-const.c (fold_binary_loc): Likewise.
9584 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9585 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9586 (bit_value_binop): Likewise.
9587 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9588 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9589 (extract_range_from_binary_expr_1): Likewise.
9590 (masked_increment): Likewise.
9591 (simplify_bit_ops_using_ranges): Likewise.
9593 2017-10-09 Martin Jambor <mjambor@suse.cz>
9596 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9597 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9598 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9600 (hsa_fixup_mov_insn_type): New function.
9601 (hsa_op_with_type::get_in_type): Use it.
9602 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9603 immediates in an assert.
9604 (hsa_op_with_type::extend_int_to_32bit): New method.
9605 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9606 types. Convert to dest type if necessary.
9607 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9608 (reg_for_gimple_ssa): Pass false as min32int to
9609 hsa_type_for_scalar_tree_type.
9610 (gen_hsa_addr): Fixup type when creating addresable temporary.
9611 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9612 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9613 necessary. Call hsa_fixup_mov_insn_type.
9614 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9615 extend operands and convert to dest type if necessary.
9616 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9617 to dest type if necessary.
9618 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9619 if conversion nt necessary and size matches.
9620 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9621 to dest type if necessary.
9622 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9623 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9625 (gen_hsa_clrsb): Likewise.
9626 (gen_hsa_ffs): Likewise.
9627 (gen_hsa_divmod): Extend operands and convert to dest type if
9629 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9631 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9633 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9634 Remove empty default arguments. Use a brace block as output
9636 (conditional return): Ditto.
9638 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9639 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9641 (group_ending_nop): Ditto.
9642 (doloop_end): Ditto.
9643 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9644 (splitters for those): Ditto.
9646 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9648 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9649 a conditional jump (and the compare for it) so that pc_rtx is the
9651 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9652 for the deleted and renamed ctr<mode>_internal[234] patterns.
9653 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9654 Delete second conditional return pattern.
9655 (ctr<mode>_internal2): Delete this second bdnz pattern.
9656 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9657 (ctr<mode>_internal4): Delete this second bdz pattern.
9659 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9661 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9662 (always_initialized_rtx_for_ssa_name_p): New predicate.
9663 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9664 (finish_out_of_ssa): Free new field of SA.
9665 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9666 * tree-ssa-coalesce.c: Include tree-ssa.h.
9667 (get_parm_default_def_partitions): Remove extern keyword.
9668 (get_undefined_value_partitions): New function.
9669 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9670 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9672 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9674 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9676 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9678 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9680 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9681 for targets that preffer 128bit.
9683 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9685 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9687 2017-10-08 Olivier Hainque <hainque@adacore.com>
9689 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9690 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9692 (thumb_set_return_address): Likewise.
9694 2017-10-08 Olivier Hainque <hainque@adacore.com>
9696 * common/config/arm/arm-common.c (arm_except_unwind_info):
9697 Handle DWARF2_UNWIND_INFO.
9699 2017-10-07 Michael Collison <michael.collison@arm.com>
9701 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9704 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9706 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9707 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9708 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9709 defined, force the creation of a new block for a dispatch label.
9711 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9713 * invoke.texi (Wsuggest-attribute=cold): Document.
9714 * common.opt (Wsuggest-attribute=cold): New
9715 * ipa-pure-const.c (warn_function_cold): New function.
9716 * predict.c (compute_function_frequency): Use it.
9717 * predict.h (warn_function_cold): Declare.
9719 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9721 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9724 2017-10-06 Martin Liska <mliska@suse.cz>
9726 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9727 keyword for member functions.
9728 (struct sanopt_tree_couple): New struct.
9729 (struct sanopt_tree_couple_hash): New function.
9730 (struct sanopt_ctx): Add new hash_map.
9731 (has_dominating_ubsan_ptr_check): New function.
9732 (record_ubsan_ptr_check_stmt): Likewise.
9733 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9734 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9735 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9737 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9740 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9741 aarch64_simd_valid_immediate on CONST_VECTORs.
9742 (aarch64_reg_or_bic_imm): Likewise.
9744 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9746 PR rtl-optimization/82396
9747 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9749 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9751 * graphite-dependences.c (scop_get_reads): Move code to...
9752 (scop_get_must_writes): Move code to...
9753 (scop_get_may_writes): Move code to...
9754 (scop_get_reads_and_writes): ... here.
9755 (scop_get_dependences): Call scop_get_reads_and_writes.
9757 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9759 PR tree-optimization/82434
9760 * fold-const.h (can_native_encode_type_p,
9761 can_native_encode_string_p): Remove.
9762 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9763 don't encode anything, just return what would be otherwise returned.
9764 (native_encode_fixed, native_encode_complex, native_encode_vector):
9766 (native_encode_string): Likewise. Inline by hand
9767 can_native_encode_string_p.
9768 (can_native_encode_type_p): Remove.
9769 (can_native_encode_string_p): Remove.
9770 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9771 STRING_CSTs using can_native_encode_string_p, test all
9772 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9773 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9774 argument from native_encode_expr.
9775 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9776 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9779 2017-10-06 Richard Biener <rguenther@suse.de>
9781 PR tree-optimization/82397
9782 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9783 operand_equal_p but rely on data_ref_compare_tree for detecting
9785 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9786 to match up with dr_group_sort_cmp.
9788 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9791 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9793 * config/s390/s390-builtin-types.def: Regenerate.
9795 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9798 * config/s390/s390-builtin-types.def: Regenerate.
9799 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9800 Change flag from B_VXE to B_VX.
9801 (s390_vec_min_dbl): Remove B_VXE flag.
9803 2017-10-06 Richard Biener <rguenther@suse.de>
9805 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9806 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9807 translate_isl_ast_to_gimple::is_valid_rename,
9808 translate_isl_ast_to_gimple::get_rename,
9809 translate_isl_ast_to_gimple::get_def_bb_for_const,
9810 translate_isl_ast_to_gimple::get_new_name,
9811 translate_isl_ast_to_gimple::collect_all_ssa_names,
9812 translate_isl_ast_to_gimple::copy_loop_phi_args,
9813 translate_isl_ast_to_gimple::collect_all_ssa_names,
9814 translate_isl_ast_to_gimple::copy_loop_phi_args,
9815 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9816 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9817 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9818 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9819 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9820 translate_isl_ast_to_gimple::copy_cond_phi_args,
9821 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9822 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9823 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9824 translate_isl_ast_to_gimple::rename_uses,
9825 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9826 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9827 (set_rename_for_each_def): Likewise.
9828 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9829 here. Handle rewriting SCEV analyzable uses here.
9830 (copy_bb_and_scalar_dependences): Generate code for PHI
9832 (graphite_regenerate_ast_isl): Adjust.
9833 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9834 (add_write, add_read): New functions.
9835 (build_cross_bb_scalars_def): Use it and simplify.
9836 (build_cross_bb_scalars_use): Likewise.
9837 (graphite_find_cross_bb_scalar_vars): Inline into...
9838 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9839 simulating out-of-SSA. Compute liveout and add dependencies.
9840 (build_scops): Force an empty entry block.
9841 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9843 (sese_build_liveouts): Declare.
9844 (sese_trivially_empty_bb_p): Likewise.
9845 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9846 compute liveout and debug_liveout.
9847 (sese_bad_liveouts_use): Remove.
9848 (sese_reset_debug_liveouts_bb): Likewise.
9849 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9850 (sese_build_liveouts): Build liveout and debug_liveout and store
9852 (new_sese_info): Adjust.
9853 (free_sese_info): Likewise.
9854 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9855 do not build liveout here.
9856 (move_sese_in_condition): Adjust region entry.
9857 (scev_analyzable_p): Match up with chrec_apply requirements.
9858 (sese_trivially_empty_bb_p): New.
9859 * tree-into-ssa.c (get_reaching_def): Properly support generating
9860 default-defs for incremental rewrite of anonymous names.
9862 2017-10-06 Richard Biener <rguenther@suse.de>
9864 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9865 precision do not perform modulo reduction.
9867 2017-10-06 Richard Biener <rguenther@suse.de>
9869 PR tree-optimization/82436
9870 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9871 conservatively choose the vectorization factor when checking
9872 whether we can perform the required load permutation.
9873 (vect_transform_slp_perm_load): Assert when we may not fail.
9875 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9877 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9878 message for incompatible -msdata=* and -mcall-* options.
9880 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9882 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9883 rate for post-reload scheduling.
9885 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9887 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9889 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9891 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9892 to improve monte carlo in scimark.
9894 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9896 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9897 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9898 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9899 pentium4_cost, nocona_cost): Set reassociation width to 1.
9900 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9901 width to 2 for fp operations and 1 otherwise.
9902 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9904 (atom_cost): Set reassociation width to 2.
9905 (slm_cost, generic_cost): Set fp reassociation width
9906 to 2 and 1 otherwise.
9907 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
9908 (core_cost): Set fp reassociation width to 4 and vector to 2.
9909 (ix86_reassociation_width): Rewrite using cost table; special case
9910 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
9911 and TARGET_AVX128_OPTIMAL.
9912 * config/i386/i386.h (processor_costs): Add
9913 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
9914 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
9915 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
9916 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
9917 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
9918 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
9920 2017-10-05 Nathan Sidwell <nathan@acm.org>
9922 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
9924 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9926 * config/arm/arm.c (arm_test_fpu_data): New.
9927 (arm_run_selftests): Call arm_test_fpu_data.
9929 2017-10-04 Nathan Sidwell <nathan@acm.org>
9931 * toplev.c (toplev::main): Remove excess parens on pretty_printer
9933 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
9935 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9937 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
9938 check type for aarch64_simd_valid_immediate.
9939 (aarch64_output_simd_mov_immediate): Update prototype.
9940 (aarch64_simd_valid_immediate): Update prototype.
9941 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
9942 support for ORR-immediate.
9943 (and<mode>3): modified pattern to add support for BIC-immediate.
9944 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
9945 now checks for valid immediate for BIC and ORR based on new enum
9947 (aarch64_output_simd_mov_immediate): Function now used to output
9948 BIC/ORR imm as well based on new enum argument.
9949 * config/aarch64/constraints.md (Do): New vector immediate constraint.
9951 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
9952 (aarch64_reg_or_bic_imm): Likewise.
9954 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9956 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
9957 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
9959 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9963 PR rtl-optimization/82396
9964 * haifa-sched.c (autopref_multipass_init): Simplify
9966 (autopref_rank_data): Simplify sort order.
9967 * sched-int.h (autopref_multipass_data_): Remove
9968 multi_mem_insn_p, min_offset and max_offset.
9970 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9972 * doc/sourcebuild.texi: Document vect_peeling_profitable.
9974 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9976 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
9979 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9981 * doc/sourcebuild.texi: Document vect_long_mult.
9983 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
9985 PR tree-optimization/82413
9986 * fold-const.c (build_range_check): Use widest_int when comparing
9987 the maximum ETYPE value with HIGH.
9989 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9991 PR rtl-optimization/82396
9992 * haifa-sched.c (autopref_multipass_init): Simplify
9994 (autopref_rank_data): Simplify sort order.
9995 * sched-int.h (autopref_multipass_data_): Remove
9996 multi_mem_insn_p, min_offset and max_offset.
9998 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10000 PR tree-optimization/82381
10001 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10002 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10005 PR tree-optimization/82374
10006 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10007 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10008 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10009 current_function_decl to the new decl.
10011 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10013 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10014 helper macro for IEEE float128 hardware built-in functions.
10015 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10017 (TRUNCF128_ODD): Likewise.
10018 (ADDF128_ODD): Likewise.
10019 (SUBF128_ODD): Likewise.
10020 (MULF128_ODD): Likewise.
10021 (DIVF128_ODD): Likewise.
10022 (FMAF128_ODD): Likewise.
10023 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10024 UNSPEC_TRUNC_ROUND_TO_ODD.
10025 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10026 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10027 floating point round to odd instructions.
10028 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10029 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10030 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10031 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10032 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10033 (trunc<mode>sf2_hw): Change the truncate with round to odd
10034 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10035 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10036 to odd hardware instructions.
10037 (sub<mode>3_odd): Likewise.
10038 (mul<mode>3_odd): Likewise.
10039 (div<mode>3_odd): Likewise.
10040 (sqrt<mode>2_odd): Likewise.
10041 (fma<mode>4_odd): Likewise.
10042 (fms<mode>4_odd): Likewise.
10043 (nfma<mode>4_odd): Likewise.
10044 (nfms<mode>4_odd): Likewise.
10045 (trunc<mode>df2_odd): Change the truncate with round to odd
10046 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10048 * doc/extend.texi (PowerPC built-in functions): Update documentation
10049 for existing IEEE float128-bit built-in functions. Add built-in
10050 functions that generate the IEEE 128-bit floating point round to
10053 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10055 PR rtl-optimization/77729
10056 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10057 to (X&(C1&~C2))|C2 transformations.
10059 2017-10-03 Martin Jambor <mjambor@suse.cz>
10061 PR tree-optimization/82363
10062 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10063 mismatch, mark lacc written regardless of racc.
10065 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10067 PR tree-optimization/82381
10068 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10069 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10070 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10071 fallthrough into reassoc_stmt_dominates_stmt_p.
10074 * combine.c (combine_instructions): Don't combine in unreachable
10077 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10080 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10081 function to not use the have_cpu variable. Do not set cpu_index,
10082 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10083 or the default cpu.
10084 (rs6000_valid_attribute_p): Remove duplicate initializations of
10085 old_optimize and func_optimize.
10086 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10087 (rs6000_activate_target_options): Make global.
10088 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10091 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10093 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10094 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10095 Set *poffset to 0 and *psize and *pmax_size to -1 if
10096 *poffset + *psize overflows in HOST_WIDE_INT.
10098 PR tree-optimization/82387
10099 PR tree-optimization/82388
10100 PR tree-optimization/82389
10101 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10102 instead of live_bytes non-NULL.
10104 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10107 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10110 2017-10-02 Richard Biener <rguenther@suse.de>
10112 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10113 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10114 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10116 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10118 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10119 requested precision matches the type's.
10120 * calls.c (alloc_max_size): Calculate the new candidate size as
10121 a widest_int and use wi::to_widest when comparing it with the
10122 current candidate size.
10123 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10124 zero rather than integer_zero_node.
10125 * match.pd: Check for a no-op conversion before using wi::add
10126 rather than after. Use tree_to_uhwi when summing small shift
10127 counts into an unsigned int.
10129 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10130 Alan Hayward <alan.hayward@arm.com>
10131 David Sherwood <david.sherwood@arm.com>
10134 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10135 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10136 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10137 POINTER_AND_FP_REGS.
10139 2017-10-02 Richard Biener <rguenther@suse.de>
10141 PR tree-optimization/82355
10142 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10143 a mapping for the enclosing loop but avoid generating one for
10144 the loop tree root.
10145 (copy_bb_and_scalar_dependences): Remove premature codegen
10146 error on PHIs in blocks duplicated into multiple places.
10147 * graphite-scop-detection.c
10148 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10149 in the region use it as loop and nest to analyze the DR in.
10150 (try_generate_gimple_bb): Likewise.
10151 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10152 (add_loop_constraints): For blocks in a loop not in the region
10153 create a dimension with a single iteration.
10154 * sese.h (gbb_loop_at_index): Remove assert.
10156 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10158 * omp-expand.c (adjust_context_scope): New function.
10159 (expand_parallel_call): Call adjust_context_scope.
10161 2017-10-01 Jeff Law <law@redhat.com>
10163 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10164 dom_opt_dom_walker class with direct access to private members.
10165 Add comments. Call test_for_singularity.
10166 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10167 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10168 m_dummy_cond anymore.
10169 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10171 (pass_dominator:execute): Build the dummy_cond here and pass it
10172 to the dom_opt_dom_walker ctor.
10173 (test_for_singularity): New function.
10175 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10176 Maya Rashish <coypu@sdf.org>
10178 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10179 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10180 (alpha*-*-netbsd*) Use nbsd_tm_file.
10181 (arm*-*-netbsdelf*) Likewise.
10182 (i[34567]86-*-netbsdelf*) Likewise.
10183 (x86_64-*-netbsd*) Likewise.
10184 (mips*-*-netbsd*) Likewise.
10185 (powerpc-*-netbsd*) Likewise.
10186 (sh*-*-netbsd*) Likewise.
10187 (sparc-*-netbsdelf*) Likewise.
10188 (sparc64-*-netbsd*) Likewise.
10189 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10191 (vax-*-netbsdelf*) Likewise.
10192 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10193 (UINT_FAST8_TYPE) Likewise.
10194 (INT_FAST16_TYPE) Check CHAR_FAST16.
10195 (UINT_FAST16_TYPE) Likewise.
10197 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10200 * config/i386/i386.md
10201 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10202 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10203 *divmodsi4_zext_2): New define_insn_and_split.
10204 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10205 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10206 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10207 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10208 New define_insn_and_split.
10209 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10210 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10211 operands[1] having DImode when mode is SImode.
10213 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10214 always SImode for DIV and MOD in REG_EQUAL notes.
10216 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10218 PR middle-end/82319
10219 * match.pd: Fix handling of NaNs in pattern.
10221 2017-09-29 Jeff Law <law@redhat.com>
10223 * sbitmap.c (bitmap_bit_in_range_p): New function.
10224 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10225 * tree-ssa-dse.c (live_bytes_read): New function.
10226 (dse_classify_store): Ignore reads of dead bytes.
10228 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10229 typos and whitespace errors.
10230 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10231 * config/s390/s390.c (s390_emit_prologue): Likewise.
10233 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10236 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10237 with a symbol for LRA.
10239 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10241 PR rtl-optimization/82338
10242 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10244 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10246 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10247 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10248 (qsort_chk): Declare.
10249 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10250 (qsort_chk): New function.
10252 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10254 PR tree-optimization/82337
10255 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10256 phi definition if the PHI result appears in an abnormal PHI.
10257 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10258 basis appears in an abnormal PHI.
10260 2017-09-29 Richard Biener <rguenther@suse.de>
10262 * graphite-isl-ast-to-gimple.c
10263 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10264 (binary_op_to_tree): Use it.
10265 (get_rename_from_scev): Likewise.
10266 (copy_loop_phi_nodes): Likewise.
10267 (copy_bb_and_scalar_dependences): Likewise.
10268 (translate_pending_phi_nodes): Likewise.
10270 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10273 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10274 for movabsq $(i32 << shift), r64.
10276 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10278 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10279 index when encoding %esp as %rsp to avoid 0x67 prefix.
10281 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10283 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10284 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10286 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10288 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10289 Extensions with more than 16 double VFP registers.
10290 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10291 to_clear_mask and all code related to it. Replace the remaining
10292 entry by a sbitmap and adapt code accordingly.
10294 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10296 * brig-builtins.def: Change pure attributes to const.
10298 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10300 * config.gcc (default_gnu_indirect_function): Default to yes for
10301 sparc*-*-linux* with glibc.
10303 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10305 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10306 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10307 when creating .init_array and .fini_array sections with priority
10310 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10313 * config/aarch64/aarch64.c
10314 (aarch64_builtin_support_vector_misalignment): Always return false
10315 when misalignment is unknown.
10317 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10319 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10320 this function to return false if the definition used by the swap
10321 instruction is artificial, or if the memory address from which the
10322 constant value is loaded is not represented by a base address held
10323 in a register or if the base address register is a frame or stack
10324 pointer. Additionally, return false if the base address of the
10325 loaded constant is a SYMBOL_REF but is not considered to be a
10327 (replace_swapped_load_constant): New function.
10328 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10329 loaded constant vector with a load of a swapped constant vector.
10331 2017-09-27 Carl Love <cel@us.ibm.com>
10333 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10334 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10335 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10338 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10340 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10341 first, always call autopref_rank_data otherwise.
10343 2017-09-27 Richard Biener <rguenther@suse.de>
10345 * graphite-scop-detection.c (find_scop_parameters): Move
10346 loop bound handling ...
10347 (gather_bbs::before_dom_children): ... here, avoiding the need
10348 to build scop_info->loop_nest.
10349 (record_loop_in_sese): Remove.
10350 * sese.h (sese_info_t::loop_nest): Remove.
10351 * sese.c (new_sese_info): Do not allocate loop_nest.
10352 (free_sese_info): Do not free loop_nest.
10354 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10357 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10358 lhs from calls if the lhs has addressable type.
10360 2017-09-27 Richard Biener <rguenther@suse.de>
10362 * graphite.h (scop::max_alias_set): New member.
10363 * graphite-scop-detection.c: Remove references to non-existing
10364 --param in comments.
10365 (build_alias_sets): Record the maximum alias set used for drs.
10366 (build_scops): Support zero as unlimited for
10367 --param graphite-max-arrays-per-scop.
10368 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10369 and inline into ...
10370 (build_poly_sr_1): ... here. Compute alias set based on the
10371 maximum alias set used for drs rather than
10372 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10374 2017-09-27 Richard Biener <rguenther@suse.de>
10376 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10377 --param loop-block-tile-size=0 to disable tiling.
10379 2017-09-27 Richard Biener <rguenther@suse.de>
10381 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10382 (graphite-max-nb-scop-params): Document special value zero.
10383 * domwalk.h (dom_walker::STOP): New symbolical constant.
10384 (dom_walker::dom_walker): Add optional parameter for bb to
10386 (dom_walker::~dom_walker): Declare.
10387 (dom_walker::before_dom_children): Document STOP return value.
10388 (dom_walker::m_user_bb_to_rpo): New member.
10389 (dom_walker::m_bb_to_rpo): Likewise.
10390 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10391 mapping here if not provided by the user.
10392 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10393 provided by the user.
10394 (dom_walker::STOP): Define.
10395 (dom_walker::walk): Do not compute bb to RPO mapping here.
10396 Support STOP return value from before_dom_children to stop
10398 * graphite-optimize-isl.c (optimize_isl): If the schedule
10399 is the same still generate code if -fgraphite-identity
10400 or -floop-parallelize-all are given.
10401 * graphite-scop-detection.c: Include cfganal.h.
10402 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10404 (gather_bbs::before_dom_children): Return STOP for BBs
10406 (build_scops): Compute bb to RPO mapping and pass it to
10407 the domwalk. Treat --param graphite-max-nb-scop-params=0
10408 as not limiting the number of params.
10409 * graphite.c (graphite_initialize): Remove limit on the
10410 number of basic-blocks in a function.
10411 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10412 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10413 default value of 10.
10415 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10417 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10418 Adjust code to eliminate needing to do the shift right 32-bits
10419 operation after XSCVDPSPN.
10421 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10423 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10424 ((X / Y) != 0 -> X >= Y): Likewise.
10426 2017-09-26 Carl Love <cel@us.ibm.com>
10428 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10429 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10430 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10431 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10432 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10433 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10434 definitions and overloading.
10435 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10436 statement for P9V_BUILTIN_XST_LEN_R.
10437 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10438 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10439 define_expand and define_insn for the instructions and builtins.
10440 * doc/extend.texi: Update the built-in documentation file for the new
10441 built-in functions.
10442 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10443 define_insn for the instructions
10445 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10448 * gcc/config/netbsd-protos.h: New file.
10449 * gcc/config/netbsd.c: New file.
10450 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10451 * gcc/config/t-netbsd: New file.
10452 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10453 (tmake_file) Add t-netbsd.
10454 (extra_objs) Add netbsd.o.
10456 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10460 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10462 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10464 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10465 sign extension from a vector register to a GPR by doing a 32-bit
10466 direct move and then an EXTSW.
10467 (extendsi<mode>2 splitter): Likewise.
10468 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10469 right or vector extract after doing XSCVDPSPN. Use
10470 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10472 (movdi_from_sf_zero_ext): Likewise.
10473 (reload_gpr_from_vsxsf): Likewise.
10474 (p8_mfvsrd_4_disf): Delete, no longer used.
10475 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10476 then needing to move the SFmode to a GPR to use the XSCVDPSP
10477 instruction instead of FRSP and XSCVDPSPN.
10478 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10479 it is adjacent to the other XSCVSPDP insns.
10480 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10481 SFmode to be in traditional Altivec registers.
10482 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10483 (vsx_xscvspdpn): Likewise.
10484 (vsx_xscvspdpn_scalar): Likewise.
10486 2017-09-26 Martin Jambor <mjambor@suse.cz>
10488 * tree-sra.c (compare_access_positions): Put integral types first,
10489 stabilize sorting of integral types, remove conditions putting
10490 non-full-precision integers last.
10491 (sort_and_splice_var_accesses): Disable scalarization if a
10492 non-integert would be represented by a non-full-precision integer.
10494 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10496 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10497 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10498 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10499 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10500 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10501 conditionals inside the function instead of around it. Call
10502 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10503 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10505 2017-09-26 Richard Biener <rguenther@suse.de>
10507 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10509 (scop_detection::build_scop_breadth): ... this. Removed.
10510 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10511 (scop_detection::harmful_stmt_in_bb): Likewise.
10512 (scop_detection::graphite_can_represent_stmt): Likewise.
10513 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10514 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10515 (scop_detection::can_represent_loop_1): ... this. Removed.
10516 (scop_detection::harmful_loop_in_region): Simplify after inlining
10517 the above and remove more quadraticness.
10518 (build_scops): Adjust.
10519 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10522 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10525 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10526 REGNO (base) == SP_REG if base is a REG.
10528 PR middle-end/35691
10529 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10530 if it is different SSA_NAME.
10531 (optimize_range_tests_cmp_bitwise): New function.
10532 (optimize_range_tests): Call it.
10534 2017-09-26 Richard Biener <rguenther@suse.de>
10536 PR tree-optimization/82321
10537 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10538 for the def being inside the loop.
10540 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10542 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10544 * config/s390/s390-builtins.def: Fix constraint on op4.
10546 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10548 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10549 independent expanders.
10550 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10551 ("vec_ordered", "vec_unordered"): New expanders.
10553 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10555 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10558 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10560 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10561 vec_unpacks_lo_v16qi.
10562 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10564 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10566 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10567 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10568 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10570 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10572 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10574 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10576 ("*vec_slb<mode>"): New insn pattern.
10577 ("vec_shr_<mode>"): New expander.
10578 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10579 and force the shift count operand to V16QImode.
10580 ("vec_srb<mode>"): Set shift count mode to V16QI.
10582 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10584 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10585 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10586 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10588 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10591 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10593 2017-09-26 Richard Biener <rguenther@suse.de>
10595 PR tree-optimization/82320
10596 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10599 2017-09-25 Jeff Law <law@redhat.com>
10601 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10602 prototype for new argument.
10603 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10604 mostly extracted from rs6000_emit_allocate_stack.
10605 (rs6000_emit_probe_stack_range_stack_clash): New function.
10606 (rs6000_emit_allocate_stack): Call
10607 rs6000_emit_probe_stack_range_stack_clash as needed.
10608 (rs6000_emit_probe_stack_range): Add additional argument
10609 to call to gen_probe_stack_range{si,di}.
10610 (output_probe_stack_range): New.
10611 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10612 (output_probe_stack_range_stack_clash): New.
10613 (rs6000_emit_prologue): Emit notes into dump file as requested.
10614 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10615 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10616 Add additional operand and pass it to output_probe_stack_range.
10618 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10620 PR tree-optimization/82163
10621 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10622 (checking_verify_loop_closed_ssa): New parameter.
10623 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10624 (check_loop_closed_ssa_stmt): Delete.
10625 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10626 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10627 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10630 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10632 * brig-builtins.def: Treat HSAIL barrier builtins as
10633 setjmp/longjump style functions.
10635 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10637 * target.def (constant_alignment): New hook.
10638 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10639 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10640 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10641 * doc/tm.texi: Regenerate.
10642 * targhooks.h (default_constant_alignment): Declare.
10643 (constant_alignment_word_strings): Likewise.
10644 * targhooks.c (default_constant_alignment): New function.
10645 (constant_alignment_word_strings): Likewise.
10646 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10647 instead of CONSTANT_ALIGNMENT.
10648 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10649 (force_const_mem): Likewise.
10650 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10651 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10652 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10653 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10654 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10656 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10657 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10658 constant_alignment_word_strings.
10659 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10660 (CONSTANT_ALIGNMENT): Likewise.
10661 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10662 (arm_constant_alignment): New function.
10663 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10664 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10665 constant_alignment_word_strings.
10666 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10667 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10668 constant_alignment_word_strings.
10669 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10670 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10671 (cris_constant_alignment): New function.
10672 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10673 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10674 (epiphany_constant_alignment): New function.
10675 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10676 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10677 constant_alignment_word_strings.
10678 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10679 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10680 constant_alignment_word_strings.
10681 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10682 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10683 constant_alignment_word_strings.
10684 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10685 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10686 * config/i386/i386.c (ix86_constant_alignment): Make static.
10687 Use the same interface as the target hook.
10688 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10689 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10690 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10691 constant_alignment_word_strings.
10692 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10693 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10694 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10695 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10696 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10697 constant_alignment_word_strings.
10698 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10699 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10700 constant_alignment_word_strings.
10701 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10702 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10703 constant_alignment_word_strings.
10704 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10705 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10707 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10708 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10709 * config/mips/mips.c (mips_constant_alignment): New function.
10710 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10711 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10712 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10713 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10714 (mmix_constant_alignment): Make static. Use the same interface
10715 as the target hook.
10716 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10717 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10718 constant_alignment_word_strings.
10719 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10720 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10721 constant_alignment_word_strings.
10722 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10723 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10724 constant_alignment_word_strings.
10725 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10726 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10727 (rs6000_constant_alignment): New function.
10728 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10729 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10730 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10731 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10732 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10733 (rs6000_constant_alignment): New function.
10734 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10735 * config/s390/s390.c (s390_constant_alignment): New function.
10736 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10737 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10738 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10739 constant_alignment_word_strings.
10740 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10741 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10742 (sparc_constant_alignment): New function.
10743 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10744 * config/spu/spu.c (spu_constant_alignment): New function.
10745 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10746 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10747 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10748 constant_alignment_word_strings.
10749 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10750 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10751 constant_alignment_word_strings.
10752 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10753 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10754 constant_alignment_word_strings.
10755 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10756 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10757 (visium_constant_alignment): New function.
10758 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10759 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10760 (xtensa_constant_alignment): New function.
10761 * system.h (CONSTANT_ALIGNMENT): Poison.
10763 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10765 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10766 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10767 (rs6000_builtin_valid_without_lhs): New helper function.
10768 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10769 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10771 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10773 * target.h (vec_perm_indices): Use unsigned short rather than
10775 (auto_vec_perm_indices): Likewise.
10776 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10777 Use unsigned int rather than unsigned char.
10778 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10780 2017-09-25 Richard Biener <rguenther@suse.de>
10782 * cfgloop.h (sort_sibling_loops): Declare.
10783 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10784 (sort_sibling_loops): New function sorting the sibling loop list
10786 * graphite.c (graphite_transform_loops): Sort sibling loops.
10788 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10790 * target.def (vec_perm_const_ok): Change sel parameter to
10792 * optabs-query.c (can_vec_perm_p): Update accordingly.
10793 * doc/tm.texi: Regenerate.
10794 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10795 auto_vec_perm_indices and remove separate nelt field.
10796 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10797 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10798 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10799 (aarch64_expand_vec_perm_const): Update accordingly.
10800 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10801 to vec_perm_indices.
10802 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10803 auto_vec_perm_indices and remove separate nelt field.
10804 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10805 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10806 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10808 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10809 to vec_perm_indices.
10810 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10811 sel to vec_perm_indices.
10812 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10813 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10814 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10816 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10819 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10822 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10823 on the FUNCTION_DECL function context if it has a DIE that is a
10826 2017-09-25 Richard Biener <rguenther@suse.de>
10828 PR tree-optimization/82285
10829 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10832 2017-09-25 Tom de Vries <tom@codesourcery.com>
10836 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10839 2017-09-25 Richard Biener <rguenther@suse.de>
10841 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10842 ISL errors other than isl_error_quota happen. Dump if the
10843 schedule is the same.
10844 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10845 errors instead of aborting inside ISL.
10847 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10850 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10851 of libgcc_eh for m64.
10852 * config/i386/darwin64.h: Likewise.
10854 2017-09-25 Richard Biener <rguenther@suse.de>
10856 PR middle-end/82144
10857 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10858 attribute for incomplete types nor twice for complete ones.
10860 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10863 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10864 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10867 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10870 * config/i386/i386.opt (mprefer-avx256): Use
10871 ix86_target_flags variable.
10872 * config/i386/i386.c (ix86_target_string): Move
10873 -mprefer-avx256 to flag2_opts.
10875 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10877 PR middle-end/35691
10878 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10879 and x != -1 | y != -1 into (x & y) != -1.
10881 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10883 * config.gcc: Add new case statement to set
10884 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10885 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10886 s390x-*-linux* case statements. Added aarch64 to the list of
10887 supported architectures.
10889 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10891 PR tree-optimization/82289
10892 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10893 STMT_VINFO_RELEVANT_P.
10895 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10896 Alan Hayward <alan.hayward@arm.com>
10897 David Sherwood <david.sherwood@arm.com>
10899 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10900 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10901 (extract_range_from_binary_expr_1): Don't call
10902 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10904 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10905 Alan Hayward <alan.hayward@arm.com>
10906 David Sherwood <david.sherwood@arm.com>
10908 * target.def (preferred_vector_alignment): New hook.
10909 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
10911 * doc/tm.texi: Regenerate.
10912 * targhooks.h (default_preferred_vector_alignment): Declare.
10913 * targhooks.c (default_preferred_vector_alignment): New function.
10914 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
10916 (DR_TARGET_ALIGNMENT): New macro.
10917 (aligned_access_p): Update commentary.
10918 (vect_known_alignment_in_bytes): New function.
10919 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
10921 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
10922 Calculate the misalignment based on the target alignment rather than
10924 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
10925 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
10926 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
10927 the target alignment, rather than masking the element misalignment
10928 with the number of elements in a vector. Also use the target
10929 alignment when calculating the maximum number of peels.
10930 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
10931 instead of TYPE_ALIGN_UNIT.
10932 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
10933 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
10934 (vect_create_addr_base_for_vector_ref): Update call accordingly.
10935 (vect_create_data_ref_ptr): Likewise.
10936 (vect_setup_realignment): Realign by ANDing with
10937 -DR_TARGET_MISALIGNMENT.
10938 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
10939 the number of peels based on DR_TARGET_ALIGNMENT.
10940 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
10941 with the guaranteed alignment boundary when deciding whether
10943 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
10944 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
10945 (ensure_base_align): Remove stmt_info parameter. Get the
10946 target base alignment from DR_TARGET_ALIGNMENT.
10947 (vectorizable_store): Update call accordingly. Interpret
10948 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
10950 (vectorizable_load): Likewise.
10952 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10953 Alan Hayward <alan.hayward@arm.com>
10954 David Sherwood <david.sherwood@arm.com>
10956 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
10957 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
10958 (vect_enhance_data_refs_alignment): Likewise.
10960 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10962 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10963 error. Only quit immediately if parsing is complete.
10964 (BEGIN): Initialize fatal_err and parse_done.
10965 (begin fpu, end fpu): Check number of arguments.
10966 (begin arch, end arch): Likewise.
10967 (begin cpu, end cpu): Likewise.
10968 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10969 (optalias): Likewise.
10971 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10973 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10974 * config/arm/arm-isa.h: Delete. Move definitions to ...
10975 * arm-cpus.in: ... here. Use new feature and fgroup values.
10976 * config/arm/arm.c (arm_option_override): Use lower case for feature
10978 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10979 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10980 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10982 (print_isa_bits_for): New function.
10983 (gen_isa): New function.
10984 (gen_comm_data): Use print_isa_bits_for.
10985 (define feature): New keyword.
10986 (define fgroup): New keyword.
10987 * config/arm/t-arm (TM_H): Remove.
10988 (GTM_H): Add arm-isa.h.
10989 (arm-isa.h): Add rule to generate file.
10990 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10991 case for feature bit names.
10993 2017-09-22 Richard Biener <rguenther@suse.de>
10995 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
10997 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
10998 print of no dependency loops ...
10999 * graphite.c (graphite_transform_loops): ... here.
11000 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11002 (same_close_phi_node, remove_duplicate_close_phi,
11003 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11004 (canonicalize_loop_closed_ssa): ... here and simplify.
11005 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11006 (optimize_isl): Use dump_printf_loc to tell when we stopped
11007 optimizing because of an ISL timeout.
11009 2017-09-22 Richard Biener <rguenther@suse.de>
11011 PR tree-optimization/82291
11012 * tree-if-conv.c (predicate_mem_writes): Make sure to
11013 remove writes in blocks predicated with false.
11015 2017-09-22 Richard Biener <rguenther@suse.de>
11017 * sese.c: Include cfganal.h.
11018 (if_region_set_false_region): Remove.
11019 (create_if_region_on_edge): Likewise.
11020 (move_sese_in_condition): Re-implement without destroying
11023 2017-09-22 Richard Biener <rguenther@suse.de>
11025 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11026 Verify both BBs contain loop PHI nodes before dispatching to
11027 copy_loop_phi_args.
11028 (graphite_regenerate_ast_isl): Do not recompute dominators,
11029 do not verify three times. Restructure for clarity.
11030 * graphite-scop-detection.c (same_close_phi_node,
11031 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11032 defined_in_loop_p, canonicalize_loop_closed_ssa,
11033 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11034 checking and SSA rewrite, move to ...
11035 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11036 (graphite_initialize): Do not pass in ctx, do not reset the
11037 SCEV cache, compute only dominators.
11038 (graphite_transform_loops): Allocate ISL ctx after
11039 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11040 Maintain post-dominators only around build_scops.
11041 * sese.c (if_region_set_false_region): Make static. Free
11042 and recompute dominators.
11043 (move_sese_in_condition): Assert we don't get called with
11044 post-dominators computed.
11045 * sese.h (if_region_set_false_region): Remove.
11047 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11049 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11050 mode attribute for TARGET_AVX512VL.
11052 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11054 * config/i386/i386.opt (mprefer-avx256): New option.
11055 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11057 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11058 for TARGET_PREFER_AVX256.
11059 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11061 2017-09-21 Jeff Law <law@redhat.com>
11063 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11064 Fix dump output if the only stack space is for pushed registers.
11066 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11068 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11071 2017-09-21 Martin Sebor <msebor@redhat.com>
11074 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11075 code (in C++) or code that triggers warnings.
11077 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11079 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11081 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11083 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11084 * sched-rgn.c: Ditto.
11085 * sel-sched-ir.c: Ditto.
11087 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11089 * toplev.h (set_random_seed): Adjust return type.
11090 * toplev.c (init_local_tick): Move eager initialization of random_seed
11091 to get_random_seed. Adjust comment.
11092 (init_random_seed): Inline to get_random_seed, delete.
11093 (get_random_seed): Initialize random_seed lazily.
11094 (set_random_seed): Do not return previous value.
11095 (print_switch_value): Do not call get_random_seed.
11097 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11099 * cgraph.c (delete_function_version): New, broken out from...
11100 (cgraph_node::delete_function_version): ...here. Rename to
11101 cgraph_node::delete_function_version_by_decl. Update all uses.
11102 (cgraph_node::remove): Call delete_function_version.
11104 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11107 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11108 VAR_DECLs to which addressable non-volatile parameters are mapped
11109 and for id->retvar after the return value assignment. Clear
11110 id->retval and id->retbnd after inlining.
11112 2017-09-21 Richard Biener <rguenther@suse.de>
11114 PR tree-optimization/82276
11115 PR tree-optimization/82244
11116 * tree-vrp.c (build_assert_expr_for): Set
11117 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11119 (remove_range_assertions): Revert earlier change.
11121 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11124 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11126 2017-09-21 Richard Biener <rguenther@suse.de>
11128 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11129 Restore valid IL after code generation errors.
11130 * graphite.c (graphite_transform_loops): Diagnose code
11131 generation issues as MSG_MISSED_OPTIMIZATION and continue
11132 with processing SCOPs.
11134 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11135 Alan Hayward <alan.hayward@arm.com>
11136 David Sherwood <david.sherwood@arm.com>
11138 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11139 rather than choosing between plus_constant and gen_rtx_<CODE>.
11140 * expr.c (emit_push_insn): Likewise.
11141 (expand_expr_real_2): Likewise.
11143 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11144 Alan Hayward <alan.hayward@arm.com>
11145 David Sherwood <david.sherwood@arm.com>
11147 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11149 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11150 Alan Hayward <alan.hayward@arm.com>
11151 David Sherwood <david.sherwood@arm.com>
11153 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11154 calling tree_to_uhwi.
11156 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11157 Alan Hayward <alan.hayward@arm.com>
11158 David Sherwood <david.sherwood@arm.com>
11160 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11161 INTEGER_CST rather than a negative test for ADDR_EXPR.
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-vrp.c (extract_range_from_binary_expr_1): Check
11168 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11170 2017-09-21 Richard Biener <rguenther@suse.de>
11172 PR tree-optimization/71351
11173 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11174 graphite_create_new_loop_guard): Remove, fold remaining parts
11176 (translate_isl_ast_node_for): ... here and simplify.
11178 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11181 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11182 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11183 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11184 alternative always use QI mode, for -Os imov (=R,R) alternative
11185 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11188 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11189 Jeff Law <law@redhat.com>
11191 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11192 (allocate_stack_space): New function, partially extracted from
11193 s390_emit_prologue.
11194 (s390_emit_prologue): Track offset to most recent stack probe.
11195 Code to allocate space moved into allocate_stack_space.
11196 Dump actions when no stack is allocated.
11197 (s390_prologue_plus_offset): New function.
11198 (s390_emit_stack_probe): Likewise.
11200 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11202 * common.opt (Wa, Wl, Wp, g, gz=): Add
11204 (gno-column-info): Remove.
11205 (gcolumn-info): Drop RejectNegative.
11206 (gno-): New prefix.
11207 (gno-record-gcc-switches): Remove.
11208 (grecord-gcc-switches): Drop RejectNegative.
11209 (gno-split-dwarf): Remove.
11210 (gsplit-dwarf): Drop RejectNegative.
11211 (gno-strict-dwarf): Remove.
11212 (gstrict-dwarf): Drop RejectNegative.
11213 * config/darwin.opt (gfull, gused): Add RejectNegative.
11214 * dwarf2out.c (gen_producer_string): Drop
11215 gno-record-gcc-switches handler.
11216 * optc-gen.awk: Add g to prefixes with negative forms.
11217 * opts-common.c (remapping_prefix_p): New.
11218 (find_opt): Check it.
11219 (generate_canonical_option): Test g prefix.
11220 (option_map): Add -gno- mapping.
11221 (add_misspelling_candidates): Check remapping_prefix_p.
11223 2017-09-20 Jeff Law <law@redhat.com>
11225 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11226 thinko in stack clash protection support.
11228 * explow.c (compute_stack_clash_protection_loop_data): Use
11229 CONST_INT_P instead of explicit test. Verify object is a
11230 CONST_INT_P before looking at INTVAL.
11231 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11232 instead of explicit test.
11234 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11237 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11238 address instead of to r1 and r11.
11240 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11242 * config.gcc: Support "knm".
11243 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11244 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11246 * config/i386/i386.c (m_KNM): Define.
11247 (processor_target_table): Add "knm".
11249 (ix86_option_override_internal): Add "knm".
11250 (ix86_issue_rate): Add PROCESSOR_KNM.
11251 (ix86_adjust_cost): Ditto.
11252 (ia32_multipass_dfa_lookahead): Ditto.
11253 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11254 (fold_builtin_cpu): Add M_INTEL_KNM.
11255 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11256 (processor_type): Add PROCESSOR_KNM.
11257 * config/i386/x86-tune.def: Add m_KNM.
11258 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11260 2017-09-20 Richard Biener <rguenther@suse.de>
11262 PR tree-optimization/80213
11263 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11264 are allowed in empty BBs as well.
11265 (canonicalize_loop_closed_ssa): Also look for other complex
11267 (scop_detection::get_sese): Include the loop-closed PHI block
11269 (scop_detection::merge_sese): Remove code adding extra blocks.
11270 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11271 (build_scops): Assert the final returned scop is invalid.
11273 2017-09-20 Richard Biener <rguenther@suse.de>
11275 PR tree-optimization/82264
11276 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11278 (vn_phi_lookup): Likewise.
11279 (vn_phi_insert): Likewise.
11281 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11283 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11284 that fits into uhwi or shwi, add DW_AT_const_value regardless
11285 of early_dwarf without going through RTL, using add_AT_unsigned
11288 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11289 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11290 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11291 (DEBUG_MACRO_SECTION): Likewise.
11292 (DEBUG_LTO_MACRO_SECTION): Likewise.
11293 (DEBUG_STR_DWO_SECTION): Likewise.
11294 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11295 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11296 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11297 (DEBUG_LTO_LINE_STR_SECTION): Define.
11298 (init_sections_and_labels): Initialize debug_line_str_section
11299 variable. Initialize debug_loc_section for -gdwarf-5 to
11300 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11302 2017-09-20 Richard Biener <rguenther@suse.de>
11304 * graphite-sese-to-poly.c (extract_affine): Properly handle
11305 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11307 2017-09-20 Richard Biener <rguenther@suse.de>
11309 PR tree-optimization/81373
11310 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11311 Force SESE live-out defs to be handled even if they are
11314 2017-09-19 Jeff Law <law@redhat.com>
11316 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11317 nothing for stack adjustments with REG_STACK_CHECK.
11318 * sched-deps.c (parse_add_or_inc): Reject insns with
11319 REG_STACK_CHECK from dependency breaking.
11320 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11321 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11322 * reg-notes.def (STACK_CHECK): New note.
11324 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11325 (ix86_expand_prologue): Dump stack clash info as needed.
11326 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11328 * function.c (dump_stack_clash_frame_info): New function.
11329 * function.h (dump_stack_clash_frame_info): Prototype.
11330 (enum stack_clash_probes): New enum.
11332 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11333 flag_stack_clash_protection.
11334 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11335 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11336 (arm_frame_pointer_required): Likewise.
11337 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11338 (ia64_expand_prologue): Likewise.
11339 * config/mips/mips.c (mips_expand_prologue): Likewise.
11340 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11341 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11342 (sparc_flat_expand_prologue): Likewise.
11343 * config/spu/spu.c (spu_expand_prologue): Likewise.
11345 * explow.c: Include "params.h".
11346 (anti_adjust_stack_and_probe_stack_clash): New function.
11347 (get_stack_check_protect): Likewise.
11348 (compute_stack_clash_protection_loop_data): Likewise.
11349 (emit_stack_clash_protection_loop_start): Likewise.
11350 (emit_stack_clash_protection_loop_end): Likewise.
11351 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11352 Use anti_adjust_stack_and_probe_stack_clash.
11353 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11354 (emit_stack_clash_protection_loop_start): Likewise.
11355 (emit_stack_clash_protection_loop_end): Likewise.
11356 * rtl.h (get_stack_check_protect): Prototype.
11357 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11358 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11359 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11361 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11363 * doc/tm.texi: Rebuilt.
11364 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11365 get_stack_check_protect.
11366 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11367 * config/arm/arm.c (arm_expand_prologue): Likewise.
11368 (arm_frame_pointer_required): Likewise.
11369 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11370 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11371 * config/mips/mips.c (mips_expand_prologue): Likewise.
11372 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11373 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11374 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11375 (sparc_flat_expand_prologue): Likewise.
11377 * common.opt (-fstack-clash-protection): New option.
11378 * flag-types.h (enum stack_check_type): Note difference between
11379 -fstack-check= and -fstack-clash-protection.
11380 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11381 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11382 * toplev.c (process_options): Issue warnings/errors for cases
11383 not handled with -fstack-clash-protection.
11384 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11385 (-fstack-check): Note additional problem with -fstack-check=generic.
11386 Note that -fstack-check is primarily for Ada and refer users
11387 to -fstack-clash-protection for stack-clash-protection.
11388 Document new params for stack clash protection.
11390 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11392 * config/i386/i386.c (ix86_split_long_move): Do not handle
11393 address used for LEA in a special way.
11395 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11397 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11399 2017-09-19 Martin Sebor <msebor@redhat.com>
11402 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11403 of incompatible types.
11405 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11407 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11408 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11409 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11410 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11412 2017-09-19 Richard Biener <rguenther@suse.de>
11414 PR tree-optimization/82244
11415 * tree-vrp.c (remove_range_assertions): Do not propagate
11416 a constant to abnormals but replace the assert with a copy.
11418 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11420 PR rtl-optimization/57878
11421 PR rtl-optimization/68988
11422 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11423 avoidance test involving non_reload_pseudos. Move frequency test
11424 below the general fragmentation avoidance test.
11426 2017-09-19 Richard Biener <rguenther@suse.de>
11428 PR tree-optimization/69728
11429 * graphite-sese-to-poly.c (schedule_error): New global.
11430 (add_loop_schedule): Handle empty domain by failing the
11432 (build_original_schedule): Handle schedule_error.
11434 2017-09-19 Richard Biener <rguenther@suse.de>
11436 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11437 Do not iterate to sibling loops but only to siblings of inner
11440 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11443 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11444 that the registers are different.
11446 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11448 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11449 to processor_model and "amdfam17h" to arch_names_table.
11450 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11452 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11455 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11457 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11458 Alan Hayward <alan.hayward@arm.com>
11459 David Sherwood <david.sherwood@arm.com>
11461 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11463 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11465 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11466 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11467 vect_schedule_slp_instance.
11468 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11469 Update call to vect_slp_analyze_node_operations. Simplify return
11471 (vect_slp_analyze_bb_1): Update call accordingly.
11472 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11473 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11474 (vect_schedule_slp): Update call accordingly.
11476 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11477 Alan Hayward <alan.hayward@arm.com>
11478 David Sherwood <david.sherwood@arm.com>
11480 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11481 with types that aren't in fact scalar.
11483 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11485 * tree-vect-slp.c (vect_record_max_nunits): New function,
11487 (vect_build_slp_tree_1): ...here.
11488 (vect_build_slp_tree_2): Call it for phis too.
11490 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11492 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11493 to vect_get_vec_def_for_operand when getting the mask operand.
11495 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11496 Alan Hayward <alan.hayward@arm.com>
11497 David Sherwood <david.sherwood@arm.com>
11499 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11502 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11503 Alan Hayward <alan.hayward@arm.com>
11504 David Sherwood <david.sherwood@arm.com>
11506 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11507 calculation for vector booleans.
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-stmts.c (can_vectorize_live_stmts): New function,
11515 (vect_transform_stmt): ...here.
11516 (vect_analyze_stmt): Use it instead of calling
11517 vectorizable_live_operation directly.
11519 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11521 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11522 non-SIMT targets in acc vector loops.
11524 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11526 * configure.ac: Add arc and check if assembler supports gdwarf2.
11527 * configure: Regenerate.
11529 2017-09-18 Richard Biener <rguenther@suse.de>
11531 PR tree-optimization/82220
11532 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11533 epilogue niters from the min_profitable_iters compute.
11535 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11538 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11539 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11540 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11542 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11545 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11546 switching to a new text section.
11548 2017-09-18 Richard Biener <rguenther@suse.de>
11550 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11552 (build_alias_set): Reject aliases with no access function.
11554 2017-09-18 Richard Biener <rguenther@suse.de>
11556 PR tree-optimization/79622
11557 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11559 (build_cross_bb_scalars_use): Likewise.
11561 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11563 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11565 2017-09-18 Alan Modra <amodra@gmail.com>
11568 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11569 stack_pointer_rtx for count 0. Update comments. Break up
11570 large rtl expression.
11572 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11574 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11575 Increase to 20 bytes.
11576 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11577 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11578 or avx version of the stub.
11580 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11583 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11584 compute the minimum stack alignment. Also update preferred stack
11585 boundary for leaf functions.
11587 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11589 PR tree-optimization/82228
11590 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11593 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11595 * common/config/nds32/nds32-common.c
11596 (nds32_option_optimization_table): Refine formatting.
11597 (nds32_option_optimization_table): Use -fsched-pressure and
11598 -fomit-frame-pointer for specific optimization level.
11600 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11602 * config/nds32/nds32.c: Refine formatting and comments.
11603 * config/nds32/nds32.h: Likewise.
11604 * config/nds32/nds32.md: Likewise.
11605 * config/nds32/nds32-cost.c: Likewise.
11606 * config/nds32/nds32-isr.c: Likewise.
11607 * config/nds32/nds32-md-auxiliary.c: Likewise.
11608 * config/nds32/nds32-multiple.md: Likewise.
11609 * config/nds32/nds32-predicates.c: Likewise.
11611 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11612 Jakub Jelinek <jakub@redhat.com>
11614 Add support for -std=c++2a.
11615 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11617 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11619 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11622 * doc/extend.texi (Common Function Attributes): Add
11623 references to ARM, AArch64, and S/390 specific attributes.
11624 (Function Specific Option Pragmas): Add AArch64 and S/390
11625 to list of back ends that support the target pragma.
11627 2017-09-15 Nathan Sidwell <nathan@acm.org>
11629 * doc/standards.texi: Fix C++17 description. Update URLs for
11632 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11634 * common.opt (Wcast-align=strict): New warning option.
11635 * doc/invoke.texi: Document -Wcast-align=strict.
11637 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11639 * cgraph.h (cgraph_thunk_info): Add comments.
11640 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11641 assert for VIRTUAL_* arguments stricter.
11643 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11645 PR tree-optimization/71026
11646 * match.pd: Move RDIV patterns from fold-const.c
11647 * fold-const.c (distribute_real_division): Removed.
11648 (fold_binary_loc): Remove calls to distribute_real_divison.
11650 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11652 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11653 c++1z and gnu++1z as deprecated. Change other references to
11654 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11655 Change -Wc++1z-compat to -Wc++17-compat.
11656 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11657 * dwarf2out.c (highest_c_language): Handle C++17.
11658 (gen_compile_unit_die): Likewise.
11660 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11662 PR rtl-optimization/82192
11663 * combine.c (make_extraction): Don't look through non-paradoxical
11664 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11667 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11668 Alan Hayward <alan.hayward@arm.com>
11669 David Sherwood <david.sherwood@arm.com>
11671 * target.def (function_arg_offset): New hook.
11672 * targhooks.h (default_function_arg_offset): Declare.
11673 * targhooks.c (default_function_arg_offset): New function.
11674 * function.c (locate_and_pad_parm): Use
11675 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11676 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11677 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11678 * doc/tm.texi: Regenerate.
11679 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11680 * config/spu/spu.c (spu_function_arg_offset): New function.
11681 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11682 * system.h (FUNCTION_ARG_OFFSET): Poison.
11684 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11685 Alan Hayard <alan.hayward@arm.com>
11686 David Sherwood <david.sherwood@arm.com>
11688 * target.def (truly_noop_truncation): New hook.
11689 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11690 than TRULY_NOOP_TRUNCATION.
11691 * hooks.h (hook_bool_uint_uint_true): Declare.
11692 * hooks.c (hook_bool_uint_uint_true): New function.
11693 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11694 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11695 * doc/tm.texi: Regenerate.
11696 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11697 rather than TRULY_NOOP_TRUNCATION in comments.
11698 (simplify_comparison): Likewise.
11699 (record_truncated_value): Likewise.
11700 * expmed.c (extract_bit_field_1): Likewise.
11701 (extract_split_bit_field): Likewise.
11702 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11703 instead of TRULY_NOOP_TRUNCATION.
11704 * function.c (assign_parm_setup_block): Likewise.
11705 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11706 * rtlhooks.c: Include target.h.
11707 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11708 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11709 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11710 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11711 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11712 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11713 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11714 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11715 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11716 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11717 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11718 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11719 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11720 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11721 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11722 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11723 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11724 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11725 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11726 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11727 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11728 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11729 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11730 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11731 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11732 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11733 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11734 rather than TRULY_NOOP_TRUNCATION in comments.
11735 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11736 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11737 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11738 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11739 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11740 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11741 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11742 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11743 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11744 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11745 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11746 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11747 rather than TRULY_NOOP_TRUNCATION in comments.
11748 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11749 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11750 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11751 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11752 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11753 TRULY_NOOP_TRUNCATION condition.
11754 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11755 (TRULY_NOOP_TRUNCATION): Delete.
11756 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11757 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11758 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11759 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11760 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11761 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11762 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11763 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11764 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11765 rather than TRULY_NOOP_TRUNCATION in comments.
11766 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11767 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11768 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11769 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11770 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11771 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11773 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11776 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11777 (*cmp_ior): Likewise.
11778 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11779 (*ior_scc_scc_cmp): Likewise.
11780 (*and_scc_scc): Likewise.
11781 (*and_scc_scc_cmp): Likewise.
11783 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11784 Alan Hayard <alan.hayward@arm.com>
11785 David Sherwood <david.sherwood@arm.com>
11787 * target.def (can_change_mode_class): New hook.
11788 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11789 (hard_regno_nregs): Likewise.
11790 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11791 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11792 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11793 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11794 (LOAD_EXTEND_OP): Update accordingly.
11795 * doc/tm.texi: Regenerate.
11796 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11797 CANNOT_CHANGE_MODE_CLASS.
11798 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11799 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11800 * combine.c (simplify_set): Update accordingly.
11801 * emit-rtl.c (validate_subreg): Likewise.
11802 * recog.c (general_operand): Likewise.
11803 * regcprop.c (mode_change_ok): Likewise.
11804 * reload1.c (choose_reload_regs): Likewise.
11805 (inherit_piecemeal_p): Likewise.
11806 * rtlanal.c (simplify_subreg_regno): Likewise.
11807 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11808 instead of CANNOT_CHANGE_MODE_CLASS.
11809 (reload_cse_simplify_operands): Likewise.
11810 * reload.c (push_reload): Use targetm.can_change_mode_class
11811 instead of CANNOT_CHANGE_MODE_CLASS.
11812 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11813 REG_CANNOT_CHANGE_MODE_P.
11814 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11815 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11816 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11817 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11818 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11819 (arm_can_change_mode_class): New function.
11820 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11821 than CANNOT_CHANGE_MODE_CLASS in comments.
11822 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11823 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11824 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11825 (ix86_can_change_mode_class): ...this new function, inverting the
11826 sense of the return value.
11827 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11828 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11829 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11830 (ia64_can_change_mode_class): New function.
11831 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11832 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11833 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11834 (m32c_can_change_mode_class): ...this new function, inverting the
11835 sense of the return value.
11836 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11837 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11838 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11839 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11840 (mips_can_change_mode_class): ...this new function, inverting the
11841 sense of the return value.
11842 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11843 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11844 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11845 (msp430_can_change_mode_class): New function.
11846 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11847 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11848 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11849 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11850 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11851 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11852 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11853 (pa_cannot_change_mode_class): Replace with...
11854 (pa_can_change_mode_class): ...this new function, inverting the
11855 sense of the return value.
11856 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11857 than CANNOT_CHANGE_MODE_CLASS in comments.
11858 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11859 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11860 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11861 (pdp11_cannot_change_mode_class): Replace with...
11862 (pdp11_can_change_mode_class): ...this new function, inverting the
11863 sense of the return value.
11864 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11865 * config/powerpcspe/powerpcspe-protos.h
11866 (rs6000_cannot_change_mode_class_ptr): Delete.
11867 * config/powerpcspe/powerpcspe.c
11868 (rs6000_cannot_change_mode_class_ptr): Delete.
11869 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11870 (rs6000_option_override_internal): Assign to
11871 targetm.can_change_mode_class instead of
11872 rs6000_cannot_change_mode_class_ptr.
11873 (rs6000_cannot_change_mode_class): Replace with...
11874 (rs6000_can_change_mode_class): ...this new function, inverting the
11875 sense of the return value.
11876 (rs6000_debug_cannot_change_mode_class): Replace with...
11877 (rs6000_debug_can_change_mode_class): ...this new function.
11878 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11879 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11880 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11881 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11882 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11884 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11885 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11886 (rs6000_option_override_internal): Assign to
11887 targetm.can_change_mode_class instead of
11888 rs6000_cannot_change_mode_class_ptr.
11889 (rs6000_cannot_change_mode_class): Replace with...
11890 (rs6000_can_change_mode_class): ...this new function, inverting the
11891 sense of the return value.
11892 (rs6000_debug_cannot_change_mode_class): Replace with...
11893 (rs6000_debug_can_change_mode_class): ...this new function.
11894 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11895 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11896 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11897 (s390_can_change_mode_class): ...this new function, inverting the
11898 sense of the return value.
11899 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11900 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11901 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11902 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11903 (sh_cannot_change_mode_class): Replace with...
11904 (sh_can_change_mode_class): ...this new function, inverting the
11905 sense of the return value.
11906 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11907 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11908 (sparc_can_change_mode_class): New function.
11909 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11910 * config/spu/spu.c (spu_can_change_mode_class): New function.
11911 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11912 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11913 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11914 (visium_can_change_mode_class): New function.
11915 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
11917 2017-09-15 Richard Biener <rguenther@suse.de>
11919 PR tree-optimization/82217
11920 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
11921 but not undefined case.
11923 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11926 * postreload.c (reload_cse_simplify_operands): Skip
11927 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
11929 2017-09-15 Richard Biener <rguenther@suse.de>
11931 PR tree-optimization/68823
11932 * graphite-scop-detection.c (build_alias_set): If we have a
11933 possible dependence check whether we can handle them by just
11934 looking at the DRs DR_ACCESS_FNs.
11935 (build_scops): If build_alias_set fails, fail the SCOP.
11937 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11939 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
11940 to support float128 built-in functions that require the ISA 3.0
11942 (BU_FLOAT128_3_HW): Likewise.
11943 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11944 built-in functions.
11945 (FMAF128): Likewise.
11947 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
11948 support for built-in functions that need the ISA 3.0 IEEE 128-bit
11949 floating point instructions.
11950 (rs6000_invalid_builtin): Likewise.
11951 (rs6000_builtin_mask_names): Likewise.
11952 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
11953 (RS6000_BTM_FLOAT128_HW): Likewise.
11954 (RS6000_BTM_COMMON): Likewise.
11955 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
11957 * doc/extend.texi (RS/6000 built-in functions): Document the
11958 IEEE 128-bit floating point square root and fused multiply-add
11959 built-in functions.
11961 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
11963 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
11964 reg (r2) isn't in the set of registers defined in the prologue.
11966 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11967 Alan Hayward <alan.hayward@arm.com>
11968 David Sherwood <david.sherwood@arm.com>
11970 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
11971 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
11972 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
11973 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
11974 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
11976 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
11977 max_vectorization_factor.
11978 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
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 (vect_min_worthwhile_factor): Delete.
11985 (vect_worthwhile_without_simd_p): Declare.
11986 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
11987 (vectorizable_reduction): Use it.
11988 * tree-vect-stmts.c (vectorizable_shift): Likewise.
11989 (vectorizable_operation): Likewise.
11991 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11992 Alan Hayward <alan.hayward@arm.com>
11993 David Sherwood <david.sherwood@arm.com>
11995 * tree-vectorizer.h (vect_get_num_copies): New function.
11996 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
11997 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11998 (vectorizable_induction): Likewise.
11999 (vectorizable_live_operation): Likewise.
12000 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12001 (vectorizable_bswap): Likewise.
12002 (vectorizable_call): Likewise.
12003 (vectorizable_conversion): Likewise.
12004 (vectorizable_assignment): Likewise.
12005 (vectorizable_shift): Likewise.
12006 (vectorizable_operation): Likewise.
12007 (vectorizable_store): Likewise.
12008 (vectorizable_load): Likewise.
12009 (vectorizable_condition): Likewise.
12010 (vectorizable_comparison): Likewise.
12011 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12013 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12014 Alan Hayward <alan.hayward@arm.com>
12015 David Sherwood <david.sherwood@arm.com>
12017 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12018 of vect_init_vector.
12020 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12021 Alan Hayward <alan.hayward@arm.com>
12022 David Sherwood <david.sherwood@arm.com>
12024 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12025 an inline wrapper that provides a location.
12026 (gimple_build_vector): Likewise.
12027 * gimple-fold.c (gimple_build_vector_from_val): New function.
12028 (gimple_build_vector): Likewise.
12029 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12030 functions to build the initial value. Always return a gimple value.
12031 (get_initial_defs_for_reduction): Likewise. Only compute
12033 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12034 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12035 (vectorizable_induction): Use gimple_build_vector rather than
12038 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12039 Alan Hayward <alan.hayward@arm.com>
12040 David Sherwood <david.sherwood@arm.com>
12042 * target.h (vec_perm_indices): New typedef.
12043 (auto_vec_perm_indices): Likewise.
12044 * optabs-query.h: Include target.h
12045 (can_vec_perm_p): Take a vec_perm_indices *.
12046 * optabs-query.c (can_vec_perm_p): Likewise.
12047 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12048 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12049 * tree-vect-generic.c (lower_vec_perm): Likewise.
12050 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12051 (vect_grouped_load_supported): Likewise.
12052 (vect_shift_permute_load_chain): Likewise.
12053 (vect_permute_store_chain): Use auto_vec_perm_indices.
12054 (vect_permute_load_chain): Likewise.
12055 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12056 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12057 Update uses of can_vec_perm_p.
12058 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12059 mode with a number of elements. Take a vec_perm_indices *.
12060 (vect_create_epilog_for_reduction): Update accordingly.
12061 Use auto_vec_perm_indices.
12062 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12063 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12064 (vect_transform_slp_perm_load): Likewise.
12065 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12066 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12067 (vect_gen_perm_mask_checked): Likewise.
12068 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12069 (vect_gen_perm_mask_checked): Likewise.
12070 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12071 (vectorizable_store): Likewise.
12072 (vectorizable_load): Likewise.
12073 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12074 (vectorizable_bswap): Likewise.
12076 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12077 Alan Hayward <alan.hayward@arm.com>
12078 David Sherwood <david.sherwood@arm.com>
12080 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12081 * tree.c (build_vector): Likewise.
12082 (build_vector_from_ctor): Update accordingly.
12083 (build_vector_from_val): Likewise.
12084 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12085 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12086 * tree-vect-generic.c (add_rshift): Likewise.
12087 (expand_vector_divmod): Likewise.
12088 (optimize_vector_constructor): Likewise.
12089 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12090 (vect_transform_slp_perm_load): Likewise.
12091 (vect_schedule_slp_instance): Likewise.
12092 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12093 (vectorizable_call): Likewise.
12094 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12095 * expmed.c (make_tree): Likewise.
12096 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12097 a vector passed to build_vector.
12098 (fold_convert_const): Likewise.
12099 (exact_inverse): Likewise.
12100 (fold_ternary_loc): Likewise.
12101 (fold_relational_const): Likewise.
12102 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12103 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12104 (const_unop): Likewise. Store the reduction accumulator in a
12105 variable rather than an array.
12106 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12107 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12108 the new vector, rather than constructing it after the input arrays.
12109 (native_interpret_vector): Use auto_vec<tree> when building
12110 a vector passed to build_vector. Add elements in order.
12111 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12112 auto_vec<tree> when building a vector passed to build_vector.
12113 (vect_create_epilog_for_reduction): Likewise.
12114 (vectorizable_induction): Likewise.
12115 (get_initial_def_for_reduction): Likewise. Fix indentation of
12117 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12119 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12120 passed to build_vector.
12122 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12123 Alan Hayward <alan.hayward@arm.com>
12124 David Sherwood <david.sherwood@arm.com>
12126 * tree-core.h (tree_base::u): Add an "nelts" field.
12127 (tree_vector): Use VECTOR_CST_NELTS as the length.
12128 * tree.c (tree_size): Likewise.
12129 (make_vector): Initialize VECTOR_CST_NELTS.
12130 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12131 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12132 TYPE_VECTOR_SUBPARTS.
12133 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12134 as the number of units, setting it from VECTOR_CST_NELTS.
12135 (const_vector_from_tree): Likewise.
12136 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12137 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12138 (fold_negate_expr_1): Likewise.
12139 (fold_convert_const): Likewise.
12140 (const_binop): Likewise. Differentiate the number of output and
12142 (const_unop): Likewise.
12143 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12144 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12145 in cases that did the opposite.
12147 2017-09-14 Richard Biener <rguenther@suse.de>
12149 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12152 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12154 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12156 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12159 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12160 if and where to split a bb, except for splitting before debug insn
12161 sequences followed by non-label real insn. Delete debug insns
12162 in between basic blocks.
12164 * combine.c (make_compound_operation_int): Formatting fixes.
12166 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12167 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12168 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12169 * config/sol2.h (LINK_EH_SPEC): Likewise.
12170 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12171 * config/s390/linux.h (LINK_SPEC): Likewise.
12172 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12173 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12174 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12175 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12176 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12177 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12178 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12179 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12181 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12183 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12185 (ENDFILE_LINUX_SPEC): Likewise.
12186 (LINK_EH_SPEC): Likewise.
12187 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12188 (LINK_OS_LINUX_SPEC32): Likewise.
12189 (LINK_OS_LINUX_SPEC64): Likewise.
12190 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12191 (LINK_OS_LINUX_SPEC): Likewise.
12193 2017-09-13 Martin Liska <mliska@suse.cz>
12195 PR middle-end/82154
12196 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12197 CASE_HIGH is NULL_TREE.
12199 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12200 Alan Hayward <alan.hayward@arm.com>
12201 David Sherwood <david.sherwood@arm.com>
12203 * target.def (secondary_memory_needed): New hook.
12204 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12205 instead of SECONDARY_MEMORY_NEEDED.
12206 (secondary_memory_needed_mode): Likewise.
12207 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12208 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12209 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12210 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12211 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12212 * doc/tm.texi: Regenerate.
12213 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12214 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12215 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12216 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12217 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12218 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12219 mode argument first and change the reg_class arguments to reg_class_t.
12220 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12221 Make static. Update the call to inline_secondary_memory_needed.
12222 (ix86_register_move_cost): Update the call to
12223 inline_secondary_memory_needed.
12224 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12225 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12227 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12228 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12230 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12231 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12232 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12233 and match hook interface. Add comment from mips.h.
12234 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12235 * config/mmix/mmix.md (truncdfsf2): Refer to
12236 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12238 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12239 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12240 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12241 (pa_secondary_memory_needed): New function.
12242 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12243 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12244 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12245 (pdp11_secondary_memory_needed): Make static and match hook interface.
12246 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12247 * config/powerpcspe/powerpcspe-protos.h
12248 (rs6000_secondary_memory_needed_ptr): Delete.
12249 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12251 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12252 (rs6000_option_override_internal): Assign to
12253 targetm.secondary_memory_needed rather than
12254 rs6000_secondary_memory_needed_ptr.
12255 (rs6000_secondary_memory_needed): Match hook interface.
12256 (rs6000_debug_secondary_memory_needed): Likewise.
12257 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12258 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12259 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12260 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12261 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12262 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12264 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
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/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12272 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12273 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12274 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12275 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12276 (sparc_secondary_memory_needed): New function.
12277 * lra-constraints.c (check_and_process_move): Refer to
12278 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12280 (curr_insn_transform): Likewise.
12281 (process_alt_operands): Use targetm.secondary_memory_needed
12282 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12283 (check_secondary_memory_needed_p): Likewise.
12284 (choose_split_class): Likewise.
12285 * reload.c: Unconditionally include code that was previously
12286 conditional on SECONDARY_MEMORY_NEEDED.
12287 (push_secondary_reload): Use targetm.secondary_memory_needed
12288 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12289 (push_reload): Likewise.
12290 * reload1.c: Unconditionally include code that was previously
12291 conditional on SECONDARY_MEMORY_NEEDED.
12292 (choose_reload_regs): Use targetm.secondary_memory_needed
12293 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12294 (gen_reload): Likewise.
12295 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12297 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12298 Alan Hayward <alan.hayward@arm.com>
12299 David Sherwood <david.sherwood@arm.com>
12301 * target.def (secondary_memory_needed_mode): New hook:
12302 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12303 * targhooks.h (default_secondary_memory_needed_mode): New function.
12304 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12305 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12306 * doc/tm.texi: Regenerate.
12307 * lra-constraints.c (check_and_process_move): Use
12308 targetm.secondary_memory_needed_mode instead of
12309 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12310 (curr_insn_transform): Likewise.
12311 * reload.c (get_secondary_mem): Likewise.
12312 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12313 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12315 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12316 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12317 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12318 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12319 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12321 * config/powerpcspe/powerpcspe-protos.h
12322 (rs6000_secondary_memory_needed_mode): Delete.
12323 * config/powerpcspe/powerpcspe.c
12324 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12325 (rs6000_secondary_memory_needed_mode): Make static.
12326 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12327 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12329 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12331 (rs6000_secondary_memory_needed_mode): Make static.
12332 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12333 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12334 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12335 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12336 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12338 (sparc_secondary_memory_needed_mode): New function.
12339 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12341 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12343 * config/aarch64/constraints.md (Umq): New constraint.
12344 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12346 (mov<mode>): Update condition.
12348 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12350 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12351 when bitposition is the same.
12353 2017-09-13 Richard Biener <rguenther@suse.de>
12355 * dwarf2out.c (output_die_symbol): Remove.
12356 (output_die): Do not output a DIEs symbol.
12358 2017-09-13 Richard Biener <rguenther@suse.de>
12360 PR middle-end/82128
12361 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12362 default-def to avoid breaking iterator update with the weird
12363 interaction with cgraph_update_edges_for_call_stmt_node.
12365 2017-09-13 Richard Biener <rguenther@suse.de>
12367 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12368 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12369 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12370 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12372 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12374 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12375 Disable pc relative literal load irrespective of
12376 TARGET_FIX_ERR_A53_84341 for default.
12378 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12380 * config/sparc/sparc.c (output_return): Output the source location of
12381 the insn in the delay slot, if any.
12382 (output_sibcall): Likewise.
12384 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12387 * common.opt (-static-pie): New alias.
12388 (shared): Negate static-pie.
12389 (-no-pie): Update help text.
12391 (static-pie): New option.
12392 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12393 -static-pie support.
12394 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12395 (LINK_EH_SPEC): Likewise.
12396 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12397 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12398 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12399 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12400 (init_gcc_specs): Likewise.
12401 (init_spec): Likewise.
12402 (display_help): Update help message for -pie.
12403 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12406 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12408 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12409 (movdi_aarch64): Likewise.
12410 (movti_aarch64): Likewise.
12412 2017-09-12 Simon Wright <simon@pushface.org>
12415 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12416 calculation of the minor version, always output as 0.
12418 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12421 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12422 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12423 on it early, rather than manual conversion late. For
12424 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12425 instead of performing manual conversion.
12427 2017-09-12 Carl Love <cel@us.ibm.com>
12429 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12430 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12432 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12433 VMULOSW): Add definitions.
12434 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12435 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12436 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12437 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12438 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12440 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12442 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12444 (movti_aarch64): Likewise.
12445 (movdf_aarch64): Likewise.
12446 (movtf_aarch64): Likewise.
12447 (load_pairdi): Likewise.
12448 (store_pairdi): Likewise.
12449 (load_pairdf): Likewise.
12450 (store_pairdf): Likewise.
12451 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12452 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12453 (ldr_got_small_<mode>): Likewise.
12454 (ldr_got_small_28k_<mode>): Likewise.
12455 (ldr_got_tiny): Likewise.
12456 * config/aarch64/iterators.md (ldst_sz): New.
12457 (ldpstp_sz): Likewise.
12458 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12460 (thunderx_load): Split load_8 to load_16.
12461 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12463 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12464 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12465 (xgene1_store_pair): Split store_8 to store_16.
12466 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12467 (falkor_st_0_st_sd): Split store_8 to store_16.
12469 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12471 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12472 and store1/2/3/4 to store_4/8/12/16.
12473 * config/aarch64/aarch64.md: Update for rename.
12474 * config/arm/arm.md: Likewise.: Likewise.
12475 * config/arm/arm.c: Likewise.
12476 * config/arm/thumb1.md: Likewise.
12477 * config/arm/thumb2.md: Likewise.
12478 * config/arm/vfp.md: Likewise.
12479 * config/arm/arm-generic.md: Likewise.
12480 * config/arm/arm1020e.md: Likewise.
12481 * config/arm/arm1026ejs.md: Likewise.
12482 * config/arm/arm1136jfs.md: Likewise.
12483 * config/arm/arm926ejs.md: Likewise.
12484 * config/arm/cortex-a15.md: Likewise.
12485 * config/arm/cortex-a17.md: Likewise.
12486 * config/arm/cortex-a5.md: Likewise.
12487 * config/arm/cortex-a53.md: Likewise.
12488 * config/arm/cortex-a57.md: Likewise.
12489 * config/arm/cortex-a7.md: Likewise.
12490 * config/arm/cortex-a8.md: Likewise.
12491 * config/arm/cortex-a9.md: Likewise.
12492 * config/arm/cortex-m4.md: Likewise.
12493 * config/arm/cortex-m7.md: Likewise.
12494 * config/arm/cortex-r4.md: Likewise.
12495 * config/arm/exynos-m1.md: Likewise.
12496 * config/arm/fa526.md: Likewise.
12497 * config/arm/fa606te.md: Likewise.
12498 * config/arm/fa626te.md: Likewise.
12499 * config/arm/fa726te.md: Likewise.
12500 * config/arm/fmp626.md: Likewise.
12501 * config/arm/iwmmxt.md: Likewise.
12502 * config/arm/ldmstm.md: Likewise.
12503 * config/arm/marvell-pj4.md: Likewise.
12504 * config/arm/xgene1.md: Likewise.
12505 * config/aarch64/thunderx.md: Likewise.
12506 * config/aarch64/thunderx2t99.md: Likewise.
12507 * config/aarch64/falkor.md: Likewise.
12509 2017-09-12 Martin Liska <mliska@suse.cz>
12511 * attribs.c (private_lookup_attribute): New function.
12512 * attribs.h (private_lookup_attribute): Declared here.
12513 (lookup_attribute): Called from this place.
12515 2017-09-12 Richard Biener <rguenther@suse.de>
12517 PR tree-optimization/82157
12518 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12519 stmts with side-effects.
12521 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12522 Alan Hayward <alan.hayward@arm.com>
12523 David Sherwood <david.sherwood@arm.com>
12525 * target.def (hard_regno_nregs): New hook.
12526 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12527 * targhooks.h (default_hard_regno_nregs): Declare.
12528 * targhooks.c (default_hard_regno_nregs): New function.
12529 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12530 (TARGET_HARD_REGNO_NREGS): ...this hook.
12531 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12532 (CLASS_MAX_NREGS): Likewise.
12533 * doc/tm.texi: Regenerate.
12534 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12535 instead of HARD_REGNO_NREGS.
12536 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12537 HARD_REGNO_NREGS in the comment.
12538 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12539 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12540 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12541 Return an unsigned int.
12542 (TARGET_HARD_REGNO_NREGS): Redefine.
12543 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12544 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12545 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12546 (arc_hard_regno_nregs): New function.
12547 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12548 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12549 (arm_hard_regno_nregs): New function.
12550 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12551 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12552 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12553 (TARGET_HARD_REGNO_NREGS): Redefine.
12554 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12555 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12556 (HARD_REGNO_NREGS): Delete.
12557 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12558 (cr16_hard_regno_nregs): New function.
12559 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12560 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12561 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12562 (cris_hard_regno_nregs): New function.
12563 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12564 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12565 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12566 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12567 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12568 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12569 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12570 (frv_hard_regno_nregs): Make static. Take and return an
12572 (frv_class_max_nregs): Remove outdated copy of documentation.
12573 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12574 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12575 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12576 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12577 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12578 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12579 (TARGET_HARD_REGNO_NREGS): Redefine.
12580 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12581 (CLASS_MAX_NREGS): Update comment.
12582 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12583 (ia64_hard_regno_nregs): New function.
12584 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12585 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12586 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12587 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12588 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12590 (m32c_hard_regno_nregs): Likewise. Make static.
12591 (TARGET_HARD_REGNO_NREGS): Redefine.
12592 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12593 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12594 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12595 (m68k_hard_regno_nregs): New function.
12596 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12597 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12598 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12599 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12600 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12601 Take and return an unsigned int.
12602 (TARGET_HARD_REGNO_NREGS): Redefine.
12603 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12604 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12605 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12606 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12607 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12608 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12609 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12610 (msp430_hard_regno_nregs): Make static. Take and return an
12612 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12613 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12614 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12615 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12616 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12617 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12618 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12619 (TARGET_HARD_REGNO_NREGS): Redefine.
12620 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12621 (PA_HARD_REGNO_NREGS): ...this.
12622 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12623 (PA_HARD_REGNO_NREGS): ...this.
12624 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12625 (pa_hard_regno_nregs): New function.
12626 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12627 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12628 (pdp11_hard_regno_nregs): New function.
12629 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12630 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12631 (rs6000_hard_regno_nregs_hook): New function.
12632 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12633 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12634 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12635 Take and return an unsigned int. Move earlier in file.
12636 (TARGET_HARD_REGNO_NREGS): Redefine.
12637 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12638 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12639 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12640 (rl78_hard_regno_nregs): Make static. Take and return an
12642 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12643 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12644 (rs6000_hard_regno_nregs_hook): New function.
12645 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12646 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12647 (TARGET_HARD_REGNO_NREGS): Redefine.
12648 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12649 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12650 instead of HARD_REGNO_NREGS.
12651 (s390_hard_regno_nregs): New function.
12652 (s390_hard_regno_mode_ok): Add comment from s390.h.
12653 (TARGET_HARD_REGNO_NREGS): Redefine.
12654 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12655 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12656 (sh_hard_regno_nregs): New function.
12657 (sh_pass_in_reg_p): Use it.
12658 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12659 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12660 (sparc_hard_regno_nregs): New function.
12661 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12662 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12663 (spu_function_arg_advance): Use it, supplying a valid register number.
12664 (TARGET_HARD_REGNO_NREGS): Redefine.
12665 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12666 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12667 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12668 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12669 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12670 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12671 (CLASS_MAX_NREGS): Remove copy of old documentation.
12672 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12673 (visium_hard_regno_nregs): New function.
12674 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12675 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12676 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12677 xtensa_hard_regno_nregs): New function.
12678 * system.h (HARD_REGNO_NREGS): Poison.
12680 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12682 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12683 hard_regno_nregs instead of HARD_REGNO_NREGS.
12684 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12685 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12686 (c6x_expand_epilogue): Likewise.
12687 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12688 (frv_read_iacc_argument): Likewise.
12689 * config/sh/sh.c: Include regs.h.
12690 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12691 (regs_used): Likewise.
12692 (output_stack_adjust): Likewise.
12693 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12694 * expmed.c: Include regs.h.
12695 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12696 * ree.c: Include regs.h.
12697 (combine_reaching_defs): Use hard_regno_nregs instead of
12699 (add_removable_extension): Likewise.
12701 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12703 * regs.h (hard_regno_nregs): Turn into a function.
12704 (end_hard_regno): Update accordingly.
12705 * caller-save.c (setup_save_areas): Likewise.
12706 (save_call_clobbered_regs): Likewise.
12707 (replace_reg_with_saved_mem): Likewise.
12708 (insert_restore): Likewise.
12709 (insert_save): Likewise.
12710 * combine.c (can_change_dest_mode): Likewise.
12711 (move_deaths): Likewise.
12712 (distribute_notes): Likewise.
12713 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12714 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12715 (rs6000_split_multireg_move): Likewise.
12716 (rs6000_register_move_cost): Likewise.
12717 (rs6000_memory_move_cost): Likewise.
12718 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12719 (rs6000_split_multireg_move): Likewise.
12720 (rs6000_register_move_cost): Likewise.
12721 (rs6000_memory_move_cost): Likewise.
12722 * cselib.c (cselib_reset_table): Likewise.
12723 (cselib_lookup_1): Likewise.
12724 * emit-rtl.c (set_mode_and_regno): Likewise.
12725 * function.c (aggregate_value_p): Likewise.
12726 * ira-color.c (setup_profitable_hard_regs): Likewise.
12727 (check_hard_reg_p): Likewise.
12728 (calculate_saved_nregs): Likewise.
12729 (assign_hard_reg): Likewise.
12730 (improve_allocation): Likewise.
12731 (calculate_spill_cost): Likewise.
12732 * ira-emit.c (modify_move_list): Likewise.
12733 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12734 (ira_hard_reg_in_set_p): Likewise.
12735 * ira.c (setup_reg_mode_hard_regset): Likewise.
12736 (clarify_prohibited_class_mode_regs): Likewise.
12737 (check_allocation): Likewise.
12738 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12739 (lra_setup_reg_renumber): Likewise.
12740 (setup_try_hard_regno_pseudos): Likewise.
12741 (spill_for): Likewise.
12742 (assign_hard_regno): Likewise.
12743 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12744 * lra-constraints.c (in_class_p): Likewise.
12745 (lra_constraint_offset): Likewise.
12746 (simplify_operand_subreg): Likewise.
12747 (lra_constraints): Likewise.
12748 (split_reg): Likewise.
12749 (split_if_necessary): Likewise.
12750 (invariant_p): Likewise.
12751 (inherit_in_ebb): Likewise.
12752 * lra-lives.c (process_bb_lives): Likewise.
12753 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12754 (get_hard_regs): Likewise.
12755 (do_remat): Likewise.
12756 * lra-spills.c (assign_spill_hard_regs): Likewise.
12757 * mode-switching.c (create_pre_exit): Likewise.
12758 * postreload.c (reload_combine_recognize_pattern): Likewise.
12759 * recog.c (peep2_find_free_register): Likewise.
12760 * regcprop.c (kill_value_regno): Likewise.
12761 (set_value_regno): Likewise.
12762 (copy_value): Likewise.
12763 (maybe_mode_change): Likewise.
12764 (find_oldest_value_reg): Likewise.
12765 (copyprop_hardreg_forward_1): Likewise.
12766 * regrename.c (check_new_reg_p): Likewise.
12767 (regrename_do_replace): Likewise.
12768 * reload.c (push_reload): Likewise.
12769 (combine_reloads): Likewise.
12770 (find_dummy_reload): Likewise.
12771 (operands_match_p): Likewise.
12772 (find_reloads): Likewise.
12773 (find_equiv_reg): Likewise.
12774 (reload_adjust_reg_for_mode): Likewise.
12775 * reload1.c (count_pseudo): Likewise.
12776 (count_spilled_pseudo): Likewise.
12777 (find_reg): Likewise.
12778 (clear_reload_reg_in_use): Likewise.
12779 (free_for_value_p): Likewise.
12780 (allocate_reload_reg): Likewise.
12781 (choose_reload_regs): Likewise.
12782 (reload_adjust_reg_for_temp): Likewise.
12783 (emit_reload_insns): Likewise.
12784 (delete_output_reload): Likewise.
12785 * rtlanal.c (subreg_get_info): Likewise.
12786 * sched-deps.c (sched_analyze_reg): Likewise.
12787 * sel-sched.c (init_regs_for_mode): Likewise.
12788 (mark_unavailable_hard_regs): Likewise.
12789 (choose_best_reg_1): Likewise.
12790 (verify_target_availability): Likewise.
12791 * valtrack.c (dead_debug_insert_temp): Likewise.
12792 * var-tracking.c (track_loc_p): Likewise.
12793 (emit_note_insn_var_location): Likewise.
12794 * varasm.c (make_decl_rtl): Likewise.
12795 * reginfo.c (choose_hard_reg_mode): Likewise.
12796 (init_reg_modes_target): Refer directly to
12797 this_target_regs->x_hard_regno_nregs.
12799 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12801 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12802 instead of hard_regno_nregs.
12804 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12806 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12807 end_hard_regno instead of hard_regno_nregs.
12808 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12809 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12810 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12811 * ira-color.c (improve_allocation): Likewise.
12812 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12813 * lra-lives.c (mark_regno_live): Likewise.
12814 (mark_regno_dead): Likewise.
12815 * lra-remat.c (operand_to_remat): Likewise.
12816 * lra.c (collect_non_operand_hard_regs): Likewise.
12817 * postreload.c (reload_combine_note_store): Likewise.
12818 (move2add_valid_value_p): Likewise.
12819 * reload.c (regno_clobbered_p): Likewise.
12821 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12823 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12825 * config/v850/v850.c (v850_reorg): Likewise.
12826 * reload.c (refers_to_regno_for_reload_p): Likewise.
12827 (find_equiv_reg): Likewise.
12828 * reload1.c (reload_reg_reaches_end_p): Likewise.
12830 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12832 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12834 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12835 * config/arm/arm.c (output_move_neon): Likewise.
12836 (arm_attr_length_move_neon): Likewise.
12837 (neon_split_vcombine): Likewise.
12838 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12839 (c6x_mark_reg_written): Likewise.
12840 (c6x_dwarf_register_span): Likewise.
12841 * config/i386/i386.c (ix86_save_reg): Likewise.
12842 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12843 (rws_access_reg): Likewise.
12844 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12845 * mode-switching.c (create_pre_exit): Likewise.
12846 * ree.c (combine_reaching_defs): Likewise.
12847 (add_removable_extension): Likewise.
12848 * regcprop.c (find_oldest_value_reg): Likewise.
12849 (copyprop_hardreg_forward_1): Likewise.
12850 * reload.c (reload_inner_reg_of_subreg): Likewise.
12851 (push_reload): Likewise.
12852 (combine_reloads): Likewise.
12853 (find_dummy_reload): Likewise.
12854 (reload_adjust_reg_for_mode): Likewise.
12855 * reload1.c (find_reload_regs): Likewise.
12856 (forget_old_reloads_1): Likewise.
12857 (reload_reg_free_for_value_p): Likewise.
12858 (reload_adjust_reg_for_temp): Likewise.
12859 (emit_reload_insns): Likewise.
12860 (delete_output_reload): Likewise.
12861 * sel-sched.c (choose_best_reg_1): Likewise.
12862 (choose_best_pseudo_reg): Likewise.
12864 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12865 Alan Hayward <alan.hayward@arm.com>
12866 David Sherwood <david.sherwood@arm.com>
12868 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12869 * target.def (slow_unaligned_access): New hook.
12870 * targhooks.h (default_slow_unaligned_access): Declare.
12871 * targhooks.c (default_slow_unaligned_access): New function.
12872 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12873 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12874 * doc/tm.texi: Regenerate.
12875 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12876 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12877 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12879 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12880 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12882 (rs6000_slow_unaligned_access): New function.
12883 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12884 (expand_block_compare): Likewise.
12885 (expand_strn_compare): Likewise.
12886 (rs6000_rtx_costs): Likewise.
12887 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12888 (riscv_slow_unaligned_access): Likewise.
12889 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12890 (riscv_slow_unaligned_access_p): ...this and make static.
12891 (riscv_option_override): Update accordingly.
12892 (riscv_slow_unaligned_access): New function.
12893 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12894 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12895 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12896 (rs6000_slow_unaligned_access): New function.
12897 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12898 (rs6000_rtx_costs): Likewise.
12899 * config/rs6000/rs6000-string.c (expand_block_compare)
12900 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12901 of SLOW_UNALIGNED_ACCESS.
12902 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12903 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12904 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12905 of SLOW_UNALIGNED_ACCESS.
12906 * expmed.c (simple_mem_bitfield_p): Likewise.
12907 * expr.c (alignment_for_piecewise_move): Likewise.
12908 (emit_group_load_1): Likewise.
12909 (emit_group_store): Likewise.
12910 (copy_blkmode_from_reg): Likewise.
12911 (emit_push_insn): Likewise.
12912 (expand_assignment): Likewise.
12913 (store_field): Likewise.
12914 (expand_expr_real_1): Likewise.
12915 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12916 * lra-constraints.c (simplify_operand_subreg): Likewise.
12917 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12918 * gimple-ssa-store-merging.c: Likewise in block comment at start
12920 * tree-ssa-strlen.c: Include target.h.
12921 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
12922 of SLOW_UNALIGNED_ACCESS.
12923 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
12925 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12927 PR rtl-optimization/82185
12928 * expmed.c (emit_store_flag_int): Only test tem if it has been
12931 2017-09-12 Richard Biener <rguenther@suse.de>
12933 PR middle-end/82149
12934 * match.pd ((FTYPE) N CMP CST): Fix typo.
12936 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
12938 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
12940 (mips_near_type_p): Add 'short_call' attribute as a synonym
12942 * doc/extend.texi (short_call): Document new function attribute.
12944 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12947 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
12948 assertion check that in the condition.
12949 (get_atomic_generic_size): Likewise. Before testing if parameter
12950 has pointer type, if it has array type, call for C++
12951 default_conversion to perform array-to-pointer conversion.
12953 2017-09-12 Richard Biener <rguenther@suse.de>
12955 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
12956 for operations we cannot scalarize.
12958 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
12960 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
12961 vectors heap vectors. Clean up comments.
12962 Make visited_bbs a reference.
12963 (profitable_jump_thread_path): Make GC
12964 vectors heap vectors. Clean up comments.
12966 (convert_and_register_jump_thread_path): Make GC vectors heap
12968 (check_subpath_and_update_thread_path): Same. Clean up comments.
12969 Make visited_bbs a reference.
12970 (handle_phi): Abstract common code to to
12971 register_jump_thread_path_if_profitable.
12972 Rename VAR_BB to DEF_BB.
12974 Make GC vectors heap vectors.
12975 Make visited_bbs a reference.
12976 (handle_assignment): Same.
12977 (register_jump_thread_path_if_profitable): New.
12978 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
12980 Make GC vectors heap vectors. Clean up comments.
12981 Make visited_bbs a reference.
12982 (find_jump_threads_backwards): Make visited_bbs live in the stack.
12983 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
12986 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
12989 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
12990 words of E_DImode object are reachable by xtensa_uimm8x4 access.
12992 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
12994 Revert r251800 and r251799.
12996 2017-09-11 Martin Jambor <mjambor@suse.cz>
12999 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13000 arguments in advance.
13001 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13002 use it to find predecessor edges.
13003 (naive_outof_ssa): Collect vector of predecessors.
13005 2017-09-08 Jason Merrill <jason@redhat.com>
13007 PR c++/70029 - ICE with ref-qualifier and -flto
13008 * langhooks.h (struct lang_hooks_for_types): Add
13009 copy_lang_qualifiers.
13010 * attribs.c (build_type_attribute_qual_variant): Use it.
13011 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13013 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13014 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13016 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13019 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13020 (*mulsi3_sp64): New instruction.
13021 (mulsi3): New expander.
13023 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13025 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13027 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13029 * sancov.c: Include memmodel.h.
13031 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13034 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13037 2017-09-07 Carl Love <cel@us.ibm.com>
13039 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13040 the sldi instruction.
13042 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13044 * sancov.c: Include tm_p.h.
13046 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13049 * output.h (switch_to_other_text_partition): New declaration.
13050 * varasm.c (switch_to_other_text_partition): New function.
13051 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13052 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13053 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13054 to the other text partition before emitting LCL label and switch back
13055 after emitting the word after it.
13057 2017-09-07 Richard Biener <rguenther@suse.de>
13059 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13060 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13061 critical edges here, after loop init.
13062 (pass_data_pre): Remove PROP_no_crit_edges flags.
13063 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13064 for valueization of call args to avoid leaking VN_TOP.
13065 (visit_use): Assert we do not visit default defs.
13066 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13067 Use error_mark_node to more easily detect leaking VN_TOP.
13068 All default-defs are varying, not VN_TOP. Mark them visited.
13069 (run_scc_vn): Make code match comment.
13071 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13073 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13074 OPTION_MASK_FLOAT128_KEYWORD.
13075 (POWERPC_MASKS): Likewise.
13076 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13077 support for the -mfloat128-type option, and make -mfloat128
13078 default on PowerPC Linux systems. Define or undefine
13079 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13080 Define __float128 to be __ieee128 if IEEE 128-bit support is
13081 enabled, or undefine it.
13082 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13083 Delete defining __FLOAT128_TYPE__.
13084 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13085 -mfloat128-type option and make -mfloat128 default on PowerPC
13087 (TARGET_FLOAT128_TYPE): Likewise.
13088 (-mfloat128-type): Likewise.
13089 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13090 Delete the -mfloat128-type option and make -mfloat128 default on
13091 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13092 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13093 128-bit floating point is enabled. Change tests from using
13094 -mfloat128-type to -mfloat128.
13095 (rs6000_mangle_type): Use the correct mangling for the __float128
13096 type even if normal long double is restricted to 64-bits.
13097 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13099 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13100 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13101 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13103 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13104 documentation for -mfloat128.
13106 2017-09-06 Olivier Hainque <hainque@adacore.com>
13108 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13110 2017-09-06 Wish Wu <wishwu007@gmail.com>
13111 Jakub Jelinek <jakub@redhat.com>
13113 * asan.c (initialize_sanitizer_builtins): Add
13114 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13115 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13116 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13117 BT_FN_VOID_UINT64_PTR variables.
13118 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13119 (BT_FN_VOID_UINT16_UINT16): Likewise.
13120 (BT_FN_VOID_UINT32_UINT32): Likewise.
13121 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13122 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13123 (BT_FN_VOID_UINT64_PTR): Likewise.
13124 * common.opt (flag_sanitize_coverage): New variable.
13125 (fsanitize-coverage=trace-pc): Remove.
13126 (fsanitize-coverage=): Add.
13127 * flag-types.h (enum sanitize_coverage_code): New enum.
13128 * fold-const.c (fold_range_test): Disable non-short-circuit
13129 optimization if flag_sanitize_coverage.
13130 (fold_truth_andor): Likewise.
13131 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13132 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13133 (coverage_sanitizer_opts): New array.
13134 (get_closest_sanitizer_option): Add OPTS argument, handle also
13135 OPT_fsanitize_coverage_.
13136 (parse_sanitizer_options): Adjusted to also handle
13137 OPT_fsanitize_coverage_.
13138 (common_handle_option): Add OPT_fsanitize_coverage_.
13139 * sancov.c (instrument_comparison, instrument_switch): New function.
13140 (sancov_pass): Add trace-cmp support.
13141 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13142 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13143 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13144 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13145 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13146 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13147 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13148 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13149 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13150 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13152 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13154 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13155 error. Only quit immediately if parsing is complete.
13156 (BEGIN): Initialize fatal_err and parse_done.
13157 (begin fpu, end fpu): Check number of arguments.
13158 (begin arch, end arch): Likewise.
13159 (begin cpu, end cpu): Likewise.
13160 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13161 (optalias): Likewise.
13163 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13165 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13166 * config/arm/arm-isa.h: Delete. Move definitions to ...
13167 * arm-cpus.in: ... here. Use new feature and fgroup values.
13168 * config/arm/arm.c (arm_option_override): Use lower case for feature
13170 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13171 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13172 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13174 (print_isa_bits_for): New function.
13175 (gen_isa): New function.
13176 (gen_comm_data): Use print_isa_bits_for.
13177 (define feature): New keyword.
13178 (define fgroup): New keyword.
13179 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13180 (arm-isa.h): Add rule to generate file.
13181 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13182 case for feature bit names.
13184 2017-09-06 Richard Biener <rguenther@suse.de>
13186 * tree-ssa-pre.c (NECESSARY): Remove.
13187 (create_expression_by_pieces): Do not touch pass-local flags.
13188 (insert_into_preds_of_block): Likewise.
13189 (do_pre_regular_insertion): Likewise.
13190 (eliminate_insert): Likewise.
13191 (eliminate_dom_walker::before_dom_children): Likewise.
13192 (fini_eliminate): Do not look at inserted_exprs.
13193 (mark_operand_necessary): Remove.
13194 (remove_dead_inserted_code): Replace with simple work-list
13195 algorithm based on inserted_exprs and SSA uses.
13196 (pass_pre::execute): Re-order fini_eliminate and
13197 remove_dead_inserted_code.
13199 2017-09-06 Olivier Hainque <hainque@adacore.com>
13201 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13202 for VxWorks 7. Adjust surrounding comments.
13204 2017-09-06 Richard Biener <rguenther@suse.de>
13206 * gimple-ssa-strength-reduction.c
13207 (find_candidates_dom_walker::before_dom_children): Also allow
13210 2017-09-06 Richard Biener <rguenther@suse.de>
13212 PR tree-optimization/82108
13213 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13214 for gap in the non-permutation SLP case.
13216 2017-09-06 Martin Jambor <mjambor@suse.cz>
13218 PR tree-optimization/82078
13219 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13220 add_access_to_work_queue...
13221 (build_accesses_from_assign): ...here.
13222 (propagate_all_subaccesses): Make sure racc is the group
13223 representative, if there is one.
13225 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13227 PR middle-end/82095
13228 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13231 2017-09-06 Richard Biener <rguenther@suse.de>
13233 * gimple-ssa-strength-reduction.c
13234 (find_candidates_dom_walker::before_doom_children): Use a
13235 type and not a mode check.
13237 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13240 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13242 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13244 2017-09-05 Jeff Law <law@redhat.com>
13246 PR tree-optimization/64910
13247 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13248 cases where we have 3 or more operands.
13250 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13252 PR middle-end/81768
13253 * omp-low.c (lower_omp_for): Recompute tree invariant if
13254 gimple_omp_for_initial/final is ADDR_EXPR.
13256 PR middle-end/81768
13257 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13258 into gimple val before gimplification fo the COND_EXPR.
13260 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13262 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13263 REGION_COPY argument.
13264 (thread_through_all_blocks): Remove unused argument to
13265 duplicate_thread_path.
13267 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13268 Alan Hayward <alan.hayward@arm.com>
13269 David Sherwood <david.sherwood@arm.com>
13271 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13272 Take a scalar_mode rather than a machine_mode.
13273 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13274 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13275 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13276 (aarch64_gen_adjusted_ldpstp): Likewise.
13277 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13279 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13280 Alan Hayward <alan.hayward@arm.com>
13281 David Sherwood <david.sherwood@arm.com>
13283 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13284 Take a scalar_int_mode instead of a machine_mode.
13285 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13286 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13287 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13288 (aarch64_simd_attr_length_rglist): Delete.
13289 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13290 a scalar_int_mode instead of a machine_mode.
13291 (aarch64_add_offset): Likewise.
13292 (aarch64_internal_mov_immediate): Likewise
13293 (aarch64_add_constant_internal): Likewise.
13294 (aarch64_add_constant): Likewise.
13295 (aarch64_movw_imm): Likewise.
13296 (aarch64_rtx_arith_op_extract_p): Likewise.
13297 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13298 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13299 Remove assert that the mode isn't a vector.
13300 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13301 (aarch64_expand_mov_immediate): Update calls after above changes.
13302 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13303 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13304 (aarch64_move_imm): Likewise.
13305 (aarch64_can_const_movi_rtx_p): Likewise.
13306 (aarch64_strip_extend): Likewise.
13307 (aarch64_extr_rtx_p): Likewise.
13308 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13309 a CONST_INT when the mode parameter is VOIDmode.
13310 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13312 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13314 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13315 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13316 (bitwise_type_for_mode): Update accordingly.
13318 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13320 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13321 * stor-layout.c (mode_for_size_tree): Likewise.
13322 (mode_for_array): Update accordingly.
13323 (layout_decl): Likewise.
13324 (compute_record_mode): Likewise. Only set the mode once.
13326 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13328 * target.def (get_mask_mode): Change return type to opt_mode.
13330 * doc/tm.texi: Regenerate.
13331 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13332 * targhooks.c (default_get_mask_mode): Likewise.
13333 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13334 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13335 targetm.get_mask_mode.
13336 * tree.c (build_truth_vector_type): Likewise.
13338 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13340 * machmode.h (mode_for_vector): Return an opt_mode.
13341 * stor-layout.c (mode_for_vector): Likewise.
13342 (mode_for_int_vector): Update accordingly.
13343 (layout_type): Likewise.
13344 * config/i386/i386.c (emit_memmov): Likewise.
13345 (ix86_expand_set_or_movmem): Likewise.
13346 (ix86_expand_vector_init): Likewise.
13347 (ix86_get_mask_mode): Likewise.
13348 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13350 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13351 * expmed.c (extract_bit_field_1): Likewise.
13352 * expr.c (expand_expr_real_2): Likewise.
13353 * optabs-query.c (can_vec_perm_p): Likewise.
13354 (can_vec_mask_load_store_p): Likewise.
13355 * optabs.c (expand_vec_perm): Likewise.
13356 * targhooks.c (default_get_mask_mode): Likewise.
13357 * tree-vect-stmts.c (vectorizable_store): Likewise.
13358 (vectorizable_load): Likewise.
13359 (get_vectype_for_scalar_type_and_size): Likewise.
13361 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13363 * machmode.h (mode_for_int_vector): New function.
13364 * stor-layout.c (mode_for_int_vector): Likewise.
13365 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13366 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13367 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13368 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13369 (s390_expand_vcond): Likewise.
13371 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13373 * machmode.h (opt_machine_mode): New type.
13374 (opt_mode<T>): Allow construction from anything that can be
13376 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13377 (mode_for_size): Return an opt_machine_mode.
13378 * stor-layout.c (mode_for_size): Likewise.
13379 (mode_for_size_tree): Update call accordingly.
13380 (bitwise_mode_for_mode): Likewise.
13381 (make_fract_type): Likewise.
13382 (make_accum_type): Likewise.
13383 * caller-save.c (replace_reg_with_saved_mem): Update call
13385 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13386 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13387 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13388 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13389 * expmed.c (extract_bit_field_1): Likewise.
13390 * reload.c (get_secondary_mem): Likewise.
13391 * varasm.c (assemble_integer): Likewise.
13392 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13395 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13397 * machmode.h (decimal_float_mode_for_size): New function.
13398 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13399 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13400 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13401 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13402 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13403 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13405 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13407 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13408 (get_builtin_sync_mode): Likewise.
13409 (expand_ifn_atomic_compare_exchange): Likewise.
13410 (expand_builtin_atomic_clear): Likewise.
13411 (expand_builtin_atomic_test_and_set): Likewise.
13412 (fold_builtin_atomic_always_lock_free): Likewise.
13413 * calls.c (compute_argument_addresses): Likewise.
13414 (emit_library_call_value_1): Likewise.
13415 (store_one_arg): Likewise.
13416 * combine.c (combine_instructions): Likewise.
13417 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13418 * config/arm/arm.c (arm_function_value): Likewise.
13419 (aapcs_allocate_return_reg): Likewise.
13420 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13421 * config/i386/i386.c (construct_container): Likewise.
13422 (ix86_gimplify_va_arg): Likewise.
13423 (ix86_expand_sse_cmp): Likewise.
13424 (emit_memmov): Likewise.
13425 (emit_memset): Likewise.
13426 (expand_small_movmem_or_setmem): Likewise.
13427 (ix86_expand_pextr): Likewise.
13428 (ix86_expand_pinsr): Likewise.
13429 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13430 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13432 * config/mips/mips.c (mips_function_value_1) Likewise.
13433 (mips_block_move_straight): Likewise.
13434 (mips_expand_ins_as_unaligned_store): Likewise.
13435 * config/powerpcspe/powerpcspe.c
13436 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13437 (rs6000_darwin64_record_arg_flush): Likewise.
13438 * config/rs6000/rs6000.c
13439 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13440 (rs6000_darwin64_record_arg_flush): Likewise.
13441 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13442 (sparc_function_value_1): Likewise.
13443 * config/spu/spu.c (adjust_operand): Likewise.
13444 (spu_emit_branch_or_set): Likewise.
13445 (arith_immediate_p): Likewise.
13446 * emit-rtl.c (gen_lowpart_common): Likewise.
13447 * expr.c (expand_expr_real_1): Likewise.
13448 * function.c (assign_parm_setup_block): Likewise.
13449 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13450 * reload1.c (alter_reg): Likewise.
13451 * stor-layout.c (mode_for_vector): Likewise.
13452 (layout_type): Likewise.
13454 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13456 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13457 (spu_convert_move): Likewise.
13458 * lower-subreg.c (resolve_simple_move): Likewise.
13460 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13463 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13464 define_insn to a define_expand.
13465 (altivec_vsum2sws_direct): New define_insn.
13466 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13468 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13470 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13473 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13478 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13479 realign stack if stack alignment needed is less than incoming
13482 2017-09-05 Marek Polacek <polacek@redhat.com>
13485 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13488 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13490 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13492 2017-09-05 Richard Biener <rguenther@suse.de>
13494 PR tree-optimization/82084
13495 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13497 2017-09-05 Richard Biener <rguenther@suse.de>
13499 PR tree-optimization/82102
13500 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13502 2017-09-05 Martin Liska <mliska@suse.cz>
13504 PR tree-optimization/82032
13505 * tree-cfg.c (generate_range_test): New function.
13506 * tree-cfg.h (generate_range_test): Declared here.
13507 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13508 (cleanup_control_expr_graph): Use it.
13509 * tree-switch-conversion.c (try_switch_expansion): Remove
13511 (emit_case_nodes): Use generate_range_test.
13513 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13516 * config/i386/i386.md (*<btsc><mode>_mask): Add
13517 TARGET_USE_BT to insn constraint.
13518 (*btr<mode>_mask): Ditto.
13520 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13522 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13523 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13525 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13528 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13529 TARGET_NEON and TARGET_IWMMXT.
13530 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13531 TARGET_NEON and TARGET_IWMMXT.
13532 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13534 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13536 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13537 (ix86_rewrite_tls_address): Ditto.
13538 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13539 (ix86_rewrite_tls_address_1): Ditto.
13540 (ix86_rewrite_tls_address): Ditto.
13541 * config/i386/predicates.md (tls_address_pattern): New predicate.
13542 * config/i386/i386.md (TLS address splitter): New splitter.
13544 2017-09-04 Richard Biener <rguenther@suse.de>
13546 PR tree-optimization/82084
13547 * fold-const.h (can_native_encode_string_p): Declare.
13548 * fold-const.c (can_native_encode_string_p): Factor out from ...
13549 (native_encode_string): ... here.
13550 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13551 vectorizing stores from constants we later cannot handle.
13553 2017-09-04 Marek Polacek <polacek@redhat.com>
13556 * doc/invoke.texi: Update -Wtautological-compare documentation.
13558 2017-09-04 Jeff Law <law@redhat.com>
13560 PR tree-optimization/64910
13561 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13562 swap the first and last operand if the last is a constant.
13564 2017-09-04 Marek Polacek <polacek@redhat.com>
13567 * convert.c (do_narrow): When sanitizing signed integer overflows,
13568 bail out for signed types.
13569 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13571 2017-09-04 Richard Biener <rguenther@suse.de>
13573 PR tree-optimization/82060
13574 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13575 Move devirtualization after stmt folding and before EH/AB/noreturn
13576 cleanup to get the stmt refs canonicalized. Use a bool instead
13577 of gimple_modified_p since that doesn't work for NOPs. Schedule
13578 NOPs generated by folding for removal.
13580 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13581 Alan Hayward <alan.hayward@arm.com>
13582 David Sherwood <david.sherwood@arm.com>
13584 * coretypes.h (pad_direction): New enum.
13585 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13586 (FUNCTION_ARG_PADDING): Likewise.
13587 * target.def (function_arg_padding): New hook.
13588 * targhooks.h (default_function_arg_padding): Declare.
13589 * targhooks.c (default_function_arg_padding): New function.
13590 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13591 (TARGET_FUNCTION_ARG_PADDING): ...this.
13592 * doc/tm.texi: Regenerate.
13593 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13594 instead of direction.
13595 (compute_argument_addresses): Likewise.
13596 (load_register_parameters): Likewise.
13597 (emit_library_call_value_1): Likewise.
13598 (store_one_arg): Use targetm.calls.function_arg_padding instead
13599 of FUNCTION_ARG_PADDING.
13600 (must_pass_in_stack_var_size_or_pad): Likewise.
13601 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13602 (emit_group_store): Likewise.
13603 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13604 instead of FUNCTION_ARG_PADDING.
13605 (emit_push_insn): Likewise, and propagate enum change throughout
13607 * function.h (direction): Delete.
13608 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13610 * function.c (assign_parm_find_stack_rtl): Likewise.
13611 (assign_parm_setup_block_p): Likewise.
13612 (assign_parm_setup_block): Likewise.
13613 (gimplify_parameters): Likewise.
13614 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13615 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13617 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13618 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13619 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13620 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13621 (aarch64_function_arg_padding): ...this new function.
13622 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13623 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13624 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13625 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13626 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13627 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13628 (arm_pad_arg_upward): Replace with...
13629 (arm_function_arg_padding): ...this new function.
13630 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13632 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13633 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13634 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13635 (ia64_hpux_function_arg_padding): Replace with...
13636 (ia64_function_arg_padding): ...this new function. Use pad_direction
13637 instead of direction. Check for TARGET_HPUX.
13638 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13639 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13640 (iq2000_function_arg_padding): New function.
13641 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13642 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13643 (mips_function_arg_padding): ...this new function.
13644 (mips_pad_reg_upward): Update accordingly.
13645 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13646 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13647 targetm.calls.function_arg_padding.
13648 (FUNCTION_ARG_PADDING): Delete.
13649 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13650 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13651 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13652 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13653 (nios2_block_reg_padding): Return pad_direction instead of direction.
13654 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13655 instead of direction.
13656 (nios2_function_arg_padding): Likewise. Make static.
13657 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13658 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13659 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13660 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13661 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13662 (pa_function_arg_padding): Make static. Return pad_direction instead
13664 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13665 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13666 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13667 instead of direction. Use targetm.calls.function_arg_padding.
13668 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13669 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13670 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13671 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13672 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13674 (function_arg_padding): Rename to...
13675 (rs6000_function_arg_padding): ...this. Make static. Return
13676 pad_direction instead of direction.
13677 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13678 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13679 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13680 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13681 instead of direction. Use targetm.calls.function_arg_padding.
13682 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13683 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13684 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13685 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13686 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13687 (function_arg_padding): Rename to...
13688 (rs6000_function_arg_padding): ...this. Make static. Return
13689 pad_direction instead of direction.
13690 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13691 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13692 * config/s390/s390.c (s390_function_arg_padding): New function.
13693 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13694 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13695 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13696 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13697 (function_arg_padding): Rename to...
13698 (sparc_function_arg_padding): ...this. Make static. Return
13699 pad_direction instead of direction.
13700 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13701 * config/spu/spu.c (spu_function_arg_padding): New function.
13702 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13703 * system.h (FUNCTION_ARG_PADDING): Poison.
13705 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13706 Alan Hayward <alan.hayward@arm.com>
13707 David Sherwood <david.sherwood@arm.com>
13709 * target.def (modes_tieable_p): New hook.
13710 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13711 (TARGET_MODES_TIEABLE_P): ...this.
13712 * doc/tm.texi.in: Regenerate.
13713 * hooks.h (hook_bool_mode_mode_true): Declare.
13714 * hooks.c (hook_bool_mode_mode_true): New function.
13715 * combine.c (subst): Use targetm.modes_tieable_p instead of
13717 * dse.c (find_shift_sequence): Likewise.
13718 * expmed.c (extract_low_bits): Likewise.
13719 * lower-subreg.c: Include target.h.
13720 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13722 * rtlanal.c (rtx_cost): Likewise.
13723 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13724 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13725 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13726 (TARGET_MODES_TIEABLE_P): Redefine.
13727 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13728 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13729 (TARGET_MODES_TIEABLE_P): Redefine.
13730 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13731 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13732 (arc_modes_tieable_p): New function.
13733 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13734 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13735 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13736 (arm_modes_tieable_p): Make static.
13737 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13738 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13739 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13740 (TARGET_MODES_TIEABLE_P): Redefine.
13741 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13742 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13743 (TARGET_MODES_TIEABLE_P): Redefine.
13744 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13745 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13746 (cr16_modes_tieable_p): New function.
13747 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13748 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13749 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13750 (TRULY_NOOP_TRUNCATION): Update comment.
13751 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13752 (TRULY_NOOP_TRUNCATION): Update comment.
13753 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13754 (frv_modes_tieable_p): New function.
13755 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13756 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13757 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13758 (TARGET_MODES_TIEABLE_P): Redefine.
13759 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13760 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13761 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13762 (TARGET_MODES_TIEABLE_P): Redefine.
13763 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13764 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13765 (ia64_modes_tieable_p): New function.
13766 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13767 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13768 (iq2000_modes_tieable_p): New function.
13769 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13770 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13771 (lm32_modes_tieable_p): New function.
13772 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13773 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13774 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13775 (TARGET_MODES_TIEABLE_P): Redefine.
13776 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13777 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13778 (m32r_modes_tieable_p): New function.
13779 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13780 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13781 (m68k_modes_tieable_p): New function.
13782 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13783 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13784 (mcore_modes_tieable_p): New function.
13785 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13786 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13788 (TARGET_MODES_TIEABLE_P): Redefine.
13789 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13790 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13791 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13792 (TARGET_MODES_TIEABLE_P): Redefine.
13793 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13794 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13795 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13796 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13797 (mn10300_modes_tieable_p): ...this and make static.
13798 (TARGET_MODES_TIEABLE_P): Redefine.
13799 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13800 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13801 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13802 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13803 (msp430_modes_tieable_p): Make static.
13804 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13805 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13806 (TARGET_MODES_TIEABLE_P): Redefine.
13807 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13808 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13809 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13810 (TARGET_MODES_TIEABLE_P): Redefine.
13811 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13812 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13813 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13814 (TARGET_MODES_TIEABLE_P): Redefine.
13815 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13816 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13817 (pdp11_modes_tieable_p): New function.
13818 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13819 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13820 (rs6000_modes_tieable_p): New function.
13821 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13822 * config/powerpcspe/powerpcspe.md: Update comment.
13823 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13824 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13825 (TARGET_MODES_TIEABLE_P): Redefine.
13826 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13827 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13828 (rl78_modes_tieable_p): New function.
13829 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13830 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13831 (rs6000_modes_tieable_p): New function.
13832 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13833 * config/rs6000/rs6000.md: Update comment.
13834 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13835 * config/rx/rx.c (rx_modes_tieable_p): New function.
13836 (TARGET_MODES_TIEABLE_P): Redefine.
13837 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13838 * config/s390/s390.c (s390_modes_tieable_p): New function.
13839 (TARGET_MODES_TIEABLE_P): Redefine.
13840 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13841 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13842 (sh_modes_tieable_p): New function.
13843 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13844 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13845 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13846 (sparc_modes_tieable_p): Make static.
13847 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13848 * config/spu/spu.c (spu_modes_tieable_p): New function.
13849 (TARGET_MODES_TIEABLE_P): Redefine.
13850 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13851 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13852 (TARGET_MODES_TIEABLE_P): Redefine.
13853 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13854 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13855 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13856 * config/v850/v850.c (v850_modes_tieable_p): New function.
13857 (TARGET_MODES_TIEABLE_P): Redefine.
13858 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13859 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13860 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13861 (visium_modes_tieable_p): New function.
13862 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13863 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13864 (xtensa_modes_tieable_p): New function.
13865 * system.h (MODES_TIEABLE_P): Poison.
13867 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13868 Alan Hayward <alan.hayward@arm.com>
13869 David Sherwood <david.sherwood@arm.com>
13871 * target.def (hard_regno_mode_ok): New hook.
13872 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13873 (TARGET_HARD_REGNO_MODE_OK): ...this.
13874 * doc/tm.texi.in: Regenerate.
13875 * hooks.h (hook_bool_uint_mode_true): Declare.
13876 * hooks.c (hook_bool_uint_mode_true): New function.
13877 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13878 HARD_REGNO_MODE_OK.
13879 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13880 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13881 instead of HARD_REGNO_MODE_OK.
13882 * caller-save.c: Include target.h.
13883 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13884 HARD_REGNO_MODE_OK.
13885 * combine.c (can_combine_p): Likewise.
13886 (combinable_i3pat): Likewise.
13887 (can_change_dest_mode): Likewise.
13888 * expr.c (init_expr_target): Likewise.
13889 (convert_move): Likewise.
13890 (convert_modes): Likewise.
13891 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13892 (setup_prohibited_mode_move_regs): Likewise.
13893 * ira.h (target_ira): Likewise.
13894 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13895 * lra-constraints.c (process_alt_operands): Likewise.
13896 (split_reg): Likewise.
13897 * recog.c (peep2_find_free_register): Likewise.
13898 * ree.c (combine_reaching_defs): Likewise.
13899 * regcprop.c (maybe_mode_change): Likewise.
13900 * reginfo.c (init_reg_sets_1): Likewise.
13901 (choose_hard_reg_mode): Likewise.
13902 (simplifiable_subregs): Likewise.
13903 * regrename.c (check_new_reg_p): Likewise.
13904 * reload.c (find_valid_class): Likewise.
13905 (find_valid_class_1): Likewise.
13906 (reload_inner_reg_of_subreg): Likewise.
13907 (push_reload): Likewise.
13908 (combine_reloads): Likewise.
13909 (find_dummy_reload): Likewise.
13910 (find_reloads): Likewise.
13911 * reload1.c (find_reg): Likewise.
13912 (set_reload_reg): Likewise.
13913 (allocate_reload_reg): Likewise.
13914 (choose_reload_regs): Likewise.
13915 (reload_adjust_reg_for_temp): Likewise.
13916 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
13917 (simplify_subreg_regno): Likewise.
13918 * sel-sched.c (init_regs_for_mode): Likewise.
13919 * varasm.c (make_decl_rtl): Likewise.
13920 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
13921 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
13922 HARD_REGNO_MODE_OK.
13923 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
13924 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
13925 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13926 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
13927 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
13928 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13929 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
13930 (arc_mode_class): Delete.
13931 (HARD_REGNO_MODE_OK): Delete.
13932 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13933 (arc_hard_regno_mode_ok): Rename old array to...
13934 (arc_hard_regno_mode_ok_modes): ...this.
13935 (arc_conditional_register_usage): Update accordingly.
13936 (arc_mode_class): Make static.
13937 (arc_hard_regno_mode_ok): New function.
13938 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
13939 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
13940 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13941 (arm_hard_regno_mode_ok): Make static.
13942 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
13943 HARD_REGNO_MODE_OK.
13944 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
13945 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
13946 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
13948 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13949 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
13950 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
13951 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
13952 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
13953 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13954 * config/bfin/predicates.md (valid_reg_operand): Use
13955 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
13956 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
13957 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
13958 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13959 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
13960 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
13961 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13962 (cr16_hard_regno_mode_ok): Make static and return a bool.
13963 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
13964 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13965 (cris_hard_regno_mode_ok): New function.
13966 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
13967 (epiphany_mode_class): Delete.
13968 (HARD_REGNO_MODE_OK): Delete.
13969 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
13970 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13971 (hard_regno_mode_ok): Rename to...
13972 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
13973 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
13974 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
13975 HARD_REGNO_MODE_OK.
13976 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
13977 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
13978 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13979 (frv_hard_regno_mode_ok): Make static and return a bool.
13980 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
13981 HARD_REGNO_MODE_OK.
13982 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
13983 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
13984 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
13985 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
13987 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13988 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
13989 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
13990 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
13992 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13993 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
13994 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13995 (ia64_hard_regno_mode_ok): New function.
13996 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
13997 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13998 (iq2000_hard_regno_mode_ok): New function.
13999 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14000 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14001 (lm32_hard_regno_mode_ok): New function.
14002 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14003 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14004 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14005 instead of HARD_REGNO_MODE_OK.
14006 (m32c_hard_regno_ok): Rename to...
14007 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14008 (m32c_cannot_change_mode_class): Update accordingly.
14009 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14010 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14011 (m32r_mode_class): Delete.
14012 (HARD_REGNO_MODE_OK): Delete.
14013 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14014 (m32r_hard_regno_mode_ok): Rename to...
14015 (m32r_hard_regno_modes): ...this.
14016 (m32r_mode_class): Make static.
14017 (m32r_hard_regno_mode_ok): New function.
14018 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14019 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14020 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14021 (m68k_hard_regno_mode_ok): Make static.
14022 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14023 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14024 (mcore_hard_regno_mode_ok): New function.
14025 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14026 (HARD_REGNO_MODE_OK): Delete.
14027 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14029 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14030 (microblaze_hard_regno_mode_ok): New function.
14031 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14032 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14033 (mips_hard_regno_mode_ok): Delete.
14034 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14035 (mips_hard_regno_mode_ok_p): ...this and make static.
14036 (mips_hard_regno_mode_ok_p): Rename to...
14037 (mips_hard_regno_mode_ok_uncached): ...this.
14038 (mips_hard_regno_mode_ok): New function.
14039 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14040 of HARD_REGNO_MODE_OK.
14041 (mips_option_override): Update after above name changes.
14042 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14043 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14044 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14045 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14046 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14047 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14048 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14049 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14050 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14051 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14052 (msp430_hard_regno_mode_ok): Make static and return a bool.
14053 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14054 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14055 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14057 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14058 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14059 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14060 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14061 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14062 (PA_HARD_REGNO_MODE_OK): ...this
14063 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14064 (PA_HARD_REGNO_MODE_OK): ...this.
14065 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14066 (pa_hard_regno_mode_ok): New function.
14067 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14068 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14069 (pdp11_hard_regno_mode_ok): New function.
14070 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14071 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14073 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14075 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14076 (rs6000_hard_regno_mode_ok): Rename to...
14077 (rs6000_hard_regno_mode_ok_uncached): ...this.
14078 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14079 (rs6000_hard_regno_mode_ok): New function.
14080 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14081 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14082 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14083 (riscv_hard_regno_mode_ok): ...this and make static.
14084 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14085 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14086 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14087 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14088 (rl78_hard_regno_mode_ok): Make static and return bool.
14089 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14090 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14092 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14093 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14094 (rs6000_hard_regno_mode_ok): Rename to...
14095 (rs6000_hard_regno_mode_ok_uncached): ...this.
14096 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14097 (rs6000_hard_regno_mode_ok): New function.
14098 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14099 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14100 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14101 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14102 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14103 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14104 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14105 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14106 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14107 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14108 (sh_hard_regno_mode_ok): Make static.
14109 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14110 instead of HARD_REGNO_MODE_OK.
14111 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14112 (sparc_mode_class): Delete.
14113 (HARD_REGNO_MODE_OK): Delete.
14114 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14115 (hard_regno_mode_classes): Make static.
14116 (sparc_mode_class): Likewise.
14117 (sparc_hard_regno_mode_ok): New function.
14118 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14119 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14120 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14122 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14123 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14124 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14125 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14126 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14127 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14128 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14129 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14130 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14131 (visium_hard_regno_mode_ok): New function.
14132 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14133 instead of HARD_REGNO_MODE_OK.
14134 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14135 (HARD_REGNO_MODE_OK): Delete.
14136 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14137 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14138 (xtensa_option_override): Update accordingly.
14139 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14140 (xtensa_hard_regno_mode_ok): New function.
14141 * system.h (HARD_REGNO_MODE_OK): Poison.
14143 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14144 Alan Hayward <alan.hayward@arm.com>
14145 David Sherwood <david.sherwood@arm.com>
14147 * target.def (hard_regno_call_part_clobbered): New hook.
14148 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14149 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14150 * doc/tm.texi: Regenerate.
14151 * hooks.h (hook_bool_uint_mode_false): Declare.
14152 * hooks.c (hook_bool_uint_mode_false): New function.
14153 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14154 * cselib.c (cselib_process_insn): Use
14155 targetm.hard_regno_call_part_clobbered instead of
14156 HARD_REGNO_CALL_PART_CLOBBERED.
14157 * ira-conflicts.c (ira_build_conflicts): Likewise.
14158 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14159 * lra-constraints.c (need_for_call_save_p): Likewise.
14160 * lra-lives.c: Include target.h.
14161 (check_pseudos_live_through_calls): Use
14162 targetm.hard_regno_call_part_clobbered instead of
14163 HARD_REGNO_CALL_PART_CLOBBERED.
14164 * regcprop.c: Include target.h.
14165 (copyprop_hardreg_forward_1): Use
14166 targetm.hard_regno_call_part_clobbered instead of
14167 HARD_REGNO_CALL_PART_CLOBBERED.
14168 * reginfo.c (choose_hard_reg_mode): Likewise.
14169 * regrename.c (check_new_reg_p): Likewise.
14170 * reload.c (find_equiv_reg): Likewise.
14171 * reload1.c (emit_reload_insns): Likewise.
14172 * sched-deps.c (deps_analyze_insn): Likewise.
14173 * sel-sched.c (init_regs_for_mode): Likewise.
14174 (mark_unavailable_hard_regs): Likewise.
14175 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14176 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14177 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14179 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14180 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14181 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14183 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14185 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14186 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14187 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14189 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14190 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14191 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14193 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14194 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14196 * config/powerpcspe/powerpcspe.c
14197 (rs6000_hard_regno_call_part_clobbered): New function.
14198 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14199 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14200 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14202 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14203 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14204 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14206 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14207 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14208 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14210 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14211 Alan Hayward <alan.hayward@arm.com>
14212 David Sherwood <david.sherwood@arm.com>
14214 * rtl.h (subreg_memory_offset): Declare.
14215 * emit-rtl.c (subreg_memory_offset): New function.
14216 * expmed.c (store_bit_field_1): Use it.
14217 * expr.c (undefined_operand_subword_p): Likewise.
14218 * simplify-rtx.c (simplify_subreg): Likewise.
14220 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14222 PR rtl-optimization/57448
14225 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14226 using atomic_load pattern.
14227 (expand_atomic_store): Likewise.
14229 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14232 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14233 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14234 handling. Use replace_call_with_value with NULL instead of
14235 gsi_replace, unlink_stmt_vdef and release_defs.
14237 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14240 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14242 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14245 * rtl.h (emit_library_call_value_1): Declare.
14246 (emit_library_call): Replace declaration with a series of overloads.
14247 Remove the parameter count argument.
14248 (emit_library_call_value): Likewise.
14249 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14250 with an "rtx_mode_t *".
14251 (emit_library_call_value): Delete.
14252 (emit_library_call): Likewise.
14253 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14254 (asan_emit_allocas_unpoison): Likewise.
14255 * builtins.c (expand_builtin_powi): Likewise.
14256 (expand_asan_emit_allocas_unpoison): Likewise.
14257 * cfgexpand.c (expand_main_function): Likewise.
14258 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14259 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14260 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14261 * config/arm/arm.c (arm_trampoline_init): Likewise.
14262 (arm_call_tls_get_addr): Likewise.
14263 (arm_expand_divmod_libfunc): Likewise.
14264 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14265 (smulsi3_highpart): Likewise.
14266 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14267 (c6x_expand_compare): Likewise.
14268 (c6x_expand_movmem): Likewise.
14269 * config/frv/frv.c (frv_trampoline_init): Likewise.
14270 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14271 (ix86_expand_divmod_libfunc): Likewise.
14272 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14273 (ia64_expand_compare): Likewise.
14274 (ia64_profile_hook): Likewise.
14275 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14276 (nonlocal_goto): Likewise.
14277 (restore_stack_nonlocal): Likewise.
14278 * config/m32r/m32r.c (block_move_call): Likewise.
14279 (m32r_trampoline_init): Likewise.
14280 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14281 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14282 (m68k_call_m68k_read_tp): Likewise.
14283 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14284 (microblaze_expand_divide): Likewise.
14285 * config/mips/mips.h (mips_args): Likewise.
14286 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14287 (MIPS_ICACHE_SYNC): Likewise.
14288 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14289 (nios2_trampoline_init): Likewise.
14290 * config/pa/pa.c (hppa_tls_call): Likewise.
14291 (pa_trampoline_init): Likewise.
14292 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14293 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14294 (expand_strn_compare): Likewise.
14295 (rs6000_generate_compare): Likewise.
14296 (rs6000_expand_float128_convert): Likewise.
14297 (output_profile_hook): Likewise.
14298 (rs6000_trampoline_init): Likewise.
14299 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14300 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14301 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14302 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14303 (rs6000_generate_compare): Likewise.
14304 (rs6000_expand_float128_convert): Likewise.
14305 (output_profile_hook): Likewise.
14306 (rs6000_trampoline_init): Likewise.
14307 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14308 * config/sh/sh.c (sh_trampoline_init): Likewise.
14309 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14310 (sparc_emit_float_lib_cmp): Likewise.
14311 (sparc32_initialize_trampoline): Likewise.
14312 (sparc64_initialize_trampoline): Likewise.
14313 (sparc_profile_hook): Likewise.
14314 * config/spu/spu.c (ea_load_store): Likewise.
14315 * config/spu/spu.md (floatunssidf2): Likewise.
14316 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14317 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14318 * config/visium/visium.c (expand_block_move_4): Likewise.
14319 (expand_block_move_2): Likewise.
14320 (expand_block_move_1): Likewise.
14321 (expand_block_set_4): Likewise.
14322 (expand_block_set_2): Likewise.
14323 (expand_block_set_1): Likewise.
14324 (visium_trampoline_init): Likewise.
14325 (visium_profile_hook): Likewise.
14326 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14327 (xtensa_setup_frame_addresses): Likewise.
14328 (xtensa_trampoline_init): Likewise.
14329 * except.c (sjlj_emit_function_enter): Likewise.
14330 (sjlj_emit_function_exit): Likewise.
14331 * explow.c (allocate_dynamic_stack_space): Likewise.
14332 (probe_stack_range): Likewise.
14333 * expr.c (convert_mode_scalar): Likewise.
14334 * optabs.c (expand_binop): Likewise.
14335 (expand_twoval_binop_libfunc): Likewise.
14336 (expand_unop): Likewise.
14337 (prepare_cmp_insn): Likewise.
14338 (prepare_float_lib_cmp): Likewise.
14339 (expand_float): Likewise.
14340 (expand_fix): Likewise.
14341 (expand_fixed_convert): Likewise.
14342 (maybe_emit_sync_lock_test_and_set): Likewise.
14343 (expand_atomic_compare_and_swap): Likewise.
14344 (expand_mem_thread_fence): Likewise.
14345 (expand_atomic_fetch_op): Likewise.
14347 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14349 * doc/generic.texi (OpenACC): Adjust URL.
14350 * doc/invoke.texi (C Dialect Options): Ditto.
14352 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14354 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14355 predicate for operand 1. Add (m,<S>) constraint.
14356 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14357 Prevent memory operand 1 with register operand 2.
14359 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14361 PR rtl-optimization/82024
14362 * combine.c (try_combine): If the combination result is a PARALLEL,
14363 and we only need to retain the SET in there that would be placed
14364 at I2, check that we can place that at I3 instead, before doing so.
14366 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14369 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14371 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14372 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14375 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14376 Jeff Law <law@redhat.com>
14378 * varasm.c (bss_initializer_p): Do not put constants into .bss
14379 (categorize_decl_for_section): Handle bss_initializer_p returning
14380 false when DECL_INITIAL is NULL.
14382 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14385 * config/s390/s390.c (s390_can_inline_p): New function.
14387 2017-09-01 Jeff Law <law@redhat.com>
14389 PR tree-optimization/82052
14390 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14391 Always initialize the returned slot after a hash table miss
14392 when INSERT is true.
14394 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14396 * config/s390/s390.md (mem_signal_fence): Remove.
14397 * doc/md.texi (mem_signal_fence): Remove.
14398 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14400 * target-insns.def (mem_signal_fence): Remove.
14402 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14405 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14408 * asan.c (create_odr_indicator): Strip name encoding from assembler
14409 name before appending it after __odr_asan_.
14411 2017-09-01 Martin Liska <mliska@suse.cz>
14413 PR tree-optimization/82059
14414 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14415 frequency only when an edge is redirected.
14417 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14419 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14420 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14421 (arc_conditional_register_usage): Remove ARC600 lp_count
14423 (arc_file_start): Emit Tag_ARC_CPU_variation.
14424 (arc_can_use_doloop_p): New conditions to use ZOLs.
14425 (hwloop_fail): New function.
14426 (hwloop_optimize): Likewise.
14427 (hwloop_pattern_reg): Likewise.
14428 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14429 (arc_reorg_loops): New function, calls reorg_loops.
14430 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14431 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14433 (arc_loop_hazard): Remove function, functionality moved into
14435 (arc_hazard): Remove arc_loop_hazard call.
14436 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14437 into hwloop_optimize.
14438 (arc_label_align): Remove ZOL handling.
14439 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14440 * config/arc/arc.md (doloop_begin): Remove pattern.
14441 (doloop_begin_i): Likewise.
14442 (doloop_end_i): Likewise.
14443 (doloop_fallback): Likewise.
14444 (doloop_fallback_m): Likewise.
14445 (doloop_end): Reimplement expand.
14446 (arc_lp): New pattern for LP instruction.
14447 (loop_end): New pattern.
14448 (loop_fail): Likewise.
14449 (decrement_and_branch_until_zero): Likewise.
14450 * config/arc/arc.opt (mlpc-width): New option.
14451 * doc/invoke.texi (mlpc-width): Document option.
14453 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14455 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14456 (arc_ccfsm_advance): Fix checking for delay slots.
14457 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14459 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14461 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14463 (movhi_insn): Update store instruction constraint which are saving
14464 6-bit short immediates.
14465 (movsi_insn): Consider also short scaled load operations.
14466 (zero_extendhisi2_i): Use Usd constraint instead of T.
14467 (extendhisi2_i): Add q constraint.
14468 (arc_clzsi2): Add type and length attributes.
14469 (arc_ctzsi2): Likewise.
14470 * config/arc/constraints.md (Usc): Update constraint, the
14471 assembler can parse two relocations for a single instruction.
14473 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14475 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14476 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14478 2017-08-31 Olivier Hainque <hainque@adacore.com>
14480 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14481 match as powerpc-wrs-vxworks*.
14483 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14485 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14486 register constraint for by-element operand.
14487 (aarch64_mls_elt_merge<mode>): Likewise.
14489 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14491 * config/arc/arc.c (arc_can_follow_jump): Check for short
14494 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14496 * config.gcc: Use g.opt for arc.
14497 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14498 functionality moved to ...
14499 (legitimate_scaled_address_p): New function, ...here.
14500 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14501 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14502 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14504 (arc_override_options): Handle G option.
14505 (arc_output_pic_addr_const): Correct function definition.
14506 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14507 (arc_decl_anon_ns_mem_p): Delete.
14508 (arc_in_small_data_p): Overhaul this function to take into
14509 consideration the value given via G option.
14510 (arc_rewrite_small_data_1): Renamed and corrected old
14511 arc_rewrite_small_data function.
14512 (arc_rewrite_small_data): New function.
14513 (small_data_pattern): Don't use pic_offset_table_rtx.
14514 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14515 * config/arc/simdext.md (movmisalignv2hi): Use
14516 prepare_move_operands function.
14518 (movmisalign*): Likewise.
14519 * doc/invoke.texi (ARC options): Document -G option.
14521 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14523 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14525 * config/arc/arc.c (arc_print_operand): Output scalled address for
14526 sdata whenever is possible.
14527 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14528 load/stores are available.
14529 (compact_sda_memory_operand): Check for the alignment required by
14530 code density instructions.
14531 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14533 * config/arc/constraints.md (Usd): Update constraint.
14534 (Us0): New constraint.
14535 (Usc): Update constraint.
14537 2017-08-31 Richard Biener <rguenther@suse.de>
14539 PR middle-end/82054
14540 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14541 function only once.
14543 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14545 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14546 Resize type_signature.
14548 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14549 Alan Hayward <alan.hayward@arm.com>
14550 David Sherwood <david.sherwood@arm.com>
14552 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14553 subregs whose inner modes can be stored in GPRs.
14554 (aarch64_classify_index): Likewise.
14556 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14557 Alan Hayward <alan.hayward@arm.com>
14558 David Sherwood <david.sherwood@arm.com>
14560 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14561 (V_INT_EQUIV): ...this.
14562 (v_cmp_result): Rename to...
14563 (v_int_equiv): ...this.
14564 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14565 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14566 (copysign<mode>3): Likewise.
14567 (aarch64_simd_bsl<mode>_internal): Likewise.
14568 (aarch64_simd_bsl<mode>): Likewise.
14569 (vec_cmp<mode><mode>): Likewise.
14570 (vcond<mode><mode>): Likewise.
14571 (vcond<v_cmp_mixed><mode>): Likewise.
14572 (vcondu<mode><v_cmp_mixed>): Likewise.
14573 (aarch64_cm<optab><mode>): Likewise.
14574 (aarch64_cmtst<mode>): Likewise.
14575 (aarch64_fac<optab><mode>): Likewise.
14576 (vec_perm_const<mode>): Likewise.
14577 (vcond_mask_<mode><v_cmp_result>): Rename to...
14578 (vcond_mask_<mode><v_int_equiv>): ...this.
14579 (vec_cmp<mode><v_cmp_result>): Rename to...
14580 (vec_cmp<mode><v_int_equiv>): ...this.
14582 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14583 Alan Hayward <alan.hayward@arm.com>
14584 David Sherwood <david.sherwood@arm.com>
14586 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14588 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14589 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14590 (UNSPEC_LD4_DREG): New unspecs.
14591 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14592 (aarch64_ld2<mode>_dreg_be): Replace with...
14593 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14595 (aarch64_ld3<mode>_dreg_le)
14596 (aarch64_ld3<mode>_dreg_be): Replace with...
14597 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14599 (aarch64_ld4<mode>_dreg_le)
14600 (aarch64_ld4<mode>_dreg_be): Replace with...
14601 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14604 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14606 PR tree-optimization/81987
14607 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14608 insert an initializer in a location not dominated by the stride
14611 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14613 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14614 on the entire header of the finally block in the fallthru case.
14616 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14618 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14620 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14622 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14623 rs6000_emit_move_from_cr and call renamed function.
14624 (rs6000_emit_prologue): Call renamed functions.
14625 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14626 movesi_from_cr, remove volatile CRs.
14628 2017-08-30 Jon Beniston <jon@beniston.com>
14629 Richard Biener <rguenther@suse.de>
14631 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14632 of VECTOR_MODE_P check.
14633 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14634 element vector types.
14636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14638 * df.h (df_read_modify_subreg_p): Remove in favor of...
14639 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14640 const_rtx instead of an rtx.
14641 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14642 * df-problems.c (df_word_lr_mark_ref): Likewise.
14643 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14644 (mark_pseudo_reg_dead): Likewise.
14645 (mark_ref_dead): Likewise.
14646 * reginfo.c (init_subregs_of_mode): Likewise.
14647 * sched-deps.c (sched_analyze_1): Likewise.
14648 * df-scan.c (df_def_record_1): Likewise.
14649 (df_uses_record): Likewise.
14650 (df_read_modify_subreg_p): Remove in favor of...
14651 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14652 const_rtx instead of an rtx.
14654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14655 Alan Hayward <alan.hayward@arm.com>
14656 David Sherwood <david.sherwood@arm.com>
14658 * rtl.h (partial_subreg_p): New function.
14659 * caller-save.c (save_call_clobbered_regs): Use it.
14660 * calls.c (expand_call): Likewise.
14661 * combine.c (combinable_i3pat): Likewise.
14662 (simplify_set): Likewise.
14663 (make_extraction): Likewise.
14664 (make_compound_operation_int): Likewise.
14665 (gen_lowpart_or_truncate): Likewise.
14666 (force_to_mode): Likewise.
14667 (make_field_assignment): Likewise.
14668 (reg_truncated_to_mode): Likewise.
14669 (record_truncated_value): Likewise.
14670 (move_deaths): Likewise.
14671 * cse.c (record_jump_cond): Likewise.
14672 (cse_insn): Likewise.
14673 * cselib.c (cselib_lookup_1): Likewise.
14674 * expmed.c (extract_bit_field_using_extv): Likewise.
14675 * function.c (assign_parm_setup_reg): Likewise.
14676 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14677 * ira-build.c (create_insn_allocnos): Likewise.
14678 * lra-coalesce.c (merge_pseudos): Likewise.
14679 * lra-constraints.c (match_reload): Likewise.
14680 (simplify_operand_subreg): Likewise.
14681 (curr_insn_transform): Likewise.
14682 * lra-lives.c (process_bb_lives): Likewise.
14683 * lra.c (new_insn_reg): Likewise.
14684 (lra_substitute_pseudo): Likewise.
14685 * regcprop.c (mode_change_ok): Likewise.
14686 (maybe_mode_change): Likewise.
14687 (copyprop_hardreg_forward_1): Likewise.
14688 * reload.c (push_reload): Likewise.
14689 (find_reloads): Likewise.
14690 (find_reloads_subreg_address): Likewise.
14691 * reload1.c (alter_reg): Likewise.
14692 (eliminate_regs_1): Likewise.
14693 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14695 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14697 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14698 back to if statements, including unpack.
14700 2017-08-30 Martin Liska <mliska@suse.cz>
14702 PR inline-asm/82001
14703 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14705 (func_checker::compare_asm_inputs_outputs): ... this function.
14706 (func_checker::compare_gimple_asm): Use the function to compare
14707 also ASM constrains.
14708 * ipa-icf-gimple.h: Rename the function.
14710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14711 Alan Hayward <alan.hayward@arm.com>
14712 David Sherwood <david.sherwood@arm.com>
14714 * coretypes.h (complex_mode): New type.
14715 * gdbhooks.py (build_pretty_printer): Handle it.
14716 * machmode.h (complex_mode): New class.
14717 (complex_mode::includes_p): New function.
14718 (is_complex_int_mode): Likewise.
14719 (is_complex_float_mode): Likewise.
14720 * genmodes.c (get_mode_class): Handle complex mode classes.
14721 * function.c (expand_function_end): Use is_complex_int_mode.
14723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14724 Alan Hayward <alan.hayward@arm.com>
14725 David Sherwood <david.sherwood@arm.com>
14727 * coretypes.h (scalar_mode_pod): New typedef.
14728 * gdbhooks.py (build_pretty_printer): Handle it.
14729 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14730 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14731 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14732 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14733 as_a <scalar_mode>.
14735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14736 Alan Hayward <alan.hayward@arm.com>
14737 David Sherwood <david.sherwood@arm.com>
14739 * machmode.h (mode_for_vector): Take a scalar_mode instead
14741 * stor-layout.c (mode_for_vector): Likewise.
14742 * explow.c (promote_mode): Use as_a <scalar_mode>.
14743 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14746 Alan Hayward <alan.hayward@arm.com>
14747 David Sherwood <david.sherwood@arm.com>
14749 * target.def (preferred_simd_mode): Take a scalar_mode
14750 instead of a machine_mode.
14751 * targhooks.h (default_preferred_simd_mode): Likewise.
14752 * targhooks.c (default_preferred_simd_mode): Likewise.
14753 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14754 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14755 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14756 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14757 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14758 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14759 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14760 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14762 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14763 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14764 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14765 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14766 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14767 * doc/tm.texi: Regenerate.
14768 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14772 Alan Hayward <alan.hayward@arm.com>
14773 David Sherwood <david.sherwood@arm.com>
14775 * target.def (scalar_mode_supported_p): Take a scalar_mode
14776 instead of a machine_mode.
14777 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14778 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14779 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14780 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14781 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14782 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14783 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14784 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14785 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14786 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14787 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14788 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14789 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14790 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14792 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14793 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14794 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14795 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14796 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14798 * doc/tm.texi: Regenerate.
14800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14801 Alan Hayward <alan.hayward@arm.com>
14802 David Sherwood <david.sherwood@arm.com>
14804 * coretypes.h (opt_scalar_mode): New typedef.
14805 * gdbhooks.py (build_pretty_printers): Handle it.
14806 * machmode.h (mode_iterator::get_2xwider): Add overload for
14808 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14810 * expr.c (convert_mode_scalar): Likewise.
14811 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14812 * optabs.c (expand_float): Likewise.
14813 (expand_fix): Likewise.
14814 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14816 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14817 Alan Hayward <alan.hayward@arm.com>
14818 David Sherwood <david.sherwood@arm.com>
14820 * optabs.c (expand_float): Explicitly check for scalars before
14821 using a branching expansion.
14822 (expand_fix): Likewise.
14824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14825 Alan Hayward <alan.hayward@arm.com>
14826 David Sherwood <david.sherwood@arm.com>
14828 * expr.c (convert_mode): Split scalar handling out into...
14829 (convert_mode_scalar): ...this new function. Treat the modes
14832 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14833 Alan Hayward <alan.hayward@arm.com>
14834 David Sherwood <david.sherwood@arm.com>
14836 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14838 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14841 Alan Hayward <alan.hayward@arm.com>
14842 David Sherwood <david.sherwood@arm.com>
14844 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14845 rather than a machine_mode.
14846 (fixed_from_string): Likewise.
14847 (fixed_convert): Likewise.
14848 (fixed_convert_from_int): Likewise.
14849 (fixed_convert_from_real): Likewise.
14850 (real_convert_from_fixed): Likewise.
14851 * fixed-value.c (fixed_from_double_int): Likewise.
14852 (fixed_from_string): Likewise.
14853 (fixed_convert): Likewise.
14854 (fixed_convert_from_int): Likewise.
14855 (fixed_convert_from_real): Likewise.
14856 (real_convert_from_fixed): Likewise.
14857 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14860 Alan Hayward <alan.hayward@arm.com>
14861 David Sherwood <david.sherwood@arm.com>
14863 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14864 of separate mode class checks. Do not allow vector modes here.
14865 (immed_wide_int_const): Use as_a <scalar_mode>.
14866 * explow.c (trunc_int_for_mode): Likewise.
14867 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14868 (wi::shwi): Likewise.
14869 (wi::min_value): Likewise.
14870 (wi::max_value): Likewise.
14871 * dwarf2out.c (loc_descriptor): Likewise.
14872 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14873 for CONST_WIDE_INT.
14875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14876 Alan Hayward <alan.hayward@arm.com>
14877 David Sherwood <david.sherwood@arm.com>
14879 * tree.h (SCALAR_TYPE_MODE): New macro.
14880 * expr.c (expand_expr_addr_expr_1): Use it.
14881 (expand_expr_real_2): Likewise.
14882 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14883 (fold_convert_const_fixed_from_int): Likewise.
14884 (fold_convert_const_fixed_from_real): Likewise.
14885 (native_encode_fixed): Likewise
14886 (native_encode_complex): Likewise
14887 (native_encode_vector): Likewise.
14888 (native_interpret_fixed): Likewise.
14889 (native_interpret_real): Likewise.
14890 (native_interpret_complex): Likewise.
14891 (native_interpret_vector): Likewise.
14892 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14893 (simd_clone_adjust_argument_types): Likewise.
14894 (simd_clone_init_simd_arrays): Likewise.
14895 (simd_clone_adjust): Likewise.
14896 * stor-layout.c (layout_type): Likewise.
14897 * tree.c (build_minus_one_cst): Likewise.
14898 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14899 * tree-inline.c (estimate_move_cost): Likewise.
14900 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14901 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14902 (vectorizable_reduction): Likewise.
14903 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14904 (vect_recog_mixed_size_cond_pattern): Likewise.
14905 (check_bool_pattern): Likewise.
14906 (adjust_bool_pattern): Likewise.
14907 (search_type_for_mask_1): Likewise.
14908 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
14909 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14910 (vectorizable_load): Likewise.
14911 (vectorizable_store): Likewise.
14912 * ubsan.c (ubsan_encode_value): Likewise.
14913 * varasm.c (output_constant): Likewise.
14915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14916 Alan Hayward <alan.hayward@arm.com>
14917 David Sherwood <david.sherwood@arm.com>
14919 * coretypes.h (scalar_mode): New class.
14920 * machmode.h (scalar_mode): Likewise.
14921 (scalar_mode::includes_p): New function.
14922 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
14923 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
14924 * genmodes.c (get_mode_class): Handle remaining scalar modes.
14925 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
14926 * expmed.c (store_bit_field_1): Likewise.
14927 (extract_bit_field_1): Likewise.
14928 * expr.c (write_complex_part): Likewise.
14929 (read_complex_part): Likewise.
14930 (emit_move_complex_push): Likewise.
14931 (expand_expr_real_2): Likewise.
14932 * function.c (assign_parm_setup_reg): Likewise.
14933 (assign_parms_unsplit_complex): Likewise.
14934 * optabs.c (expand_binop): Likewise.
14935 * rtlanal.c (subreg_get_info): Likewise.
14936 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14937 * varasm.c (output_constant_pool_2): Likewise.
14939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14940 Alan Hayward <alan.hayward@arm.com>
14941 David Sherwood <david.sherwood@arm.com>
14943 * expmed.c (extract_high_half): Use scalar_int_mode and remove
14945 (expmed_mult_highpart_optab): Likewise.
14946 (expmed_mult_highpart): Likewise.
14948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14949 Alan Hayward <alan.hayward@arm.com>
14950 David Sherwood <david.sherwood@arm.com>
14952 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
14953 instead of a machine_mode.
14954 (builtin_memset_read_str): Likewise.
14955 * builtins.c (c_readstr): Likewise.
14956 (builtin_memcpy_read_str): Likewise.
14957 (builtin_strncpy_read_str): Likewise.
14958 (builtin_memset_read_str): Likewise.
14959 (builtin_memset_gen_str): Likewise.
14960 (expand_builtin_signbit): Use scalar_int_mode for local variables.
14961 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
14962 instead of a machine_mode.
14963 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
14965 (make_extraction): Likewise.
14966 (try_widen_shift_mode): Take and return scalar_int_modes instead
14968 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
14969 a scalar_int_mode instead of a machine_mode.
14970 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
14971 (avr_addr_space_pointer_mode): Likewise.
14972 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
14973 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
14974 (msp430_unwind_word_mode): Likewise.
14975 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
14976 (spu_addr_space_pointer_mode): Likewise.
14977 (spu_addr_space_address_mode): Likewise.
14978 (spu_libgcc_cmp_return_mode): Likewise.
14979 (spu_libgcc_shift_count_mode): Likewise.
14980 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
14981 (rl78_addr_space_pointer_mode): Likewise.
14982 (fl78_unwind_word_mode): Likewise.
14983 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
14985 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
14986 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
14987 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
14988 (mips_valid_pointer_mode): Likewise.
14989 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
14990 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
14991 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
14993 (ft32_addr_space_address_mode): Likewise.
14994 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
14995 scalar_int_mode instead of a machine_mode.
14996 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
14998 (m32c_addr_space_address_mode): Likewise.
14999 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15000 (rs6000_eh_return_filter_mode): Likewise.
15001 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15002 (rs6000_eh_return_filter_mode): Likewise.
15003 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15004 (s390_libgcc_shift_count_mode): Likewise.
15005 (s390_unwind_word_mode): Likewise.
15006 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15008 * target.def (mode_rep_extended): Likewise.
15009 (valid_pointer_mode): Likewise.
15010 (addr_space.valid_pointer_mode): Likewise.
15011 (eh_return_filter_mode): Return a scalar_int_mode rather than
15013 (libgcc_cmp_return_mode): Likewise.
15014 (libgcc_shift_count_mode): Likewise.
15015 (unwind_word_mode): Likewise.
15016 (addr_space.pointer_mode): Likewise.
15017 (addr_space.address_mode): Likewise.
15018 * doc/tm.texi: Regenerate.
15019 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15021 (do_jump): Use scalar_int_mode for local variables.
15022 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15023 rather than a machine_mode.
15024 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15025 (scompare_loc_descriptor_wide): Likewise.
15026 (scompare_loc_descriptor_narrow): Likewise.
15027 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15029 * except.c (sjlj_emit_dispatch_table): Likewise.
15030 (expand_builtin_eh_copy_values): Likewise.
15031 * explow.c (convert_memory_address_addr_space_1): Likewise.
15032 Take a scalar_int_mode rather than a machine_mode.
15033 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15034 than a machine_mode.
15035 (memory_address_addr_space): Use scalar_int_mode for local variables.
15036 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15037 rather than a machine_mode.
15038 * expmed.c (mask_rtx): Likewise.
15039 (init_expmed_one_conv): Likewise.
15040 (expand_mult_highpart_adjust): Likewise.
15041 (extract_high_half): Likewise.
15042 (expmed_mult_highpart_optab): Likewise.
15043 (expmed_mult_highpart): Likewise.
15044 (expand_smod_pow2): Likewise.
15045 (expand_sdiv_pow2): Likewise.
15046 (emit_store_flag_int): Likewise.
15047 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15049 (extract_low_bits): Likewise.
15050 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15052 * expr.c (pieces_addr::adjust): Likewise.
15053 (can_store_by_pieces): Likewise.
15054 (store_by_pieces): Likewise.
15055 (clear_by_pieces_1): Likewise.
15056 (expand_expr_addr_expr_1): Likewise.
15057 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15058 (expand_expr_real_1): Likewise.
15059 (try_casesi): Likewise.
15060 * final.c (shorten_branches): Likewise.
15061 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15062 type of "mode" to machine_mode.
15063 * internal-fn.c (expand_arith_overflow_result_store): Take a
15064 scalar_int_mode rather than a machine_mode.
15065 (expand_mul_overflow): Use scalar_int_mode for local variables.
15066 * loop-doloop.c (doloop_modify): Likewise.
15067 (doloop_optimize): Likewise.
15068 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15069 than a machine_mode.
15070 (expand_doubleword_shift_condmove): Likewise.
15071 (expand_doubleword_shift): Likewise.
15072 (expand_doubleword_clz): Likewise.
15073 (expand_doubleword_popcount): Likewise.
15074 (expand_doubleword_parity): Likewise.
15075 (expand_absneg_bit): Use scalar_int_mode for local variables.
15076 (prepare_float_lib_cmp): Likewise.
15077 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15078 rather than a machine_mode.
15079 (convert_memory_address_addr_space): Likewise.
15080 (get_mode_bounds): Likewise.
15081 (get_address_mode): Return a scalar_int_mode rather than a
15083 * rtlanal.c (get_address_mode): Likewise.
15084 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15085 than a machine_mode.
15086 * targhooks.c (default_mode_rep_extended): Likewise.
15087 (default_valid_pointer_mode): Likewise.
15088 (default_addr_space_valid_pointer_mode): Likewise.
15089 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15090 than a machine_mode.
15091 (default_libgcc_cmp_return_mode): Likewise.
15092 (default_libgcc_shift_count_mode): Likewise.
15093 (default_unwind_word_mode): Likewise.
15094 (default_addr_space_pointer_mode): Likewise.
15095 (default_addr_space_address_mode): Likewise.
15096 * targhooks.h (default_eh_return_filter_mode): Likewise.
15097 (default_libgcc_cmp_return_mode): Likewise.
15098 (default_libgcc_shift_count_mode): Likewise.
15099 (default_unwind_word_mode): Likewise.
15100 (default_addr_space_pointer_mode): Likewise.
15101 (default_addr_space_address_mode): Likewise.
15102 (default_mode_rep_extended): Take a scalar_int_mode rather than
15104 (default_valid_pointer_mode): Likewise.
15105 (default_addr_space_valid_pointer_mode): Likewise.
15106 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15108 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15109 rather than a machine_mode.
15110 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15111 for local variables.
15112 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15113 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15114 than a machine_mode.
15116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15117 Alan Hayward <alan.hayward@arm.com>
15118 David Sherwood <david.sherwood@arm.com>
15120 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15121 the mode argument to scalar_int_mode.
15122 (do_jump_by_parts_zero_rtx): Likewise.
15123 (do_jump_by_parts_equality_rtx): Likewise.
15124 (do_jump_by_parts_greater): Take a mode argument.
15125 (do_jump_by_parts_equality): Likewise.
15126 (do_jump_1): Update calls accordingly.
15128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15129 Alan Hayward <alan.hayward@arm.com>
15130 David Sherwood <david.sherwood@arm.com>
15132 * is-a.h (safe_dyn_cast): New function.
15133 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15134 (jump_table_for_label): Likewise.
15135 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15136 instead of an rtx_insn *.
15137 (shorten_branches): Use dyn_cast instead of LABEL_P and
15138 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15139 rtx_jump_table_data::get_data_mode.
15140 (final_scan_insn): Likewise.
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 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15147 trying to combine a full-register integer set with a subreg
15150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15151 Alan Hayward <alan.hayward@arm.com>
15152 David Sherwood <david.sherwood@arm.com>
15154 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15155 that is always either address_mode or pointer_mode.
15157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15158 Alan Hayward <alan.hayward@arm.com>
15159 David Sherwood <david.sherwood@arm.com>
15161 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15162 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
15169 is_a <scalar_int_mode> instead of checking for a nonzero
15170 precision. Forcibly convert op_mode to a scalar_int_mode
15171 in that case. More clearly differentiate the operand and
15172 result modes and use the former when deciding what the value
15173 of a count-bits operation should be. Use is_int_mode instead
15174 of checking for a MODE_INT. Remove redundant check for whether
15175 this mode has a zero precision.
15177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15178 Alan Hayward <alan.hayward@arm.com>
15179 David Sherwood <david.sherwood@arm.com>
15181 * optabs.c (widen_leading): Change the type of the mode argument
15182 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15183 (widen_bswap): Likewise.
15184 (expand_parity): Likewise.
15185 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15186 (expand_ffs): Likewise.
15187 (epand_unop): Check for scalar integer modes before calling the
15190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15191 Alan Hayward <alan.hayward@arm.com>
15192 David Sherwood <david.sherwood@arm.com>
15194 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15196 (expand_expr_real_1): Update call accordingly.
15198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15199 Alan Hayward <alan.hayward@arm.com>
15200 David Sherwood <david.sherwood@arm.com>
15202 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15203 value_mode arguments. Use scalar_int_mode internally.
15204 (store_bit_field_1): Rename the new integer mode from imode
15205 to op0_mode and use it instead of GET_MODE (op0). Update calls
15206 to store_split_bit_field, store_bit_field_using_insv and
15207 store_fixed_bit_field.
15208 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15209 Use scalar_int_mode internally. Use a bit count rather than a mode
15210 when calculating the largest bit size for get_best_mode.
15211 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15212 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15213 Remove assertion that OP0 has a scalar integer mode.
15214 (store_split_bit_field): Add op0_mode and value_mode arguments.
15215 Update calls to extract_fixed_bit_field.
15216 (extract_bit_field_using_extv): Add an op0_mode argument.
15217 Use scalar_int_mode internally.
15218 (extract_bit_field_1): Rename the new integer mode from imode to
15219 op0_mode and use it instead of GET_MODE (op0). Update calls to
15220 extract_split_bit_field, extract_bit_field_using_extv and
15221 extract_fixed_bit_field.
15222 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15223 to extract_split_bit_field and extract_fixed_bit_field_1.
15224 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15225 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15226 on the target mode.
15227 (extract_split_bit_field): Add an op0_mode argument. Update call
15228 to extract_fixed_bit_field.
15230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15231 Alan Hayward <alan.hayward@arm.com>
15232 David Sherwood <david.sherwood@arm.com>
15234 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15235 * explow.c (hard_function_value): Likewise.
15236 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15237 convert_to_mode call outside the loop.
15238 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15239 for the mode iterator. Require the mode specified by max_pieces
15241 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15243 (copy_blkmode_to_reg): Likewise.
15244 (set_storage_via_setmem): Likewise.
15245 * optabs.c (prepare_cmp_insn): Likewise.
15246 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15247 * stor-layout.c (finish_bitfield_representative): Likewise.
15249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15250 Alan Hayward <alan.hayward@arm.com>
15251 David Sherwood <david.sherwood@arm.com>
15253 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15254 * expr.c (convert_move): Use them.
15255 (convert_modes): Likewise.
15256 (store_expr_with_bounds): Likewise.
15258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15259 Alan Hayward <alan.hayward@arm.com>
15260 David Sherwood <david.sherwood@arm.com>
15262 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15263 parameter for the mode of "x". Remove the "known_x", "known_mode"
15264 and "known_ret" arguments. Change the type of the mode argument
15265 to scalar_int_mode.
15266 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15267 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15268 (reg_num_sign_bit_copies_for_combine): Likewise.
15269 * rtlanal.c (nonzero_bits1): Likewise.
15270 (num_sign_bit_copies1): Likewise.
15271 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15272 (reg_num_sign_bit_copies_general): Likewise.
15273 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15274 (reg_nonzero_bits_general): Likewise.
15276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15277 Alan Hayward <alan.hayward@arm.com>
15278 David Sherwood <david.sherwood@arm.com>
15280 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15281 than in subroutines. Return 1 for non-integer modes.
15282 (cached_num_sign_bit_copies): Change the type of the mode parameter
15283 to scalar_int_mode.
15284 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15285 classes. Handle CONST_INT_P first and then check whether X also
15286 has a scalar integer mode. Check the same thing for inner registers
15287 of a SUBREG and for values that are being extended or truncated.
15289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15290 Alan Hayward <alan.hayward@arm.com>
15291 David Sherwood <david.sherwood@arm.com>
15293 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15294 in subroutines. Return the mode mask for non-integer modes.
15295 (cached_nonzero_bits): Change the type of the mode parameter
15296 to scalar_int_mode.
15297 (nonzero_bits1): Likewise. Remove early exit for other mode
15298 classes. Handle CONST_INT_P first and then check whether X
15299 also has a scalar integer mode.
15301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15302 Alan Hayward <alan.hayward@arm.com>
15303 David Sherwood <david.sherwood@arm.com>
15305 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15306 Return a scalar_int_mode and assert that the size is greater than
15308 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15309 (op_by_pieces_d::op_by_pieces_d): Likewise.
15310 (op_by_pieces_d::run): Likewise.
15311 (can_store_by_pieces): Likewise.
15313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15314 Alan Hayward <alan.hayward@arm.com>
15315 David Sherwood <david.sherwood@arm.com>
15317 * combine.c (extract_left_shift): Add a mode argument and update
15319 (make_compound_operation_int): Change the type of the mode parameter
15320 to scalar_int_mode and update the call to extract_left_shift.
15322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15323 Alan Hayward <alan.hayward@arm.com>
15324 David Sherwood <david.sherwood@arm.com>
15326 * combine.c (simplify_and_const_int): Change the type of the mode
15327 parameter to scalar_int_mode.
15328 (simplify_and_const_int_1): Likewise. Update recursive call.
15330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15331 Alan Hayward <alan.hayward@arm.com>
15332 David Sherwood <david.sherwood@arm.com>
15334 * combine.c (simplify_compare_const): Check that the mode is a
15335 scalar_int_mode (rather than VOIDmode) before testing its
15337 (simplify_comparison): Move COMPARISON_P handling out of the
15338 loop and restrict the latter part of the loop to scalar_int_modes.
15339 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15340 and when considering SUBREG_REGs. Use is_int_mode instead of
15341 checking GET_MODE_CLASS against MODE_INT.
15343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15344 Alan Hayward <alan.hayward@arm.com>
15345 David Sherwood <david.sherwood@arm.com>
15347 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15348 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15349 shift_unit_mode and for modes involved in scalar shifts.
15351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15352 Alan Hayward <alan.hayward@arm.com>
15353 David Sherwood <david.sherwood@arm.com>
15355 * combine.c (force_int_to_mode): New function, split out from...
15356 (force_to_mode): ...here. Keep xmode up-to-date and use it
15357 instead of GET_MODE (x).
15359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15360 Alan Hayward <alan.hayward@arm.com>
15361 David Sherwood <david.sherwood@arm.com>
15363 * optabs-query.h (extraction_insn::struct_mode): Change type to
15364 opt_scalar_int_mode and update comment.
15365 (extraction_insn::field_mode): Change type to scalar_int_mode.
15366 (extraction_insn::pos_mode): Likewise.
15367 * combine.c (make_extraction): Update accordingly.
15368 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15369 (get_optab_extraction_insn): Likewise.
15370 * recog.c (simplify_while_replacing): Likewise.
15371 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15372 parameter to opt_scalar_int_mode.
15374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15375 Alan Hayward <alan.hayward@arm.com>
15376 David Sherwood <david.sherwood@arm.com>
15378 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15379 to a scalar_int_mode instead of a machine_mode.
15380 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15381 (get_best_mode): Return a boolean and use a pointer argument to store
15382 the selected mode. Replace the limit mode parameter with a bit limit.
15383 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15384 for the values returned by bit_field_mode_iterator::next_mode.
15385 (store_bit_field): Update call to get_best_mode.
15386 (store_fixed_bit_field): Likewise.
15387 (extract_fixed_bit_field): Likewise.
15388 * expr.c (optimize_bitfield_assignment_op): Likewise.
15389 * fold-const.c (optimize_bit_field_compare): Likewise.
15390 (fold_truth_andor_1): Likewise.
15391 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15392 Update for new type of m_mode.
15393 (get_best_mode): As above.
15395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15396 Alan Hayward <alan.hayward@arm.com>
15397 David Sherwood <david.sherwood@arm.com>
15399 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15400 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15401 (store_bit_field): Check is_a <scalar_int_mode> before calling
15402 strict_volatile_bitfield_p.
15403 (extract_bit_field): Likewise.
15405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15406 Alan Hayward <alan.hayward@arm.com>
15407 David Sherwood <david.sherwood@arm.com>
15409 * target.def (cstore_mode): Return a scalar_int_mode.
15410 * doc/tm.texi: Regenerate.
15411 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15412 * targhooks.h (default_cstore_mode): Likewise.
15413 * targhooks.c (default_cstore_mode): Likewise, using a forced
15415 * expmed.c (emit_cstore): Expect the target of the cstore to be
15418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15419 Alan Hayward <alan.hayward@arm.com>
15420 David Sherwood <david.sherwood@arm.com>
15422 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15424 (niter_desc): Likewise mode.
15425 (iv_analyze): Add a mode parameter.
15427 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15428 and change its type to scalar_int_mode.
15429 * loop-iv.c: Update commentary at head of file.
15430 (iv_constant): Pass the mode paraeter before the rtx it describes
15431 and change its type to scalar_int_mode. Remove VOIDmode handling.
15432 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15433 (iv_extend): Likewise.
15434 (shorten_into_mode): Likewise.
15435 (iv_add): Use scalar_int_mode.
15436 (iv_mult): Likewise.
15437 (iv_shift): Likewise.
15438 (canonicalize_iv_subregs): Likewise.
15439 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15440 it describes and change its mode to scalar_int_mode. Also change
15441 the type of the returned inner_mode to scalar_int_mode.
15442 (get_biv_step): Likewise, turning outer_mode from a pointer
15443 into a direct parameter. Update call to get_biv_step_1.
15444 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15445 iv_constant and get_biv_step.
15446 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15447 and change its type to scalar_int_mode. Don't initialise iv->mode
15448 to VOIDmode and remove later checks for its still being VOIDmode.
15449 Update calls to iv_analyze_op and iv_analyze_expr. Check
15450 is_a <scalar_int_mode> when changing the mode under consideration.
15451 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15452 Update call to iv_analyze_expr.
15453 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15454 inner register is not also a scalar_int_mode. Update call to
15456 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15457 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15458 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15459 separate mode class checks. Update calls to iv_analyze. Remove
15460 fix-up of VOIDmodes after iv_analyze_biv.
15461 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15462 don't have a scalar_int_mode. Update call to biv_p.
15464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15465 Alan Hayward <alan.hayward@arm.com>
15466 David Sherwood <david.sherwood@arm.com>
15468 * cfgexpand.c (convert_debug_memory_address): Use
15469 as_a <scalar_int_mode>.
15470 * combine.c (expand_compound_operation): Likewise.
15471 (make_extraction): Likewise.
15472 (change_zero_ext): Likewise.
15473 (simplify_comparison): Likewise.
15474 * cse.c (cse_insn): Likewise.
15475 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15476 (mem_loc_descriptor): Likewise.
15477 (loc_descriptor): Likewise.
15478 * expmed.c (init_expmed_one_mode): Likewise.
15479 (synth_mult): Likewise.
15480 (emit_store_flag_1): Likewise.
15481 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15482 of a comparison with size.
15483 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15484 (reduce_to_bit_field_precision): Likewise.
15485 * function.c (expand_function_end): Likewise.
15486 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15487 * loop-doloop.c (doloop_modify): Likewise.
15488 * optabs.c (expand_binop): Likewise.
15489 (expand_unop): Likewise.
15490 (expand_copysign_absneg): Likewise.
15491 (prepare_cmp_insn): Likewise.
15492 (maybe_legitimize_operand): Likewise.
15493 * recog.c (const_scalar_int_operand): Likewise.
15494 * rtlanal.c (get_address_mode): Likewise.
15495 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15496 (simplify_cond_clz_ctz): Likewise.
15497 * tree-nested.c (get_nl_goto_field): Likewise.
15498 * tree.c (build_vector_type_for_mode): Likewise.
15499 * var-tracking.c (use_narrower_mode): Likewise.
15501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15502 Alan Hayward <alan.hayward@arm.com>
15503 David Sherwood <david.sherwood@arm.com>
15505 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15506 * builtins.c (expand_builtin_signbit): Use it.
15507 * cfgexpand.c (expand_debug_expr): Likewise.
15508 * dojump.c (do_jump): Likewise.
15509 (do_compare_and_jump): Likewise.
15510 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15511 * expmed.c (make_tree): Likewise.
15512 * expr.c (expand_expr_real_2): Likewise.
15513 (expand_expr_real_1): Likewise.
15514 (try_casesi): Likewise.
15515 * fold-const-call.c (fold_const_call_ss): Likewise.
15516 * fold-const.c (unextend): Likewise.
15517 (extract_muldiv_1): Likewise.
15518 (fold_single_bit_test): Likewise.
15519 (native_encode_int): Likewise.
15520 (native_encode_string): Likewise.
15521 (native_interpret_int): Likewise.
15522 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15523 * internal-fn.c (expand_addsub_overflow): Likewise.
15524 (expand_neg_overflow): Likewise.
15525 (expand_mul_overflow): Likewise.
15526 (expand_arith_overflow): Likewise.
15527 * match.pd: Likewise.
15528 * stor-layout.c (layout_type): Likewise.
15529 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15530 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15531 * tree-ssanames.c (get_range_info): Likewise.
15532 * tree-switch-conversion.c (array_value_type) Likewise.
15533 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15534 (vect_recog_divmod_pattern): Likewise.
15535 (vect_recog_mixed_size_cond_pattern): Likewise.
15536 * tree-vrp.c (extract_range_basic): Likewise.
15537 (simplify_float_conversion_using_ranges): Likewise.
15538 * tree.c (int_fits_type_p): Likewise.
15539 * ubsan.c (instrument_bool_enum_load): Likewise.
15540 * varasm.c (mergeable_string_section): Likewise.
15541 (narrowing_initializer_constant_valid_p): Likewise.
15542 (output_constant): Likewise.
15544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15545 Alan Hayward <alan.hayward@arm.com>
15546 David Sherwood <david.sherwood@arm.com>
15548 * machmode.h (NARROWEST_INT_MODE): New macro.
15549 * expr.c (alignment_for_piecewise_move): Use it instead of
15550 GET_CLASS_NARROWEST_MODE (MODE_INT).
15551 (push_block): Likewise.
15552 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15554 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15557 Alan Hayward <alan.hayward@arm.com>
15558 David Sherwood <david.sherwood@arm.com>
15560 * postreload.c (move2add_valid_value_p): Change the type of the
15561 mode parameter to scalar_int_mode.
15562 (move2add_use_add2_insn): Add a mode parameter and use it instead
15564 (move2add_use_add3_insn): Likewise.
15565 (reload_cse_move2add): Update accordingly.
15567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15568 Alan Hayward <alan.hayward@arm.com>
15569 David Sherwood <david.sherwood@arm.com>
15571 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15573 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15574 * optabs.c (expand_unop): Likewise.
15576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15577 Alan Hayward <alan.hayward@arm.com>
15578 David Sherwood <david.sherwood@arm.com>
15580 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15581 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15582 (popcount_loc_descriptor): Likewise.
15583 (bswap_loc_descriptor): Likewise.
15584 (rotate_loc_descriptor): Likewise.
15585 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15586 calling the functions above.
15588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15589 Alan Hayward <alan.hayward@arm.com>
15590 David Sherwood <david.sherwood@arm.com>
15592 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15594 (try_combine): Likewise.
15595 (simplify_if_then_else): Likewise.
15596 * cse.c (cse_insn): Likewise.
15597 * dwarf2out.c (mem_loc_descriptor): Likewise.
15598 * emit-rtl.c (gen_lowpart_common): Likewise.
15599 * simplify-rtx.c (simplify_truncation): Likewise.
15600 (simplify_binary_operation_1): Likewise.
15601 (simplify_const_relational_operation): Likewise.
15602 (simplify_ternary_operation): Likewise.
15603 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15606 Alan Hayward <alan.hayward@arm.com>
15607 David Sherwood <david.sherwood@arm.com>
15609 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15610 * reload.c (push_reload): Likewise.
15611 (find_reloads): Likewise.
15613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15614 Alan Hayward <alan.hayward@arm.com>
15615 David Sherwood <david.sherwood@arm.com>
15617 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15618 (make_compound_operation_int): Likewise.
15619 (change_zero_ext): Likewise.
15620 * expr.c (convert_move): Likewise.
15621 (convert_modes): Likewise.
15622 * fwprop.c (forward_propagate_subreg): Likewise.
15623 * loop-iv.c (get_biv_step_1): Likewise.
15624 * optabs.c (widen_operand): Likewise.
15625 * postreload.c (move2add_valid_value_p): Likewise.
15626 * recog.c (simplify_while_replacing): Likewise.
15627 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15628 (simplify_binary_operation_1): Likewise. Remove redundant
15629 mode equality check.
15631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15632 Alan Hayward <alan.hayward@arm.com>
15633 David Sherwood <david.sherwood@arm.com>
15635 * combine.c (combine_simplify_rtx): Add checks for
15636 is_a <scalar_int_mode>.
15637 (simplify_if_then_else): Likewise.
15638 (make_field_assignment): Likewise.
15639 (simplify_comparison): Likewise.
15640 * ifcvt.c (noce_try_bitop): Likewise.
15641 * loop-invariant.c (canonicalize_address_mult): Likewise.
15642 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15645 Alan Hayward <alan.hayward@arm.com>
15646 David Sherwood <david.sherwood@arm.com>
15648 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15649 is_a <scalar_int_mode> instead of != BLKmode.
15651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15652 Alan Hayward <alan.hayward@arm.com>
15653 David Sherwood <david.sherwood@arm.com>
15655 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15656 instead of != VOIDmode.
15657 * combine.c (if_then_else_cond): Likewise.
15658 (change_zero_ext): Likewise.
15659 * dwarf2out.c (mem_loc_descriptor): Likewise.
15660 (loc_descriptor): Likewise.
15661 * rtlanal.c (canonicalize_condition): Likewise.
15662 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15665 Alan Hayward <alan.hayward@arm.com>
15666 David Sherwood <david.sherwood@arm.com>
15668 * simplify-rtx.c (simplify_binary_operation_1): Use
15669 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15672 Alan Hayward <alan.hayward@arm.com>
15673 David Sherwood <david.sherwood@arm.com>
15675 * wide-int.h (int_traits<unsigned char>) New class.
15676 (int_traits<unsigned short>) Likewise.
15677 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15678 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15680 * combine.c (set_nonzero_bits_and_sign_copies): Use
15681 is_a <scalar_int_mode>.
15682 (find_split_point): Likewise.
15683 (combine_simplify_rtx): Likewise.
15684 (simplify_logical): Likewise.
15685 (expand_compound_operation): Likewise.
15686 (expand_field_assignment): Likewise.
15687 (make_compound_operation): Likewise.
15688 (extended_count): Likewise.
15689 (change_zero_ext): Likewise.
15690 (simplify_comparison): Likewise.
15691 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15692 (ucompare_loc_descriptor): Likewise.
15693 (minmax_loc_descriptor): Likewise.
15694 (mem_loc_descriptor): Likewise.
15695 (loc_descriptor): Likewise.
15696 * expmed.c (init_expmed_one_mode): Likewise.
15697 * lra-constraints.c (lra_constraint_offset): Likewise.
15698 * optabs.c (prepare_libcall_arg): Likewise.
15699 * postreload.c (move2add_note_store): Likewise.
15700 * reload.c (operands_match_p): Likewise.
15701 * rtl.h (load_extend_op): Likewise.
15702 * rtlhooks.c (gen_lowpart_general): Likewise.
15703 * simplify-rtx.c (simplify_truncation): Likewise.
15704 (simplify_unary_operation_1): Likewise.
15705 (simplify_binary_operation_1): Likewise.
15706 (simplify_const_binary_operation): Likewise.
15707 (simplify_const_relational_operation): Likewise.
15708 (simplify_subreg): Likewise.
15709 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15710 * var-tracking.c (adjust_mems): Likewise.
15711 (prepare_call_arguments): Likewise.
15713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15714 Alan Hayward <alan.hayward@arm.com>
15715 David Sherwood <david.sherwood@arm.com>
15717 * machmode.h (is_int_mode): New fuction.
15718 * combine.c (find_split_point): Use it.
15719 (combine_simplify_rtx): Likewise.
15720 (simplify_if_then_else): Likewise.
15721 (simplify_set): Likewise.
15722 (simplify_shift_const_1): Likewise.
15723 (simplify_comparison): Likewise.
15724 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15725 * cse.c (notreg_cost): Likewise.
15726 (cse_insn): Likewise.
15727 * cselib.c (cselib_lookup_1): Likewise.
15728 * dojump.c (do_jump_1): Likewise.
15729 (do_compare_rtx_and_jump): Likewise.
15730 * dse.c (get_call_args): Likewise.
15731 * dwarf2out.c (rtl_for_decl_init): Likewise.
15732 (native_encode_initializer): Likewise.
15733 * expmed.c (emit_store_flag_1): Likewise.
15734 (emit_store_flag): Likewise.
15735 * expr.c (convert_modes): Likewise.
15736 (store_field): Likewise.
15737 (expand_expr_real_1): Likewise.
15738 * fold-const.c (fold_read_from_constant_string): Likewise.
15739 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15740 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15741 * optabs.c (expand_binop): Likewise.
15742 (expand_unop): Likewise.
15743 (expand_abs_nojump): Likewise.
15744 (expand_one_cmpl_abs_nojump): Likewise.
15745 * simplify-rtx.c (mode_signbit_p): Likewise.
15746 (val_signbit_p): Likewise.
15747 (val_signbit_known_set_p): Likewise.
15748 (val_signbit_known_clear_p): Likewise.
15749 (simplify_relational_operation_1): Likewise.
15750 * tree.c (vector_type_mode): Likewise.
15752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15753 Alan Hayward <alan.hayward@arm.com>
15754 David Sherwood <david.sherwood@arm.com>
15756 * machmode.h (smallest_mode_for_size): Fix formatting.
15757 (smallest_int_mode_for_size): New function.
15758 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15759 instead of smallest_mode_for_size.
15760 * combine.c (make_extraction): Likewise.
15761 * config/arc/arc.c (arc_expand_movmem): Likewise.
15762 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15763 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15764 * config/s390/s390.c (s390_expand_insv): Likewise.
15765 * config/sparc/sparc.c (assign_int_registers): Likewise.
15766 * config/spu/spu.c (spu_function_value): Likewise.
15767 (spu_function_arg): Likewise.
15768 * coverage.c (get_gcov_type): Likewise.
15769 (get_gcov_unsigned_t): Likewise.
15770 * dse.c (find_shift_sequence): Likewise.
15771 * expmed.c (store_bit_field_1): Likewise.
15772 * expr.c (convert_move): Likewise.
15773 (store_field): Likewise.
15774 * internal-fn.c (expand_arith_overflow): Likewise.
15775 * optabs-query.c (get_best_extraction_insn): Likewise.
15776 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15777 * stor-layout.c (layout_type): Likewise.
15778 (initialize_sizetypes): Likewise.
15779 * targhooks.c (default_get_mask_mode): Likewise.
15780 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15783 Alan Hayward <alan.hayward@arm.com>
15784 David Sherwood <david.sherwood@arm.com>
15786 * machmode.h (opt_mode::else_blk): New function.
15787 (int_mode_for_mode): Declare.
15788 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15789 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15791 * cfgexpand.c (expand_debug_expr): Likewise.
15792 * combine.c (gen_lowpart_or_truncate): Likewise.
15793 (gen_lowpart_for_combine): Likewise.
15794 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15795 * config/avr/avr.c (avr_to_int_mode): Likewise.
15796 (avr_out_plus_1): Likewise.
15797 (avr_out_plus): Likewise.
15798 (avr_out_round): Likewise.
15799 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15800 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15801 (s390_expand_vcond): Likewise.
15802 * config/spu/spu.c (spu_split_immediate): Likewise.
15803 (spu_expand_mov): Likewise.
15804 * dse.c (get_stored_val): Likewise.
15805 * expmed.c (store_bit_field_1): Likewise.
15806 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15808 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15809 (extract_low_bits): Likewise.
15810 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15811 handling rather than repeating the check.
15812 (emit_group_store): Likewise.
15813 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15814 * optabs.c (expand_absneg_bit): Likewise.
15815 (expand_copysign_absneg): Likewise.
15816 (expand_copysign_bit): Likewise.
15817 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15818 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15819 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15820 * var-tracking.c (prepare_call_arguments): Likewise.
15821 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15822 int_mode_for_mode instead of mode_for_size.
15823 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15826 Alan Hayward <alan.hayward@arm.com>
15827 David Sherwood <david.sherwood@arm.com>
15829 * machmode.h (int_mode_for_size): New function.
15830 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15831 instead of mode_for_size.
15832 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15834 * combine.c (expand_field_assignment): Use int_mode_for_size
15835 instead of mode_for_size.
15836 (make_extraction): Likewise.
15837 (simplify_shift_const_1): Likewise.
15838 (simplify_comparison): Likewise.
15839 * dojump.c (do_jump): Likewise.
15840 * dwarf2out.c (mem_loc_descriptor): Likewise.
15841 * emit-rtl.c (init_derived_machine_modes): Likewise.
15842 * expmed.c (flip_storage_order): Likewise.
15843 (convert_extracted_bit_field): Likewise.
15844 * expr.c (copy_blkmode_from_reg): Likewise.
15845 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15846 * internal-fn.c (expand_mul_overflow): Likewise.
15847 * lower-subreg.c (simple_move): Likewise.
15848 * optabs-libfuncs.c (init_optabs): Likewise.
15849 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15850 * tree.c (vector_type_mode): Likewise.
15851 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15852 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15853 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15854 * tree-vect-stmts.c (vectorizable_load): Likewise.
15855 (vectorizable_store): Likewise.
15857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15858 Alan Hayward <alan.hayward@arm.com>
15859 David Sherwood <david.sherwood@arm.com>
15861 * coretypes.h (pod_mode): New type.
15862 (scalar_int_mode_pod): New typedef.
15863 * machmode.h (pod_mode): New class.
15864 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15865 * genmodes.c (emit_mode_int_n): Update accordingly.
15866 * lower-subreg.h (target_lower_subreg): Change type to
15867 scalar_int_mode_pod.
15868 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15869 scalar_int_mode_pod.
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 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15876 machine_mode to scalar_int_mode.
15877 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15878 (rs6000_option_override_internal): Remove cast to int.
15879 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15880 machine_mode to scalar_int_mode.
15881 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15882 (rs6000_option_override_internal): Remove cast to int.
15883 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15884 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15886 * config/s390/s390.c (s390_expand_builtin): Likewise.
15887 * coretypes.h (scalar_int_mode): New type.
15888 (opt_scalar_int_mode): New typedef.
15889 * machmode.h (scalar_int_mode): New class.
15890 (scalar_int_mode::includes_p): New function.
15891 (byte_mode): Change type to scalar_int_mode.
15892 (word_mode): Likewise.
15893 (ptr_mode): Likewise.
15894 * emit-rtl.c (byte_mode): Likewise.
15895 (word_mode): Likewise.
15896 (ptr_mode): Likewise.
15897 (init_derived_machine_modes): Update accordingly.
15898 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15899 and MODE_PARTIAL_INT.
15900 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15901 opt_scalar_int_mode.
15903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15904 Alan Hayward <alan.hayward@arm.com>
15905 David Sherwood <david.sherwood@arm.com>
15907 * target.def (libgcc_floating_mode_supported_p): Take a
15909 * doc/tm.texi: Regenerate.
15910 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
15912 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15913 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
15916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15917 Alan Hayward <alan.hayward@arm.com>
15918 David Sherwood <david.sherwood@arm.com>
15920 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
15921 * doc/tm.texi: Regenerate.
15922 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
15923 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
15924 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
15925 * targhooks.h (default_floatn_mode): Likewise.
15926 * targhooks.c (default_floatn_mode): Likewise.
15927 * tree.c (build_common_tree_nodes): Update accordingly.
15929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15930 Alan Hayward <alan.hayward@arm.com>
15931 David Sherwood <david.sherwood@arm.com>
15933 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
15934 (mode_iterator::iterate_p): Likewise.
15935 (mode_iterator::get_wider): Likewise.
15936 * expr.c (init_expr_target): Use opt_scalar_float_mode.
15938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15939 Alan Hayward <alan.hayward@arm.com>
15940 David Sherwood <david.sherwood@arm.com>
15942 * coretypes.h (opt_scalar_float_mode): New typedef.
15943 * machmode.h (float_mode_for_size): New function.
15944 * emit-rtl.c (double_mode): Delete.
15945 (init_emit_once): Use float_mode_for_size.
15946 * stor-layout.c (layout_type): Likewise.
15947 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
15949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15950 Alan Hayward <alan.hayward@arm.com>
15951 David Sherwood <david.sherwood@arm.com>
15953 * output.h (assemble_real): Take a scalar_float_mode.
15954 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
15955 * config/arm/arm.md (consttable_4): Likewise.
15956 (consttable_8): Likewise.
15957 (consttable_16): Likewise.
15958 * config/mips/mips.md (consttable_float): Likewise.
15959 * config/s390/s390.c (s390_output_pool_entry): Likewise.
15960 * varasm.c (assemble_real): Take a scalar_float_mode.
15961 (output_constant_pool_2): Update accordingly.
15962 (output_constant): Likewise.
15964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15965 Alan Hayward <alan.hayward@arm.com>
15966 David Sherwood <david.sherwood@arm.com>
15968 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
15969 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
15970 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
15971 (native_encode_real): Likewise.
15972 (native_interpret_real): Likewise.
15973 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
15974 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15977 Alan Hayward <alan.hayward@arm.com>
15978 David Sherwood <david.sherwood@arm.com>
15980 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
15981 <scalar_float_mode>. Simplify.
15982 (gen_extend_conv_libfunc): Likewise.
15984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15985 Alan Hayward <alan.hayward@arm.com>
15986 David Sherwood <david.sherwood@arm.com>
15988 * coretypes.h (scalar_float_mode): New type.
15989 * machmode.h (mode_traits::from_int): Use machine_mode if
15990 USE_ENUM_MODES is defined.
15991 (is_a): New function.
15993 (dyn_cast): Likewise.
15994 (scalar_float_mode): New class.
15995 (scalar_float_mode::includes_p): New function.
15996 (is_float_mode): Likewise.
15997 * gdbhooks.py (MachineModePrinter): New class.
15998 (build_pretty_printer): Use it for scalar_float_mode.
15999 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16000 (format_helper::format_helper): Turn into a template.
16001 * genmodes.c (get_mode_class): New function.
16002 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16003 or machine_mode if none.
16004 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16005 as_a <scalar_float_mode>.
16006 * dwarf2out.c (mem_loc_descriptor): Likewise.
16007 (insert_float): Likewise.
16008 (add_const_value_attribute): Likewise.
16009 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16010 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16011 (expand_unop): Update accordingly.
16012 (expand_abs_nojump): Likewise.
16013 (expand_copysign_absneg): Take a scalar_float_mode.
16014 (expand_copysign_bit): Likewise.
16015 (expand_copysign): Update accordingly.
16017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16018 Alan Hayward <alan.hayward@arm.com>
16019 David Sherwood <david.sherwood@arm.com>
16021 * coretypes.h (opt_mode): New class.
16022 * machmode.h (opt_mode): Likewise.
16023 (opt_mode::else_void): New function.
16024 (opt_mode::require): Likewise.
16025 (opt_mode::exists): Likewise.
16026 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16027 (GET_MODE_2XWIDER_MODE): Likewise.
16028 (mode_iterator::get_wider): Update accordingly.
16029 (mode_iterator::get_2xwider): Likewise.
16030 (mode_iterator::get_known_wider): Likewise, turning into a template.
16031 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16032 forcing a wider mode to exist.
16033 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16034 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16035 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16036 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16037 * lower-subreg.c (init_lower_subreg): Likewise.
16038 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16039 on the final iteration.
16040 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16041 a wider mode exists before asking for a move pattern.
16042 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16043 forcing a wider mode to exist.
16044 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16045 returning false if no such mode exists.
16046 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16047 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16048 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16049 Avoid checking for a MODE_INT if we already know the mode is not a
16051 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16052 forcing a wider mode to exist.
16053 (expmed_mult_highpart_optab): Likewise.
16054 (expmed_mult_highpart): Likewise.
16055 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16057 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16058 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16059 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16060 * internal-fn.c (expand_mul_overflow): Update use of
16061 GET_MODE_2XWIDER_MODE.
16062 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16063 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16064 GET_MODE_WIDER_MODE.
16065 (convert_plusminus_to_widen): Likewise.
16066 * tree-switch-conversion.c (array_value_type): Likewise.
16067 * var-tracking.c (emit_note_insn_var_location): Likewise.
16068 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16069 Return false inside rather than outside the loop if no wider mode
16071 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16072 and GET_MODE_2XWIDER_MODE
16073 (can_compare_p): Use else_void.
16074 * gdbhooks.py (OptMachineModePrinter): New class.
16075 (build_pretty_printer): Use it for opt_mode.
16077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16078 Alan Hayward <alan.hayward@arm.com>
16079 David Sherwood <david.sherwood@arm.com>
16081 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16082 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16085 Alan Hayward <alan.hayward@arm.com>
16086 David Sherwood <david.sherwood@arm.com>
16088 * machmode.h (mode_traits): New structure.
16089 (get_narrowest_mode): New function.
16090 (mode_iterator::start): Likewise.
16091 (mode_iterator::iterate_p): Likewise.
16092 (mode_iterator::get_wider): Likewise.
16093 (mode_iterator::get_known_wider): Likewise.
16094 (mode_iterator::get_2xwider): Likewise.
16095 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16096 (FOR_EACH_MODE): Likewise.
16097 (FOR_EACH_MODE_FROM): Likewise.
16098 (FOR_EACH_MODE_UNTIL): Likewise.
16099 (FOR_EACH_WIDER_MODE): Likewise.
16100 (FOR_EACH_2XWIDER_MODE): Likewise.
16101 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16102 * combine.c (simplify_comparison): Likewise
16103 * config/i386/i386.c (type_natural_mode): Likewise.
16104 * cse.c (cse_insn): Likewise.
16105 * dse.c (find_shift_sequence): Likewise.
16106 * emit-rtl.c (init_derived_machine_modes): Likewise.
16107 (init_emit_once): Likewise.
16108 * explow.c (hard_function_value): Likewise.
16109 * expmed.c (extract_fixed_bit_field_1): Likewise.
16110 (extract_bit_field_1): Likewise.
16111 (expand_divmod): Likewise.
16112 (emit_store_flag_1): Likewise.
16113 * expr.c (init_expr_target): Likewise.
16114 (convert_move): Likewise.
16115 (alignment_for_piecewise_move): Likewise.
16116 (widest_int_mode_for_size): Likewise.
16117 (emit_block_move_via_movmem): Likewise.
16118 (copy_blkmode_to_reg): Likewise.
16119 (set_storage_via_setmem): Likewise.
16120 (compress_float_constant): Likewise.
16121 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16122 * optabs-query.c (get_best_extraction_insn): Likewise.
16123 * optabs.c (expand_binop): Likewise.
16124 (expand_twoval_unop): Likewise.
16125 (expand_twoval_binop): Likewise.
16126 (widen_leading): Likewise.
16127 (widen_bswap): Likewise.
16128 (expand_parity): Likewise.
16129 (expand_unop): Likewise.
16130 (prepare_cmp_insn): Likewise.
16131 (prepare_float_lib_cmp): Likewise.
16132 (expand_float): Likewise.
16133 (expand_fix): Likewise.
16134 (expand_sfix_optab): Likewise.
16135 * postreload.c (move2add_use_add2_insn): Likewise.
16136 * reg-stack.c (reg_to_stack): Likewise.
16137 * reginfo.c (choose_hard_reg_mode): Likewise.
16138 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16139 * stor-layout.c (mode_for_size): Likewise.
16140 (smallest_mode_for_size): Likewise.
16141 (mode_for_vector): Likewise.
16142 (finish_bitfield_representative): Likewise.
16143 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16144 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16145 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16146 * var-tracking.c (prepare_call_arguments): Likewise.
16148 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16149 Alan Hayward <alan.hayward@arm.com>
16150 David Sherwood <david.sherwood@arm.com>
16152 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16153 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16154 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16155 * machmode.h (mode_size): Move earlier in file.
16156 (mode_precision): Likewise.
16157 (mode_inner): Likewise.
16158 (mode_nunits): Likewise.
16159 (mode_unit_size): Likewise.
16160 (unit_unit_precision): Likewise.
16161 (mode_wider): Likewise.
16162 (mode_2xwider): Likewise.
16163 (machine_mode): New class.
16164 (mode_to_bytes): New function.
16165 (mode_to_bits): Likewise.
16166 (mode_to_precision): Likewise.
16167 (mode_to_inner): Likewise.
16168 (mode_to_unit_size): Likewise.
16169 (mode_to_unit_precision): Likewise.
16170 (mode_to_nunits): Likewise.
16171 (GET_MODE_SIZE): Use mode_to_bytes.
16172 (GET_MODE_BITSIZE): Use mode_to_bits.
16173 (GET_MODE_PRECISION): Use mode_to_precision.
16174 (GET_MODE_INNER): Use mode_to_inner.
16175 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16176 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16177 (GET_MODE_NUNITS): Use mode_to_nunits.
16178 * system.h (ALWAYS_INLINE): New macro.
16179 * config/powerpcspe/powerpcspe-c.c
16180 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16181 int for arg1_mode and arg2_mode.
16183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16184 Alan Hayward <alan.hayward@arm.com>
16185 David Sherwood <david.sherwood@arm.com>
16187 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16188 Prefix mode names with E_ in case statements.
16189 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16190 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16191 (aarch64_split_simd_move): Likewise.
16192 (aarch64_gen_storewb_pair): Likewise.
16193 (aarch64_gen_loadwb_pair): Likewise.
16194 (aarch64_gen_store_pair): Likewise.
16195 (aarch64_gen_load_pair): Likewise.
16196 (aarch64_get_condition_code_1): Likewise.
16197 (aarch64_constant_pool_reload_icode): Likewise.
16198 (get_rsqrte_type): Likewise.
16199 (get_rsqrts_type): Likewise.
16200 (get_recpe_type): Likewise.
16201 (get_recps_type): Likewise.
16202 (aarch64_gimplify_va_arg_expr): Likewise.
16203 (aarch64_simd_container_mode): Likewise.
16204 (aarch64_emit_load_exclusive): Likewise.
16205 (aarch64_emit_store_exclusive): Likewise.
16206 (aarch64_expand_compare_and_swap): Likewise.
16207 (aarch64_gen_atomic_cas): Likewise.
16208 (aarch64_emit_bic): Likewise.
16209 (aarch64_emit_atomic_swap): Likewise.
16210 (aarch64_emit_atomic_load_op): Likewise.
16211 (aarch64_evpc_trn): Likewise.
16212 (aarch64_evpc_uzp): Likewise.
16213 (aarch64_evpc_zip): Likewise.
16214 (aarch64_evpc_ext): Likewise.
16215 (aarch64_evpc_rev): Likewise.
16216 (aarch64_evpc_dup): Likewise.
16217 (aarch64_gen_ccmp_first): Likewise.
16218 (aarch64_gen_ccmp_next): Likewise.
16219 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16220 (alpha_emit_xfloating_libcall): Likewise.
16221 (emit_insxl): Likewise.
16222 (alpha_arg_type): Likewise.
16223 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16224 (arc_preferred_simd_mode): Likewise.
16225 (arc_secondary_reload): Likewise.
16226 (get_arc_condition_code): Likewise.
16227 (arc_print_operand): Likewise.
16228 (arc_legitimate_constant_p): Likewise.
16229 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16230 * config/arc/arc.md (casesi_load): Likewise.
16231 (casesi_compact_jump): Likewise.
16232 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16233 (cc_use_register): Likewise.
16234 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16235 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16236 (arm_init_iwmmxt_builtins): Likewise.
16237 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16238 (neon_expand_vector_init): Likewise.
16239 (arm_attr_length_move_neon): Likewise.
16240 (maybe_get_arm_condition_code): Likewise.
16241 (arm_emit_vector_const): Likewise.
16242 (arm_preferred_simd_mode): Likewise.
16243 (arm_output_iwmmxt_tinsr): Likewise.
16244 (thumb1_output_casesi): Likewise.
16245 (thumb2_output_casesi): Likewise.
16246 (arm_emit_load_exclusive): Likewise.
16247 (arm_emit_store_exclusive): Likewise.
16248 (arm_expand_compare_and_swap): Likewise.
16249 (arm_evpc_neon_vuzp): Likewise.
16250 (arm_evpc_neon_vzip): Likewise.
16251 (arm_evpc_neon_vrev): Likewise.
16252 (arm_evpc_neon_vtrn): Likewise.
16253 (arm_evpc_neon_vext): Likewise.
16254 (arm_validize_comparison): Likewise.
16255 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16256 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16257 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16258 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16259 (c6x_preferred_simd_mode): Likewise.
16260 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16261 (epiphany_rtx_costs): Likewise.
16262 * config/epiphany/predicates.md (proper_comparison_operator):
16264 * config/frv/frv.c (condexec_memory_operand): Likewise.
16265 (frv_emit_move): Likewise.
16266 (output_move_single): Likewise.
16267 (output_condmove_single): Likewise.
16268 (frv_hard_regno_mode_ok): Likewise.
16269 (frv_matching_accg_mode): Likewise.
16270 * config/h8300/h8300.c (split_adds_subs): Likewise.
16271 (h8300_rtx_costs): Likewise.
16272 (h8300_print_operand): Likewise.
16273 (compute_mov_length): Likewise.
16274 (output_logical_op): Likewise.
16275 (compute_logical_op_length): Likewise.
16276 (compute_logical_op_cc): Likewise.
16277 (h8300_shift_needs_scratch_p): Likewise.
16278 (output_a_shift): Likewise.
16279 (compute_a_shift_length): Likewise.
16280 (compute_a_shift_cc): Likewise.
16281 (expand_a_rotate): Likewise.
16282 (output_a_rotate): Likewise.
16283 * config/i386/i386.c (classify_argument): Likewise.
16284 (function_arg_advance_32): Likewise.
16285 (function_arg_32): Likewise.
16286 (function_arg_64): Likewise.
16287 (function_value_64): Likewise.
16288 (ix86_gimplify_va_arg): Likewise.
16289 (ix86_legitimate_constant_p): Likewise.
16290 (put_condition_code): Likewise.
16291 (split_double_mode): Likewise.
16292 (ix86_avx256_split_vector_move_misalign): Likewise.
16293 (ix86_expand_vector_logical_operator): Likewise.
16294 (ix86_split_idivmod): Likewise.
16295 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16296 (ix86_build_const_vector): Likewise.
16297 (ix86_build_signbit_mask): Likewise.
16298 (ix86_match_ccmode): Likewise.
16299 (ix86_cc_modes_compatible): Likewise.
16300 (ix86_expand_branch): Likewise.
16301 (ix86_expand_sse_cmp): Likewise.
16302 (ix86_expand_sse_movcc): Likewise.
16303 (ix86_expand_int_sse_cmp): Likewise.
16304 (ix86_expand_vec_perm_vpermi2): Likewise.
16305 (ix86_expand_vec_perm): Likewise.
16306 (ix86_expand_sse_unpack): Likewise.
16307 (ix86_expand_int_addcc): Likewise.
16308 (ix86_split_to_parts): Likewise.
16309 (ix86_vectorize_builtin_gather): Likewise.
16310 (ix86_vectorize_builtin_scatter): Likewise.
16311 (avx_vpermilp_parallel): Likewise.
16312 (inline_memory_move_cost): Likewise.
16313 (ix86_tieable_integer_mode_p): Likewise.
16314 (x86_maybe_negate_const_int): Likewise.
16315 (ix86_expand_vector_init_duplicate): Likewise.
16316 (ix86_expand_vector_init_one_nonzero): Likewise.
16317 (ix86_expand_vector_init_one_var): Likewise.
16318 (ix86_expand_vector_init_concat): Likewise.
16319 (ix86_expand_vector_init_interleave): Likewise.
16320 (ix86_expand_vector_init_general): Likewise.
16321 (ix86_expand_vector_set): Likewise.
16322 (ix86_expand_vector_extract): Likewise.
16323 (emit_reduc_half): Likewise.
16324 (ix86_emit_i387_round): Likewise.
16325 (ix86_mangle_type): Likewise.
16326 (ix86_expand_round_sse4): Likewise.
16327 (expand_vec_perm_blend): Likewise.
16328 (canonicalize_vector_int_perm): Likewise.
16329 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16330 (expand_vec_perm_1): Likewise.
16331 (expand_vec_perm_interleave3): Likewise.
16332 (expand_vec_perm_even_odd_pack): Likewise.
16333 (expand_vec_perm_even_odd_1): Likewise.
16334 (expand_vec_perm_broadcast_1): Likewise.
16335 (ix86_vectorize_vec_perm_const_ok): Likewise.
16336 (ix86_expand_vecop_qihi): Likewise.
16337 (ix86_expand_mul_widen_hilo): Likewise.
16338 (ix86_expand_sse2_abs): Likewise.
16339 (ix86_expand_pextr): Likewise.
16340 (ix86_expand_pinsr): Likewise.
16341 (ix86_preferred_simd_mode): Likewise.
16342 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16343 * config/i386/sse.md (*andnot<mode>3): Likewise.
16344 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16345 (*<code><mode>3): Likewise.
16346 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16347 (ia64_expand_atomic_op): Likewise.
16348 (ia64_arg_type): Likewise.
16349 (ia64_mode_to_int): Likewise.
16350 (ia64_scalar_mode_supported_p): Likewise.
16351 (ia64_vector_mode_supported_p): Likewise.
16352 (expand_vec_perm_broadcast): Likewise.
16353 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16354 (iq2000_function_arg_advance): Likewise.
16355 (iq2000_function_arg): Likewise.
16356 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16357 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16358 (m68k_libcall_value): Likewise.
16359 (m68k_function_value): Likewise.
16360 (sched_attr_op_type): Likewise.
16361 * config/mcore/mcore.c (mcore_output_move): Likewise.
16362 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16364 (microblaze_function_arg): Likewise.
16365 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16366 (mips_print_operand): Likewise.
16367 (mips_mode_ok_for_mov_fmt_p): Likewise.
16368 (mips_vector_mode_supported_p): Likewise.
16369 (mips_preferred_simd_mode): Likewise.
16370 (mips_expand_vpc_loongson_even_odd): Likewise.
16371 (mips_expand_vec_unpack): Likewise.
16372 (mips_expand_vi_broadcast): Likewise.
16373 (mips_expand_vector_init): Likewise.
16374 (mips_expand_vec_reduc): Likewise.
16375 (mips_expand_msa_cmp): Likewise.
16376 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16377 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16378 (cc_flags_for_mode): Likewise.
16379 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16380 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16381 (nds32_output_casesi_pc_relative): Likewise.
16382 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16383 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16384 (nvptx_gen_unpack): Likewise.
16385 (nvptx_gen_pack): Likewise.
16386 (nvptx_gen_shuffle): Likewise.
16387 (nvptx_gen_wcast): Likewise.
16388 (nvptx_preferred_simd_mode): Likewise.
16389 * config/pa/pa.c (pa_secondary_reload): Likewise.
16390 * config/pa/predicates.md (base14_operand): Likewise.
16391 * config/powerpcspe/powerpcspe-c.c
16392 (altivec_resolve_overloaded_builtin): Likewise.
16393 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16395 (rs6000_preferred_simd_mode): Likewise.
16396 (output_vec_const_move): Likewise.
16397 (rs6000_expand_vector_extract): Likewise.
16398 (rs6000_split_vec_extract_var): Likewise.
16399 (reg_offset_addressing_ok_p): Likewise.
16400 (rs6000_legitimate_offset_address_p): Likewise.
16401 (rs6000_legitimize_address): Likewise.
16402 (rs6000_emit_set_const): Likewise.
16403 (rs6000_const_vec): Likewise.
16404 (rs6000_emit_move): Likewise.
16405 (spe_build_register_parallel): Likewise.
16406 (rs6000_darwin64_record_arg_recurse): Likewise.
16407 (swap_selector_for_mode): Likewise.
16408 (spe_init_builtins): Likewise.
16409 (paired_init_builtins): Likewise.
16410 (altivec_init_builtins): Likewise.
16411 (do_load_for_compare): Likewise.
16412 (rs6000_generate_compare): Likewise.
16413 (rs6000_expand_float128_convert): Likewise.
16414 (emit_load_locked): Likewise.
16415 (emit_store_conditional): Likewise.
16416 (rs6000_output_function_epilogue): Likewise.
16417 (rs6000_handle_altivec_attribute): Likewise.
16418 (rs6000_function_value): Likewise.
16419 (emit_fusion_gpr_load): Likewise.
16420 (emit_fusion_p9_load): Likewise.
16421 (emit_fusion_p9_store): Likewise.
16422 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16423 (fusion_gpr_mem_load): Likewise.
16424 (fusion_addis_mem_combo_load): Likewise.
16425 (fusion_addis_mem_combo_store): Likewise.
16426 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16427 (fusion_gpr_mem_load): Likewise.
16428 (fusion_addis_mem_combo_load): Likewise.
16429 (fusion_addis_mem_combo_store): Likewise.
16430 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16432 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16433 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16434 (rs6000_preferred_simd_mode): Likewise.
16435 (output_vec_const_move): Likewise.
16436 (rs6000_expand_vector_extract): Likewise.
16437 (rs6000_split_vec_extract_var): Likewise.
16438 (reg_offset_addressing_ok_p): Likewise.
16439 (rs6000_legitimate_offset_address_p): Likewise.
16440 (rs6000_legitimize_address): Likewise.
16441 (rs6000_emit_set_const): Likewise.
16442 (rs6000_const_vec): Likewise.
16443 (rs6000_emit_move): Likewise.
16444 (rs6000_darwin64_record_arg_recurse): Likewise.
16445 (swap_selector_for_mode): Likewise.
16446 (paired_init_builtins): Likewise.
16447 (altivec_init_builtins): Likewise.
16448 (rs6000_expand_float128_convert): Likewise.
16449 (emit_load_locked): Likewise.
16450 (emit_store_conditional): Likewise.
16451 (rs6000_output_function_epilogue): Likewise.
16452 (rs6000_handle_altivec_attribute): Likewise.
16453 (rs6000_function_value): Likewise.
16454 (emit_fusion_gpr_load): Likewise.
16455 (emit_fusion_p9_load): Likewise.
16456 (emit_fusion_p9_store): Likewise.
16457 * config/rx/rx.c (rx_gen_move_template): Likewise.
16458 (flags_from_mode): Likewise.
16459 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16460 (s390_slb_comparison): Likewise.
16461 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16462 (s390_vector_mode_supported_p): Likewise.
16463 (s390_cc_modes_compatible): Likewise.
16464 (s390_match_ccmode_set): Likewise.
16465 (s390_canonicalize_comparison): Likewise.
16466 (s390_emit_compare_and_swap): Likewise.
16467 (s390_branch_condition_mask): Likewise.
16468 (s390_rtx_costs): Likewise.
16469 (s390_secondary_reload): Likewise.
16470 (__SECONDARY_RELOAD_CASE): Likewise.
16471 (s390_expand_cs): Likewise.
16472 (s390_preferred_simd_mode): Likewise.
16473 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16474 * config/sh/sh.c (sh_print_operand): Likewise.
16475 (dump_table): Likewise.
16476 (sh_secondary_reload): Likewise.
16477 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16478 * config/sh/sh.md (casesi_worker_1): Likewise.
16479 (casesi_worker_2): Likewise.
16480 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16481 (fcc_comparison_operator): Likewise.
16482 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16483 (emit_soft_tfmode_cvt): Likewise.
16484 (sparc_preferred_simd_mode): Likewise.
16485 (output_cbranch): Likewise.
16486 (sparc_print_operand): Likewise.
16487 (sparc_expand_vec_perm_bmask): Likewise.
16488 (vector_init_bshuffle): Likewise.
16489 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16490 (spu_vector_mode_supported_p): Likewise.
16491 (spu_expand_insv): Likewise.
16492 (spu_emit_branch_or_set): Likewise.
16493 (spu_handle_vector_attribute): Likewise.
16494 (spu_builtin_splats): Likewise.
16495 (spu_builtin_extract): Likewise.
16496 (spu_builtin_promote): Likewise.
16497 (spu_expand_sign_extend): Likewise.
16498 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16499 (tilegx_simd_int): Likewise.
16500 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16501 (tilepro_simd_int): Likewise.
16502 * config/v850/v850.c (const_double_split): Likewise.
16503 (v850_print_operand): Likewise.
16504 (ep_memory_offset): Likewise.
16505 * config/vax/vax.c (vax_rtx_costs): Likewise.
16506 (vax_output_int_move): Likewise.
16507 (vax_output_int_add): Likewise.
16508 (vax_output_int_subtract): Likewise.
16509 * config/visium/predicates.md (visium_branch_operator): Likewise.
16510 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16511 (visium_print_operand_address): Likewise.
16512 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16513 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16514 (xtensa_expand_conditional_branch): Likewise.
16515 (xtensa_copy_incoming_a7): Likewise.
16516 (xtensa_output_literal): Likewise.
16517 * dfp.c (decimal_real_maxval): Likewise.
16518 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16521 Alan Hayward <alan.hayward@arm.com>
16522 David Sherwood <david.sherwood@arm.com>
16524 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16525 (mode_nunits_inline): Likewise.
16526 (mode_inner_inline): Likewise.
16527 (mode_unit_size_inline): Likewise.
16528 (mode_unit_precision_inline): Likewise.
16529 (emit_insn_modes_h): Likewise. Also emit a #define of the
16531 (emit_mode_wider): Add an E_ prefix to mode names.
16532 (emit_mode_complex): Likewise.
16533 (emit_mode_inner): Likewise.
16534 (emit_mode_adjustments): Likewise.
16535 (emit_mode_int_n): Likewise.
16536 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16537 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16538 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16539 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16540 (CRC32_BUILTIN, ENTRY): Likewise.
16541 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16542 (aarch64_pop_regs): Likewise.
16543 (aarch64_process_components): Likewise.
16544 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16545 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16546 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16547 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16548 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16549 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16551 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16552 (pcmpistr): Likewise.
16553 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16554 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16555 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16557 * config/rl78/rl78.c (mduc_regs): Likewise.
16558 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16559 (htm_expand_builtin): Likewise.
16560 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16561 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16562 * config/xtensa/xtensa.c (print_operand): Likewise.
16563 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16564 (NUM_MODE_VECTOR_INT): Likewise.
16565 * genoutput.c (null_operand): Likewise.
16566 (output_operand_data): Likewise.
16567 * genrecog.c (print_parameter_value): Likewise.
16568 * lra.c (debug_operand_data): Likewise.
16570 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16572 * dwarf2out.c (scompare_loc_descriptor_wide)
16573 (scompare_loc_descriptor_narrow): New functions, split out from...
16574 (scompare_loc_descriptor): ...here.
16575 * expmed.c (emit_store_flag_int): New function, split out from...
16576 (emit_store_flag): ...here.
16578 2017-08-30 Richard Biener <rguenther@suse.de>
16580 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16581 (dwarf2out_early_finish): Move setting of AT_pubnames from
16582 early debug output to early finish.
16584 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16586 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16587 and -mdata-region to the assembler.
16589 2017-08-30 Richard Biener <rguenther@suse.de>
16591 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16593 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16594 (dwarf2out_early_global_decl): Only generate a DIE for the
16595 abstract origin if it doesn't already exist or is a declaration DIE.
16596 (resolve_addr): Do not add the linkage name twice when
16597 generating a stub DIE for the DW_TAG_GNU_call_site target.
16599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16601 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16602 Use machine_mode rather than int for arg1_mode.
16604 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16607 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16608 that the second argument of the built-in functions to unpack
16609 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16610 switch statement instead a lot of if statements.
16611 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16612 Allow 64-bit values to be in Altivec registers as well as
16613 traditional floating point registers.
16614 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16616 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16618 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16619 MAX_REGS_PER_ADDRESS == 1.
16621 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16623 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16624 * config/i386/i386.c (ix86_option_override_internal): Simplify
16625 setting of opts->x_flag_entry.
16627 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16628 Jakub Jelinek <jakub@redhat.com>
16629 Richard Biener <rguenther@suse.de>
16631 PR tree-optimization/81503
16632 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16633 folded constant fits in the target type; reorder tests for clarity.
16635 2017-08-29 Martin Liska <mliska@suse.cz>
16637 * passes.def: Include pass_lower_switch.
16638 * stmt.c (dump_case_nodes): Remove and move to
16639 tree-switch-conversion.
16640 (case_values_threshold): Likewise.
16641 (expand_switch_as_decision_tree_p): Likewise.
16642 (emit_case_decision_tree): Likewise.
16643 (expand_case): Likewise.
16644 (balance_case_nodes): Likewise.
16645 (node_has_low_bound): Likewise.
16646 (node_has_high_bound): Likewise.
16647 (node_is_bounded): Likewise.
16648 (emit_case_nodes): Likewise.
16649 (struct simple_case_node): New struct.
16650 (add_case_node): Remove.
16651 (emit_case_dispatch_table): Use vector instead of case_list.
16652 (reset_out_edges_aux): Remove.
16653 (compute_cases_per_edge): Likewise.
16654 (expand_case): Build list of simple_case_node.
16655 (expand_sjlj_dispatch_table): Use it.
16656 * tree-switch-conversion.c (struct case_node): Moved from
16657 stmt.c and adjusted.
16658 (emit_case_nodes): Likewise.
16659 (node_has_low_bound): Likewise.
16660 (node_has_high_bound): Likewise.
16661 (node_is_bounded): Likewise.
16662 (case_values_threshold): Likewise.
16663 (reset_out_edges_aux): Likewise.
16664 (compute_cases_per_edge): Likewise.
16665 (add_case_node): Likewise.
16666 (dump_case_nodes): Likewise.
16667 (balance_case_nodes): Likewise.
16668 (expand_switch_as_decision_tree_p): Likewise.
16669 (emit_jump): Likewise.
16670 (emit_case_decision_tree): Likewise.
16671 (try_switch_expansion): Likewise.
16672 (do_jump_if_equal): Likewise.
16673 (emit_cmp_and_jump_insns): Likewise.
16674 (fix_phi_operands_for_edge): New function.
16675 (record_phi_operand_mapping): Likewise.
16676 (class pass_lower_switch): New pass.
16677 (pass_lower_switch::execute): New function.
16678 (make_pass_lower_switch): Likewise.
16679 (conditional_probability):
16680 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16681 * tree-pass.h: Add make_pass_lower_switch.
16683 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16686 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16689 2017-08-29 Richard Biener <rguenther@suse.de>
16691 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16692 we do not add a DW_AT_inline attribute twice.
16693 (gen_subprogram_die): Remove code setting DW_AT_inline on
16694 DECL_ABSTRACT_P nodes.
16696 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16698 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16699 calls to internal functions.
16700 (gimplify_modify_expr): Likewise.
16701 * tree-call-cdce.c (use_internal_fn): Likewise.
16702 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16703 (convert_to_divmod): Set the nothrow flag.
16704 * tree-if-conv.c (predicate_mem_writes): Likewise.
16705 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16706 (vectorizable_call): Likewise.
16707 (vectorizable_store): Likewise.
16708 (vectorizable_load): Likewise.
16709 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16710 (vect_recog_mask_conversion_pattern): Likewise.
16712 2017-08-29 Martin Liska <mliska@suse.cz>
16715 * gcc.c (driver_handle_option): Add new argument.
16716 * opts-common.c (handle_option): Pass
16717 target_option_override_hook.
16718 * opts-global.c (lang_handle_option): Add new option.
16719 (set_default_handlers): Add new argument.
16720 (decode_options): Likewise.
16721 * opts.c (target_handle_option): Likewise.
16722 (common_handle_option): Call target_option_override_hook.
16723 * opts.h (struct cl_option_handler_func): Add hook for
16724 target option override.
16725 (struct cl_option_handlers): Likewise.
16726 (set_default_handlers): Add new argument.
16727 (decode_options): Likewise.
16728 (common_handle_option): Likewise.
16729 (target_handle_option): Likewise.
16730 * toplev.c (toplev::main): Pass targetm.target_option.override
16733 2017-08-29 Richard Biener <rguenther@suse.de>
16734 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16736 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16737 life to the active subtree.
16739 2017-08-28 Jeff Law <law@redhat.com>
16741 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16742 derive_equivalences.
16743 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16744 Code moved into....
16745 (edge_info::derive_equivalences): New private member function
16747 * tree-ssa-dom.c (class edge_info): Changed from a struct
16748 to a class. Add ctor/dtor, methods and data members.
16749 (edge_info::edge_info): Renamed from allocate_edge_info.
16750 Initialize additional members.
16751 (edge_info::~edge_info): New.
16752 (free_dom_edge_info): Delete the edge info.
16753 (record_edge_info): Use new class & associated member functions.
16754 Tighten forms for testing for edge equivalences.
16755 (record_temporary_equivalences): Iterate over the simple
16756 equivalences rather than assuming there's only one per edge.
16757 (cprop_into_successor_phis): Iterate over the simple
16758 equivalences rather than assuming there's only one per edge.
16759 (optimize_stmt): Use operand_equal_p rather than pointer
16760 equality for mini-DSE code.
16762 2017-08-28 Nathan Sidwell <nathan@acm.org>
16764 * gcc.c (execute): Fold SIGPIPE handling into switch
16765 statement. Adjust internal error message.
16767 2017-08-28 Richard Biener <rguenther@suse.de>
16770 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16771 Do nothing for removed DIEs.
16773 2017-08-28 Richard Biener <rguenther@suse.de>
16775 PR tree-optimization/81977
16776 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16779 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16782 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16783 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16784 using targetm.gen_mem_thread_fence.
16786 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16789 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16790 predicate to register_operand. Reorder operands.
16791 (*btr<mode>): Ditto.
16792 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16793 (*btr<mode>_mask): Ditto.
16795 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16797 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16798 * config/rs6000/xmmintrin.h: New file.
16799 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16801 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16804 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16805 parameter and_insn and return it.
16806 (recombine_lvx_pattern): Insert a copy to ensure availability of
16807 the base register of the copied masking operation at the point of
16808 the instruction replacement.
16809 (recombine_stvx_pattern): Likewise.
16811 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16813 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16814 undocumented switches.
16815 (-mpower9-dform-vector): Likewise.
16816 (-mpower9-dform): Likewise.
16817 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16818 comments to delete references to -mpower9-dform* switches.
16819 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16820 Delete reference to -mpower9-dform* switches, test for
16821 -mpower9-vector instead.
16822 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16823 (OTHER_P9_VECTOR_MASKS): Likewise.
16824 (POWERPC_MASKS): Likewise.
16825 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16826 tests against -mpower9-dform* to -mpower9-vector. Delete code
16827 that checked for -mpower9-dform* consistancy with other options.
16828 Add test for -mpower9-misc to enable other power9 switches.
16829 (rs6000_init_hard_regno_mode_ok): Likewise.
16830 (rs6000_option_override_internal): Likewise.
16831 (rs6000_emit_prologue): Likewise.
16832 (rs6000_emit_epilogue): Likewise.
16833 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16834 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16835 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16837 (emit_fusion_p9_store): Likewise.
16838 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16839 resetting these macros if the assembler does not support ISA 3.0
16841 (TARGET_P9_DFORM_VECTOR): Likewise.
16842 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16843 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16845 2017-08-25 Alan Modra <amodra@gmail.com>
16847 PR rtl-optimization/81747
16848 * cse.c (cse_extended_basic_block): Don't attempt to record
16849 equivalences for degenerate conditional branches that branch
16850 to their fall-through.
16852 2017-08-24 Martin Sebor <msebor@redhat.com>
16854 PR middle-end/81908
16855 * gimple-fold.c (size_must_be_zero_p): New function.
16856 (gimple_fold_builtin_memory_op): Call it.
16858 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16860 * config/rs6000/mm_malloc.h: New file.
16862 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16864 PR tree-optimization/81913
16865 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16866 analysis when either IVs in condition can wrap.
16868 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16870 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16871 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16873 2017-08-24 Richard Biener <rguenther@suse.de>
16876 * targhooks.c (default_target_can_inline_p): Properly
16877 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16878 is present and always compare.
16879 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16880 infer -mfpmath=sse from TARGET_SSE_P.
16881 (ix86_can_inline_p): Properly use target_option_default_node when
16882 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16884 2017-08-24 Richard Biener <rguenther@suse.de>
16887 * dwarf2out.c (output_die): Handle flag_generate_offload like
16889 (output_comp_unit): Likewise.
16890 (gen_array_type_die): Likewise.
16891 (dwarf2out_early_finish): Likewise.
16892 (note_variable_value_in_expr): Likewise.
16893 (dwarf2out_finish): Likewise. Adjust assert.
16894 * cgraphunit.c (symbol_table::compile): Move setting of
16895 flag_generate_offload earlier ...
16896 (symbol_table::finalize_compilation_unit): ... here, before
16897 early debug finalization.
16899 2017-08-24 Richard Biener <rguenther@suse.de>
16901 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16902 and ipa-fnsummary.h.
16903 (ix86_can_inline_p): When ix86_fpmath flags do not match
16904 check whether the callee uses FP math at all.
16906 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
16908 PR middle-end/81931
16909 * tree-ssanames.c (get_nonzero_bits): Use element_precision
16910 instead of TYPE_PRECISION.
16912 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
16913 Alan Hayward <alan.hayward@arm.com>
16914 David Sherwood <david.sherwood@arm.com>
16916 * combine.c (make_extraction): Use subreg_offset_from_lsb.
16918 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
16920 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
16922 (ix86_frame::stack_realign_allocate): New field.
16923 (struct machine_frame_state): Modify comments.
16924 (machine_frame_state::sp_realigned_fp_end): New field.
16925 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
16926 layout calculation.
16927 (sp_valid_at): Add assertion to assure no attempt to access invalid
16928 offset of a realigned stack.
16929 (fp_valid_at): Likewise.
16930 (choose_baseaddr): Modify comments.
16931 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
16932 ix86_expand_prologue.
16933 (ix86_expand_prologue): Modify stack realignment and allocation.
16934 (ix86_expand_epilogue): Modify comments.
16935 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
16936 avx2_runtime, avx512f, and avx512f_runtime.
16938 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
16940 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
16941 (mstackrealign): Do not init to -1.
16942 * config/i386/i386.c (ix86_option_override_internal):
16943 Check opts_set, not opts when setting default value of
16944 opts->x_ix86_force_align_arg_pointer.
16946 2017-08-23 Richard Biener <rguenther@suse.de>
16948 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
16949 lang_hooks.decl_printable_name.
16950 * print-rtl-function.c (print_rtx_function): Likewise.
16951 * tree-pretty-print.c (dump_function_header): Likewise.
16953 2017-08-23 Richard Biener <rguenther@suse.de>
16956 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
16957 -g0 at compile-time.
16959 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16961 PR middle-end/19706
16962 * doc/sourcebuild.texi (Other hardware attributes):
16965 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16967 PR middle-end/19706
16968 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
16969 Fix single-use check.
16971 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
16973 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
16975 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16977 * doc/install.texi: Modify to add more details on running selected
16980 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16982 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
16983 is combined with -mabi=ms.
16984 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
16987 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16989 PR tree-optimization/81488
16990 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
16991 and cached_basis fields.
16992 (MAX_SPREAD): New constant.
16993 (alloc_cand_and_find_basis): Initialize new fields.
16994 (clear_visited): New function.
16995 (create_phi_basis_1): Rename from create_phi_basis, set visited
16996 and cached_basis fields.
16997 (create_phi_basis): New wrapper function.
16998 (phi_add_costs_1): Rename from phi_add_costs, add spread
16999 parameter, set visited field, short-circuit when limits reached.
17000 (phi_add_costs): New wrapper function.
17001 (record_phi_increments_1): Rename from record_phi_increments, set
17003 (record_phi_increments): New wrapper function.
17004 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17005 (phi_incr_cost): New wrapper function.
17006 (all_phi_incrs_profitable_1): Rename from
17007 all_phi_incrs_profitable, set visited field.
17008 (all_phi_incrs_profitable): New wrapper function.
17010 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17011 Alan Hayward <alan.hayward@arm.com>
17012 David Sherwood <david.sherwood@arm.com>
17014 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17015 that takes the outer and inner modes.
17016 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17017 comparison as the canonical test for a paradoxical subreg.
17018 * combine.c (simplify_set): Use paradoxical_subreg_p.
17019 (make_extraction): Likewise.
17020 (force_to_mode): Likewise.
17021 (rtx_equal_for_field_assignment_p): Likewise.
17022 (gen_lowpart_for_combine): Likewise.
17023 (simplify_comparison): Likewise.
17024 * cse.c (equiv_constant): Likewise.
17025 * expmed.c (store_bit_field_1): Likewise.
17026 * final.c (alter_subreg): Likewise.
17027 * fwprop.c (propagate_rtx): Likewise.
17028 (forward_propagate_subreg): Likewise.
17029 * ira-conflicts.c (ira_build_conflicts): Likewise.
17030 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17031 * lra-constraints.c (curr_insn_transform): Likewise.
17032 (split_reg): Likewise.
17033 * lra-eliminations.c (move_plus_up): Likewise.
17034 (lra_eliminate_regs_1): Likewise.
17035 * recog.c (general_operand): Likewise.
17036 * ree.c (combine_reaching_defs): Likewise.
17037 * reload.c (push_reload): Likewise.
17038 (find_reloads): Likewise.
17039 * reload1.c (elimination_effects): Likewise.
17040 (compute_reload_subreg_offset): Likewise.
17041 (choose_reload_regs): Likewise.
17042 * rtlanal.c (subreg_lsb_1): Likewise.
17043 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17044 (simplify_subreg): Likewise.
17045 * var-tracking.c (track_loc_p): Likewise.
17046 * emit-rtl.c (byte_lowpart_offset): Likewise.
17047 (paradoxical_subreg_p): Delete out-of-line definition.
17049 2017-08-22 Jeff Law <law@redhat.com>
17051 PR tree-optimization/81741
17052 PR tree-optimization/71947
17053 * tree-ssa-dom.c: Include tree-inline.h.
17054 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17055 equivalences if one is more expensive to compute than the other.
17056 * tree-ssa-scopedtables.h (class const_or_copies): Make
17057 record_const_or_copy_raw method private.
17058 (class avail_exprs_stack): New method simplify_binary_operation.
17059 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17060 avail_exprs_stack::simplify_binary_operation as needed.
17061 (avail_exprs_stack::simplify_binary_operation): New function.
17063 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17065 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17066 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17067 (DOT_SYMBOLS): Likewise.
17068 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17069 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17070 (RS6000_ABI_NAME): Likewise.
17071 (TARGET_CMODEL): Likewise.
17072 (TOC_SECTION_ASM_OP): Likewise.
17073 (SET_CMODEL): New macro.
17074 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17076 2017-08-22 Richard Biener <rguenther@suse.de>
17078 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17079 to allow for free-lang-data replacements similar to verify_type_variant.
17081 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17082 Alan Hayward <alan.hayward@arm.com>
17083 David Sherwood <david.sherwood@arm.com>
17085 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17086 VOIDmode for the LABEL_REF.
17088 2017-08-22 Richard Biener <rguenther@suse.de>
17090 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17092 2017-08-22 Richard Biener <rguenther@suse.de>
17094 * common.opt (feliminate-dwarf2-dups): Ignore.
17095 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17096 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17097 same_die_p_wrap, compute_section_prefix,
17098 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17099 (comdat_symbol_id, comdat_symbol_number): Likewise.
17100 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17102 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17103 (output_die): Mark unreachable path unreachable.
17104 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17105 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17106 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17107 (dwarf2out_early_finish): Likewise.
17109 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17111 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17113 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17116 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17117 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17118 (avr_attribute_table) <io, io_low, address>: Initialize
17119 .decl_required with true.
17121 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17123 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17124 undocumented debugging options.
17125 (-mvsx-scalar-double): Likewise.
17126 (-mallow-df-permute): Likewise.
17127 (-mvectorize-builtins): Likewise.
17128 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17129 (rs6000_builtin_vectorized_function): Likewise.
17130 (rs6000_builtin_md_vectorized_function): Likewise.
17131 (rs6000_opt_vars): Likewise.
17133 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17136 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17137 (*btrq_imm): Rename from *btrq.
17138 (*btcq_imm): Rename from *btcq.
17139 (btsc): New code attribute.
17140 (*<btsc><mode>): New insn pattern.
17141 (*btr<mode>): Ditto.
17142 (*<btsc><mode>_mask): New insn_and_split pattern.
17143 (*btr<mode>_mask): Ditto.
17145 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17146 Alan Hayward <alan.hayward@arm.com>
17147 David Sherwood <david.sherwood@arm.com>
17149 * function.c (pad_below): Simplify padding calculation.
17151 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17152 Alan Hayward <alan.hayward@arm.com>
17153 David Sherwood <david.sherwood@arm.com>
17155 * target.def (function_prologue): Remove frame size argument.
17156 (function_epilogue): Likewise.
17157 * doc/tm.texi: Regenerate.
17158 * final.c (final_start_function): Update call to function_prologue.
17159 (final_end_function): Update call to function_epilogue.
17160 (default_function_pro_epilogue): Remove frame size argument.
17161 * output.h (default_function_pro_epilogue): Likewise.
17162 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17163 (arm_output_function_prologue): Likewise.
17164 * config/frv/frv.c (frv_function_prologue): Likewise.
17165 (frv_function_epilogue): Likewise.
17166 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17167 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17168 (ia64_output_function_epilogue): Likewise.
17169 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17170 (m32r_output_function_epilogue): Likewise.
17171 * config/microblaze/microblaze.c (microblaze_function_prologue)
17172 (microblaze_function_epilogue): Likewise.
17173 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17174 (mips_output_function_epilogue): Likewise.
17175 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17176 (mmix_target_asm_function_epilogue): Likewise.
17177 * config/msp430/msp430.c (msp430_start_function): Likewise.
17178 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17179 (nds32_asm_function_epilogue): Likewise.
17180 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17181 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17182 (pa_output_function_epilogue): Likewise.
17183 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17184 (rs6000_output_function_epilogue): Likewise.
17185 * config/rl78/rl78.c (rl78_start_function): Likewise.
17186 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17187 (rs6000_output_function_epilogue): Likewise.
17188 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17189 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17190 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17191 (sparc_asm_function_epilogue): Likewise.
17193 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17195 * tree.h (type_has_mode_precision_p): New function.
17196 * convert.c (convert_to_integer_1): Use it.
17197 * expr.c (expand_expr_real_2): Likewise.
17198 (expand_expr_real_1): Likewise.
17199 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17200 * match.pd: Likewise.
17201 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17202 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17203 * tree-tailcall.c (process_assignment): Likewise.
17204 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17205 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17206 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17207 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17208 (vectorizable_assignment): Likewise.
17209 (vectorizable_shift): Likewise.
17210 (vectorizable_operation): Likewise.
17211 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17213 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17215 * match.pd: Add pow (C, x) simplification.
17217 2017-08-21 Richard Biener <rguenther@suse.de>
17219 PR tree-optimization/81900
17220 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17221 for blocks with abnormal predecessors.
17222 (compute_antic): Do not set visited flag prematurely.
17224 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17227 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17229 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17231 * stor-layout.h (vector_type_mode): Move to...
17232 * tree.h (vector_type_mode): ...here.
17233 * stor-layout.c (vector_type_mode): Move to...
17234 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17236 2017-08-21 Richard Biener <rguenther@suse.de>
17238 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17239 register_external_die hooks.
17240 (debug_false_tree_charstarstar_uhwistar): Declare.
17241 (debug_nothing_tree_charstar_uhwi): Likewise.
17242 * debug.c (do_nothing_debug_hooks): Adjust.
17243 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17244 (debug_nothing_tree_charstar_uhwi): Likewise.
17245 * dbxout.c (dbx_debug_hooks): Adjust.
17246 (xcoff_debug_hooks): Likewise.
17247 * sdbout.c (sdb_debug_hooks): Likewise.
17248 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17249 * dwarf2out.c (macinfo_label_base): New global.
17250 (dwarf2out_register_external_die): New function for the
17251 register_external_die hook.
17252 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17253 (dwarf2_debug_hooks): Use them.
17254 (dwarf2_lineno_debug_hooks): Adjust.
17255 (struct die_struct): Add with_offset flag.
17256 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17257 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17258 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17259 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17260 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17261 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17262 defining section names for the early LTO debug variants.
17263 (reset_indirect_string): New helper.
17264 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17265 (print_dw_val): Add support for offsetted symbol references.
17266 (get_ultimate_context): Split out from is_cxx.
17267 (is_cxx): Use get_ultimate_context.
17268 (is_fortran): Add decl overload.
17269 (compute_comp_unit_symbol): Split out worker from
17270 compute_section_prefix.
17271 (compute_section_prefix): Call compute_comp_unit_symbol and
17272 set comdat_type_p here.
17273 (output_die): Skip DIE symbol output for the LTO added one.
17274 Handle DIE symbol references with offset.
17275 (output_comp_unit): Guard section name mangling properly.
17276 For LTO debug sections emit a symbol at the section beginning
17277 which we use to refer to its DIEs.
17278 (add_abstract_origin_attribute): For DIEs registered via
17279 dwarf2out_register_external_die directly refer to the early
17280 DIE rather than indirectly through the shadow one we created.
17281 Remove obsolete call to dwarf2out_abstract_function for
17282 non-function/block origins.
17283 (gen_array_type_die): When generating early LTO debug do
17284 not emit DW_AT_string_length.
17285 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17286 late when in LTO. As suggested place a gcc_unreachable for
17287 the DECL_ABSTRACT_P case.
17288 (gen_subprogram_die): Avoid another specification DIE
17289 for early built declarations/definitions for the late LTO case.
17290 (gen_variable_die): Add type references for late duplicated VLA dies
17292 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17293 we have the abstract instance already.
17294 (process_scope_var): Adjust decl DIE contexts in LTO which
17295 first puts them in limbo.
17296 (gen_decl_die): Do not generate type DIEs late apart from
17297 types for VLAs or for decls we do not yet have a DIE. Do not
17298 call dwarf2out_abstract_function late.
17299 (dwarf2out_early_global_decl): Make sure to create DIEs
17300 for abstract instances of a decl first.
17301 (dwarf2out_late_global_decl): Adjust comment.
17302 (output_macinfo_op): With multiple macro sections use
17303 macinfo_label_base to distinguish labels.
17304 (output_macinfo): Likewise. Update macinfo_label_base.
17305 Pass in the line info label.
17306 (note_variable_value_in_expr): When generating LTO resolve
17307 all variable values here by generating DIEs as needed.
17308 (init_sections_and_labels): Add early LTO debug flag parameter
17309 and generate different sections and names if set. Add generation
17310 counter for the labels so we can have multiple of them.
17311 (reset_dies): Helper to allow DIEs to be output multiple times.
17312 (dwarf2out_finish): When outputting DIEs to the fat part of an
17313 LTO object first reset DIEs.
17314 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17315 (modified_type_die): Check for decl_ultimate_origin being self
17317 (gen_type_die_with_usage): Likewise.
17318 (gen_typedef_die): Allow decl_ultimate_origin being self.
17319 (set_decl_abstract_flags): Remove.
17320 (set_block_abstract_flags): Likewise.
17321 (dwarf2out_abstract_function): Treat the early generated DIEs
17322 as the abstract copy and only add DW_AT_inline and
17323 DW_AT_artificial here and call set_decl_origin_self.
17324 If the DIE has an abstract origin don't do anything.
17325 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17326 if we have none yet (Go fails to build one, PR78628).
17327 (variably_modified_type_p): Prevent endless recursion for Ada
17328 cyclic pointer types.
17329 * lto-streamer-in.c: Include debug.h.
17330 (dref_queue): New global.
17331 (lto_read_tree_1): Stream in DIE references.
17332 (lto_input_tree): Register DIE references.
17333 (input_function): Stream DECL_DEBUG_ARGS.
17334 * lto-streamer-out.c: Include debug.h.
17335 (lto_write_tree_1): Output DIE references.
17336 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17337 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17338 (output_function): Stream DECL_DEBUG_ARGS.
17339 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17340 Stream DECL_ABSTRACT_ORIGIN.
17341 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17342 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17343 DECL_CONTEXT for file-scope decls.
17344 * lto-streamer.h (struct dref_entry): Declare.
17345 (dref_queue): Likewise.
17346 * cfgexpand.c (pass_expand::execute): Do not call the
17347 outlining_inline_function hook here.
17348 * lto-wrapper.c (debug_obj): New global.
17349 (tool_cleanup): Unlink it if required.
17350 (debug_objcopy): New function.
17351 (run_gcc): Handle early debug sections in the IL files by
17352 extracting them to separate files, partially linkin them and
17353 feeding the result back as result to the linker.
17354 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17355 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17356 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17357 sections into a separate segment.
17358 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17360 (darwin_asm_dwarf_section): Likewise.
17361 (darwin_asm_output_dwarf_offset): Likewise.
17362 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17364 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17365 Alan Hayward <alan.hayward@arm.com>
17366 David Sherwood <david.sherwood@arm.com>
17368 * read-md.h (md_reader::record_potential_iterator_use): Replace
17369 pointer argument with an rtx and an index.
17370 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17371 (apply_mode_iterator): Likewise.
17372 (apply_code_iterator): Likewise.
17373 (apply_int_iterator): Likewise.
17374 (apply_subst_iterator): Likewise.
17375 (record_iterator_use): Likewise.
17376 (record_attribute_use): Likewise.
17377 (md_reader::record_potential_iterator_use): Likewise. Update calls
17378 to record_iterator_use and apply_iterator.
17379 (iterator_use): Replace ptr with x and index.
17380 (attribute_use): Likewise.
17381 (apply_attribute_uses): Update calls to apply_iterator.
17382 (apply_iterators): Likewise. Update initialization of iterator_use.
17383 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17384 and record_potential_iterator_use.
17385 (rtx_reader::read_rtx_operand): Likewise.
17387 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17388 Alan Hayward <alan.hayward@arm.com>
17389 David Sherwood <david.sherwood@arm.com>
17391 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17394 2017-08-21 Richard Biener <rguenther@suse.de>
17396 PR middle-end/81884
17397 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17398 at struct end conservatively when comparing common bases.
17400 2017-08-21 Richard Biener <rguenther@suse.de>
17402 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17403 (mem_ref_in_stmt): Remove.
17404 (determine_max_movement): Use ref index to get at the reference.
17405 (invariantness_dom_walker::before_dom_children): Deal with
17406 lim data already initialized.
17407 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17409 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17411 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17412 (TARGET_ISA_ROUND): Ditto.
17413 (TARGET_ROUND): Ditto.
17414 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17415 * config/i386/i386.md: Ditto.
17416 * config/i386/sse.md: Ditto.
17417 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17418 with OPTION_MASK_ISA_SSE4_1.
17420 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17423 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17424 __builtin_ia32_lzcnt_u16.
17426 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17429 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17430 (rs6000_set_current_function): Rewrite function to use it.
17432 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17435 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17436 and TYPE_WARN_IF_NOT_ALIGN.
17437 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17438 (handle_warn_if_not_align): New.
17439 (place_union_field): Call handle_warn_if_not_align.
17440 (place_field): Call handle_warn_if_not_align.
17441 Copy TYPE_WARN_IF_NOT_ALIGN.
17442 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17443 (layout_type): Likewise.
17444 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17446 (tree_decl_common): Add warn_if_not_align.
17447 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17448 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17449 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17450 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17451 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17452 * doc/extend.texi: Document warn_if_not_aligned attribute.
17453 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17455 2017-08-17 Martin Liska <mliska@suse.cz>
17458 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17459 (get_data_dependence): Use it as pointer type.
17460 (distribute_loop): Likewise.
17462 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17464 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17465 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17466 (p8_vmrgow_<mode>_direct): New define_insn.
17467 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17468 handle endianness for vmrgew and vmrgow permute patterns.
17470 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17472 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17473 * config/rs6000/rs6000-cpus.def: Remove comment.
17474 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17475 (POWERPC_MASKS): Likewise.
17476 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17477 use of TARGET_VSX_TIMODE.
17478 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17479 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17480 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17481 (rs6000_option_override_internal): Remove dead code.
17482 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17483 (rs6000_legitimize_reload_address): Likewise.
17484 (rs6000_legitimate_address_p): Likewise.
17485 (rs6000_opt_masks): Delete "vsx-timode".
17486 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17487 from function comment.
17488 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17489 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17490 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17492 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17493 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17494 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17495 (VSX_TI): Likewise.
17497 (define_peephole2): Likewise.
17499 2017-08-17 Martin Sebor <msebor@redhat.com>
17502 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17503 past the end of an array.
17504 (test_pp_format): Add test cases.
17506 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17508 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17509 missing ECF_NOTHROW flags.
17511 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17514 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17515 operands residing in integer registers.
17516 (*vsx_le_perm_load_<mode>): Likewise.
17517 (*vsx_le_perm_store_<mode>): Likewise.
17518 (define_peephole2): Add peepholes to optimize the above.
17520 2017-08-17 Marek Polacek <polacek@redhat.com>
17522 PR middle-end/81814
17523 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17524 to mimic what shorten_compare did. Change the return type to bool.
17525 (fold_cond_expr_with_comparison): Update call to
17526 operand_equal_for_comparison_p.
17527 (fold_ternary_loc): Likewise.
17529 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17531 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17533 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17535 2017-08-17 Richard Biener <rguenther@suse.de>
17537 * tree-ssa-structalias.c (solve_graph): When propagating
17538 to successors update the graphs succ edges and avoid duplicate work.
17540 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17543 * config/i386/i386.c (ix86_option_override_internal): Save target
17544 specific options after ix86_stack_protector_guard_reg was changed.
17546 2017-08-17 Richard Biener <rguenther@suse.de>
17548 PR tree-optimization/81827
17549 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17550 (new_var_info): Initialize it conservatively.
17551 (get_call_vi): Mark register vars.
17552 (new_scalar_tmp_constraint_exp): Likewise.
17553 (handle_rhs_call): Likewise.
17554 (handle_const_call): Likewise.
17555 (create_function_info_for): Likewise.
17556 (solve_constraints): Sort varinfos to separate register from
17557 non-register vars to pack points-to solution bitmaps during
17560 2017-08-17 Marek Polacek <polacek@redhat.com>
17562 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17564 2017-08-17 Richard Biener <rguenther@suse.de>
17566 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17567 to true when overflow is undefined and we saturated the result.
17569 2017-08-17 Alan Modra <amodra@gmail.com>
17572 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17573 Don't use store multiple if only one reg needs saving.
17574 (interesting_frame_related_regno): New function.
17575 (rs6000_frame_related): Don't emit frame info for regs that
17577 (rs6000_emit_epilogue): Likewise.
17579 2017-08-16 Nathan Sidwell <nathan@acm.org>
17581 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17582 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17583 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17584 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17585 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17586 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17587 (verify_type): Adjust for TYPE_BINFO move.
17588 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17589 process TYPE_BINFO directly.
17590 (hash_tree): Likewise.
17591 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17593 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17596 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17598 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17600 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17603 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17604 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17605 Add ix86_binary_operator_ok to insn constraint.
17606 (*iordi_1_bts): Ditto.
17607 (*xordi_1_btc): Ditto.
17608 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17609 Update corresponding peephole2 pattern.
17613 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17615 PR tree-optimization/81832
17616 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17617 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17619 2017-08-16 Marek Polacek <polacek@redhat.com>
17622 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17623 regarding size_zero_node.
17625 2017-08-16 Martin Liska <mliska@suse.cz>
17628 * config.gcc: Respect previously set extra_objs in case
17631 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17633 PR tree-optimization/81835
17634 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17635 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17636 not depend on the phi.
17638 2017-08-16 Alan Modra <amodra@gmail.com>
17640 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17643 2017-08-16 Alan Modra <amodra@gmail.com>
17645 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17646 (save_reg_p): ..into this. Update all callers.
17647 (first_reg_to_save): Simplify.
17649 2017-08-16 Alan Modra <amodra@gmail.com>
17651 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17654 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17658 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17659 constant count if that count is less than 32.
17661 2017-08-15 Nathan Sidwell <nathan@acm.org>
17663 * gcc.c (execute): Emit friendlier message if inferior is killed
17664 by an external cause.
17666 2017-08-15 Richard Biener <rguenther@suse.de>
17668 PR tree-optimization/81790
17669 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17670 CONSTRUCTORs from simplifying and VN.
17672 2017-08-14 Martin Sebor <msebor@redhat.com>
17674 * builtin-attrs.def: Add comments.
17676 2017-08-14 Martin Sebor <msebor@redhat.com>
17679 * doc/extend.texi (attribute nonstring): Document new attribute.
17681 2017-08-14 Martin Sebor <msebor@redhat.com>
17684 * tree-diagnostic.c (default_tree_printer): Handle %G.
17685 * gimple-pretty-print.h (percent_G_format): Declare new function.
17686 * gimple-pretty-print.c (percent_G_format): Define.
17687 * tree-pretty-print.c (percent_K_format): Add argument.
17689 2017-08-14 Martin Sebor <msebor@redhat.com>
17691 PR translation/79998
17692 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17693 Remove a stray space.
17695 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17698 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17699 (*iordi_1_bts): Ditto.
17700 (*xordi_1_btc): Ditto.
17702 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17705 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17706 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17708 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17709 quoted strings, and make more translator-friendly.
17710 (darwin_rs6000_override_options): Likewise.
17711 (rs6000_option_override_internal): Likewise.
17712 (rs6000_return_in_memory): Fix overlong line.
17713 (init_cmulative_args): Use quoted strings, and make more
17714 translator-friendly.
17715 (rs6000_pass_by_reference): Fix overlong line.
17716 (def_builtin): Use quoted strings.
17717 (altivec_expand_predicate_builtin): Use quoted strings, and make
17718 more translator-friendly.
17719 (htm_expand_builtin): Use quoted strings.
17720 (cpu_expand_builtin): Use quoted strings, and make more
17721 translator-friendly.
17722 (altivec_expand_builtin): Likewise.
17723 (paired_expand_predicate_builtin): Likewise.
17724 (rs6000_invalid_builtin): Likewise.
17725 (builtin_function_type): Use quoted strings.
17726 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17727 more translator-friendly.
17728 (rs6000_trampoline_init): Likewise.
17729 (rs6000_handle_altivec_attribute): Likewise.
17730 (rs6000_inner_target_options): Use quoted strings.
17731 (rs6000_disable_incompatible_switches): Likewise.
17732 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17733 strings, and make more translator-friendly.
17734 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17736 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17738 PR tree-optimization/81799
17739 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17740 cond_expr to simple gimple operand.
17742 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17744 PR middle-end/46932
17745 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17747 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17751 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17752 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17753 TARGET_GASISR_PROLOGUES.
17754 * config/avr/avr.c (avr_option_override): Same.
17755 (avr_pass_pre_proep::execute): Same.
17757 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17760 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17761 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17763 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17765 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17766 define_insn_and_split. Split to a memory load from 0 in
17767 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17768 using PTR mode iterator.
17769 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17770 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17771 (*add_tp_<mode>): Redefine as define_insn_and_split.
17772 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17773 address space. Merge with *add_tp_x32 using PTR mode iterator.
17774 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17775 Split to an add with a memory load from 0 in
17776 DEFAULT_TLS_SEG_REG address space.
17778 2017-08-12 Andrew Pinski <apinski@cavium.com>
17780 * config/aarch64/aarch64-option-extensions.def (rdma):
17781 Fix feature string to what Linux prints out in /proc/cpuinfo.
17783 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17786 * dwarf2out.c (modified_type_die): For C typedef types that have
17787 an ultimate origin, process the ultimate origin instead of the
17789 (gen_typedef_die): Assert that input DECLs have no ultimate
17791 (gen_type_die_with_usage): For typedef variants that have an
17792 ultimate origin, just call gen_decl_die on the original DECL.
17793 (process_scope_var): Avoid creating DIEs for local typedefs and
17794 concrete static variables.
17796 2017-08-12 Alan Modra <amodra@gmail.com>
17800 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17801 match gnu-user.h startfile.
17802 (ENDFILE_LINUX_SPEC): Similarly.
17804 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17807 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17809 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17811 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17812 * config/aarch64/aarch64.md (mov<mode>): Change.
17813 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17814 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17815 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17817 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17819 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17820 for storage order barriers.
17822 2017-08-11 Martin Liska <mliska@suse.cz>
17825 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17826 variables of void type.
17828 2017-08-11 Martin Liska <mliska@suse.cz>
17830 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17831 TARGET_SUPPORTS_ALIASES.
17832 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17833 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17834 (optimize_weakref): Likewise.
17835 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17836 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17837 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17839 2017-08-11 Martin Liska <mliska@suse.cz>
17842 * config/i386/i386.c (make_resolver_func): Do complete
17843 refactoring of the function.
17845 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17848 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17849 * config/i386/i386.c (ix86_stack_protect_guard): Use
17850 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17851 * doc/invoke.texi (x86 Options): Document
17852 -mstack-protector-guard-symbol= option.
17854 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17856 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17857 * config/i386/i386.c (ix86_split_stack_guard): New function.
17858 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17859 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17860 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17861 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17862 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17863 (split_stack_space_check): Call ix86_split_stack_guard.
17865 2017-08-10 Martin Sebor <msebor@redhat.com>
17867 * print-tree.c (print_node): Print location using the established
17869 Replace spaces with colons.
17870 (debug_raw, debug): Ditto.
17872 2017-08-10 Martin Sebor <msebor@redhat.com>
17875 * pretty-print.c (pp_format): Correct the handling of %s precision.
17877 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17880 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17882 (ix86_finalize_stack_frame_flags): This. Also clear
17883 frame_pointer_needed if -fno-omit-frame-pointer is used without
17885 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17886 with ix86_finalize_stack_frame_flags.
17887 (ix86_expand_epilogue): Likewise.
17888 (ix86_expand_split_stack_prologue): Likewise.
17889 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17891 2017-08-10 Martin Liska <mliska@suse.cz>
17894 * c-attribs.c (handle_target_attribute):
17895 Report warning for an empty string argument of target attribute.
17897 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17900 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17902 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17903 or DECL_NONLOCAL labels.
17904 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17905 or DECL_NONLOCAL labels here.
17907 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
17909 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
17910 to indicate when early gimple folding has been disabled.
17911 (rs6000_gimple_fold_builtin): Add debug content.
17912 (rs6000_invalid_builtin): Fix whitespace.
17913 (rs6000_expand_builtin): Fix whitespace.
17914 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
17916 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
17919 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
17920 SAVE_MULTIPLE if not all the registers that saves, should be saved.
17922 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
17924 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
17925 (qdf24xx): Likewise.
17926 * config/aarch64/aarch64.md: Include falkor.md.
17927 * config/aarch64/falkor.md: New.
17929 2017-08-09 Marek Polacek <polacek@redhat.com>
17932 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
17933 * diagnostic.c (emit_diagnostic): Add a comment.
17934 (emit_diagnostic_valist): New function.
17936 2017-08-09 Marek Polacek <polacek@redhat.com>
17939 * input.c (make_location): New overload.
17940 * input.h (make_location): Declare.
17942 2017-08-08 Alan Modra <amodra@gmail.com>
17943 H.J. Lu <hongjiu.lu@intel.com>
17946 * gcc.c (NO_PIE_SPEC): Delete.
17947 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
17949 (LINK_PIE_SPEC): ..to here.
17950 (LINK_COMMAND_SPEC): Support -no-pie.
17951 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
17952 chain of crtbegin*.o selection, update for PIE_SPEC changes and
17954 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
17955 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
17956 (ENDFILE_CRTEND_SPEC): Similarly.
17958 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
17961 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
17962 (mstack-protector-guard-offset=): Ditto.
17963 * config/i386/i386.c (ix86_option_override): Handle
17964 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
17966 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
17967 ix86_stack_protect_guard_offset variables.
17968 (TARGET_STACK_PROTECT_GUARD): Always define.
17969 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
17970 and -mstack-protector-guard-offset= options.
17972 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17974 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
17975 boundary case for the last candidate.
17977 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17979 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
17980 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
17982 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17984 PR middle-end/19706
17985 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
17986 * config/aarch64/aarch64-builtins.c
17987 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
17988 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
17989 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
17991 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17992 Andrew Pinski <pinskia@gmail.com>
17994 PR middle-end/19706
17995 * internal-fn.def (XORSIGN): New.
17996 * optabs.def (xorsign_optab): New.
17997 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
17998 (convert_expand_mult_copysign): New.
17999 (pass_optimize_widening_mul::execute): Call
18000 convert_expand_mult_copysign.
18002 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18004 PR tree-optimization/81354
18005 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18006 Insert on edges rather than explicitly creating landing pads.
18007 (analyze_candidates_and_replace): Commit edge inserts.
18009 2017-08-08 Richard Biener <rguenther@suse.de>
18011 PR middle-end/81719
18012 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18013 (expand_simple_operations): Also look through ADDR_EXPRs with
18014 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18016 2017-08-08 Richard Biener <rguenther@suse.de>
18018 PR tree-optimization/81723
18019 * tree-vect-slp.c (struct bst_traits): New hash traits.
18020 (bst_fail): New global.
18021 (vect_build_slp_tree_2): New worker, split out from ...
18022 (vect_build_slp_tree): ... this now wrapping it with using
18023 bst_fail set to cache SLP tree build fails. Properly handle
18025 (vect_analyze_slp_instance): Allocate and free bst_fail.
18027 2017-08-08 Martin Liska <mliska@suse.cz>
18030 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18031 LABEL_DECLs that can be from a different function.
18033 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18035 PR tree-optimization/81744
18036 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18037 loop's number of iterations.
18039 2017-08-08 Martin Liska <mliska@suse.cz>
18041 * asan.c: Include header files.
18042 * attribs.c (build_decl_attribute_variant): New function moved
18044 (build_type_attribute_qual_variant): Likewise.
18045 (cmp_attrib_identifiers): Likewise.
18046 (simple_cst_list_equal): Likewise.
18047 (omp_declare_simd_clauses_equal): Likewise.
18048 (attribute_value_equal): Likewise.
18049 (comp_type_attributes): Likewise.
18050 (build_type_attribute_variant): Likewise.
18051 (lookup_ident_attribute): Likewise.
18052 (remove_attribute): Likewise.
18053 (merge_attributes): Likewise.
18054 (merge_type_attributes): Likewise.
18055 (merge_decl_attributes): Likewise.
18056 (merge_dllimport_decl_attributes): Likewise.
18057 (handle_dll_attribute): Likewise.
18058 (attribute_list_equal): Likewise.
18059 (attribute_list_contained): Likewise.
18060 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18061 (lookup_attribute_by_prefix): Likewise.
18062 * bb-reorder.c: Include header files.
18063 * builtins.c: Likewise.
18064 * calls.c: Likewise.
18065 * cfgexpand.c: Likewise.
18066 * cgraph.c: Likewise.
18067 * cgraphunit.c: Likewise.
18068 * convert.c: Likewise.
18069 * dwarf2out.c: Likewise.
18070 * final.c: Likewise.
18071 * fold-const.c: Likewise.
18072 * function.c: Likewise.
18073 * gimple-expr.c: Likewise.
18074 * gimple-fold.c: Likewise.
18075 * gimple-pretty-print.c: Likewise.
18076 * gimple.c: Likewise.
18077 * gimplify.c: Likewise.
18078 * hsa-common.c: Likewise.
18079 * hsa-gen.c: Likewise.
18080 * internal-fn.c: Likewise.
18081 * ipa-chkp.c: Likewise.
18082 * ipa-cp.c: Likewise.
18083 * ipa-devirt.c: Likewise.
18084 * ipa-fnsummary.c: Likewise.
18085 * ipa-inline.c: Likewise.
18086 * ipa-visibility.c: Likewise.
18088 * lto-cgraph.c: Likewise.
18089 * omp-expand.c: Likewise.
18090 * omp-general.c: Likewise.
18091 * omp-low.c: Likewise.
18092 * omp-offload.c: Likewise.
18093 * omp-simd-clone.c: Likewise.
18094 * opts-global.c: Likewise.
18095 * passes.c: Likewise.
18096 * predict.c: Likewise.
18097 * sancov.c: Likewise.
18098 * sanopt.c: Likewise.
18099 * symtab.c: Likewise.
18100 * toplev.c: Likewise.
18101 * trans-mem.c: Likewise.
18102 * tree-chkp.c: Likewise.
18103 * tree-eh.c: Likewise.
18104 * tree-into-ssa.c: Likewise.
18105 * tree-object-size.c: Likewise.
18106 * tree-parloops.c: Likewise.
18107 * tree-profile.c: Likewise.
18108 * tree-ssa-ccp.c: Likewise.
18109 * tree-ssa-live.c: Likewise.
18110 * tree-ssa-loop.c: Likewise.
18111 * tree-ssa-sccvn.c: Likewise.
18112 * tree-ssa-structalias.c: Likewise.
18113 * tree-ssa.c: Likewise.
18114 * tree-streamer-in.c: Likewise.
18115 * tree-vectorizer.c: Likewise.
18116 * tree-vrp.c: Likewise.
18117 * tsan.c: Likewise.
18118 * ubsan.c: Likewise.
18119 * varasm.c: Likewise.
18120 * varpool.c: Likewise.
18121 * tree.c: Remove functions moved to attribs.[ch].
18122 * tree.h: Likewise.
18123 * config/aarch64/aarch64.c: Add attrs.h header file.
18124 * config/alpha/alpha.c: Likewise.
18125 * config/arc/arc.c: Likewise.
18126 * config/arm/arm.c: Likewise.
18127 * config/avr/avr.c: Likewise.
18128 * config/bfin/bfin.c: Likewise.
18129 * config/c6x/c6x.c: Likewise.
18130 * config/cr16/cr16.c: Likewise.
18131 * config/cris/cris.c: Likewise.
18132 * config/darwin.c: Likewise.
18133 * config/epiphany/epiphany.c: Likewise.
18134 * config/fr30/fr30.c: Likewise.
18135 * config/frv/frv.c: Likewise.
18136 * config/ft32/ft32.c: Likewise.
18137 * config/h8300/h8300.c: Likewise.
18138 * config/i386/winnt.c: Likewise.
18139 * config/ia64/ia64.c: Likewise.
18140 * config/iq2000/iq2000.c: Likewise.
18141 * config/lm32/lm32.c: Likewise.
18142 * config/m32c/m32c.c: Likewise.
18143 * config/m32r/m32r.c: Likewise.
18144 * config/m68k/m68k.c: Likewise.
18145 * config/mcore/mcore.c: Likewise.
18146 * config/microblaze/microblaze.c: Likewise.
18147 * config/mips/mips.c: Likewise.
18148 * config/mmix/mmix.c: Likewise.
18149 * config/mn10300/mn10300.c: Likewise.
18150 * config/moxie/moxie.c: Likewise.
18151 * config/msp430/msp430.c: Likewise.
18152 * config/nds32/nds32-isr.c: Likewise.
18153 * config/nds32/nds32.c: Likewise.
18154 * config/nios2/nios2.c: Likewise.
18155 * config/nvptx/nvptx.c: Likewise.
18156 * config/pa/pa.c: Likewise.
18157 * config/pdp11/pdp11.c: Likewise.
18158 * config/powerpcspe/powerpcspe.c: Likewise.
18159 * config/riscv/riscv.c: Likewise.
18160 * config/rl78/rl78.c: Likewise.
18161 * config/rx/rx.c: Likewise.
18162 * config/s390/s390.c: Likewise.
18163 * config/sh/sh.c: Likewise.
18164 * config/sol2.c: Likewise.
18165 * config/sparc/sparc.c: Likewise.
18166 * config/spu/spu.c: Likewise.
18167 * config/stormy16/stormy16.c: Likewise.
18168 * config/tilegx/tilegx.c: Likewise.
18169 * config/tilepro/tilepro.c: Likewise.
18170 * config/v850/v850.c: Likewise.
18171 * config/vax/vax.c: Likewise.
18172 * config/visium/visium.c: Likewise.
18173 * config/xtensa/xtensa.c: Likewise.
18175 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18178 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18179 constraints since the -mupper-regs-* switches have been
18181 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18182 into a vector from a double word element that was extracted from
18183 another vector, and eliminate extra XXPERMDI instructions.
18184 (vsx_concat_<mode>_2): Likewise.
18185 (vsx_concat_<mode>_3): Likewise.
18186 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18187 concat to allow optimizing inserts from previous extracts.
18189 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18191 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18192 memory reference to a SSP offset in TLS address space.
18193 (ix86_print_operand) <case '@'>: Remove.
18194 (ix86_print_operand_punct_valid_p): Remove '@' code.
18195 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18196 UNSPEC_SP_TLS_TEST.
18197 (stack_tls_protect_set_<mode>): Remove.
18198 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18199 (stack_tls_protect_test_<mode>): Remove.
18200 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18202 2017-08-07 Olivier Hainque <hainque@adacore.com>
18205 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18207 2017-08-07 Douglas Rupp <rupp@adacore.com>
18209 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18210 variable was referenced as multidir in command.
18212 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18215 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18218 2017-08-07 Martin Liska <mliska@suse.cz>
18220 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18221 * config/rl78/rl78.c: Add include of attribs.h.
18222 * config/sh/sh.c: Likewise.
18223 * config/v850/v850.c: Likewise.
18225 2017-08-07 Tom de Vries <tom@codesourcery.com>
18227 PR middle-end/78266
18228 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18230 2017-08-07 Martin Liska <mliska@suse.cz>
18232 * config/mips/mips.c: Include attribs.h.
18234 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18237 * doc/invoke.texi: Document change in behvaior for -Ofast for
18240 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18242 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18244 (aarch64_pop_regs): Likewise.
18245 (aarch64_gen_load_pair): Likewise.
18246 (aarch64_save_callee_saves): Likewise.
18247 (aarch64_restore_callee_saves): Likewise.
18249 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18251 * config/i386/i386.c: Revert the last change.
18253 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18256 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18258 (ix86_finalize_stack_frame_flags): This. Also clear
18259 frame_pointer_needed if -fno-omit-frame-pointer is used without
18261 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18262 with ix86_finalize_stack_frame_flags.
18263 (ix86_expand_epilogue): Likewise.
18264 (ix86_expand_split_stack_prologue): Likewise.
18266 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18269 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18270 to P_AES for Westmere.
18272 2017-08-07 Jonathan Yong <10walls@gmail.com>
18274 * config/i386/mingw.opt (fset-stack-executable): Removed.
18275 * config/i386/cygming.opt (fset-stack-executable): Moved
18277 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18279 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18281 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18283 2017-08-07 Marek Polacek <polacek@redhat.com>
18285 PR middle-end/81737
18286 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18288 2017-08-07 Martin Liska <mliska@suse.cz>
18290 * attribs.h (canonicalize_attr_name): New function.
18291 (cmp_attribs): Move from c-format.c and adjusted.
18292 (is_attribute_p): Moved from tree.h.
18293 * tree-inline.c: Add new includes.
18294 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18295 (private_is_attribute_p): Remove.
18296 (private_lookup_attribute): Likewise.
18297 (private_lookup_attribute_by_prefix): Simplify.
18298 (remove_attribute): Use is_attribute_p.
18299 * tree.h: Remove removed declarations.
18301 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18303 PR middle-end/81698
18304 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18305 instead of computing it in the function. Formatting fix.
18306 (expand_case): Don't rely on default_edge being the first edge,
18307 clear it if removing it, pass default_edge to
18308 emit_case_dispatch_table.
18309 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18312 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18314 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18315 insn in the function, emit NOP after the insn.
18317 2017-08-06 Tom de Vries <tom@codesourcery.com>
18319 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18322 2017-08-06 Tom de Vries <tom@codesourcery.com>
18324 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18327 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18329 PR tree-optimization/57371
18330 * match.pd: New pattern.
18332 2017-08-04 Marek Polacek <polacek@redhat.com>
18334 PR middle-end/81695
18335 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18336 perform the computation in offset_int.
18338 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18340 PR tree-optimization/81136
18341 * tree-vectorizer.h: Include tree-hash-traits.h.
18342 (vec_base_alignments): New typedef.
18343 (vec_info): Add a base_alignments field.
18344 (vect_record_base_alignments): Declare.
18345 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18347 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18348 (create_data_ref): Add an is_conditional_in_stmt argument.
18349 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18350 the is_conditional_in_stmt field.
18351 (data_ref_loc): Add an is_conditional_in_stmt field.
18352 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18353 (find_data_references_in_stmt): Update call to create_data_ref.
18354 (graphite_find_data_references_in_stmt): Likewise.
18355 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18356 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18357 (vect_record_base_alignment): New function.
18358 (vect_record_base_alignments): Likewise.
18359 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18360 for nested statements even if we fail to compute a misalignment.
18361 Use pooled base alignments for unconditional references.
18362 (vect_find_same_alignment_drs): Compare base addresses instead
18364 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18365 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18367 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18369 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18370 Add an explicit name for the enum. Use auto_vec for slp_instances
18371 and grouped_stores.
18372 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18374 (_bb_vec_info): Add a constructor and destructor.
18375 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18376 (destroy_loop_vec_info): Delete.
18377 (vect_destroy_datarefs): Likewise.
18378 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18379 (vec_info::vec_info): New function.
18380 (vec_info::~vec_info): Likewise.
18381 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18382 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18383 destroy_loop_vec_info.
18384 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18385 (_loop_vec_info::_loop_vec_info): ...this.
18386 (destroy_loop_vec_info): Replace with...
18387 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18388 the stmt_vec_infos. Leave handling of vec_info information to its
18389 destructor. Remove explicit vector releases.
18390 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18391 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18392 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18393 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18394 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18395 (destroy_bb_vec_info): Replace with...
18396 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18397 information to its destructor.
18398 (vect_slp_analyze_bb_1): Use new and delete instead of
18399 new_bb_vec_info and destroy_bb_vec_info.
18400 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18403 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18405 * tree-data-ref.h (subscript): Add access_fn field.
18406 (data_dependence_relation): Add could_be_independent_p.
18407 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18408 (same_access_functions): Move to tree-data-ref.c.
18409 * tree-data-ref.c (ref_contains_union_access_p): New function.
18410 (access_fn_component_p): Likewise.
18411 (access_fn_components_comparable_p): Likewise.
18412 (dr_analyze_indices): Add a reference to access_fn_component_p.
18413 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18415 (constant_access_functions): Likewise.
18416 (add_other_self_distances): Likewise.
18417 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18418 (initialize_data_dependence_relation): Use XCNEW and remove
18419 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18420 of access functions that have the same type. Allow the
18421 subsequence to end with different bases in some circumstances.
18422 Record the chosen access functions in SUB_ACCESS_FN.
18423 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18424 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18425 (subscript_dependence_tester_1): Likewise dra and drb.
18426 (build_classic_dist_vector): Update calls accordingly.
18427 (subscript_dependence_tester): Likewise.
18428 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18429 DDR_COULD_BE_INDEPENDENT_P.
18430 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18431 comp_alias_ddrs instead of may_alias_ddrs.
18432 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18434 (vect_analyze_data_ref_dependence): Use it if
18435 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18436 distance vectors if that fails.
18437 (dependence_distance_ge_vf): New function.
18438 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18439 LOOP_VINFO_MAY_ALIAS_DDRS.
18441 2017-08-04 Richard Biener <rguenther@suse.de>
18443 PR middle-end/81705
18444 * fold-const.c (fold_binary_loc): Properly restrict
18445 minus_var0 && minus_var1 case when associating undefined overflow
18448 2017-08-04 Tom de Vries <tom@codesourcery.com>
18450 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18452 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18454 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18455 Don't start diagnostic messages with a capital letter.
18456 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18458 (rs6000_invalid_builtin): Likewise.
18459 (rs6000_trampoline_init): Likewise.
18461 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18464 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18465 after setting changeable df flags.
18467 2017-08-03 Richard Biener <rguenther@suse.de>
18469 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18470 up if the use is in USE - X.
18472 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18474 * toplev.c (dumpfile.h): New include.
18475 (internal_error_reentered): New static function. Use it...
18476 (internal_error_function): ...here to handle reentered internal_error.
18478 2017-08-03 Richard Biener <rguenther@suse.de>
18480 PR middle-end/81148
18481 * fold-const.c (split_tree): Add minus_var and minus_con
18482 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18483 here but always use minus_*.
18484 (associate_trees): Assert we never associate with MINUS_EXPR
18485 and NULL first operand. Do not recurse for PLUS_EXPR operands
18486 when associating as MINUS_EXPR either.
18487 (fold_binary_loc): Track minus_var and minus_con.
18489 2017-08-03 Tom de Vries <tom@codesourcery.com>
18492 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18493 ACCEL_COMPILER, apply finish_options on
18494 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18496 2017-08-03 Tom de Vries <tom@codesourcery.com>
18499 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18500 function_entry_patch_area_size > 0.
18502 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18505 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18506 instead of 10??LU, perform unit multiplication in wide_int,
18507 don't change alloc_object_size_limit if the limit is larger
18510 PR tree-optimization/81655
18511 PR tree-optimization/81588
18512 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18513 the case when ranges[i].low and high are 1 for unsigned type with
18516 PR middle-end/81052
18517 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18518 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18520 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18522 * tree-vrp.h: Add include guard.
18524 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18527 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18528 (ud2): New insn pattern.
18529 * config/i386/i386.c (ix86_expand_epilogue):
18530 For naked functions, generate ud2 instead of trap insn.
18532 2017-08-02 Marek Polacek <polacek@redhat.com>
18535 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18537 2017-08-02 Tom de Vries <tom@codesourcery.com>
18538 Cesar Philippidis <cesar@codesourcery.com>
18540 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18541 Add missing edge probabilities.
18543 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18545 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18546 Correct endianness.
18548 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18550 PR middle-end/79499
18551 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18552 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18553 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18554 prologue_seq sequences - if any.
18556 2017-08-02 Richard Biener <rguenther@suse.de>
18558 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18559 via vectors if supported, integer extracts via punning if supported
18560 or otherwise vector extracts.
18562 2017-08-02 Richard Biener <rguenther@suse.de>
18564 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18566 (bitmap_insert_into_set): ... this.
18568 2017-08-02 Richard Biener <rguenther@suse.de>
18570 PR tree-optimization/81633
18572 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18574 PR tree-optimization/71752
18575 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18577 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18579 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18580 (machine_function::call_ms2sysv_pad_out): Remove field.
18581 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18582 (ix86_compute_frame_layout): Likewise.
18584 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18587 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18588 attribute with interrupt attribute.
18590 2017-08-01 Andrew Pinski <apinski@cavium.com>
18592 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18593 BIT_INSERT_EXPR's operand 1
18594 to see if the types precision matches.
18596 2017-08-01 Martin Liska <mliska@suse.cz>
18598 PR middle-end/70140
18599 * builtins.c (expand_builtin_memcpy_args): Remove.
18600 (expand_builtin_memcpy): Call newly added function
18601 expand_builtin_memory_copy_args.
18602 (expand_builtin_memcpy_with_bounds): Likewise.
18603 (expand_builtin_mempcpy): Remove last argument.
18604 (expand_builtin_mempcpy_with_bounds): Likewise.
18605 (expand_builtin_memory_copy_args): New function created from
18606 expand_builtin_mempcpy_args with small modifications.
18607 (expand_builtin_mempcpy_args): Remove.
18608 (expand_builtin_stpcpy): Remove unused argument.
18609 (expand_builtin): Likewise.
18610 (expand_builtin_with_bounds): Likewise.
18612 2017-08-01 Martin Liska <mliska@suse.cz>
18615 Make mempcpy more optimal (PR middle-end/70140).
18617 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18620 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18621 __builtin_vec_cmpne verify both arguments are compatible vectors
18622 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18623 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18624 move computation of aligned to after checking the argument types.
18628 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18629 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18632 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18633 Doug Rupp <rupp@adacore.com>
18634 Olivier Hainque <hainque@adacore.com>
18636 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18637 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18639 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18640 for TARGET_OS_CPP_BUILTIN.
18641 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18642 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18644 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18645 passing required abi options to the assembler for EABI configurations.
18646 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18647 of .text.hot and .text.unlikely sections for kernel modules when
18648 using ARM style exceptions.
18649 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18650 options. Add EXTRA_CC1_SPEC.
18651 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18653 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18655 (ARM_TARGET2_DWARF_FORMAT): Define.
18656 * config/arm/t-vxworks: Adjust multilib control to removal of the
18657 Diab command line options.
18659 2017-08-01 Martin Liska <mliska@suse.cz>
18661 PR gcov-profile/81561
18662 * gcov.c (unblock): Make unblocking safe as we need to preserve
18663 index correspondence of blocks and block_lists.
18665 2017-08-01 Richard Biener <rguenther@suse.de>
18667 PR tree-optimization/81181
18668 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18669 (compute_antic): ... end of iteration here.
18671 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18673 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18674 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18675 (ftree-slp-vectorize): Likewise.
18676 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18677 can no longer be set independent of flag_tree_loop_vectorize.
18678 * omp-general.c (emp_max_vf): Likewise.
18679 * opts.c (enable_fdo_optimizations): Remove references to
18680 flag_tree_vectorize, these are now implicit.
18681 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18682 and leave it for the options machinery.
18684 2017-08-01 Martin Liska <mliska@suse.cz>
18686 PR middle-end/70140
18687 * builtins.c (expand_builtin_memcpy_args): Remove.
18688 (expand_builtin_memcpy): Call newly added function
18689 expand_builtin_memory_copy_args.
18690 (expand_builtin_memcpy_with_bounds): Likewise.
18691 (expand_builtin_mempcpy): Remove last argument.
18692 (expand_builtin_mempcpy_with_bounds): Likewise.
18693 (expand_builtin_memory_copy_args): New function created from
18694 expand_builtin_mempcpy_args with small modifications.
18695 (expand_builtin_mempcpy_args): Remove.
18696 (expand_builtin_stpcpy): Remove unused argument.
18697 (expand_builtin): Likewise.
18698 (expand_builtin_with_bounds): Likewise.
18700 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18703 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18704 print "ds:" only for immediates in generic address space.
18706 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18709 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18710 (ix86_function_ok_for_sibcall): Return false for naked functions.
18712 2017-08-01 Richard Biener <rguenther@suse.de>
18714 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18715 (compute_antic): Seed worklist with exit block predecessors.
18716 * cfganal.c (dfs_find_deadend): For a cycle return the source
18717 of the edge closing it.
18719 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18721 * config/aarch64/aarch64.c
18722 (aarch64_can_const_movi_rtx_p): Move 0 check.
18724 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18726 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18727 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18729 * match.pd: Ditto in address comparison pattern.
18731 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18733 PR tree-optimization/81627
18734 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18735 closed ssa form for store-store chain.
18737 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18739 PR tree-optimization/81620
18740 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18741 for store-store chain.
18743 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18745 PR tree-optimization/81588
18746 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18747 ranges[i].in_p, invert comparison code ccode. For >/>=,
18748 swap rhs1 and rhs2 and comparison code unconditionally,
18749 for </<= don't do that. Don't swap rhs1/rhs2 again if
18750 ranges[i].in_p, instead invert comparison code ccode if
18751 opcode or oe->rank is BIT_IOR_EXPR.
18754 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18755 a direct optab to conversion optab.
18756 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18757 with GET_MODE_INNER as last argument instead of optab_handler.
18758 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18759 vector extraction if possible and optab is available.
18760 * expr.c (store_constructor): Use convert_optab_handler instead
18761 of optab_handler. Use vector initialization from smaller
18762 vectors if possible and optab is available.
18763 * tree-vect-stmts.c (vectorizable_load): Likewise.
18764 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18765 now have two modes.
18766 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18767 of vec_init from half-sized vectors with the same element mode.
18768 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18769 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18770 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18771 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18772 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18773 after mode in gen_vec_extract* calls.
18774 (vec_extract<mode>): Renamed to ...
18775 (vec_extract<mode><ssescalarmodelower>): ... this.
18776 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18777 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18778 element mode after mode in gen_vec_init* calls.
18779 (VEC_INIT_HALF_MODE): New mode iterator.
18780 (vec_init<mode>): Renamed to ...
18781 (vec_init<mode><ssescalarmodelower>): ... this.
18782 (vec_init<mode><ssehalfvecmodelower>): New expander.
18783 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18784 (vec_extractv2sfsf): ... this.
18785 (vec_initv2sf): Renamed to ...
18786 (vec_initv2sfsf): ... this.
18787 (vec_extractv2si): Renamed to ...
18788 (vec_extractv2sisi): ... this.
18789 (vec_initv2si): Renamed to ...
18790 (vec_initv2sisi): ... this.
18791 (vec_extractv4hi): Renamed to ...
18792 (vec_extractv4hihi): ... this.
18793 (vec_initv4hi): Renamed to ...
18794 (vec_initv4hihi): ... this.
18795 (vec_extractv8qi): Renamed to ...
18796 (vec_extractv8qiqi): ... this.
18797 (vec_initv8qi): Renamed to ...
18798 (vec_initv8qiqi): ... this.
18799 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18800 (vec_init<mode>): Renamed to ...
18801 (vec_init<mode><VEC_base_l>): ... this.
18802 (vec_extract<mode>): Renamed to ...
18803 (vec_extract<mode><VEC_base_l>): ... this.
18804 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18805 (vec_initv2sfsf): ... this.
18806 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18807 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18808 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18809 element mode after mode in gen_vec_init* calls.
18810 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18811 (vec_init<mode><Vel>): ... this.
18812 (vec_extract<mode>): Renamed to ...
18813 (vec_extract<mode><Vel>): ... this.
18814 * config/aarch64/iterators.md (Vel): New mode attribute.
18815 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18816 Add element mode after mode in gen_vec_extract* calls.
18817 * config/s390/vector.md (non_vec_l): New mode attribute.
18818 (vec_extract<mode>): Renamed to ...
18819 (vec_extract<mode><non_vec_l>): ... this.
18820 (vec_init<mode>): Renamed to ...
18821 (vec_init<mode><non_vec_l>): ... this.
18822 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18823 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18825 * config/arm/iterators.md (V_elem_l): New mode attribute.
18826 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18827 (vec_extract<mode><V_elem_l>): ... this.
18828 (vec_extractv2di): Renamed to ...
18829 (vec_extractv2didi): ... this.
18830 (vec_init<mode>): Renamed to ...
18831 (vec_init<mode><V_elem_l>): ... this.
18832 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18833 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18834 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18835 Add element mode after gen_vec_extract* calls.
18836 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18837 (vec_init<mode><unitmode>): ... this.
18838 (vec_extract<mode>): Renamed to ...
18839 (vec_extract<mode><unitmode>): ... this.
18840 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18841 (vec_init<mode><unitmode>): ... this.
18842 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18843 (vec_initv2sfsf): ... this.
18844 (vec_extractv2sf): Renamed to ...
18845 (vec_extractv2sfsf): ... this.
18846 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18847 Add element mode after gen_vec_extract* calls.
18848 * config/mips/mips.md (unitmode): New mode iterator.
18849 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18850 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18851 * config/spu/spu.md (inner_l): New mode attribute.
18852 (vec_init<mode>): Renamed to ...
18853 (vec_init<mode><inner_l>): ... this.
18854 (vec_extract<mode>): Renamed to ...
18855 (vec_extract<mode><inner_l>): ... this.
18856 * config/sparc/sparc.md (veltmode): New mode iterator.
18857 (vec_init<VMALL:mode>): Renamed to ...
18858 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18859 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18860 (vec_initv2sisi): ... this.
18861 (vec_initv2sf): Renamed to ...
18862 (vec_initv2sfsf): ... this.
18863 (vec_extractv2sf): Renamed to ...
18864 (vec_extractv2sfsf): ... this.
18865 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18866 (vec_init<mode>): Renamed to ...
18867 (vec_init<mode><VEC_base_l>): ... this.
18868 (vec_extract<mode>): Renamed to ...
18869 (vec_extract<mode><VEC_base_l>): ... this.
18870 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18871 (vec_initv2sfsf): ... this.
18872 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18873 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18874 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18875 gen_vec_init* calls.
18877 2017-08-01 Richard Biener <rguenther@suse.de>
18879 PR tree-optimization/81297
18880 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18881 TREE_OVERFLOW from INTEGER_CSTs.
18883 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18885 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18887 2017-07-31 Carl Love <cel@us.ibm.com>
18889 * config/rs6000/rs6000-c: Add support for built-in functions
18890 vector signed char vec_xl_be (signed long long, signed char *);
18891 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18892 vector signed int vec_xl_be (signed long long, signed int *);
18893 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18894 vector signed long long vec_xl_be (signed long long, signed long long *);
18895 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18896 vector signed short vec_xl_be (signed long long, signed short *);
18897 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18898 vector double vec_xl_be (signed long long, double *);
18899 vector float vec_xl_be (signed long long, float *);
18900 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18901 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18902 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18904 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18905 (altivec_expand_builtin): Add switch statement to call
18906 altivec_expand_xl_be for each builtin.
18907 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
18908 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
18909 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
18910 __builtin_vsx_le_be_v16qi.
18911 * doc/extend.texi: Update the built-in documentation file for the
18912 new built-in functions.
18914 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18917 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
18919 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18921 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18923 * config.gcc: Add z14.
18924 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
18925 CPU model numbers for z13s and z14.
18926 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
18928 * config/s390/s390-opts.h (enum processor_type): Rename
18929 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18930 * config/s390/s390.c (processor_table): Add field for CPU name to
18931 be passed to Binutils.
18932 (s390_asm_output_machine_for_arch): Use the new field in
18933 processor_table for Binutils.
18934 (s390_expand_builtin): Replace arch12 with z14.
18935 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18936 (s390_get_sched_attrmask): Likewise.
18937 (s390_get_unit_mask): Likewise.
18938 * config/s390/s390.opt: Add z14 to processor_type enum.
18940 2017-07-31 Martin Jambor <mjambor@suse.cz>
18943 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
18944 regardless of optimization level.
18946 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
18947 Martin Liska <mliska@suse.cz>
18949 * predict.def: Remove old comment and adjust probability.
18950 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
18951 PREDICT statements.
18953 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18956 * config/i386/i386.c (ix86_function_naked): New function.
18957 (ix86_can_use_return_insn_p): Return false for naked functions.
18958 (ix86_expand_prologue): Skip prologue for naked functions.
18959 (ix86_expand_epilogue): Skip epilogue for naked functions
18960 and emit trap instruction.
18961 (ix86_warn_func_return): New function.
18962 (ix86_attribute_table): Add "naked" attribute specification.
18963 (TARGET_WARN_FUNC_RETURN): Define.
18964 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
18966 2017-07-31 Martin Liska <mliska@suse.cz>
18968 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
18969 (dump_gimple_bb_header): Always dump BB info.
18970 (pp_cfg_jump): Do not append info about BB when dumping a jump.
18972 2017-07-31 Martin Liska <mliska@suse.cz>
18975 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
18976 also with current_function_decl non-null equality.
18978 2017-07-31 Jakub Jelinek <jakub@redhat.com>
18981 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
18982 change type to the element type, instead add eltype variable and
18983 use it where we are interested in the element type.
18985 PR tree-optimization/81603
18986 * ipa-polymorphic-call.c
18987 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
18988 offset arithmetic in offset_int, bail out if the resulting bit offset
18989 doesn't fit into shwi.
18991 2017-07-31 Martin Liska <mliska@suse.cz>
18993 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
18994 (gimplify_save_expr): Fix comment.
18996 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
18999 * config/i386/i386.c (ix86_function_arg): Update arguments for
19001 (ix86_compute_frame_layout): Set the initial stack offset to
19002 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19003 INCOMING_FRAME_SP_OFFSET.
19004 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19005 stack before exception handler returns.
19006 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19007 the 'ERROR_CODE' for exception handler.
19009 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19011 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19012 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19013 (ASM_OUTPUT_REG_POP): Ditto.
19014 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19015 instead of asm_fprintf to output pure string.
19017 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19019 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19020 to imported_module_or_decl hook.
19021 (debug_nothing_tree_tree_tree_bool): Remove.
19022 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19023 * debug.c (do_nothing_debug_hooks): Use
19024 debug_nothing_tree_tree_tree_bool_bool instead of
19025 debug_nothing_tree_tree_tree_bool.
19026 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19027 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19028 * sdbout.c (sdb_debug_hooks): Likewise.
19029 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19030 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19032 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19033 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19034 attribute, don't add anything.
19036 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19038 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19039 (fold_build2_stat_loc): Likewise.
19040 (fold_build3_stat_loc): Likewise.
19041 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19042 (fold_build1_loc): Remove macro.
19043 (fold_build2_loc): Likewise.
19044 (fold_build3_loc): Likewise.
19046 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19048 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19049 (gimple_build_debug_bind_source_stat): Likewise.
19050 * gimple.h (gimple_build_debug_bind): Remove macro.
19051 (gimple_build_debug_bind_source): Likewise.
19053 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19055 * bitmap.c (bitmap_alloc): Adjust.
19056 (bitmap_gc_alloc): Likewise.
19057 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19059 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19061 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19062 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19063 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19064 (bitmap_gc_alloc_stat): Likewise.
19065 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19067 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19069 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19070 * rtl.h (shallow_copy_rtx): Remove macro.
19072 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19074 * emit-rtl.c (gen_raw_REG): Adjust.
19075 * gengenrtl.c (gendef): Likewise.
19076 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19077 * rtl.h (rtx_alloc): Remove macro.
19079 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19081 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19082 (build_tree_list_stat): Likewise.
19083 * tree.h (build_tree_list): Remove macro.
19084 (build_tree_list_vec): Likewise.
19086 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19088 * tree.c (make_vector_stat): Remove _stat from name.
19089 (build_vector_stat): Likewise.
19090 * tree.h (make_vector_stat): Remove macro.
19091 (build_vector_stat): Likewise.
19093 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19095 * tree.h (build_var_debug_value): Remove prototype.
19097 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19099 * tree.c (tree_cons_stat): Remove _stat from name.
19100 * tree.h (tree_cons): Remove macro.
19102 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19104 * tree.c (build_vl_exp_stat): Remove _stat from name.
19105 * tree.h (build_vl_exp): Remove macro.
19107 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19109 * tree.c (build_decl_stat): Remove _stat from name.
19110 * tree.h (build_decl): Remove macro.
19112 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19114 * gimple.c (gimple_build_with_ops_stat): Adjust.
19115 (gimple_alloc_stat): Remove _stat from name.
19116 * gimple.h (gimple_alloc): Remove macro.
19118 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19120 * tree.c (make_tree_vec_stat): Remove _stat from name.
19121 (grow_tree_vec_stat): Likewise.
19122 * tree.h (make_tree_vec_stat): Adjust prototype.
19123 (grow_tree_vec_stat): Likewise.
19124 (make_tree_vec): Remove macro.
19125 (grow_tree_vec): Likewise.
19127 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19129 * fold-const.c (fold_build1_stat_loc): Adjust.
19130 (fold_build2_stat_loc): Likewise.
19131 (fold_build3_stat_loc): Likewise.
19132 * tree.c (build0_stat): Remove _stat from name.
19133 (build1_stat): Likewise.
19134 (build2_stat): Likewise.
19135 (build3_stat): Likewise.
19136 (build4_stat): Likewise.
19137 (build5_stat): Likewise.
19138 * tree.h (build1_loc): Remove macro, and rename _stat function
19140 (build2_loc): Likewise.
19141 (build3_loc): Likewise.
19142 (build4_loc): Likewise.
19143 (build5_loc): Likewise.
19145 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19147 * tree.c (make_int_cst_stat): Remove _stat from name.
19148 * tree.h (make_int_cst_stat): Adjust prototype.
19149 (make_int_cst): Remove macro.
19151 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19153 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19154 * tree.h (make_tree_binfo_stat): Adjust prototype.
19155 (make_tree_binfo): Remove.
19157 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19159 * tree.c (copy_node_stat): Rename to copy_node.
19160 (build_distinct_type_copy): Adjust.
19161 * tree.h (copy_node_stat): Adjust prototype.
19162 (copy_node): Remove macro.
19164 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19166 * tree.c (make_node_stat): rename to make_node.
19167 (build_tree_list_stat): Adjust.
19168 (build0_stat): Likewise.
19169 (build2_stat): Likewise.
19170 (build3_stat): Likewise.
19171 (build4_stat): Likewise.
19172 (build5_stat): Likewise.
19173 (build_decl_stat): Likewise.
19174 * tree.h (make_node_stat): Adjust prototype.
19175 (make_node): remove macro.
19177 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19179 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19180 (PPC_FEATURE2_SCV): Likewise.
19181 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19183 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19185 * config/aarch64/aarch64.c
19186 (aarch64_internal_mov_immediate): Add new special pattern.
19187 * config/aarch64/aarch64.md (*movdi_aarch64):
19188 Add reg/32bit const mov case.
19190 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19191 Richard Sandiford <richard.sandiford@linaro.org>
19193 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19194 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19195 Add integer and movi cases.
19196 (movi-split-hf-df-sf split, fp16): New.
19197 (enabled): Added TARGET_FP_F16INST.
19198 * config/aarch64/iterators.md (GPF_HF): New.
19199 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19201 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19203 * config/aarch64/aarch64.c
19204 (aarch64_simd_container_mode): Add prototype.
19205 (aarch64_expand_mov_immediate): Add HI support.
19206 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19207 (aarch64_can_const_movi_rtx_p): New.
19208 (aarch64_preferred_reload_class):
19209 Remove restrictions of using FP registers for certain SIMD operations.
19210 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19211 (aarch64_valid_floating_const): Add integer move validation.
19212 (aarch64_simd_imm_scalar_p): Remove.
19213 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19214 (aarch64_legitimate_constant_p): Expand list of supported cases.
19215 * config/aarch64/aarch64-protos.h
19216 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19217 (aarch64_reinterpret_float_as_int): New.
19218 (aarch64_simd_imm_scalar_p): Remove.
19219 * config/aarch64/constraints.md (Uvi): New.
19220 (Dd): Split into Ds and new Dd.
19221 * config/aarch64/aarch64.md (*movsi_aarch64):
19223 (*movdi_aarch64): Add SIMD mov case.
19225 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19227 * tree-predcom.c: (struct chain): Handle store-store chain in which
19228 stores for elimination only store loop invariant values.
19229 (execute_pred_commoning_chain): Ditto.
19230 (prepare_initializers_chain_store_elim): Ditto.
19231 (prepare_finalizers): Ditto.
19232 (is_inv_store_elimination_chain): New function.
19233 (initialize_root_vars_store_elim_1): New function.
19235 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19237 * tree-predcom.c: Revise general description of the pass.
19238 (enum chain_type): New enum type for store elimination.
19239 (struct chain): New field supporting store elimination.
19240 (struct component): Ditto.
19241 (dump_chain): Dump store-stores chain.
19242 (release_chain): Release resources.
19243 (split_data_refs_to_components): Compute and create component
19244 contains only stores for elimination.
19245 (get_chain_last_ref_at): New function.
19246 (make_invariant_chain): Initialization.
19247 (make_rooted_chain): Specify chain type in parameter and record it.
19248 (add_looparound_copies): Skip for store-stores chain.
19249 (determine_roots_comp): Compute type of chain and pass it to
19251 (initialize_root_vars_store_elim_2): New function.
19252 (finalize_eliminated_stores): New function.
19253 (remove_stmt): Handle store for elimination.
19254 (execute_pred_commoning_chain): Execute predictive commoning on
19255 store-store chains.
19256 (determine_unroll_factor): Skip unroll for store-stores chain.
19257 (prepare_initializers_chain_store_elim): New function.
19258 (prepare_initializers_chain): Hanlde store-store chain.
19259 (prepare_finalizers_chain, prepare_finalizers): New function.
19260 (tree_predictive_commoning_loop): Return integer value indicating
19261 if loop is unrolled or lcssa form is corrupted.
19262 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19264 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19266 * tree-predcom.c (initialize_root): Delete.
19267 (execute_pred_commoning_chain): Initialize root vars and replace
19268 reference of non-combined chain directly, rather than call above
19271 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19273 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19274 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19276 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19278 * tree-predcom.c (struct chain): New field init_seq.
19279 (release_chain): Release init_seq.
19280 (prepare_initializers_chain): Record intialization stmts in above
19282 (insert_init_seqs): New function.
19283 (tree_predictive_commoning_loop): Call insert_init_seqs.
19285 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19287 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19289 2017-07-28 Richard Biener <rguenther@suse.de>
19291 * match.pd: Remove superfluous :c.
19292 * genmatch.c (simplify::id): Add member.
19293 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19295 (current_id): New global.
19296 (dt_node::parent): Move from ...
19297 (dt_operand::parent): ... here. Add for_id member.
19298 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19299 (decision_tree::find_node): Relax order requirement when
19300 merging DT_TRUE nodes to ones inbetween the current simplify
19301 and the one we try to merge with. Add diagnostic whenever
19302 we need to enforce pattern order by not merging.
19303 (decision_tree::insert): Set current_id.
19304 (decision_tree::print_node): Dump parent node and for_id.
19305 (parser::last_id): Add member.
19306 (parser::push_simplify): Assign unique id.
19307 (parser::parser): Initialize last_id.
19309 2017-07-28 Martin Liska <mliska@suse.cz>
19312 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19313 gimple_build_debug_bind.
19315 2017-07-28 Richard Biener <rguenther@suse.de>
19317 PR tree-optimization/81502
19318 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19320 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19322 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19323 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19324 for BIT_FIELD_REF args.
19325 * fold-const.c (make_bit_field_ref): Likewise.
19326 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19328 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19331 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19332 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19333 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19334 Or it into SANITIZER_UNDEFINED.
19335 * ubsan.c: Include gimple-fold.h and varasm.h.
19336 (ubsan_expand_ptr_ifn): New function.
19337 (instrument_pointer_overflow): New function.
19338 (maybe_instrument_pointer_overflow): New function.
19339 (instrument_object_size): Formatting fix.
19340 (pass_ubsan::execute): Call instrument_pointer_overflow
19341 and maybe_instrument_pointer_overflow.
19342 * internal-fn.c (expand_UBSAN_PTR): New function.
19343 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19344 * sanitizer.def (__ubsan_handle_pointer_overflow,
19345 __ubsan_handle_pointer_overflow_abort): New builtins.
19346 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19347 * internal-fn.def (UBSAN_PTR): New internal function.
19348 * opts.c (sanitizer_opts): Add pointer-overflow.
19349 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19350 * fold-const.c (build_range_check): Compute pointer range check in
19351 integral type if pointer arithmetics would be needed. Formatting
19354 2017-07-28 Martin Liska <mliska@suse.cz>
19357 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19358 parameters that are of a variable-length.
19360 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19362 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19364 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19365 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19366 (CRT_CALL_STATIC_FUNCTION): Likewise.
19367 (ASM_DEFAULT_SPEC): New define.
19368 (ASM_SPEC32): Likewise.
19369 (ASM_SPEC64): Likewise.
19370 (ASM_SPEC_COMMON): Likewise.
19371 (ASM_SPEC): Likewise.
19372 (INVALID_64BIT): Likewise.
19373 (LINK_OS_DEFAULT_SPEC): Likewise.
19374 (LINK_OS_SPEC32): Likewise.
19375 (LINK_OS_SPEC64): Likewise.
19376 (POWERPC_LINUX): Likewise.
19377 (PTRDIFF_TYPE): Likewise.
19378 (RESTORE_FP_PREFIX): Likewise.
19379 (RESTORE_FP_SUFFIX): Likewise.
19380 (SAVE_FP_PREFIX): Likewise.
19381 (SAVE_FP_SUFFIX): Likewise.
19382 (SIZE_TYPE): Likewise.
19383 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19384 (TARGET_64BIT): Likewise.
19385 (TARGET_64BIT): Likewise.
19386 (TARGET_AIX): Likewise.
19387 (WCHAR_TYPE_SIZE): Likewise.
19388 (WCHAR_TYPE): Undefine.
19389 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19390 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19391 (CPP_OS_RTEMS_SPEC): Delete.
19392 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19393 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19395 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19397 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19399 PR tree-optimization/81578
19400 * tree-parloops.c (build_new_reduction): Bail out if
19401 reduction_code isn't one of the standard OpenMP reductions.
19402 Move the details printing after that decision.
19404 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19406 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19407 related to reload_in_progress.
19408 (splat_input_operand): Likewise.
19409 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19411 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19413 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19414 (TARGET_INSTANTIATE_DECLS): Likewise.
19415 (legitimate_indexed_address_p): Delete reload_in_progress code.
19416 (rs6000_debug_legitimate_address_p): Likewise.
19417 (rs6000_eliminate_indexed_memrefs): Likewise.
19418 (rs6000_emit_le_vsx_store): Likewise.
19419 (rs6000_emit_move_si_sf_subreg): Likewise.
19420 (rs6000_emit_move): Likewise.
19421 (register_to_reg_type): Likewise.
19422 (rs6000_pre_atomic_barrier): Likewise.
19423 (rs6000_machopic_legitimize_pic_address): Likewise.
19424 (rs6000_allocate_stack_temp): Likewise.
19425 (rs6000_address_for_fpconvert): Likewise.
19426 (rs6000_address_for_altivec): Likewise.
19427 (rs6000_secondary_memory_needed_rtx): Delete function.
19428 (rs6000_check_sdmode): Likewise.
19429 (rs6000_alloc_sdmode_stack_slot): Likewise.
19430 (rs6000_instantiate_decls): Likewise.
19431 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19432 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19433 Delete reload_in_progress.
19434 (*vec_reload_and_plus_<mptrsize>): Likewise.
19435 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19436 (vsx_div_v2di): Likewise.
19437 (vsx_udiv_v2di): Likewise.
19439 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19441 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19442 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19443 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19444 (rs6000_debug_reg_global): Delete print of LRA status.
19445 (rs6000_option_override_internal): Delete dead LRA related code.
19446 (rs6000_lra_p): Delete function.
19447 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19449 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19451 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19452 * config/riscv/rtems.h: New file.
19454 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19455 Sudakshina Das <sudi.das@arm.com>
19457 * config/aarch64/aarch64.md
19458 (define_split for and<mode>3nr_compare): Move
19459 non aarch64_logical_operand to a register.
19460 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19461 register immediate operand to a register.
19462 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19464 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19466 PR middle-end/81564
19467 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19469 2017-07-27 Richard Biener <rguenther@suse.de>
19471 PR tree-optimization/81573
19472 PR tree-optimization/81494
19473 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19474 multi defuse cycle case.
19476 2017-07-27 Richard Biener <rguenther@suse.de>
19478 PR tree-optimization/81571
19479 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19482 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19484 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19485 earlier and only if MASK_FPU is set. Adjust formatting.
19487 2017-07-27 Martin Liska <mliska@suse.cz>
19489 * opt-functions.awk: Add validation of value of Init.
19490 * optc-gen.awk: Pass new argument.
19492 2017-07-27 Martin Liska <mliska@suse.cz>
19494 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19495 Fix wrong condition.
19497 2017-07-27 Martin Liska <mliska@suse.cz>
19499 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19500 BBs and edges seen by autoFDO.
19502 2017-07-27 Richard Biener <rguenther@suse.de>
19504 PR tree-optimization/81502
19505 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19506 with incompatible but same sized type.
19507 (execute_update_addresses_taken): Likewise.
19509 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19511 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19512 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19514 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19517 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19518 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19519 Change s_operand to memory_operand.
19521 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19523 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19524 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19525 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19526 Emit instructions rather than returning an expression. Handle TFmode
19527 and KFmode by casting to TImode.
19528 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19529 (rs6000_emit_le_vsx_store): Likewise.
19530 * config/rs6000/vsx.md (VSX_TI): New iterator.
19531 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19532 (*vsx_le_undo_permute_<mode>): Likewise.
19533 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19534 emit the split sequence.
19535 (*vsx_le_perm_store_<mode>): Likewise.
19537 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19539 PR tree-optimization/81555
19540 PR tree-optimization/81556
19541 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19542 if true, force CHANGED for the recursive invocation.
19543 (reassociate_bb): Remember original length of ops array, pass
19544 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19546 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19547 attributes for noipa attribute. For naked attribute use
19548 lookup_attribute first before lookup_attribute_spec.
19549 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19551 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19552 for functions with noipa attribute.
19553 (cgraph_externally_visible_p): Return true for functions with noipa
19555 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19556 for functions with noipa attribute.
19557 * doc/extend.texi: Document noipa function attribute.
19558 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19559 also for functions with noipa attribute.
19560 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19562 2017-07-26 Andrew Pinski <apinski@cavium.com>
19564 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19565 vec_unalign_load_cost and vec_unalign_store_cost.
19567 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19569 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19570 -mvsx-small-integer option.
19571 (ISA_3_0_MASKS_IEEE): Likewise.
19572 (OTHER_VSX_VECTOR_MASKS): Likewise.
19573 (POWERPC_MASKS): Likewise.
19574 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19575 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19576 code, only testing for DImode being allowed in non-VSX floating
19578 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19579 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19581 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19582 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19583 TARGET_P8_VECTOR test.
19584 (rs6000_secondary_reload_simple_move): Likewise.
19585 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19586 since TARGET_P9_VECTOR was already tested.
19587 (rs6000_opt_masks): Remove -mvsx-small-integer.
19588 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19589 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19591 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19592 test for TARGET_VEXTRACTUB was used, and that uses
19594 (p9 extract splitter): Likewise.
19595 (vsx_extract_<mode>_di_p9): Likewise.
19596 (vsx_extract_<mode>_store_p9): Likewise.
19597 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19598 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19599 the elimination of TARGET_VSX_SMALL_INTEGER.
19600 (vsx_extract_<mode>_p8): Likewise.
19601 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19602 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19603 (vsx_set_<mode>_p9): Likewise.
19604 (vsx_set_v4sf_p9): Likewise.
19605 (vsx_set_v4sf_p9_zero): Likewise.
19606 (vsx_insert_extract_v4sf_p9): Likewise.
19607 (vsx_insert_extract_v4sf_p9_2): Likewise.
19608 * config/rs6000/rs6000.md (sign extend splitter): Change
19609 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19610 (floatsi<mode>2_lfiwax_mem): Likewise.
19611 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19612 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19613 since a test for TARGET_P9_VECTOR was used.
19614 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19615 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19616 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19617 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19618 TARGET_P8_VECTOR test.
19619 (fix_trunc<mode>si2_stfiwx): Likewise.
19620 (fix_trunc<mode>si2_internal): Likewise.
19621 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19622 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19624 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19625 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19626 TARGET_P8_VECTOR test.
19627 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19628 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19629 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19631 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19632 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19633 since a test for TARGET_P9_VECTOR was used.
19634 (splitter for loading small constants): Likewise.
19636 2017-07-26 Andrew Pinski <apinski@cavium.com>
19638 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19641 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19644 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19645 (fp_valid_at): Likewise.
19647 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19649 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19650 (qdf24xx_addrcost_table): Likewise.
19651 (cortexa57_tunings): Update to use generic_branch_cost.
19652 (cortexa72_tunings): Likewise.
19653 (cortexa73_tunings): Likewise.
19654 (qdf24xx_tunings): Likewise.
19656 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19658 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19659 (thunderx2t99_branch_cost): Likewise.
19660 (cortexa35_tunings): Update to use generic_branch_cost.
19661 (cortexa53_tunings): Likewise.
19662 (cortexa57_tunings): Likewise.
19663 (cortexa72_tunings): Likewise.
19664 (cortexa73_tunings): Likewise.
19665 (thunderx2t99_tunings): Likewise.
19667 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19669 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19670 (sparc_option_override): Honour MASK_FSMULD.
19671 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19672 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19673 * config/sparc/sparc.opt (mfsmuld): New option.
19674 * doc/invoke.texi (mfsmuld): Document option.
19676 2017-07-26 Marek Polacek <polacek@redhat.com>
19678 PR middle-end/70992
19679 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19681 2017-07-26 Richard Biener <rguenther@suse.de>
19683 * gimple-match-head.c (do_valueize): Return OP if valueize
19685 (get_def): New helper to get at the def stmt of a SSA name
19686 if valueize allows.
19687 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19688 do_valueize to get at the def stmt.
19689 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19691 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19693 PR middle-end/46932
19694 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19696 2017-07-26 Martin Liska <mliska@suse.cz>
19699 * function.c (expand_function_start): Make expansion of
19700 nonlocal_goto_save_area after parm_birth_insn.
19702 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19704 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19705 from all CPU target flags enable members.
19707 2017-07-26 Richard Biener <rguenther@suse.de>
19709 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19710 (decision_tree::gen): Make 'type' const.
19711 (write_predicate): Likewise.
19713 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19715 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19716 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19717 (rs6000_option_override_internal): Likewise.
19718 (rs6000_expand_vector_set): Likewise.
19719 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19720 (TARGET_UPPER_REGS_SF): Likewise.
19721 (TARGET_UPPER_REGS_DI): Likewise.
19722 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19723 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19724 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19725 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19726 (Splitters for DI constants in Altivec registers): Likewise.
19727 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19728 (vsx_set_v4sf_p9): Likewise.
19729 (vsx_set_v4sf_p9_zero): Likewise.
19730 (vsx_insert_extract_v4sf_p9): Likewise.
19731 (vsx_insert_extract_v4sf_p9_2): Likewise.
19733 2017-07-25 Carl Love <cel@us.ibm.com>
19735 * doc/extend.texi: Update the built-in documentation file for the
19736 existing built-in functions
19737 vector signed char vec_cnttz (vector signed char);
19738 vector unsigned char vec_cnttz (vector unsigned char);
19739 vector signed short vec_cnttz (vector signed short);
19740 vector unsigned short vec_cnttz (vector unsigned short);
19741 vector signed int vec_cnttz (vector signed int);
19742 vector unsigned int vec_cnttz (vector unsigned int);
19743 vector signed long long vec_cnttz (vector signed long long);
19744 vector unsigned long long vec_cnttz (vector unsigned long long);
19746 2017-07-25 Andrew Pinski <apinski@cavium.com>
19748 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19749 accesses where the use is for the first operand of a BIT_INSERT.
19751 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19754 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19755 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19757 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19759 * config/i386/gstabs.h: Delete.
19760 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19762 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19764 * config/i386/i386.c (ix86_decompose_address): Do not check for
19765 register RTX when looking at index_reg or base_reg.
19766 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19768 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19770 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19771 to update EH info here.
19773 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19775 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19777 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19779 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19781 2017-07-25 Torsten Duwe <duwe@suse.de>
19783 * common.opt: Introduce -fpatchable-function-entry
19784 command line option, and its variables function_entry_patch_area_size
19785 and function_entry_patch_area_start.
19786 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19787 including a two-value parser.
19788 * target.def (print_patchable_function_entry): New target hook.
19789 * targhooks.h (default_print_patchable_function_entry): New function.
19790 * targhooks.c (default_print_patchable_function_entry): Likewise.
19791 * toplev.c (process_options): Switch off IPA-RA if
19792 patchable function entries are being generated.
19793 * varasm.c (assemble_start_function): Look at the
19794 patchable-function-entry command line switch and current
19795 function attributes and maybe generate NOP instructions by
19796 calling the print_patchable_function_entry hook.
19797 * doc/extend.texi: Document patchable_function_entry attribute.
19798 * doc/invoke.texi: Document -fpatchable_function_entry
19799 command line option.
19800 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19802 * doc/tm.texi: Re-generate.
19804 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19807 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19808 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19810 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19812 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19814 2017-07-25 Richard Biener <rguenther@suse.de>
19816 PR tree-optimization/81455
19817 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19818 not walk in cycles when looking for guards.
19820 2017-07-25 Richard Biener <rguenther@suse.de>
19822 PR tree-optimization/81529
19823 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19824 when optimizing backedge uses.
19826 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19828 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19830 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19831 to dl_section_ref. On AIX, append an expression to subtract
19832 the size of the section length to dl_section_ref.
19834 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19836 * configure.ac: If any of the config.* scripts fail, exit 1.
19837 * configure: Regenerate.
19839 2017-07-25 Richard Biener <rguenther@suse.de>
19841 PR middle-end/81546
19842 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19843 of immediate uses, be more verbose on errors.
19845 2017-07-25 Richard Biener <rguenther@suse.de>
19847 PR tree-optimization/81510
19848 * tree-vect-loop.c (vect_is_simple_reduction): When the
19849 reduction stmt is not inside the loop bail out.
19851 2017-07-25 Richard Biener <rguenther@suse.de>
19853 PR tree-optimization/81303
19854 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19855 profitability check against LOOP_VINFO_NITERSM1.
19857 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19859 * domwalk.c (cmp_bb_postorder): Simplify.
19860 (sort_bbs_postorder): New function. Use it...
19861 (dom_walker::walk): ...here to optimize common cases.
19863 2017-07-25 Martin Liska <mliska@suse.cz>
19866 * ipa-visibility.c (function_and_variable_visibility): Make the
19867 redirection just on target that supports aliasing.
19868 Fix GNU coding style.
19870 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19873 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19875 * config/arm/rtems.h: Likewise.
19876 * config/bfin/rtems.h: Likewise.
19877 * config/i386/rtemself.h: Likewise.
19878 * config/lm32/rtems.h: Likewise.
19879 * config/m32c/rtems.h: Likewise.
19880 * config/m68k/rtemself.h: Likewise.
19881 * config/microblaze/rtems.h: Likewise.
19882 * config/mips/rtems.h: Likewise.
19883 * config/moxie/rtems.h: Likewise.
19884 * config/nios2/rtems.h: Likewise.
19885 * config/powerpcspe/rtems.h: Likewise.
19886 * config/rs6000/rtems.h: Likewise.
19887 * config/rtems.h: Likewise.
19888 * config/sh/rtems.h: Likewise.
19889 * config/sh/rtemself.h: Likewise.
19890 * config/sparc/rtemself.h: Likewise.
19892 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19895 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19896 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19897 * tree-ssa-structalias.c (alias_get_name): Same.
19899 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19902 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19903 instructions if no du chain is found.
19905 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19907 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
19909 2017-07-25 Richard Biener <rguenther@suse.de>
19911 PR middle-end/81505
19912 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
19915 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19917 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
19918 upper-regs options.
19919 (ISA_2_7_MASKS_SERVER): Likewise.
19920 (ISA_3_0_MASKS_IEEE): Likewise.
19921 (OTHER_P8_VECTOR_MASKS): Likewise.
19922 (OTHER_VSX_VECTOR_MASKS): Likewise.
19923 (POWERPC_MASKS): Likewise.
19924 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
19925 duplicate list of options.
19926 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
19927 explicit -mupper-regs options.
19928 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
19929 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
19930 alias for -mupper-regs-df.
19931 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19932 (rs6000_init_hard_regno_mode_ok): Likewise.
19933 (rs6000_option_override_internal): Likewise.
19934 (rs6000_opt_masks): Likewise.
19935 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
19936 options in terms of whether -mvsx or -mpower8-vector was used.
19937 (TARGET_UPPER_REGS_DI): Likewise.
19938 (TARGET_UPPER_REGS_SF): Likewise.
19939 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
19940 -mupper-regs-* options.
19942 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19944 * passes.c (emergency_dump_function): Print some empty lines and a
19945 header before the RTL dump.
19947 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19949 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
19951 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
19954 * config/aarch64/aarch64.c (aarch64_classify_symbol):
19955 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
19957 2017-07-24 Carl Love <cel@us.ibm.com>
19959 * config/rs6000/rs6000-c.c: Add support for built-in functions
19960 vector float vec_extract_fp32_from_shorth (vector unsigned short);
19961 vector float vec_extract_fp32_from_shortl (vector unsigned short);
19962 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
19963 vec_extract_fp_from_shortl): Add defines for the two builtins.
19964 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
19965 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
19967 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
19968 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
19969 * doc/extend.texi: Update the built-in documentation file for the
19970 new built-in function.
19972 2017-07-24 Jakub Jelinek <jakub@redhat.com>
19975 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
19977 * doc/generic.texi: Likewise.
19978 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19979 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19981 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
19983 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
19984 (aarch64_mls_elt_merge<mode>): Likewise.
19986 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
19988 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19989 having __cxa_atexit.
19991 2017-07-23 Michael Collison <michael.collison@arm.com>
19993 * config/arm/arm.c (arm_option_override): Deprecate
19994 use of -mstructure-size-boundary.
19995 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
19996 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
19998 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20001 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20002 Reduce cost estimate for direct moves.
20004 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20007 * config/i386/i386.c (ix86_option_override_internal): Disable
20008 BMI, BMI2 and TBM instructions for -m16.
20010 2017-07-21 Carl Love <cel@us.ibm.com>
20012 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20013 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20014 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20015 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20016 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20017 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20018 VMULOSW): New enum "unspec" values.
20019 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20020 altivec_vmulosw): New patterns.
20021 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20022 VMULOSW): Add definitions.
20024 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20026 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20027 (qdf24xx): Likewise.
20028 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20029 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20030 (AARCH64_FL_V8_1): Renumber.
20031 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20032 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20033 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20034 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20035 rdma to feature modifiers list.
20037 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20039 PR middle-end/56727
20040 * ipa-visibility (function_and_variable_visibility): Convert
20041 recursive PLT call to direct call if appropriate.
20043 2017-07-21 Andrew Pinski <apinski@cavium.com>
20045 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20046 operand 1 to see if the types precision matches.
20047 * fold-const.c (operand_equal_p): Likewise.
20049 2017-07-21 Richard Biener <rguenther@suse.de>
20051 PR tree-optimization/81303
20052 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20053 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20054 (vect_peeling_hash_get_lowest_cost): Adjust.
20055 (vect_enhance_data_refs_alignment): Likewise. Use
20056 vect_get_peeling_costs_all_drs to compute the penalty for no
20057 peeling to match up costs.
20059 2017-07-21 Richard Biener <rguenther@suse.de>
20061 PR tree-optimization/81500
20062 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20063 we didn't identify a reduction path.
20065 2017-07-21 Tom de Vries <tom@codesourcery.com>
20066 Cesar Philippidis <cesar@codesourcery.com>
20068 PR gcov-profile/81442
20069 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20072 2017-07-21 Tom de Vries <tom@codesourcery.com>
20075 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20077 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20078 nvptx_override_options_after_change.
20080 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20082 * dwarf2out.c (output_file_names): Avoid double testing for
20083 dwarf_version >= 5.
20085 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20087 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20089 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20091 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20093 (try_crossjump_to_edge): Do not punt on partitioned functions.
20095 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20097 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20098 Put all BBs reachable only via paths crossing cold region to cold
20100 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20102 2016-07-21 Richard Biener <rguenther@suse.de>
20104 PR tree-optimization/81303
20105 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20106 into account prologue and epilogue iterations when raising
20107 min_profitable_iters to sth at least covering one vector iteration.
20109 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20111 * config/arm/arm.c (arm_test_cpu_arch_dat):
20114 2017-07-20 Nathan Sidwell <nathan@acm.org>
20116 Remove TYPE_METHODS.
20117 * tree.h (TYPE_METHODS): Delete.
20118 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20119 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20120 (dbxout_type_methods): Scan TYPE_FIELDS.
20121 (dbxout_type): Don't check TYPE_METHODS here.
20122 * function.c (use_register_for_decl): Always ignore register for
20123 class types when not optimizing.
20124 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20125 * tree.c (free_lang_data_in_type): Stitch out member functions and
20126 templates from TYPE_FIELDS.
20127 (build_distinct_type_copy, verify_type_variant,
20128 verify_type): Member fns are on TYPE_FIELDS.
20129 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20130 * tree-pretty-print.c (dump_generic_node): Likewise.
20132 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20135 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20136 V2TImode and V4TImode.
20137 (ix86_expand_vector_extract): Likewise.
20138 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20139 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20140 (ssescalarmode): Handle V4TImode and V2TImode.
20141 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20142 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20143 (VEXTRACTI128_MODE): New mode iterator.
20144 (splitter for *vec_extractv?ti first element): New.
20145 (VEC_INIT_MODE): New mode iterator.
20146 (vec_init<mode>): Consolidate 3 expanders into one using
20147 VEC_INIT_MODE mode iterator.
20149 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20151 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20152 non_spilled_static_chain_regno_p.
20154 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20156 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20158 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20160 * bb-reorder.c (connect_traces): Allow copying of blocks within
20163 2017-07-20 Richard Biener <rguenther@suse.de>
20165 * gimple.h (gimple_phi_result): Add gphi * overload.
20166 (gimple_phi_result_ptr): Likewise.
20167 (gimple_phi_arg): Likewise. Adjust index assert to only
20168 allow actual argument accesses rather than all slots available
20170 (gimple_phi_arg_def): Add gphi * overload.
20171 * tree-phinodes.c (make_phi_node): Initialize only actual
20173 (resize_phi_node): Clear memory not covered by old node,
20174 do not initialize excess argument slots.
20175 (reserve_phi_args_for_new_edge): Initialize new argument slot
20178 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20180 PR tree-optimization/81388
20182 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20184 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20185 by removing computation of may_be_zero.
20187 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20188 Tom de Vries <tom@codesourcery.com>
20190 PR middle-end/81030
20191 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20192 when gimple level profile disagrees with what RTL expander did.
20194 2017-07-20 Richard Biener <rguenther@suse.de>
20196 PR tree-optimization/61171
20197 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20198 (vect_analyze_stmt): Add slp instance parameter.
20199 (vectorizable_reduction): Likewise.
20200 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20201 (vect_is_simple_reduction): Deal with chains not detected
20202 as SLP reduction chain, specifically not properly associated
20203 chains containing a mix of plus/minus.
20204 (get_reduction_op): Remove.
20205 (get_initial_defs_for_reduction): Simplify, pass in whether
20206 this is a reduction chain, pass in the SLP node for the PHIs.
20207 (vect_create_epilog_for_reduction): Get the SLP instance as
20209 (vectorizable_reduction): Get the SLP instance as arg.
20210 During analysis remember the SLP node with the PHIs in the
20211 instance. Simplify getting at the vectorized reduction PHIs.
20212 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20213 through SLP instance.
20214 (vect_slp_analyze_operations): Likewise.
20215 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20216 (vect_transform_stmt): Likewise.
20218 2017-07-20 Tom de Vries <tom@codesourcery.com>
20220 PR tree-optimization/81489
20221 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20222 read of phi arg location to before loop that modifies phi.
20224 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20226 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20229 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20231 PR middle-end/81331
20232 * except.c (execute): Fix ordering issue.
20234 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20236 PR rtl-optimization/81423
20237 * combine.c (make_compound_operation_int): Don't try to optimize
20238 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20240 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20242 PR rtl-optimization/81423
20243 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20244 with a constant that is -1 in the truncated to mode.
20246 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20248 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20249 (determine_unlikely_bbs): ... here.
20250 * predict.h (propagate_unlikely_bbs_forward): Declare.
20251 * cfgexpand.c (pass_expand::execute): Use it.
20252 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20254 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20255 propagate_unlikely_bbs_forward.
20257 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20259 PR middle-end/81331
20260 * except.c (maybe_add_nop_after_section_switch): New function.
20263 2017-07-19 Tom de Vries <tom@codesourcery.com>
20265 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20267 2017-07-19 Tom de Vries <tom@codesourcery.com>
20269 * gimple.h (gimple_phi_arg): Make assert more strict.
20271 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20273 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20274 * config/rs6000/mmintrin.h: New file.
20275 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20277 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20279 PR tree-optimization/81346
20280 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20282 2017-07-19 Tom de Vries <tom@codesourcery.com>
20284 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20286 2017-07-19 Tom de Vries <tom@codesourcery.com>
20288 * config/nvptx/nvptx-modes.def: Add V2DImode.
20289 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20290 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20291 (nvptx_output_mov_insn): Handle lack of mov.b128.
20292 (nvptx_print_operand): Handle 'H' and 'L' codes.
20293 (nvptx_vector_mode_supported): Allow V2DImode.
20294 (nvptx_preferred_simd_mode): New function.
20295 (nvptx_data_alignment): New function.
20296 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20297 nvptx_preferred_simd_mode.
20298 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20300 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20302 2017-07-19 Tom de Vries <tom@codesourcery.com>
20304 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20305 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20306 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20307 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20308 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20309 (mov<VECIM>_insn): New define_insn.
20310 (define_expand "mov<VECIM>): New define_expand.
20312 2017-07-19 Tom de Vries <tom@codesourcery.com>
20314 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20316 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20318 PR tree-optimization/81346
20319 * fold-const.h (fold_div_compare, range_check_type): Declare.
20320 * fold-const.c (range_check_type): New function.
20321 (build_range_check): Use range_check_type.
20322 (fold_div_compare): No longer static, rewritten into
20323 a match.pd helper function.
20324 (fold_comparison): Don't call fold_div_compare here.
20325 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20326 as helper function.
20328 2017-07-19 Nathan Sidwell <nathan@acm.org>
20330 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20331 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20332 * tree.c (find_decls_types_r, verify_type): Use
20333 TYPE_{MIN,MAX}_VALUE_RAW.
20334 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20335 (hash_tree): Likewise.
20336 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20338 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20341 2017-07-18 Tom de Vries <tom@codesourcery.com>
20343 PR middle-end/81464
20344 * omp-expand.c (expand_omp_for_static_chunk): Handle
20345 equal-argument loop exit phi.
20347 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20350 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20351 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20352 operand 2 predicate.
20353 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20354 operand 2 predicate.
20355 (ror,rol -> rorx splitters): Use const_int_operand as
20356 operand 2 predicate.
20358 2017-06-18 Richard Biener <rguenther@suse.de>
20360 PR tree-optimization/81410
20361 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20362 the gap in the ! slp_perm SLP case after each group.
20364 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20366 PR middle-end/81463
20367 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20370 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20372 PR middle-end/81462
20373 * predict.c (set_even_probabilities): Cleanup; do not affect
20374 probabilities that are already known.
20375 (combine_predictions_for_bb): Call even when count is set.
20377 2017-07-18 Nathan Sidwell <nathan@acm.org>
20379 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20382 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20385 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20386 optimization for loop niter analysis.
20388 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20391 * config/avr/avr.c (avr_optimize_casesi): Don't use
20392 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20394 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20396 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20397 body_cost_vec from _vect_peel_extended_info.
20398 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20399 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20402 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20404 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20406 2017-07-18 Richard Biener <rguenther@suse.de>
20408 PR tree-optimization/80620
20409 PR tree-optimization/81403
20410 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20411 info when re-using a VN table entry.
20413 2017-07-18 Richard Biener <rguenther@suse.de>
20415 PR tree-optimization/81418
20416 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20417 vectype_in. Verify that with lane-reducing reduction operations
20418 we have a single def-use cycle.
20420 2017-07-17 Carl Love <cel@us.ibm.com>
20422 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20424 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20425 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20426 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20427 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20428 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20429 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20430 VMULOSW): New enum "unspec" values.
20431 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20432 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20433 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20434 altivec_vmulosw): New patterns.
20435 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20436 VMULOSW): Add definitions.
20438 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20440 * config/alpha/alpha.c: Include predict.h.
20442 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20444 * tree-vrp.c (compare_assert_loc): Fix comparison function
20445 to return predictable results.
20447 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20449 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20451 (subdi3): Likewise.
20452 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20453 * doc/invoke.texi (mexpand-adddi): Update text.
20455 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20457 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20458 that also clobbers the CC register. The old expand code is moved
20460 (*arc_clzsi2): ... here.
20461 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20462 the CC register. The old expand code is moved to ...
20463 (arc_ctzsi2): ... here.
20465 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20467 * config/arc/arc.opt (mindexed-loads): Use initial value
20468 TARGET_INDEXED_LOADS_DEFAULT.
20469 (mauto-modify-reg): Use initial value
20470 TARGET_AUTO_MODIFY_REG_DEFAULT.
20471 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20472 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20473 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20474 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20476 2017-07-17 Martin Liska <mliska@suse.cz>
20479 * opts.c (finish_options): Do not allow -fgnu-tm
20480 w/ -fsanitize={kernel-,}address. Say sorry.
20482 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20485 * tree-loop-distribution.c (classify_partition): Only assert on
20486 numer of iterations.
20487 (merge_dep_scc_partitions): Delete prameter. Update function call.
20488 (distribute_loop): Remove code handling loop with unknown niters.
20489 (pass_loop_distribution::execute): Skip loop with unknown niters.
20491 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20494 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20495 function sort_partitions_by_post_order.
20497 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20499 PR tree-optimization/81374
20500 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20501 the max index of basic blocks, rather than number of basic blocks.
20503 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20505 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20507 (arc_legitimate_pic_operand_p): Likewise.
20508 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20510 (arc_needs_pcl_p): Likewise.
20511 (arc_legitimate_pc_offset_p): Likewise.
20512 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20513 function is also used in constrains.md.
20514 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20515 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20516 PLUS. Only return true/false in known cases, otherwise assert.
20517 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20518 is already called in arc_legitimate_constant_p.
20519 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20521 (LEGITIMATE_PIC_OPERAND_P): Use
20522 arc_raw_symbolic_reference_mentioned_p function.
20523 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20528 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20529 Andrew Burgess <andrew.burgess@embecosm.com>
20531 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20532 (arc_return_address_register): New function.
20533 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20534 (arc_handle_fndecl_attribute): Add naked attribute.
20535 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20536 (TARGET_WARN_FUNC_RETURN): Likewise.
20537 (arc_allocate_stack_slots_for_args): New function.
20538 (arc_warn_func_return): Likewise.
20539 (machine_function): Change type fn_type.
20540 (arc_compute_function_type): Consider new naked function type,
20541 change function return type.
20542 (arc_must_save_register): Adapt to handle new
20543 arc_compute_function_type's return type.
20544 (arc_expand_prologue): Likewise.
20545 (arc_expand_epilogue): Likewise.
20546 (arc_return_address_regs): Delete.
20547 (arc_return_address_register): New function.
20548 (arc_epilogue_uses): Use above function.
20549 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20550 (arc_function_type): Change encoding, add naked type.
20551 (ARC_INTERRUPT_P): Change to handle the new encoding.
20552 (ARC_FAST_INTERRUPT_P): Likewise.
20553 (ARC_NORMAL_P): Define.
20554 (ARC_NAKED_P): Likewise.
20555 (arc_compute_function_type): Delete prototype.
20556 * config/arc/arc.md (in_ret_delay_slot): Use
20557 arc_return_address_register function.
20558 (simple_return): Likewise.
20559 (p_return_i): Likewise.
20561 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20563 PR tree-optimization/81428
20564 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20565 can't be built for those types.
20567 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20569 Remove stuff dead since r239246.
20571 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20572 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20573 (avr_inform_devices): Remove dead stuff.
20575 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20577 * config/arm/arm_neon.h: Fix softp typo.
20579 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20581 PR tree-optimization/81365
20582 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20583 aggregate moves onto bb predecessor edges, make sure there are no
20584 loads that could alias the lhs in between the start of bb and the
20587 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20590 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20591 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20592 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20594 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20596 PR tree-optimization/81396
20597 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20598 (init_symbolic_number): Initialize it to 1.
20599 (perform_symbolic_merge): Add n_ops from both operands into the new
20601 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20602 without base_addr as useless if they need more than one operation.
20603 (bswap_replace): Handle !bswap case for NULL base_addr.
20605 2017-07-17 Tom de Vries <tom@codesourcery.com>
20608 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20611 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20613 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20614 conditional builtin define __FIX_LEON3FT_B2BST.
20616 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20618 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20619 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20622 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20624 PR rtl-optimization/81424
20625 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20626 to remove potential trapping from operands if -fnon-call-exceptions.
20628 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20630 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20631 profile_proability for scalling.
20632 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20634 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20636 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20638 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20640 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20641 fixpoint arithmetics.
20643 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20645 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20646 fixpoint arithmetics.
20648 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20650 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20651 fixpoint arithmetics.
20653 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20655 * profile-count.h (profile_probability::from_reg_br_prob_note,
20656 profile_probability::to_reg_br_prob_note): New functions.
20657 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20658 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20659 * predict.c (probability_reliable_p): Update.
20660 (edge_probability_reliable_p): Update.
20661 (br_prob_note_reliable_p): Update.
20662 (invert_br_probabilities): Update.
20663 (add_reg_br_prob_note): New function.
20664 (combine_predictions_for_insn): Update.
20665 * asan.c (asan_clear_shadow): Update.
20666 * cfgbuild.c (compute_outgoing_frequencies): Update.
20667 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20668 (update_br_prob_note): Update.
20669 (rtl_verify_edges): Update.
20670 (purge_dead_edges): Update.
20671 (fixup_reorder_chain): Update.
20672 * emit-rtl.c (try_split): Update.
20673 * ifcvt.c (cond_exec_process_insns): Update.
20674 (cond_exec_process_if_block): Update.
20675 (dead_or_predicable): Update.
20676 * internal-fn.c (expand_addsub_overflow): Update.
20677 (expand_neg_overflow): Update.
20678 (expand_mul_overflow): Update.
20679 * loop-doloop.c (doloop_modify): Update.
20680 * loop-unroll.c (compare_and_jump_seq): Update.
20681 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20682 * predict.h: Update.
20683 * reorg.c (mostly_true_jump): Update.
20685 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20686 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20687 * config/arc/arc.c: (emit_unlikely_jump): Update.
20688 * config/arm/arm.c: (emit_unlikely_jump): Update.
20689 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20690 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20691 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20692 (ix86_print_operand): Update.
20693 (ix86_split_fp_branch): Update.
20694 (predict_jump): Update.
20695 * config/ia64/ia64.c (ia64_print_operand): Update.
20696 * config/mmix/mmix.c (mmix_print_operand): Update.
20697 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20698 (rs6000_expand_split_stack_prologue): Update.
20699 * config/rs6000/rs6000.c: Update.
20700 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20701 (s390_expand_vec_movstr): Update.
20702 (s390_expand_cs_tdsi): Update.
20703 (s390_expand_split_stack_prologue): Update.
20704 * config/sh/sh.c (sh_print_operand): Update.
20705 (expand_cbranchsi4): Update.
20706 (expand_cbranchdi4): Update.
20707 * config/sparc/sparc.c (output_v9branch): Update.
20708 * config/spu/spu.c (get_branch_target): Update.
20709 (ea_load_store_inline): Update.
20710 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20711 * config/tilepro/tilepro.c: Update.
20713 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20715 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20716 (gimplify_save_expr): Likewise.
20718 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20720 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20722 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20724 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20726 * timevar.def (TV_IPA_FNSUMMARY): Define.
20728 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20730 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20731 to back store errata sensitive sequence from being generated.
20732 (sqrtdf2_fix): Likewise.
20734 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20736 * tree-ssa-threadupdate.c (compute_path_counts,
20737 update_joiner_offpath_counts): Use profile_probability.
20739 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20742 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20744 * config/arm/arm-c.c (arm_cpu_builtins): Define
20745 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20747 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20749 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20750 array entries to represent __ieee128 versions of the
20751 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20752 scalar_extract_sig, and scalar_insert_exp built-in functions.
20753 (altivec_resolve_overloaded_builtin): Add special case handling
20754 for the __builtin_scalar_insert_exp function, as represented by
20755 the P9V_BUILTIN_VEC_VSIEDP constant.
20756 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20757 exponent support for __ieee128 argument.
20758 (VSESQP): Add scalar extract signature support for __ieee128
20760 (VSTDCNQP): Add scalar test negative support for __ieee128
20762 (VSIEQP): Add scalar insert exponent support for __int128 argument
20763 with __ieee128 result.
20764 (VSIEQPF): Add scalar insert exponent support for __ieee128
20765 argument with __ieee128 result.
20766 (VSTDCQP): Add scalar test data class support for __ieee128
20768 (VSTDCNQP): Add overload support for scalar test negative with
20769 __ieee128 argument.
20770 (VSTDCQP): Add overload support for scalar test data class
20771 __ieee128 argument.
20772 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20773 UNSPEC_VSX_SXSIGDP.
20774 (UNSPEC_VSX_SIEXPQP): New constant.
20775 (xsxexpqp): New insn for VSX scalar extract exponent quad
20777 (xsxsigqp): New insn for VSX scalar extract significand quad
20779 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20780 precision with floating point argument.
20781 (xststdcqp): New expand for VSX scalar test data class quad
20783 (xststdcnegqp): New expand for VSX scalar test negative quad
20785 (xststdcqp): New insn to match expansions for VSX scalar test data
20786 class quad precision and VSX scalar test negative quad precision.
20787 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20788 special case operand checking to enforce that second operand of
20789 VSX scalar test data class with quad precision argument is a 7-bit
20791 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20792 prototypes and descriptions of __ieee128 versions of
20793 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20794 scalar_test_data_class, and scalar_test_neg built-in functions.
20796 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20798 PR tree-optimization/81162
20799 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20800 replace a negate with an add.
20802 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20804 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20806 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20808 * config/arm/arm-c.c (arm_cpu_builtins): Define
20809 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20811 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20813 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20814 (armv8-r): Set ARM Cortex-R52 as default CPU.
20815 * config/arm/arm-tables.opt: Regenerate.
20816 * config/arm/arm-tune.md: Regenerate.
20817 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20819 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20820 extension for -mcpu=cortex-r52.
20822 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20824 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20825 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20826 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20827 (fp-armv8): Define it as FP_ARMv8 only.
20828 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20829 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20831 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20832 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20833 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20834 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20835 than TARGET_FPU_ARMV8.
20836 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20837 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20838 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20840 * config/arm/neon.md (neon_vrint): Likewise.
20841 (neon_vcvt): Likewise.
20842 (neon_<fmaxmin_op><mode>): Likewise.
20843 (<fmaxmin><mode>3): Likewise.
20844 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20845 * config/arm/predicates.md (arm_cond_move_operator): Check against
20846 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20848 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20850 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20851 to top of function.
20853 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20855 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20856 loop in comment with memset.
20858 2017-07-14 Martin Liska <mliska@suse.cz>
20860 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20861 * dwarf2out.c (is_java): Remove the function.
20862 (output_pubname): Remove usage of the function.
20863 (lower_bound_default): Remove usage of DW_LANG_Java.
20864 (gen_compile_unit_die): Likewise.
20865 * gcc.c: Remove compiler defaults for .java and .zip files.
20866 * gimple-expr.c (remove_suffix): Change as there's no longer
20867 extension than 4-letter one.
20868 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20869 (gimplify_save_expr): Likewise.
20870 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20871 as it's possible even for other languages than Java.
20872 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20873 * lto-opts.c (lto_write_options): Remove reference to Java.
20874 * opts.c (strip_off_ending): Update file extension handling.
20875 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20876 * tree-eh.c (lower_resx): Likewise.
20877 * tree.c (free_lang_data_in_type): Remove dead code.
20878 (find_decls_types_r): Likewise.
20879 (build_common_builtin_nodes): Remove Java from a comment.
20880 (verify_type): Remove dead code.
20881 * varasm.c (assemble_external): Remove Java from a comment.
20883 2017-07-14 Martin Liska <mliska@suse.cz>
20885 * opts.c (finish_options): Add quotes.
20886 (common_handle_option): Likewise.
20888 2017-07-14 Martin Liska <mliska@suse.cz>
20890 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20891 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20892 Remove N_SO_PASCAL.
20893 * dwarf2out.c (lower_bound_default): Do not handle
20895 (gen_compile_unit_die): Likewise.
20896 * gcc.c: Remove default extension binding for GNU Pascal.
20897 * stmt.c: Remove Pascal language from a comment.
20898 * xcoffout.c: Likewise.
20900 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20903 * diagnostic-show-locus.c (fixit_cmp): New function.
20904 (layout::layout): Sort m_fixit_hints.
20905 (column_range::column_range): Assert that the values are valid.
20906 (struct char_span): New struct.
20907 (correction::overwrite): New method.
20908 (struct source_line): New struct.
20909 (line_corrections::add_hint): Add assertions. Reimplement memcpy
20910 calls in terms of classes source_line and char_span, and
20911 correction::overwrite.
20912 (selftest::test_overlapped_fixit_printing_2): New function.
20913 (selftest::diagnostic_show_locus_c_tests): Call it.
20915 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
20917 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
20918 early if there is no lhs.
20920 2017-07-13 Martin Liska <mliska@suse.cz>
20922 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
20923 (gen_reference_type_die): Likewise.
20924 * stor-layout.c: Remove Pascal-related comment.
20926 2017-07-13 Martin Liska <mliska@suse.cz>
20928 * opts.c (finish_options): Add quotes to error messages.
20929 (parse_sanitizer_options): Likewise.
20931 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
20933 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
20935 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
20937 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
20939 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
20941 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
20943 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
20945 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
20948 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
20949 provides the hardware capability bits, define the macro
20950 __BUILTIN_CPU_SUPPORTS__.
20951 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
20952 if GLIBC does not provide the hardware capability bits. Add a
20953 gcc_unreachable call if the built-in cpu function is neither
20954 __builtin_cpu_is nor __builtin_cpu_supports.
20955 (rs6000_get_function_versions_dispatcher): Change the warning
20956 that an old GLIBC is used which does not export the capability
20957 bits to be an error.
20958 * doc/extend.texi (target_clones attribute): Document the
20959 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
20960 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
20961 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
20962 the macros defined by GCC if the newer GLIBC is available.
20964 2017-07-12 Jeff Law <law@redhat.com>
20966 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
20967 remaining includes slightly.
20968 * config/riscv/riscv-builtins.c: Include profile-count.h.
20970 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20973 * config/avr/avr.c (avr_set_current_function): In diagnostic
20974 messages: Quote keywords and (parts of) identifiers.
20975 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
20978 2017-07-12 Carl Love <cel@us.ibm.com>
20980 * config/rs6000/rs6000-c.c: Add support for built-in functions
20981 vector bool char vec_revb (vector bool char);
20982 vector bool short vec_revb (vector short char);
20983 vector bool int vec_revb (vector bool int);
20984 vector bool long long vec_revb (vector bool long long);
20985 * doc/extend.texi: Update the built-in documentation file for the
20986 new built-in functions.
20988 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20990 * config/s390/s390.md: Remove movcc splitter.
20992 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20994 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
20995 load/store on condition.
20997 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21000 * config/avr/avr.c (avr_encode_section_info)
21001 [progmem && !TREE_READONLY]: Error if progmem object needs
21004 2017-07-11 Michael Collison <michael.collison@arm.com>
21006 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21009 2017-07-11 Carl Love <cel@us.ibm.com>
21011 * config/rs6000/rs6000-c.c: Add support for builtins
21012 vector unsigned int vec_parity_lsbb (vector signed int);
21013 vector unsigned int vec_parity_lsbb (vector unsigned int);
21014 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21015 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21016 vector unsigned long long vec_parity_lsbb (vector signed long long);
21017 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21018 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21019 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21020 * doc/extend.texi: Update the built-in documentation file for the
21021 new built-in functions.
21023 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21025 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21026 (layout::m_primary_loc): New field.
21027 (layout::layout): Initialize new field. Move location filtering
21028 logic from here to...
21029 (layout::maybe_add_location_range): ...this new method. Add
21030 support for filtering to just the lines already specified by other
21032 (layout::will_show_line_p): New method.
21033 (gcc_rich_location::add_location_if_nearby): New method.
21034 (selftest::test_add_location_if_nearby): New test function.
21035 (selftest::diagnostic_show_locus_c_tests): Call it.
21036 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21039 2017-07-11 Tom de Vries <tom@codesourcery.com>
21041 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21042 (bb_first_real_insn): New function.
21043 (nvptx_single): Add extra initialization of broadcasted condition
21046 2017-07-11 Nathan Sidwell <nathan@acm.org>
21048 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21050 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21052 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21053 Binutils doc as TEXI will mess them up.
21054 * doc/invoke.texi (AVR Options): Same here.
21056 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21058 * config/sparc/sparc.opt (mfix-ut700): New option.
21059 (mfix-gr712rc): Likewise.
21060 (sparc_fix_b2bst): New variable.
21061 * doc/invoke.texi (SPARC options): Document them.
21062 (ARM options): Fix warnings.
21063 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21064 instructions to prevent sequences that can trigger the store-store
21065 errata for certain LEON3FT processors.
21066 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21067 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21068 * config/sparc/sparc.md (fix_b2bst): New attribute.
21069 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21071 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21074 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21076 (*rsqrtsf2_sse): Ditto.
21078 (div<mode>3): Macroize insn from divdf3 and divsf3
21079 using MODEF mode iterator.
21081 2017-07-10 Martin Sebor <msebor@redhat.com>
21083 PR tree-optimization/80397
21084 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21085 instead of testing for equality to INTEGER_TYPE.
21087 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21089 * config.gcc: Remove uclibc from arc target spec.
21091 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21093 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21095 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21098 * lto-wrapper.c (remove_option): New function.
21099 (merge_and_complain): Merge PIC/PIE options more realistically.
21101 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21103 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21107 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21108 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21109 * config.in: Regenerate.
21110 * configure: Regenerate.
21111 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21112 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21113 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21114 (TARGET_GASISR_PROLOGUES): ...target mask.
21115 * common/config/avr/avr-common.c
21116 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21117 Set -mgas-isr-prologues.
21118 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21119 INSERT_PASS_BEFORE for it.
21120 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21121 * config/avr/avr.c (avr_option_override)
21122 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21123 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21124 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21125 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21126 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21128 (make_avr_pass_pre_proep): New function.
21129 (emit_push_sfr) <treg>: Add argument to function and use it
21130 instead of TMP_REG.
21131 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21132 and set machine->gasisr.yes.
21133 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21134 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21135 __gcc_isr.n_pushed to .L__stack_usage.
21136 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21137 (avr_asm_final_postscan_insn): ...this new static function.
21138 * config/avr/avr.h (machine_function)
21139 <is_no_gccisr, use_L__stack_usage>: New fields.
21140 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21141 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21142 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21143 (gasisr, *gasisr): New expander and insn.
21144 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21145 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21146 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21148 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21150 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21153 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21155 Move jump-tables out of .text again.
21158 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21159 (ASM_OUTPUT_ADDR_VEC): New function.
21160 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21161 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21162 INSN_ADDRESSes as asm comment.
21163 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21164 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21165 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21166 * config/avr/avr.md (*tablejump): Adjust comment.
21167 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21168 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21170 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21171 (avr_output_addr_vec): New proto.
21172 (avr_log_t) <insn_addresses>: New field.
21174 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21177 * config/i386/i386.c (ix86_function_arg_advance): Set
21178 outgoing_args_on_stack to true if there are outgoing arguments
21180 (ix86_function_arg): Likewise.
21181 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21182 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21183 * config/i386/i386.h (machine_function): Add
21184 outgoing_args_on_stack.
21186 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21188 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21189 supporting pthreds.
21190 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21192 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21194 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21195 (REAL_H): Remove $(MACHMODE_H).
21196 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21198 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21199 $(MACHMODE_H) and double-int.h.
21200 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21202 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21205 2017-07-07 Andrew Pinski <apinski@cavium.com>
21207 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21208 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21210 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21212 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21213 Add warning if GCC was not configured to link against a GLIBC that
21214 exports the hardware capability bits.
21215 (make_resolver_func): Make resolver function private and not a
21216 COMDAT function. Create the name with clone_function_name instead
21217 of make_unique_name.
21220 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21221 correct operand in doing the split.
21223 2017-07-07 Carl Love <cel@us.ibm.com>
21225 * config/rs6000/rs6000-c: Add support for built-in function
21226 vector unsigned short vec_pack_to_short_fp32 (vector float,
21228 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21229 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21230 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21231 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21232 (convert_4f32_8i16): Add define_expand.
21233 * doc/extend.texi: Update the built-in documentation file for the
21234 new built-in function.
21236 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21238 * config/sparc/m8.md: New file.
21239 * config/sparc/sparc.md: Include m8.md.
21241 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21243 * config/sparc/sparc.opt: New option -mvis4b.
21244 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21245 (sparc_option_override): Handle VIS4B.
21246 (enum sparc_builtins): Define
21247 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21248 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21249 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21250 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21251 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21252 (check_constant_argument): New function.
21253 (sparc_vis_init_builtins): Define builtins
21254 __builtin_vis_dictunpack{8,16,32},
21255 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21256 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21257 __builtin_vis_fpcmpde{8,16,32}shl and
21258 __builtin_vis_fpcmpur{8,16,32}shl.
21259 (sparc_expand_builtin): Check that the constant operands to
21260 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21261 constant and in range.
21262 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21264 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21265 (SPARC_IMM5_P): Likewise.
21266 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21267 (enabled): Handle vis4b.
21268 (UNSPEC_DICTUNPACK): New unspec.
21269 (UNSPEC_FPCMPSHL): Likewise.
21270 (UNSPEC_FPUCMPSHL): Likewise.
21271 (UNSPEC_FPCMPDESHL): Likewise.
21272 (UNSPEC_FPCMPURSHL): Likewise.
21273 (cpu_feature): New CPU feature `vis4b'.
21274 (dictunpack{8,16,32}): New insns.
21275 (FPCSMODE): New mode iterator.
21276 (fpcscond): New code iterator.
21277 (fpcsucond): Likewise.
21278 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21279 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21280 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21281 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21282 * config/sparc/constraints.md: Define constraints `q' for unsigned
21283 2-bit integer constants and `t' for unsigned 5-bit integer
21285 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21287 (imm5_operand_dictunpack16): Likewise.
21288 (imm5_operand_dictunpack32): Likewise.
21289 (imm2_operand): Likewise.
21290 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21291 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21292 ditunpack* and fpcmp*shl builtins.
21294 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21296 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21297 * config.in: Add HAVE_AS_SPARC6 define.
21298 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21300 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21302 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21303 (CPP_CPU_SPEC): Handle m8.
21304 (ASM_CPU_SPEC): Likewise.
21305 * config/sparc/sparc-opts.h (enum processor_type): Add
21307 * config/sparc/sparc.c (m8_costs): New struct.
21308 (sparc_option_override): Handle TARGET_CPU_m8.
21309 (sparc32_initialize_trampoline): Likewise.
21310 (sparc64_initialize_trampoline): Likewise.
21311 (sparc_issue_rate): Likewise.
21312 (sparc_register_move_cost): Likewise.
21313 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21314 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21315 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21316 (CPP_CPU_SPEC): Handle M8.
21317 (ASM_CPU_SPEC): Likewise.
21318 (AS_M8_FLAG): Define.
21319 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21320 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21321 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21323 * configure: Regenerate.
21324 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21327 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21329 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21331 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21332 ("*movdi_insn_sp32"): Do not set v3pipe.
21333 ("*movsi_insn"): Likewise.
21334 ("*movdi_insn_sp64"): Likewise.
21335 ("*movsf_insn"): Likewise.
21336 ("*movdf_insn_sp32"): Likewise.
21337 ("*movdf_insn_sp64"): Likewise.
21338 ("*zero_extendsidi2_insn_sp64"): Likewise.
21339 ("*sign_extendsidi2_insn"): Likewise.
21340 ("*mov<VM32:mode>_insn"): Likewise.
21341 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21342 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21343 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21344 ("<vlop:code><VL:mode>3"): Likewise.
21345 ("*not_<vlop:code><VL:mode>3"): Likewise.
21346 ("*nand<VL:mode>_vis"): Likewise.
21347 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21348 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21349 ("one_cmpl<VL:mode>2"): Likewise.
21350 ("faligndata<VM64:mode>_vis"): Likewise.
21351 ("alignaddrsi_vis"): Likewise.
21352 ("alignaddrdi_vis"): Likweise.
21353 ("alignaddrlsi_vis"): Likewise.
21354 ("alignaddrldi_vis"): Likewise.
21355 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21356 ("bmaskdi_vis"): Likewise.
21357 ("bmasksi_vis"): Likewise.
21358 ("bshuffle<VM64:mode>_vis"): Likewise.
21359 ("cmask8<P:mode>_vis"): Likewise.
21360 ("cmask16<P:mode>_vis"): Likewise.
21361 ("cmask32<P:mode>_vis"): Likewise.
21362 ("pdistn<P:mode>_vis"): Likewise.
21363 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21365 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21367 * config/sparc/sparc.md ("subtype"): New insn attribute.
21368 ("*wrgsr_sp64"): Set insn subtype.
21369 ("*rdgsr_sp64"): Likewise.
21370 ("alignaddrsi_vis"): Likewise.
21371 ("alignaddrdi_vis"): Likewise.
21372 ("alignaddrlsi_vis"): Likewise.
21373 ("alignaddrldi_vis"): Likewise.
21374 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21375 ("fexpand_vis"): Likewise.
21376 ("fpmerge_vis"): Likewise.
21377 ("faligndata<VM64:mode>_vis"): Likewise.
21378 ("bshuffle<VM64:mode>_vis"): Likewise.
21379 ("cmask8<P:mode>_vis"): Likewise.
21380 ("cmask16<P:mode>_vis"): Likewise.
21381 ("cmask32<P:mode>_vis"): Likewise.
21382 ("fchksm16_vis"): Likewise.
21383 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21384 ("fmean16_vis"): Likewise.
21385 ("fp<plusminus_insn>64_vis"): Likewise.
21386 ("<plusminus_insn>v8qi3"): Likewise.
21387 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21388 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21389 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21390 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21391 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21392 ("*movqi_insn"): Likewise.
21393 ("*movhi_insn"): Likewise.
21394 ("*movsi_insn"): Likewise.
21395 ("movsi_pic_gotdata_op"): Likewise.
21396 ("*movdi_insn_sp32"): Likewise.
21397 ("*movdi_insn_sp64"): Likewise.
21398 ("movdi_pic_gotdata_op"): Likewise.
21399 ("*movsf_insn"): Likewise.
21400 ("*movdf_insn_sp32"): Likewise.
21401 ("*movdf_insn_sp64"): Likewise.
21402 ("*zero_extendhisi2_insn"): Likewise.
21403 ("*zero_extendqihi2_insn"): Likewise.
21404 ("*zero_extendqisi2_insn"): Likewise.
21405 ("*zero_extendqidi2_insn"): Likewise.
21406 ("*zero_extendhidi2_insn"): Likewise.
21407 ("*zero_extendsidi2_insn_sp64"): Likewise.
21408 ("ldfsr"): Likewise.
21409 ("prefetch_64"): Likewise.
21410 ("prefetch_32"): Likewise.
21411 ("tie_ld32"): Likewise.
21412 ("tie_ld64"): Likewise.
21413 ("*tldo_ldub_sp32"): Likewise.
21414 ("*tldo_ldub1_sp32"): Likewise.
21415 ("*tldo_ldub2_sp32"): Likewise.
21416 ("*tldo_ldub_sp64"): Likewise.
21417 ("*tldo_ldub1_sp64"): Likewise.
21418 ("*tldo_ldub2_sp64"): Likewise.
21419 ("*tldo_ldub3_sp64"): Likewise.
21420 ("*tldo_lduh_sp32"): Likewise.
21421 ("*tldo_lduh1_sp32"): Likewise.
21422 ("*tldo_lduh_sp64"): Likewise.
21423 ("*tldo_lduh1_sp64"): Likewise.
21424 ("*tldo_lduh2_sp64"): Likewise.
21425 ("*tldo_lduw_sp32"): Likewise.
21426 ("*tldo_lduw_sp64"): Likewise.
21427 ("*tldo_lduw1_sp64"): Likewise.
21428 ("*tldo_ldx_sp64"): Likewise.
21429 ("*mov<VM32:mode>_insn"): Likewise.
21430 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21431 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21433 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21435 * config/sparc/sparc.md ("type"): New insn type viscmp.
21436 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21438 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21439 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21440 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21441 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21443 ("n7_vis_logical_11cycle"): Likewise.
21444 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21445 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21446 * config/sparc/niagara.md ("niag_vis"): Likewise.
21447 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21448 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21450 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21452 * config/sparc/sparc.md: New instruction type `bmask'.
21453 (bmaskdi_vis): Use the `bmask' type.
21454 (bmasksi_vis): Likewise.
21455 * config/sparc/ultra3.md (us3_array): Likewise.
21456 * config/sparc/niagara7.md (n7_array): Likewise.
21457 * config/sparc/niagara4.md (n4_array): Likewise.
21458 * config/sparc/niagara2.md (niag2_vis): Likewise.
21459 (niag3_vis): Likewise.
21460 * config/sparc/niagara.md (niag_vis): Likewise.
21462 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21464 * ipa-comdats.c: Remove optimize check from gate.
21465 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21466 for functions not optimized.
21467 (ipa_fn_summary_read): Skip optimize check.
21468 (ipa_fn_summary_write): Likewise.
21469 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21471 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21473 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21474 (check_callers): Give up on caller which is not optimized.
21475 (inline_small_functions): Likewise.
21476 (ipa_inline): Do not give up when not optimizing.
21477 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21478 away unoptimizes cdtors.
21479 (whole_program_function_and_variable_visibility): Do
21480 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21481 * ipa.c (process_references): Do not check optimize.
21482 (symbol_table::remove_unreachable_nodes): Update optimize check.
21483 (set_writeonly_bit): Update optimize check.
21484 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21485 (pass_ipa_single_use::gate): Remove.
21487 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21489 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21490 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21491 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21492 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21493 permute_load, permute_store, adjust_extract, adjust_splat,
21494 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21495 replace_swap_with_copy, dump_swap_insn_table,
21496 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21497 recombine_lvx_pattern, recombine_stvx_pattern,
21498 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21499 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21500 to file rs6000-p8swap.c.
21501 * config/rs6000/rs6000-p8swap.c: New file.
21502 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21503 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21504 and rs6000*-*-* targets.
21506 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21508 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21510 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21512 * lto-wrapper.c (merge_and_complain): Do not merge
21513 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21514 fsigned_zeros, ftrapping_math, fwrapv.
21515 (append_compiler_options): Do not track these options.
21516 (append_linker_options): Likewie
21518 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21520 * cgraphunit.c (cgraph_node::finalize_function): When
21521 !flag_toplevel_reorde set no_reorder flag.
21522 (varpool_node::finalize_decl): Likewise.
21523 (symbol_table::compile): Drop no toplevel reorder path.
21525 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21527 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21528 edges; zero probability is not better than uninitialized.
21530 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21532 * asan.h (asan_sanitize_allocas_p): Declare.
21533 * asan.c (asan_sanitize_allocas_p): New function.
21534 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21535 (handle_builtin_alloca): Likewise.
21536 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21537 if !asan_sanitize_allocas_p.
21538 * params.def (asan-instrument-allocas): Add new option.
21539 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21540 * opts.c (common_handle_option): Disable allocas sanitization for
21543 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21545 * asan.c: Include gimple-fold.h.
21546 (get_last_alloca_addr): New function.
21547 (handle_builtin_stackrestore): Likewise.
21548 (handle_builtin_alloca): Likewise.
21549 (asan_emit_allocas_unpoison): Likewise.
21550 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21551 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21552 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21553 (instrument_builtin_call): Pass gimple iterator to
21554 get_mem_refs_of_builtin_call.
21555 (last_alloca_addr): New global.
21556 * asan.h (asan_emit_allocas_unpoison): Declare.
21557 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21558 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21559 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21560 if function calls alloca.
21561 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21562 * gimple-fold.h (replace_call_with_value): Declare.
21563 * internal-fn.c: Include asan.h.
21564 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21565 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21567 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21569 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21570 (C_SELFTEST_FLAGS): New.
21571 (CPP_SELFTEST_FLAGS): New.
21572 (SELFTEST_DEPS): New, from deps of s-selftest.
21573 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21574 (CPP_SELFTEST_DEPS): New.
21575 (selftest): Add dependency on s-selftest-c++.
21576 (s-selftest): Rename to...
21577 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21578 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21579 than SELFTEST_FLAGS.
21580 (selftest-gdb): Rename to...
21581 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21583 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21584 (selftest-valgrind): Rename to...
21585 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21587 (selftest-valgrind): Reintroduce as an alias for
21588 selftest-c-valgrind.
21589 (s-selftest-c++): New.
21590 (selftest-c++-gdb): New.
21591 (selftest-c++-valgrind): New.
21593 2017-07-06 Olivier Hainque <hainque@adacore.com>
21595 * gcc.c (process_command): When deciding if undefined variables
21596 should be ignored when processing specs, accept "gcc -v" as well.
21598 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21600 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21601 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21603 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21605 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21606 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21607 * config/arm/arm-tables.opt: Regenerate.
21608 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21610 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21612 2017-07-06 Carl Love <cel@us.ibm.com>
21614 * ChangeLog: Clean up from mid air collision
21616 2017-07-06 Carl Love <cel@us.ibm.com>
21618 * config/rs6000/rs6000-c.c: Add support for built-in functions
21619 vector signed int vec_subc (vector signed int, vector signed int);
21620 vector signed __int128 vec_subc (vector signed __int128,
21621 vector signed __int128);
21622 vector unsigned __int128 vec_subc (vector unsigned __int128,
21623 vector unsigned __int128);
21624 vector signed int vec_sube (vector signed int, vector signed int,
21625 vector signed int);
21626 vector unsigned int vec_sube (vector unsigned int,
21627 vector unsigned int,
21628 vector unsigned int);
21629 vector signed __int128 vec_sube (vector signed __int128,
21630 vector signed __int128,
21631 vector signed__int128);
21632 vector unsigned __int128 vec_sube (vector unsigned __int128,
21633 vector unsigned __int128,
21634 vector unsigned __int128);
21635 vector signed int vec_subec (vector signed int, vector signed int,
21636 vector signed int);
21637 vector unsigned int vec_subec (vector unsigned int,
21638 vector unsigned int,
21639 vector unsigned int);
21640 vector signed __int128 vec_subec (vector signed __int128,
21641 vector signed __int128,
21642 vector signed__int128);
21643 vector unsigned __int128 vec_subec (vector unsigned __int128,
21644 vector unsigned __int128,
21645 vector unsigned __int128);
21646 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21647 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21648 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21649 BU_ALTIVEC_OVERLOAD_X definitions.
21650 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21651 * doc/extend.texi: Update the built-in documentation file for the new
21652 built-in functions.
21654 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21657 * diagnostic-show-locus.c (layout::layout): Use start and finish
21658 spelling location for the start and finish of each range.
21659 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21660 Add unused aspect param.
21661 * input.c (expand_location_1): Add "aspect" param, and use it
21662 to access the correct part of the location.
21663 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21665 (expand_location_to_spelling_point): Likewise.
21666 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21667 param, and pass it to expand_location_1.
21669 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21671 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21672 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21673 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21674 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21675 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21676 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21677 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21678 _mm_maskz_getmant_ss): New intrinsics.
21679 (__builtin_ia32_getexpss128_mask): Changed to ...
21680 __builtin_ia32_getexpss128_round ... this.
21681 (__builtin_ia32_getexpsd128_mask): Changed to ...
21682 __builtin_ia32_getexpsd128_round ... this.
21683 * config/i386/i386-builtin-types.def
21684 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21685 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21686 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21687 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21688 __builtin_ia32_getmantss_mask_round): New builtins.
21689 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21690 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21691 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21692 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21693 * config/i386/sse.md
21694 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21695 avx512f_sgetexp<mode><mask_scalar_name>
21696 <round_saeonly_scalar_name> ... this.
21697 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21698 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21699 vgetexp<ssescalarmodesuffix>
21700 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21701 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21702 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21703 avx512f_vgetmant<mode><mask_scalar_name>
21704 <round_saeonly_scalar_name> ... this.
21705 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21706 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21707 vgetmant<ssescalarmodesuffix>
21708 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21709 %0<mask_scalar_operand4>, %1, %2
21710 <round_saeonly_scalar_mask_op4>, %3} ... this.
21711 * config/i386/subst.md (mask_scalar_operand4,
21712 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21713 round_saeonly_scalar_nimm_predicate): New subst attributes.
21715 2017-07-06 Julia Koval <julia.koval@intel.com>
21717 * config/i386/i386.c (ix86_erase_embedded_rounding):
21718 Remove code for old rounding pattern.
21720 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21722 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21724 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21726 * doc/sourcebuild.texi (Test Directives, Variants of
21727 dg-require-support): Add documentation for dg-require-stack-check.
21729 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21731 * config/i386/subst.md (mask_scalar, round_scalar,
21732 round_saeonly_scalar): New meta-templates.
21733 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21734 round_scalar_mask_operand3, round_scalar_mask_op3,
21735 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21736 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21737 round_saeonly_scalar_constraint,
21738 round_saeonly_scalar_prefix): New subst attribute.
21739 * config/i386/sse.md
21740 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21741 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21742 <round_scalar_name> ... this.
21743 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21744 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21745 <round_scalar_name> ... this.
21746 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21747 <sse>_vm<code><mode>3<mask_scalar_name>
21748 <round_saeonly_scalar_name> ... this.
21749 (v<plusminus_mnemonic><ssescalarmodesuffix>
21750 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21751 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21752 v<plusminus_mnemonic><ssescalarmodesuffix>
21753 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21754 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21755 (v<multdiv_mnemonic><ssescalarmodesuffix>
21756 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21757 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21758 v<multdiv_mnemonic><ssescalarmodesuffix>
21759 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21760 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21761 (v<maxmin_float><ssescalarmodesuffix>
21762 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21763 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21764 v<maxmin_float><ssescalarmodesuffix>
21765 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21766 %0<mask_scalar_operand3>, %1, %<iptr>2
21767 <round_saeonly_scalar_mask_op3>} ... this.
21769 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21771 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21772 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21774 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21775 Alan Hayward <alan.hayward@arm.com>
21776 David Sherwood <david.sherwood@arm.com>
21778 * combine.c (simplify_if_then_else): Remove "enum" before
21780 * compare-elim.c (can_eliminate_compare): Likewise.
21781 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21783 (aarch64_lookup_simd_builtin_type): Likewise.
21784 (aarch64_simd_builtin_type): Likewise.
21785 (aarch64_init_simd_builtin_types): Likewise.
21786 (aarch64_simd_expand_args): Likewise.
21787 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21789 (aarch64_reverse_mask): Likewise.
21790 (aarch64_simd_emit_reg_reg_move): Likewise.
21791 (aarch64_gen_adjusted_ldpstp): Likewise.
21792 (aarch64_ccmp_mode_to_code): Likewise.
21793 (aarch64_operands_ok_for_ldpstp): Likewise.
21794 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21795 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21797 (aarch64_min_divisions_for_recip_mul): Likewise.
21798 (aarch64_reassociation_width): Likewise.
21799 (aarch64_get_condition_code_1): Likewise.
21800 (aarch64_simd_emit_reg_reg_move): Likewise.
21801 (aarch64_simd_attr_length_rglist): Likewise.
21802 (aarch64_reverse_mask): Likewise.
21803 (aarch64_operands_ok_for_ldpstp): Likewise.
21804 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21805 (aarch64_gen_adjusted_ldpstp): Likewise.
21806 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21808 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21809 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21810 (arm_lookup_simd_builtin_type): Likewise.
21811 (arm_simd_builtin_type): Likewise.
21812 (arm_init_simd_builtin_types): Likewise.
21813 (arm_expand_builtin_args): Likewise.
21814 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21815 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21816 (ft32_setup_incoming_varargs): Likewise.
21817 (ft32_function_arg): Likewise.
21818 (ft32_function_arg_advance): Likewise.
21819 (ft32_pass_by_reference): Likewise.
21820 (ft32_arg_partial_bytes): Likewise.
21821 (ft32_valid_pointer_mode): Likewise.
21822 (ft32_addr_space_pointer_mode): Likewise.
21823 (ft32_addr_space_legitimate_address_p): Likewise.
21824 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21826 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21827 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21828 (iamcu_alignment): Likewise.
21829 (canonicalize_vector_int_perm): Likewise.
21830 (ix86_noce_conversion_profitable_p): Likewise.
21831 (ix86_mpx_bound_mode): Likewise.
21832 (ix86_operands_ok_for_move_multiple): Likewise.
21833 * config/microblaze/microblaze-protos.h
21834 (microblaze_expand_conditional_branch_reg): Likewise.
21835 * config/microblaze/microblaze.c
21836 (microblaze_expand_conditional_branch_reg): Likewise.
21837 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21839 (rs6000_reassociation_width): Likewise.
21840 (rs6000_invalid_binary_op): Likewise.
21841 (fusion_p9_p): Likewise.
21842 (emit_fusion_p9_load): Likewise.
21843 (emit_fusion_p9_store): Likewise.
21844 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21846 (riscv_hard_regno_mode_ok_p): Likewise.
21847 (riscv_address_insns): Likewise.
21848 (riscv_split_symbol): Likewise.
21849 (riscv_legitimize_move): Likewise.
21850 (riscv_function_value): Likewise.
21851 (riscv_hard_regno_nregs): Likewise.
21852 (riscv_expand_builtin): Likewise.
21853 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21854 (riscv_build_integer): Likewise.
21855 (riscv_split_integer): Likewise.
21856 (riscv_legitimate_constant_p): Likewise.
21857 (riscv_cannot_force_const_mem): Likewise.
21858 (riscv_regno_mode_ok_for_base_p): Likewise.
21859 (riscv_valid_base_register_p): Likewise.
21860 (riscv_valid_offset_p): Likewise.
21861 (riscv_valid_lo_sum_p): Likewise.
21862 (riscv_classify_address): Likewise.
21863 (riscv_legitimate_address_p): Likewise.
21864 (riscv_address_insns): Likewise.
21865 (riscv_load_store_insns): Likewise.
21866 (riscv_force_binary): Likewise.
21867 (riscv_split_symbol): Likewise.
21868 (riscv_force_address): Likewise.
21869 (riscv_legitimize_address): Likewise.
21870 (riscv_move_integer): Likewise.
21871 (riscv_legitimize_const_move): Likewise.
21872 (riscv_legitimize_move): Likewise.
21873 (riscv_address_cost): Likewise.
21874 (riscv_subword): Likewise.
21875 (riscv_output_move): Likewise.
21876 (riscv_canonicalize_int_order_test): Likewise.
21877 (riscv_emit_int_order_test): Likewise.
21878 (riscv_function_arg_boundary): Likewise.
21879 (riscv_pass_mode_in_fpr_p): Likewise.
21880 (riscv_pass_fpr_single): Likewise.
21881 (riscv_pass_fpr_pair): Likewise.
21882 (riscv_get_arg_info): Likewise.
21883 (riscv_function_arg): Likewise.
21884 (riscv_function_arg_advance): Likewise.
21885 (riscv_arg_partial_bytes): Likewise.
21886 (riscv_function_value): Likewise.
21887 (riscv_pass_by_reference): Likewise.
21888 (riscv_setup_incoming_varargs): Likewise.
21889 (riscv_print_operand): Likewise.
21890 (riscv_elf_select_rtx_section): Likewise.
21891 (riscv_save_restore_reg): Likewise.
21892 (riscv_for_each_saved_reg): Likewise.
21893 (riscv_register_move_cost): Likewise.
21894 (riscv_hard_regno_mode_ok_p): Likewise.
21895 (riscv_hard_regno_nregs): Likewise.
21896 (riscv_class_max_nregs): Likewise.
21897 (riscv_memory_move_cost): Likewise.
21898 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21899 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21900 (rl78_addr_space_address_mode): Likewise.
21901 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21903 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21904 (rs6000_reassociation_width): Likewise.
21905 (rs6000_invalid_binary_op): Likewise.
21906 (fusion_p9_p): Likewise.
21907 (emit_fusion_p9_load): Likewise.
21908 (emit_fusion_p9_store): Likewise.
21909 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
21910 (ok_for_simple_move_operands): Likewise.
21911 (ok_for_simple_move_strict_operands): Likewise.
21912 (ok_for_simple_arith_logic_operands): Likewise.
21913 (visium_legitimize_reload_address): Likewise.
21914 (visium_select_cc_mode): Likewise.
21915 (output_cbranch): Likewise.
21916 (visium_split_double_move): Likewise.
21917 (visium_expand_copysign): Likewise.
21918 (visium_expand_int_cstore): Likewise.
21919 (visium_expand_fp_cstore): Likewise.
21920 * config/visium/visium.c (visium_pass_by_reference): Likewise.
21921 (visium_function_arg): Likewise.
21922 (visium_function_arg_advance): Likewise.
21923 (visium_libcall_value): Likewise.
21924 (visium_setup_incoming_varargs): Likewise.
21925 (visium_legitimate_constant_p): Likewise.
21926 (visium_legitimate_address_p): Likewise.
21927 (visium_legitimize_address): Likewise.
21928 (visium_secondary_reload): Likewise.
21929 (visium_register_move_cost): Likewise.
21930 (visium_memory_move_cost): Likewise.
21931 (prepare_move_operands): Likewise.
21932 (ok_for_simple_move_operands): Likewise.
21933 (ok_for_simple_move_strict_operands): Likewise.
21934 (ok_for_simple_arith_logic_operands): Likewise.
21935 (visium_function_value_1): Likewise.
21936 (rtx_ok_for_offset_p): Likewise.
21937 (visium_legitimize_reload_address): Likewise.
21938 (visium_split_double_move): Likewise.
21939 (visium_expand_copysign): Likewise.
21940 (visium_expand_int_cstore): Likewise.
21941 (visium_expand_fp_cstore): Likewise.
21942 (visium_split_cstore): Likewise.
21943 (visium_select_cc_mode): Likewise.
21944 (visium_split_cbranch): Likewise.
21945 (output_cbranch): Likewise.
21946 (visium_print_operand_address): Likewise.
21947 * expmed.c (flip_storage_order): Likewise.
21948 * expmed.h (emit_cstore): Likewise.
21949 (flip_storage_order): Likewise.
21950 * genrecog.c (validate_pattern): Likewise.
21951 * hsa-gen.c (gen_hsa_addr): Likewise.
21952 * internal-fn.c (expand_arith_overflow): Likewise.
21953 * ira-color.c (allocno_copy_cost_saving): Likewise.
21954 * lra-assigns.c (find_hard_regno_for_1): Likewise.
21955 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
21956 (process_invariant_for_inheritance): Likewise.
21957 * lra-eliminations.c (move_plus_up): Likewise.
21958 * omp-low.c (lower_oacc_reductions): Likewise.
21959 * simplify-rtx.c (simplify_subreg): Likewise.
21960 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
21961 (TARGET_CHKP_BOUND_MODE): Likewise..
21962 * targhooks.c (default_chkp_bound_mode): Likewise.
21963 (default_setup_incoming_vararg_bounds): Likewise.
21964 * targhooks.h (default_chkp_bound_mode): Likewise.
21965 (default_setup_incoming_vararg_bounds): Likewise.
21966 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
21967 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
21968 (have_whole_vector_shift): Likewise.
21969 * tree-vect-stmts.c (vectorizable_load): Likewise.
21970 * doc/tm.texi: Regenerate.
21972 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21974 Graceful degrade if Binutils PR21472 is not available.
21977 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
21978 .rodata in flash test fails.
21979 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
21980 * confgure: Regenerate.
21981 * config.in: Regenerate.
21982 * config/avr/avr.c (avr_asm_named_section)
21983 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
21984 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
21985 (avr_asm_init_sections): Same.
21987 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21989 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
21990 (fma<VH:mode>4_intrinsic): Likewise.
21991 (*fmsub<VCVTF:mode>4): Likewise.
21992 (*fmsub<VH:mode>4_intrinsic): Likewise.
21994 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21997 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
21998 Don't depend on "optimize > 0".
21999 (out_movhi_r_mr, out_movqi_mr_r): Same.
22000 (out_movhi_mr_r, out_movqi_r_mr): Same.
22001 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22002 io_address_operand on "optimize > 0".
22004 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22006 * tree-loop-distribution.c: Add general explanantion on the pass.
22007 (generate_loops_for_partition): Mark distributed loop.
22008 (pg_add_dependence_edges): New parameter. Handle alias data
22009 dependence specially and record it in the parameter if asked.
22010 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22011 (init_partition_graph_vertices, add_partition_graph_edge): New.
22012 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22013 (free_partition_graph_vdata, build_partition_graph): New.
22014 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22015 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22016 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22017 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22018 (version_for_distribution_p, finalize_partitions): New.
22019 (distribute_loop): Handle alias data dependence specially. Factor
22020 out loop fusion code as functions and call these functions.
22022 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22024 * tree-loop-distribution.c (classify_partition): New parameter and
22025 better handle reduction statement.
22026 (rdg_build_partitions): Revise comment.
22027 (distribute_loop): Compute statements in all partitions and pass it
22028 to classify_partition.
22030 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22032 * tree-loop-distribution.c (enum partition_type): New.
22033 (struct partition): New field type.
22034 (partition_merge_into): Add parameter. Update partition type.
22035 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22036 (build_rdg_partition_for_vertex): Compute partition type.
22037 (rdg_build_partitions): Dump partition type.
22038 (distribute_loop): Update calls to partition_merge_into.
22040 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22042 * tree-loop-distribution.c (struct ddr_hasher): New.
22043 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22045 (classify_partition): Call get_data_dependence.
22046 (pg_add_dependence_edges): Ditto.
22047 (distribute_loop): Release data dependence hash table.
22049 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22051 * tree-loop-distribution.c (ref_base_address): Delete.
22052 (similar_memory_accesses): Rename ...
22053 (share_memory_accesses): ... to this. Check if partitions access
22054 the same memory reference.
22055 (distribute_loop): Call share_memory_accesses.
22057 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22059 * tree-loop-distribution.c (struct partition): New field recording
22060 its data reference.
22061 (partition_alloc, partition_free): Init and release data refs.
22062 (partition_merge_into): Merge data refs.
22063 (build_rdg_partition_for_vertex): Collect data refs for partition.
22064 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22066 (distribute_loop): Remve data refs from vertice data of partition
22069 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22071 * tree-loop-distribution.c (params.h): Include header file.
22072 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22073 (datarefs_vec): New global var.
22074 (create_rdg_vertices): Use datarefs_vec directly.
22075 (free_rdg): Don't free data references.
22076 (build_rdg): Update use. Don't free data references.
22077 (distribute_loop): Compute global variable for data references.
22078 Bail out if there are too many data references.
22080 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22082 * tree-loop-distribution.c (loop_nest): New global var.
22083 (build_rdg): Use loop directly, rather than loop nest.
22084 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22086 (distribute_loop): Compute global variable loop nest. Update use.
22088 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22090 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22091 (partition_merge_into): New parameter. Dump reason for fusion.
22092 (distribute_loop): Update use of partition_merge_into.
22094 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22096 * tree-loop-distribution.c (bb_top_order_index): New.
22097 (bb_top_order_index_size, bb_top_order_cmp): New.
22098 (stmts_from_loop): Use topological order.
22099 (pass_loop_distribution::execute): Compute and release topological
22100 order for basic blocks.
22102 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22104 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22107 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22109 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22110 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22111 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22112 * internal-fn.def (LOOP_DIST_ALIAS): New.
22113 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22114 (fold_loop_internal_call): ... this.
22115 (vect_loop_dist_alias_call): New function.
22116 (set_uid_loop_bbs): Call fold_loop_internal_call.
22117 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22120 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22123 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22124 Require dead FLAGS_REG at the beginning of a peephole.
22126 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22129 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22130 arguments in the call to __builtin_ia32_sbb_u32.
22131 (_subborrow_u64): Swap _X and _Y arguments in the call to
22132 __builtin_ia32_sbb_u64.
22134 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22137 * tree-vrp.c (compare_assert_loc): Turn into a function template
22138 with stable template parameter. Only test if a->e is NULL,
22139 !a->e == !b->e has been verified already. Use e == NULL or
22140 e != NULL instead of e or ! e tests. If stable is true, don't use
22141 iterative_hash_expr, on the other side allow a or b or both NULL
22142 and sort the NULLs last.
22143 (process_assert_insertions): Sort using compare_assert_loc<false>
22144 instead of compare_assert_loc, later sort using
22145 compare_assert_loc<true> before calling process_assert_insertions_for
22146 in a loop. Use break instead of continue once seen NULL pointer.
22148 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22150 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22151 Cortex-R7 and Cortex-R8 processors.
22153 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22155 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22156 uninitialized while src is not.
22158 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22160 * common/config/arm/arm-common.c: Adjust include path for
22162 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22163 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22164 (arm-cpu-data.h): Likewise.
22165 (arm-cpu-cdata.h): Likewise.
22166 * config/arm/arm-cpu.h: Delete.
22167 * config/arm/arm-cpu-cdata.h: Delete.
22168 * config/arm/arm-cpu-data.h: Delete.
22170 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22172 * config/arm/arm-cpus.in (cortex-a55): New.
22173 (cortex-a75): Likewise.
22174 (cortex-a75.cortex-a55): Likewise.
22175 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22177 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22178 * config/arm/arm-cpu-cdata.h: Regenerate.
22179 * config/arm/arm-cpu-data.h: Regenerate.
22180 * config/arm/arm-cpu.h: Regenerate.
22181 * config/arm/arm-tables.opt: Regenerate.
22182 * config/arm/arm-tune.md: Regenerate.
22184 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22186 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22188 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22190 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22193 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22195 PR tree-optimization/81292
22196 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22197 full_string_p, also call adjust_related_strinfos if the adjustment
22198 is simple, otherwise invalidate related strinfos.
22200 2017-07-04 Martin Liska <mliska@suse.cz>
22203 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22204 newly created variable as DECL_IGNORED_P.
22206 2017-07-04 Martin Liska <mliska@suse.cz>
22209 * ipa-inline.c (inline_small_functions):
22210 Use xstrdup_for_dump.
22212 2017-07-04 Tom de Vries <tom@codesourcery.com>
22214 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22216 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22219 * config/darwin.c (darwin_function_switched_text_sections):
22222 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22224 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22226 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22228 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22230 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22232 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22233 min_profitable_iters, and th as inclusive lower bounds.
22234 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22235 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22236 for min_profitable_iters and min_profitable_estimate.
22237 (vect_transform_loop): Treat th as an inclusive lower bound.
22238 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22240 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22243 * config/darwin.c (darwin_function_switched_text_sections):
22244 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22245 in two pieces, and suppress the use of buf.
22247 2017-07-03 Nathan Sidwell <nathan@acm.org>
22249 * hash-table.h (hash_table_mod1): Fix indentation.
22251 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22253 PR middle-end/81290
22254 * predict.c (force_edge_cold): Be more careful about propagation
22256 * profile-count.h (profile_probability::guessed,
22257 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22259 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22261 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22263 * doc/invoke.texi (rcpc architecture extension): Document it.
22265 2017-07-03 Richard Biener <rguenther@suse.de>
22267 PR tree-optimization/60510
22268 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22269 the scalar reduction PHI and use it.
22270 (vectorizable_reduction): Properly guard the single_defuse_cycle
22271 path for non-SLP reduction chains where we cannot use it.
22272 Rework reduc_def/index and vector type deduction. Rework
22273 vector operand gathering during reduction op code-gen.
22274 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22275 chains dissolve the chain and leave it to non-SLP reduction
22278 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22280 * tree-data-ref.h (dr_alignment): Declare.
22281 * tree-data-ref.c (dr_alignment): New function.
22282 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22283 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22285 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22287 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22289 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22290 and base_misalignment fields.
22291 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22292 * tree-data-ref.c: Include builtins.h.
22293 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22294 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22295 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22296 * tree-vect-data-refs.c: Include tree-cfg.h.
22297 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22298 fields instead of calculating an alignment here.
22299 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22300 innermost_loop_behavior fields.
22302 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22304 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22306 (DR_STEP_ALIGNMENT): New macro.
22307 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22308 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22309 (create_data_ref): Print it.
22310 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22311 to tell whether the step preserves vector (mis)alignment.
22312 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22313 Move the check for an integer step and generalise to all INTEGER_CST.
22314 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22315 Print the outer step alignment.
22317 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22319 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22320 with offset_alignment.
22321 (DR_ALIGNED_TO): Delete.
22322 (DR_OFFSET_ALIGNMENT): New macro.
22323 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22324 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22325 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22326 (create_data_ref): Likewise.
22327 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22328 (vect_analyze_data_refs): Likewise.
22329 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22330 creating dummy innermost behavior.
22332 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22334 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22335 with a "innermost_loop_behavior *" and refeence tree.
22336 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22337 (create_data_ref): Update call accordingly.
22338 * tree-predcom.c (find_looparound_phi): Likewise.
22340 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22342 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22343 fields with dr_wrt_vec_loop.
22344 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22345 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22346 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22347 (vect_dr_behavior): New function.
22348 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22349 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22350 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22351 track whether the step preserves the misalignment.
22352 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22353 Use vect_dr_behavior.
22354 (vect_setup_realignment): Update call accordingly.
22355 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22356 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22357 call to vect_create_addr_base_for_vector_ref.
22358 (vect_create_cond_for_align_checks): Likewise.
22359 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22360 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22361 (vect_recog_mask_conversion_pattern): Likewise.
22362 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22363 (new_stmt_vec_info): Remove redundant zeroing.
22365 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22367 * common/config/arm/arm-common.c (arm_be8_option): New function.
22368 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22369 (ISA_ARMv6): Add isa_bit_be8.
22370 * config/arm/arm.h (arm_be8_option): Add prototype.
22371 (BE8_SPEC_FUNCTION): New define.
22372 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22373 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22374 (mlittle-endian): Similarly.
22375 (mbe8, mbe32): New options.
22376 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22377 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22379 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22381 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22383 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22385 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22386 (cleanup_tree_cfg_bb): Use it.
22387 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22389 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22391 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22394 * loop-doloop.c (add_test): Update profile.
22396 2017-07-03 Martin Liska <mliska@suse.cz>
22399 * sanopt.c (rewrite_usage_of_param): New function.
22400 (sanitize_rewrite_addressable_params): Likewise.
22401 (pass_sanopt::execute): Call rewrite_usage_of_param.
22403 2017-07-03 Richard Biener <rguenther@suse.de>
22405 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22406 back to using VIEW_CONVERT_EXPR.
22408 2017-07-03 Martin Liska <mliska@suse.cz>
22411 * doc/extend.texi: Document when a resolver function is
22412 generated for target_clones.
22414 2017-07-03 Martin Liska <mliska@suse.cz>
22416 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22417 and shadow memory of auto variables which are subject of
22418 use-after-scope sanitization.
22419 (asan_expand_mark_ifn): Add do set only when is_poison.
22421 2016-07-03 Richard Biener <rguenther@suse.de>
22423 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22425 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22426 (vectorizable_reduction): Perform reduction PHI creation when
22427 visiting a reduction PHI and adjust and simplify code generation
22428 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22429 (vect_transform_loop): Visit reduction PHIs.
22430 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22431 defs into the SLP tree.
22432 (vect_build_slp_tree): Reduction defs terminate the recursion.
22433 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22435 (vect_get_vec_defs_for_stmt_copy): Export.
22436 (vect_get_vec_defs): Likewise.
22437 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22439 (vect_get_vec_defs_for_stmt_copy): Declare.
22440 (vect_get_vec_defs): Likewise.
22442 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22444 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22445 parameter with a "loop" parameter and use it instead of the
22446 loop containing DR_STMT. Don't check simple_iv when doing
22447 BB analysis. Describe the two analysis modes in the comment.
22449 2017-07-03 Tom de Vries <tom@codesourcery.com>
22451 PR tree-optimization/69468
22452 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22453 (find_same_succ_bb): Handle ignore_edge_flags.
22455 2017-07-03 Tom de Vries <tom@codesourcery.com>
22457 PR tree-optimization/81192
22458 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22460 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22462 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22464 2017-07-03 Tom de Vries <tom@codesourcery.com>
22466 PR tree-optimization/81192
22467 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22468 BB_SAME_SUCC (bb) == NULL.
22470 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22472 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22475 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22477 * dumpfile.c: Include profile-count.h
22478 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22480 (insert_cond_bb): Update profile.
22481 * tree-cfg.h (insert_cond_bb): Update prototype.
22482 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22483 * tree-dump.c: Do not include tree-cfg.
22485 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22487 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22489 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22491 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22494 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22496 * tree-complex.c (expand_complex_div_wide): update profile.
22498 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22499 Alan Hayward <alan.hayward@arm.com>
22500 David Sherwood <david.sherwood@arm.com>
22502 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22503 (CORETYPES_H): New define.
22504 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22505 (insn-modes-inline.h, s-modes-inline-h): New rules.
22506 (generated_files): Add insn-modes-inline.h.
22507 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22508 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22509 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22510 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22511 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22512 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22513 (build/gencodes.o, build/genconditions.o): Likewise.
22514 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22515 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22516 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22517 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22518 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22519 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22520 * coretypes.h: Include everything up to real.h for generators.
22521 Include insn-modes.h first. Include wide-int-print.h after
22522 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22523 * machmode.h: Don't include insn-modes.h here.
22524 * function-tests.c: Remove includes of signop.h, machmode.h,
22525 double-int.h and wide-int.h.
22527 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22529 * optc-save-gen.awk: Likewise.
22530 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22531 * godump.c: Remove include of wide-int-print.h.
22532 * pretty-print.h: Likewise.
22533 * wide-int-print.cc: Likewise.
22534 * wide-int.cc: Likewise.
22535 * hash-map-tests.c: Remove include of signop.h.
22536 * hash-set-tests.c: Likewise.
22537 * rtl-tests.c: Likewise.
22538 * mkconfig.sh: Remove include of machmode.h.
22539 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22541 (emit_insn_modes_inline_h): ...this new function. Emit the code
22542 into an insn-modes-inline.h header file, adding appropriate
22543 include guards and end comments.
22544 (emit_insn_modes_c_header): Remove include of machmode.h.
22545 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22547 (main): Handle -i flag and call emit_insn_modes_inline_h when
22550 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22552 * tree-ssa-strlen.c (strinfo): Rename the length field to
22553 nonzero_chars. Add a full_string_p field.
22554 (compare_nonzero_chars, zero_length_string_p): New functions.
22555 (get_addr_stridx): Add an offset_out parameter.
22556 Use compare_nonzero_chars.
22557 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22558 (new_strinfo): Update after above changes to strinfo.
22559 (set_endptr_and_length): Set full_string_p.
22560 (get_string_length): Update after above changes to strinfo.
22561 (unshare_strinfo): Update call to new_strinfo.
22562 (maybe_invalidate): Likewise.
22563 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22564 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22565 as a uhwi instead of an shwi. Update after above changes to
22566 strinfo and new_strinfo.
22567 (zero_length_string): Assert that chainsi contains full strings.
22568 Use zero_length_string_p. Update call to new_strinfo.
22569 (adjust_related_strinfos): Update after above changes to strinfo.
22570 Copy full_string_p from origsi.
22571 (adjust_last_stmt): Use zero_length_string_p.
22572 (handle_builtin_strlen): Update after above changes to strinfo and
22573 new_strinfo. Install the lhs as the string length if the previous
22574 entry didn't describe a full string.
22575 (handle_builtin_strchr): Update after above changes to strinfo
22577 (handle_builtin_strcpy): Likewise.
22578 (handle_builtin_strcat): Likewise.
22579 (handle_builtin_malloc): Likewise.
22580 (handle_pointer_plus): Likewise.
22581 (handle_builtin_memcpy): Likewise. Track nonzero characters
22582 that aren't necessarily followed by a nul terminator.
22583 (handle_char_store): Likewise.
22585 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22587 PR tree-optimization/80769
22588 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22589 for malloc and calloc. Document the new invariant that all related
22590 strinfos have delayed lengths or none do.
22591 (verify_related_strinfos): Move earlier in file.
22592 (set_endptr_and_length): New function, split out from...
22593 (get_string_length): ...here. Also set the lengths of related
22595 (zero_length_string): Assert that chainsi has known (rather than
22597 (adjust_related_strinfos): Likewise.
22599 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22601 PR tree-optimization/81136
22602 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22603 assert that two references with the same misalignment have the same
22604 compile-time misalignment if those compile-time misalignments
22607 2017-07-01 Andi Kleen <ak@linux.intel.com>
22609 * print-tree.c (print_node): Print all attributes.
22611 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22613 * cfg.c (scale_bbs_frequencies): New function.
22614 * cfg.h (scale_bbs_frequencies): Declare it.
22615 * cfgloopanal.c (single_likely_exit): Cleanup.
22616 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22618 (scale_loop_profile): Likewise.
22619 (loop_version): Likewise.
22620 (create_empty_loop_on_edge): Update.
22621 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22622 scale_loop_frequencies, scale_loop_profile, loopify,
22623 loop_version): Update prototypes.
22624 * modulo-sched.c (sms_schedule): Update.
22625 * predict.c (unlikely_executed_edge_p): Also check probability.
22626 (probably_never_executed_edge_p): Fix typo.
22627 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22628 * tree-parloops.c (gen_parallel_loop): Update.
22629 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22630 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22631 * tree-ssa-loop-split.c (split_loop): Update.
22632 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22633 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22634 (vect_loop_versioning): Update.
22635 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22637 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22639 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22641 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22643 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22644 to keep profile consistent.
22646 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22648 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22649 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22650 * profile-count.h (max_safe_multiplier): Make unsigned.
22651 (profile_count::guessed_zero): New.
22653 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22655 * bb-reorder.c (fix_up_crossing_landing_pad,
22656 fix_crossing_conditional_branches): Use make_single_succ_edge
22657 to keep profile consistent.
22659 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22661 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22664 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22667 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22668 the right scopes, make sure cond_jump isn't preserved between multiple
22669 iterations. Search for fallthru edge whenever there are 3+ edges and
22670 use find_fallthru_edge for it.
22672 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22674 Patch by Alexander Monakov <amonakov@ispras.ru>
22675 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22676 probabilities consistently.
22678 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22680 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22681 emit_cmp_and_jump_insns.
22683 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22686 * tree-inline.c (expand_call_inline): Combine profile statuses.
22688 2017-06-30 Andrew Pinski <apinski@cavium.com>
22690 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22691 fold_stmt returned true.
22693 2017-06-30 Nathan Sidwell <nathan@acm.org>
22695 * ggc.h (empty_string): Delete.
22696 * cfgexpand.c (expand_asm_stmt): Use plain "".
22697 * optabs.c (expand_asm_memory_barrier): Likewise.
22698 * stringpool.c (empty_string): Delete.
22699 (digit_vector, digit_string): Delete.
22700 (ggc_alloc_string): Use plain "", don't optimize single digit
22701 strings. Use ggc_alloc_atomic.
22703 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22705 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22706 comparison set and one other set, use the cost of the non-comparison
22709 2017-06-30 Nathan Sidwell <nathan@acm.org>
22711 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22714 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22716 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22717 loops. Remove now unneeded calls to gimple_switch_set_label() that
22718 just set removed labels to NULL_TREE.
22720 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22722 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22723 (set_range_info): ...here. Only call set_range_info_raw if domain
22725 (set_nonzero_bits): Call set_range_info_raw.
22726 * tree-ssanames.h (set_range_info_raw): New.
22728 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22731 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22732 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22733 of nonimmediate_operand and <store_mask_constraint> instead of m
22734 for the input operand. For V8FI iterator, always split if input
22735 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22736 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22737 <store_mask_predicate> instead of register_operand and
22738 <store_mask_constraint> instead of v for the input operand. Make
22739 sure both operands aren't MEMs for if not <mask_applied>.
22741 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22743 * lto-wrapper.c (copy_file) Close both file descriptors before
22746 2017-06-30 Martin Liska <mliska@suse.cz>
22749 * multiple_target.c (create_dispatcher_calls): Make ifunc
22750 also for function that don't have calls or are not referenced.
22752 2017-06-30 Richard Biener <rguenther@suse.de>
22754 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22755 analyze the first scalar stmt. Move vector type computation
22756 for the BB case here from ...
22757 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22758 live operation processing in the SLP case properly.
22760 2017-06-30 Richard Biener <rguenther@suse.de>
22762 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22764 2017-06-30 Martin Liska <mliska@suse.cz>
22767 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22768 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22770 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22772 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22775 2017-06-30 Richard Biener <rguenther@suse.de>
22777 PR tree-optimization/81249
22778 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22779 condition reduction result to original scalar type.
22781 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22783 * profile-count.h (enum profile_quality): Fix typos and whitespace
22786 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22788 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22789 type for branch probabilities.
22791 2017-06-29 Julian Brown <julian@codesourcery.com>
22792 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22794 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22795 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22796 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22797 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22799 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22801 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22802 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22803 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22804 CC usage from generic code to here.
22805 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22806 CC usage into the target macros.
22808 2017-06-29 Maya Rashish <coypu@sdf.org>
22810 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22813 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22815 * arm/arm-builtins.c: Include profile-count.h
22816 * except.c (sjlj_emit_function_enter): Use
22817 profile_probability::unlikely.
22819 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22821 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22822 and tocrel_offset be pointer args rather than implicitly using
22824 (legitimate_constant_pool_address_p, rs6000_emit_move,
22825 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22826 tocrel_offset and use in toc_relative_expr_p call.
22827 (print_operand, print_operand_address): Use static tocrel_base_oac
22828 and tocrel_offset_oac.
22829 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22832 2017-06-29 Maya Rashish <coypu@sdf.org>
22834 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22836 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22838 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22839 objects, take into account only the alignment of 'op0' and 'mode1' if
22842 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22844 * ccmp.c (ccmp_tree_comparison_p): New function.
22845 (ccmp_candidate_p): Update to use above function.
22846 (get_compare_parts): New function.
22847 (expand_ccmp_next): Update to use new functions.
22848 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22850 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22851 take mode as argument.
22852 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22853 * expr.c (expand_expr_real_1): Pass mode as argument.
22855 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22857 * combine.c (combine_instructions): Print insns to dump_file, together
22860 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22862 * asan.c (asan_emit_stack_protection): Update.
22863 (create_cond_insert_point): Update.
22864 * auto-profile.c (afdo_propagate_circuit): Update.
22865 * basic-block.h (struct edge_def): Turn probability to
22866 profile_probability.
22867 (EDGE_FREQUENCY): Update.
22868 * bb-reorder.c (find_traces_1_round): Update.
22869 (better_edge_p): Update.
22870 (sanitize_hot_paths): Update.
22871 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22872 (make_single_succ_edge): Update.
22873 (check_bb_profile): Update.
22874 (dump_edge_info): Update.
22875 (update_bb_profile_for_threading): Update.
22876 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22877 probabilitycount to 0.
22878 * cfgbuild.c (compute_outgoing_frequencies): Update.
22879 * cfgcleanup.c (try_forward_edges): Update.
22880 (outgoing_edges_match): Update.
22881 (try_crossjump_to_edge): Update.
22882 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22883 (expand_gimple_tailcall): Update.
22884 (construct_init_block): Use make_single_succ_edge.
22885 (construct_exit_block): Use make_single_succ_edge.
22886 * cfghooks.c (verify_flow_info): Update.
22887 (redirect_edge_succ_nodup): Update.
22888 (split_edge): Update.
22889 (account_profile_record): Update.
22890 * cfgloopanal.c (single_likely_exit): Update.
22891 * cfgloopmanip.c (scale_loop_profile): Update.
22892 (set_zero_probability): Remove.
22893 (duplicate_loop_to_header_edge): Update.
22894 * cfgloopmanip.h (loop_version): Update prototype.
22895 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22896 (force_nonfallthru_and_redirect): Update.
22897 (update_br_prob_note): Update.
22898 (rtl_verify_edges): Update.
22899 (purge_dead_edges): Update.
22900 (rtl_lv_add_condition_to_bb): Update.
22901 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22902 * cgraphunit.c (init_lowered_empty_function): Update.
22903 (cgraph_node::expand_thunk): Update.
22904 * cilk-common.c: Include profile-count.h
22905 * dojump.c (inv): Remove.
22906 (jumpifnot): Update.
22907 (jumpifnot_1): Update.
22908 (do_jump_1): Update.
22910 (do_jump_by_parts_greater_rtx): Update.
22911 (do_compare_rtx_and_jump): Update.
22912 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
22913 do_jump_1. do_compare_rtx_and_jump): Update prototype.
22914 * dwarf2cfi.c: Include profile-count.h
22915 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
22916 (sjlj_emit_dispatch_table): Likewise.
22917 * explow.c: Include profile-count.h
22918 * expmed.c (emit_store_flag_force): Update.
22919 (do_cmp_and_jump): Update.
22920 * expr.c (compare_by_pieces_d::generate): Update.
22921 (compare_by_pieces_d::finish_mode): Update.
22922 (emit_block_move_via_loop): Update.
22923 (store_expr_with_bounds): Update.
22924 (store_constructor): Update.
22925 (expand_expr_real_2): Update.
22926 (expand_expr_real_1): Update.
22927 * expr.h (try_casesi, try_tablejump): Update prototypes.
22928 * gimple-pretty-print.c (dump_probability): Update.
22929 (dump_profile): New.
22930 (dump_gimple_label): Update.
22931 (dump_gimple_bb_header): Update.
22932 * graph.c (draw_cfg_node_succ_edges): Update.
22933 * hsa-gen.c (convert_switch_statements): Update.
22934 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
22935 (find_if_case_1): Update.
22936 (find_if_case_2): Update.
22937 * internal-fn.c (expand_arith_overflow_result_store): Update.
22938 (expand_addsub_overflow): Update.
22939 (expand_neg_overflow): Update.
22940 (expand_mul_overflow): Update.
22941 (expand_vector_ubsan_overflow): Update.
22942 * ipa-cp.c (good_cloning_opportunity_p): Update.
22943 * ipa-split.c (split_function): Use make_single_succ_edge.
22944 * ipa-utils.c (ipa_merge_profiles): Update.
22945 * loop-doloop.c (add_test): Update.
22946 (doloop_modify): Update.
22947 * loop-unroll.c (compare_and_jump_seq): Update.
22948 (unroll_loop_runtime_iterations): Update.
22949 * lra-constraints.c (lra_inheritance): Update.
22950 * lto-streamer-in.c (input_cfg): Update.
22951 * lto-streamer-out.c (output_cfg): Update.
22952 * mcf.c (adjust_cfg_counts): Update.
22953 * modulo-sched.c (sms_schedule): Update.
22954 * omp-expand.c (expand_omp_for_init_counts): Update.
22955 (extract_omp_for_update_vars): Update.
22956 (expand_omp_ordered_sink): Update.
22957 (expand_omp_for_ordered_loops): Update.
22958 (expand_omp_for_generic): Update.
22959 (expand_omp_for_static_nochunk): Update.
22960 (expand_omp_for_static_chunk): Update.
22961 (expand_cilk_for): Update.
22962 (expand_omp_simd): Update.
22963 (expand_omp_taskloop_for_outer): Update.
22964 (expand_omp_taskloop_for_inner): Update.
22965 * omp-simd-clone.c (simd_clone_adjust): Update.
22966 * optabs.c (expand_doubleword_shift): Update.
22967 (expand_abs): Update.
22968 (emit_cmp_and_jump_insn_1): Update.
22969 (expand_compare_and_swap_loop): Update.
22970 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
22971 * predict.c (predictable_edge_p): Update.
22972 (edge_probability_reliable_p): Update.
22973 (set_even_probabilities): Update.
22974 (combine_predictions_for_insn): Update.
22975 (combine_predictions_for_bb): Update.
22976 (propagate_freq): Update.
22977 (estimate_bb_frequencies): Update.
22978 (force_edge_cold): Update.
22979 * profile-count.c (profile_count::dump): Add missing space into dump.
22980 (profile_count::debug): Add newline.
22981 (profile_count::differs_from_p): Explicitly convert to unsigned.
22982 (profile_count::stream_in): Update.
22983 (profile_probability::dump): New member function.
22984 (profile_probability::debug): New member function.
22985 (profile_probability::differs_from_p): New member function.
22986 (profile_probability::differs_lot_from_p): New member function.
22987 (profile_probability::stream_in): New member function.
22988 (profile_probability::stream_out): New member function.
22989 * profile-count.h (profile_count_quality): Rename to ...
22990 (profile_quality): ... this one.
22991 (profile_probability): New.
22992 (profile_count): Update.
22993 * profile.c (compute_branch_probabilities): Update.
22994 * recog.c (peep2_attempt): Update.
22995 * sched-ebb.c (schedule_ebbs): Update.
22996 * sched-rgn.c (find_single_block_region): Update.
22997 (compute_dom_prob_ps): Update.
22998 (schedule_region): Update.
22999 * sel-sched-ir.c (compute_succs_info): Update.
23000 * stmt.c (struct case_node): Update.
23001 (do_jump_if_equal): Update.
23002 (get_outgoing_edge_probs): Update.
23003 (conditional_probability): Update.
23004 (emit_case_dispatch_table): Update.
23005 (expand_case): Update.
23006 (expand_sjlj_dispatch_table): Update.
23007 (emit_case_nodes): Update.
23008 * targhooks.c: Update.
23009 * tracer.c (better_p): Update.
23010 (find_best_successor): Update.
23011 * trans-mem.c (expand_transaction): Update.
23012 * tree-call-cdce.c: Update.
23013 * tree-cfg.c (gimple_split_edge): Upate.
23014 (move_sese_region_to_fn): Upate.
23015 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23016 * tree-eh.c (lower_resx): Upate.
23017 (cleanup_empty_eh_move_lp): Upate.
23018 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23019 * tree-inline.c (copy_edges_for_bb): Update.
23020 (copy_cfg_body): Update.
23021 * tree-parloops.c (gen_parallel_loop): Update.
23022 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23023 (gimple_gen_time_profiler): Update.
23024 * tree-ssa-dce.c (remove_dead_stmt): Update.
23025 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23026 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23027 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23028 (unloop_loops): Update.
23029 (try_peel_loop): Update.
23030 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23031 * tree-ssa-loop-split.c (connect_loops): Update.
23032 (split_loop): Update.
23033 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23034 (hoist_guard): Update.
23035 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23036 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23037 (value_replacement): Update.
23038 * tree-ssa-reassoc.c (branch_fixup): Update.
23039 * tree-ssa-tail-merge.c (replace_block_by): Update.
23040 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23041 (create_edge_and_update_destination_phis): Update.
23042 (compute_path_counts): Update.
23043 (recompute_probabilities): Update.
23044 (update_joiner_offpath_counts): Update.
23045 (freqs_to_counts_path): Update.
23046 (duplicate_thread_path): Update.
23047 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23048 (struct switch_conv_info): Update.
23049 (gen_inbound_check): Update.
23050 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23051 (vect_do_peeling): Update.
23052 (vect_loop_versioning): Update.
23053 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23054 (optimize_mask_stores): Update.
23055 * ubsan.c (ubsan_expand_null_ifn): Update.
23056 * value-prof.c (gimple_divmod_fixed_value): Update.
23057 (gimple_divmod_fixed_value_transform): Update.
23058 (gimple_mod_pow2): Update.
23059 (gimple_mod_pow2_value_transform): Update.
23060 (gimple_mod_subtract): Update.
23061 (gimple_mod_subtract_transform): Update.
23062 (gimple_ic): Update.
23063 (gimple_stringop_fixed_value): Update.
23064 (gimple_stringops_transform): Update.
23065 * value-prof.h: Update.
23067 2017-06-29 Carl Love <cel@us.ibm.com>
23069 * config/rs6000/rs6000-c.c: Add support for built-in functions
23070 vector signed int vec_signed (vector float);
23071 vector signed long long vec_signed (vector double);
23072 vector signed int vec_signed2 (vector double, vector double);
23073 vector signed int vec_signede (vector double);
23074 vector signed int vec_signedo (vector double);
23075 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23076 instruction generator.
23077 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23078 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23079 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23081 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23082 vunsignede_v2df): Add define_expands.
23083 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23084 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23085 VEC_UNSIGNEDO): Add definitions.
23086 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23087 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23088 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23089 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23090 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23091 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23092 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23093 vec_unsignede, vec_unsignedo): Add builtin defines.
23094 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23096 * doc/extend.texi: Update the built-in documentation file for the
23097 new built-in functions.
23099 2017-06-29 Richard Biener <rguenther@suse.de>
23101 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23102 reduction chains to LOOP_VINFO_REDUCTIONS.
23103 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23104 SLP reductions after processing reduction chains.
23106 2017-06-29 Nathan Sidwell <nathan@acm.org>
23108 * builtins.c (fold_builtin_FUNCTION): Use
23109 lang_hooks.decl_printable_name.
23111 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23113 PR middle-end/81194
23114 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23115 with only one label.
23116 * stmt.c (expand_case): Assert NCASES is greater than one.
23118 2017-06-29 Richard Biener <rguenther@suse.de>
23120 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23122 (group_case_labels): Likewise.
23123 (find_taken_edge): Push sanity checking on val to workers...
23124 (find_taken_edge_cond_expr): ... here
23125 (find_taken_edge_switch_expr): ... and here, handle cases
23126 with just a default label.
23127 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23128 (group_case_labels): Likewise.
23129 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23130 group_case_labels does anything cleanup the CFG again.
23132 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23134 PR tree-optimization/81196
23135 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23136 exit condition comparing two IVs.
23138 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23140 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23141 profile to the dummy entry at the end of the list of architectures.
23142 * config/arm/arm-cpu-cdata.h: Regenerated.
23144 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23145 Michael Collison <michael.collison@arm.com>
23148 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23150 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23151 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23152 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23153 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23154 of shift when the shift amount is masked with constant equal to
23155 the size of the mode.
23156 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23159 2017-06-29 Martin Liska <mliska@suse.cz>
23161 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23163 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23166 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23169 (edge_growth_cache_entry::edge_growth_cache_entry): New
23171 (reset_edge_growth_cache): Update to use constructor.
23173 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23175 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23176 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23177 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23179 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23181 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23182 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23184 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23186 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23187 (*-linux-uclibc*): Add t-uclibc tmake_file.
23188 * config/t-musl: New.
23189 * config/t-uclibc: New.
23191 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23193 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23195 (gen_comm_data): Emit architectural setting of arch_prof.
23196 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23198 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23199 (armv8-m.base, armv8-m.main): Likewise.
23200 * arm-protos.h (arm_build_target): Add profile field.
23201 (arch_option): Likewise.
23202 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23204 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23205 arm_active_target.profile.
23207 2017-06-28 Richard Biener <rguenther@suse.de>
23209 PR middle-end/81227
23210 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23211 TYPE_OVERFLOW_WRAPS.
23212 * match.pd (negate_expr_p): Likewise.
23213 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23214 fold_build2, not fold_binary.
23216 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23218 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23219 Convert memory address to Pmode.
23220 (aarch64_print_operand): Assert MEM operands are always Pmode.
23222 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23225 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23226 Remove redundant if.
23227 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23228 * config/arm/aarch-common-protos.h
23229 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23230 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23232 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23235 * multiple_target.c (create_dispatcher_calls): Set the default
23236 clone to be static, not public.
23238 2017-06-28 Richard Biener <rguenther@suse.de>
23240 * tree-vect-loop.c (vectorizable_reduction): Move special
23241 cond reduction IV var creation ...
23242 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23243 parameter. Use STMT_VINFO_VECTYPE.
23244 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23247 2017-06-28 Martin Liska <mliska@suse.cz>
23250 * ipa-visibility.c (non_local_p): Handle visibility.
23252 2017-06-28 Martin Liska <mliska@suse.cz>
23255 * common.opt: Add IntegerRange to various options.
23256 * opt-functions.awk (integer_range_info): New function.
23257 * optc-gen.awk: Add integer_range_info to cl_options struct.
23258 * opts-common.c (decode_cmdline_option): Handle
23259 CL_ERR_INT_RANGE_ARG.
23260 (cmdline_handle_error): Likewise.
23261 * opts.c (print_filtered_help): Show valid interval in
23262 when --help is provided.
23263 * opts.h (struct cl_option): Add range_min and range_max fields.
23264 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23266 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23268 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23269 (x * C EQ/NE y * C): New transformation.
23271 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23273 * genmultilib (combination_space): Accept '+' in option names.
23275 2017-06-28 Martin Liska <mliska@suse.cz>
23278 * asan.c (instrument_derefs): Bail out inner references
23279 that are hard register variables.
23281 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23284 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23285 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23287 2017-06-28 Richard Biener <rguenther@suse.de>
23289 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23290 (vect_get_slp_defs): Adjust.
23291 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23293 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23295 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23296 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23297 (vectorizable_reduction): Adjust.
23298 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23300 (vect_get_slp_defs): Likewise.
23301 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23302 (vectorizable_bswap): Adjust.
23303 (vectorizable_call): Likewise.
23304 (vectorizable_conversion): Likewise.
23305 (vectorizable_assignment): Likewise.
23306 (vectorizable_shift): Likewise.
23307 (vectorizable_operation): Likewise.
23308 (vectorizable_store): Likewise.
23309 (vectorizable_condition): Likewise.
23310 (vectorizable_comparison): Likewise.
23312 2017-06-28 Michael Collison <michael.collison@arm.com>
23315 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23317 (arm_gen_movmemqi): Removed unused variable 'i'.
23318 Convert 'for' loop into 'while' loop.
23319 (arm_expand_prologue): Remove last unnecessary set of insn.
23320 (thumb_pop): Remove unused variable 'pushed_words'.
23321 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23323 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23325 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23326 * config/s390/s390-protos.h: Add prototype of
23327 s390_rel_address_ok_p.
23328 * config/s390/s390.c (s390_got_symbol): New function.
23329 (s390_rel_address_ok_p): New function.
23330 (legitimize_pic_address): Use s390_rel_address_ok_p.
23331 (s390_load_got): Use s390_got_symbol.
23332 (s390_option_override): Issue error if
23333 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23334 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23336 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23338 2017-06-27 Andrew Pinski <apinski@cavium.com>
23340 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23341 (X * copysign (1.0, X)): New pattern.
23342 (X * copysign (1.0, -X)): New pattern.
23343 (copysign (-1.0, CST)): New pattern.
23345 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23347 * genmultilib (combination_space): Remove variable.
23348 Validate reuse rules against regular expression for any sequence
23349 of multilib options in any order.
23351 2017-06-27 Michael Collison <michael.collison@arm.com>
23353 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23354 call aarch64_split_simd_combine.
23355 * (aarch64_combine_internal<mode>): Delete pattern.
23356 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23357 Allow register and subreg operands.
23359 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23361 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23362 specific need, just fallback on defaults.
23363 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23365 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23366 Olivier Hainque <hainque@adacore.com>
23368 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23370 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23371 targets. Pick a default if no particular attempt applied.
23372 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23375 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23377 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23378 (x86_64-wrs-vxworks7): Likewise.
23380 2017-06-27 Marek Polacek <polacek@redhat.com>
23383 * ubsan.c (instrument_null): Check get_base_address's result for null.
23385 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23387 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23389 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23391 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23392 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23393 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23394 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23395 New function types.
23396 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23397 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23398 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23399 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23400 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23401 BUILT_IN_FEUPDATEENV): New builtins.
23402 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23403 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23404 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23405 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23407 (builtin_structptr_types): Adjust size.
23408 * tree.c (builtin_structptr_types): Add four entries.
23410 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23411 Olivier Hainque <hainque@adacore.com>
23413 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23414 (TLS_SYM): New local macro, forcing reference to __tls__ on
23415 link command lines for VxWorks 7 RTPs, triggering initialization
23417 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23418 OS features TLS support, true for RTPs on VxWorks 7.
23419 * config/vxworks.c (vxworks_override_options): Setup emutls
23422 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23424 * predict.c (test_prediction_value_range): Use -1U instead of -1
23425 to avoid narrowing conversion warning.
23426 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23427 to avoid narrowing conversion warning.
23428 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23430 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23432 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23434 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23435 64bit configurations.
23436 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23437 (SIZE_TYPE): Likewise.
23438 * config/vxworks.c (vxworks_emutls_var_fields): Use
23439 long_unsigned_type_node instead of unsigned_type_node as the offset
23440 field type, which is "pointer" mode in emutls.c.
23442 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23445 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23447 PR middle-end/81207
23448 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23449 gimple_vuse copying separately from gimple_vdef copying.
23451 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23453 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23456 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23457 Olivier Hainque <hainque@adacore.com>
23459 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23460 for all vxworks7 targets.
23461 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23462 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23463 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23464 variations for VX6/VX7 and 32/64bits later on in ...
23465 (VXWORKS_LIB_SPEC): Leverage new macros.
23466 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23467 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23469 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23471 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23472 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23474 2017-06-26 Carl Love <cel@us.ibm.com>
23476 * config/rs6000/rs6000-c.c: Add support for built-in functions
23477 vector bool char vec_reve (vector bool char);
23478 vector signed char vec_reve (vector signed char);
23479 vector unsigned char vec_reve (vector unsigned char);
23480 vector bool int vec_reve (vector bool int);
23481 vector signed int vec_reve (vector signed int);
23482 vector unsigned int vec_reve (vector unsigned int);
23483 vector bool long long vec_reve (vector bool long long);
23484 vector signed long long vec_reve (vector signed long long);
23485 vector unsigned long long vec_reve (vector unsigned long long);
23486 vector bool short vec_reve (vector bool short);
23487 vector signed short vec_reve (vector signed short);
23488 vector double vec_reve (vector double);
23489 vector float vec_reve (vector float);
23490 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23491 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23492 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23493 (altivec_vreve): New pattern.
23494 * config/rs6000/altivec.h (vec_reve): New define.
23495 * doc/extend.texi (vec_rev): Update the built-in documentation file
23496 for the new built-in functions.
23498 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23500 PR tree-optimization/71815
23501 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23503 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23505 (slsr_process_phi): Likewise.
23506 (replace_uncond_cands_and_profitable_phis): Don't replace a
23507 multiply candidate with a stride of 1 (copy or cast).
23508 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23510 (lowest_cost_path): Likewise.
23511 (total_savings): Likewise.
23513 2017-06-26 Richard Biener <rguenther@suse.de>
23516 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23517 Use def_builtin_pure for all gather builtins.
23519 2017-06-26 Richard Biener <rguenther@suse.de>
23521 PR tree-optimization/81203
23522 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23523 non-dominating BBs.
23525 2017-06-26 Marek Polacek <polacek@redhat.com>
23528 * doc/invoke.texi: Document -Wmultistatement-macros.
23530 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23532 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23533 arm_neon_ok_no_float_abi effective target.
23535 2017-06-26 Richard Biener <rguenther@suse.de>
23537 PR tree-optimization/80928
23538 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23539 (copy_bbs): Set BB_DUPLICATED flag early.
23540 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23542 (execute_on_shrinking_pred): Likewise.
23543 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23544 BB_DUPLICATED blocks.
23545 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23546 iterate over all PHIs considering removal of *gsi.
23548 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23550 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23553 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23555 * config/rs6000/rs6000-string.c: (expand_block_clear,
23556 do_load_for_compare, select_block_compare_mode,
23557 compute_current_alignment, expand_block_compare,
23558 expand_strncmp_align_check, expand_strn_compare,
23559 expand_block_move, rs6000_output_load_multiple)
23560 Move functions related to string/block move/compare
23561 to a separate file.
23562 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23563 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23564 for this function which is now used in two files.
23565 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23566 * config.gcc: Add rs6000-string.o to extra_objs for
23567 targets powerpc*-*-* and rs6000*-*-*.
23569 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23572 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23573 32-bit, since indexed is not valid for DImode.
23574 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23575 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23576 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23577 (define_peephole2 for Altivec d-form store): Likewise.
23580 * multiple_target.c (create_dispatcher_calls): Only create the
23581 dispatcher call if the function is the default clone of a
23582 versioned function.
23584 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23586 PR middle-end/80902
23587 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23588 a call, force the call to not be a tail call.
23590 2017-06-23 Jeff Law <law@redhat.com>
23592 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23595 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23597 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23598 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23599 handling for early expansion of vector shifts (sl,sr,sra,rl).
23600 (builtin_function_type): Add vector shift right instructions
23601 to the unsigned argument list.
23603 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23605 rtl-optimizatoin/79286
23606 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23607 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23608 trap. PIC register plus a const unspec without offset can never trap.
23610 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23612 * tree.h (builtin_structptr_type): New type.
23613 (builtin_structptr_types): Declare new array.
23614 * tree.c (builtin_structptr_types): New array.
23615 (free_lang_data, build_common_tree_nodes): Use it.
23617 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23620 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23623 2017-06-22 Matt Turner <mattst88@gmail.com>
23625 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23626 Lake models to skylake case. Assume skylake for unknown
23627 models with clflushopt.
23629 2017-06-22 Jeff Law <law@redhat.com>
23631 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23632 frame sizes that do not satisfy aarch64_uimm12_shift.
23634 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23636 * profile-count.h (apply_probability,
23637 apply_scale, probability_in): Fix checks for zero.
23639 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23641 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23642 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23644 2016-06-22 Richard Biener <rguenther@suse.de>
23646 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23647 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23648 REDUC_MAX_EXPR support.
23649 (vectorizable_reduction): Likewise.
23650 (vect_create_epilog_for_reduction): Likewise.
23652 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23654 * match.pd (A / (1 << B) -> A >> B): New.
23655 * generic-match-head.c: Include optabs-tree.h.
23656 * gimple-match-head.c: Likewise.
23657 * optabs-tree.h (target_supports_op_p): New.
23658 * optabs-tree.c (target_supports_op_p): New.
23660 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23662 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23663 $gcc_cv_ld --help output.
23664 (gcc_cv_ld_demangle): Likewise.
23665 (gcc_cv_ld_eh_frame_hdr): Likewise.
23666 (gcc_cv_ld_pie): Likewise.
23667 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23668 (gcc_cv_ld_buildid): Likewise.
23669 (gcc_cv_ld_sysroot): Likewise.
23670 (ld_bndplt_support): Likewise.
23671 (ld_pushpopstate_support): Likewise.
23672 * configure: Regenerate.
23673 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23675 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23678 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23679 operands indexes to avoid gap between operands and match_dups.
23681 2017-06-21 Andrew Pinski <apinski@cavium.com>
23683 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23684 Increment Arith_shift and Arith_shift_reg by 1.
23685 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23687 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23688 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23689 (aarch64_strip_extend): Add new argument and test for it.
23690 (aarch64_cheap_mult_shift_p): New function.
23691 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23692 add a cost if it is true.
23693 Update calls to aarch64_strip_extend.
23694 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23696 2017-06-21 Andrew Pinski <apinski@cavium.com>
23698 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23700 (thunderxt88): Likewise.
23701 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23702 (thunderx_prefetch_tune): New variable.
23703 (thunderx2t99_prefetch_tune): Update for the correct values.
23704 (thunderxt88_tunings): New variable.
23705 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23706 generic_prefetch_tune.
23707 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23709 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23711 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23712 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23713 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23714 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23715 (aarch64_atomic_cas<mode>, GPI): Likewise.
23717 2017-06-21 Martin Liska <mliska@suse.cz>
23719 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23720 statements on cold and hot labels.
23721 * predict.c (tree_estimate_probability_bb): Remove the
23722 prediction from this place.
23724 2017-06-21 Martin Liska <mliska@suse.cz>
23726 PR tree-optimization/79489
23727 * gimplify.c (maybe_add_early_return_predict_stmt): New
23729 (gimplify_return_expr): Call the function.
23730 * predict.c (tree_estimate_probability_bb): Remove handling
23732 * predict.def: Update comment about early return predictor.
23733 * gimple-predict.h (is_gimple_predict): New function.
23734 * predict.def: Change default value of early return to 66.
23735 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23737 * passes.def: Put pass_strip_predict_hints to the beginning of
23740 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23742 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23743 FUNCTION_DECL declarations.
23744 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23746 (dwaf2out_decl): Likewise.
23747 * godump.c (go_early_global_decl): Skip call to the real debug hook
23748 for FUNCTION_DECL declarations.
23749 * passes.c (rest_of_decl_compilation): Skip call to the
23750 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23751 -fdump-go-spec is passed.
23753 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23755 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23756 Reorder for compactness.
23757 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23758 (def_builtin_pure, def_builtin_pure2): New functions.
23759 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23761 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23763 * match.pd (nop_convert): New predicate.
23764 ((A +- CST1) +- CST2): Allow some NOP conversions.
23766 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23769 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23770 with ctors/dtors if GOVD_SHARED is set.
23772 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23774 * config/aarch64/aarch64.md (movti_aarch64):
23775 Emit mov rather than orr.
23776 (movtf_aarch64): Likewise.
23777 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23778 Emit mov rather than orr.
23780 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23782 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23783 Swap alternatives, make integer dup more expensive.
23785 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23787 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23788 Return true for non-tls symbols.
23790 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23792 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23793 (cortex-a75): Likewise.
23794 (cortex-a75.cortex-a55): Likewise.
23795 * config/aarch64/aarch64-tune.md: Regenerate.
23796 * doc/invoke.texi (-mtune): Document new values for -mtune.
23798 2017-06-21 Tom de Vries <tom@codesourcery.com>
23800 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23801 stack_size feature.
23802 (Effective-Target Keywords, Other attributes): Suggest using
23803 dg-add-options stack_size feature to get stack limit in stack_size
23804 effective target documentation.
23806 2017-06-21 Julian Brown <julian@codesourcery.com>
23807 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23809 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23810 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23811 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23813 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23814 attribute type list for neon_multiply.
23815 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23816 attribute type list for neon_multiply.
23817 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23818 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23819 attribute type list for neon_multiply.
23820 * config/arm/types.md (crypto_pmull): Add.
23821 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23822 attribute type list.
23824 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23826 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23829 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23831 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23832 to make sure not to dereference a NULL cost_classes_ptr pointer.
23834 2017-06-20 Carl Love <cel@us.ibm.com>
23836 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23837 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23838 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23839 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23840 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23841 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23842 VMULOSW): New enum "unspec" values.
23843 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23844 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23845 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23846 altivec_vmulosw): New patterns.
23847 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23848 VMULOSW): Add definitions.
23850 2017-06-20 Julia Koval <julia.koval@intel.com>
23852 * config/i386/i386.c: Fix rounding expand for new pattern.
23853 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23855 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23857 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23858 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23860 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23862 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23865 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23867 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23868 architecture, then by implementer ID.
23869 * config/aarch64/aarch64-tune.md: Regenerate.
23871 2017-06-20 Richard Biener <rguenther@suse.de>
23873 PR middle-end/81097
23874 * fold-const.c (split_tree): Fold to type before negating.
23876 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23878 * diagnostic-show-locus.c
23879 (selftest::test_fixit_deletion_affecting_newline): New function.
23880 (selftest::diagnostic_show_locus_c_tests): Call it.
23882 2017-06-20 Andreas Schwab <schwab@suse.de>
23885 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23888 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23890 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23891 __ARM_FEATURE_COPROC according to support.
23893 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23895 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23896 Rewritten to avoid overflow for > 32-bit pointers.
23899 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23900 by removing enum keyword.
23901 (ubsan_type_descriptor): Likewise. Formatting fix.
23904 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23905 splitter): Require TARGET_SSE2 in the condition.
23907 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
23910 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
23911 for doing vector set of SFmode on ISA 3.0.
23912 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
23913 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
23915 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
23916 SFmode value into a V4SF variable that was extracted from another
23917 V4SF variable without converting the element to double precision
23918 and back to single precision vector format.
23919 (vsx_insert_extract_v4sf_p9_2): Likewise.
23921 2017-06-19 Jakub Jelinek <jakub@redhat.com>
23923 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
23924 in UWHI to avoid undefined overflow.
23927 * ubsan.h (enum ubsan_encode_value_phase): New.
23928 (ubsan_encode_value): Change second argument to
23929 enum ubsan_encode_value_phase with default value of
23930 UBSAN_ENCODE_VALUE_GENERIC.
23931 * ubsan.c (ubsan_encode_value): Change second argument to
23932 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
23933 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
23934 create_tmp_var_raw instead of create_tmp_var and use a
23936 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
23937 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
23938 ubsan_encode_value callers.
23941 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
23942 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
23943 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
23945 2017-06-19 Richard Biener <rguenther@suse.de>
23947 PR middle-end/81118
23948 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
23949 estimates if we changed anything.
23951 2017-06-19 Richard Biener <rguenther@suse.de>
23953 PR tree-optimization/80887
23954 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
23955 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
23956 simplified lookups, then reset mprts_hook.
23957 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
23959 (try_to_simplify): Likewise.
23961 2017-06-19 Martin Liska <mliska@suse.cz>
23964 * gimplify.c (gimplify_switch_expr):
23965 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
23967 2017-06-19 Martin Liska <mliska@suse.cz>
23969 * doc/install.texi: Document that PGO runs in 4 stages.
23971 2017-06-19 Martin Liska <mliska@suse.cz>
23974 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
23975 to dispatcher function name.
23976 * multiple_target.c (replace_function_decl): New function.
23977 (create_dispatcher_calls): Redirect both edges and references.
23979 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
23981 * profile-count.c (profile_count::dump): Dump quality.
23982 (profile_count::differs_from_p): Update for unsigned val.
23983 * profile-count.h (profile_count_quality): New enum.
23984 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
23986 2017-06-19 Richard Biener <rguenther@suse.de>
23988 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
23989 struct function as arg.
23990 (estimate_numbers_of_iterations): Export overload with loop arg.
23991 (free_numbers_of_iterations_estimates_loop): Use an overload of
23992 free_numbers_of_iterations_estimates instead.
23993 * tree-cfg.c (remove_bb): Adjust.
23994 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
23995 * tree-parloops.c (gen_parallel_loop): Likewise.
23996 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
23998 (tree_unroll_loops_completely): Likewise.
23999 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24000 Use an overload instead and export.
24001 (estimated_loop_iterations): Adjust.
24002 (max_loop_iterations): Likewise.
24003 (likely_max_loop_iterations): Likewise.
24004 (estimate_numbers_of_iterations): Take struct function as arg
24006 (loop_exits_before_overflow): Adjust.
24007 (free_numbers_of_iterations_estimates_loop): Use an overload.
24008 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24009 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24011 2017-06-19 Richard Biener <rguenther@suse.de>
24014 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24015 RANGE_EXPR conservatively.
24017 2017-06-16 Carl Love <cel@us.ibm.com>
24019 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24020 definitions for vec_float, vec_float2, vec_floato,
24021 vec_floate built-ins.
24022 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24023 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24025 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24026 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24027 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24028 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24029 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24030 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24031 vec_floato): Add builtin defines.
24032 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24033 Update the built-in documentation file for the new built-in
24036 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24038 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24039 (mthumb): Mark as the negative of -marm.
24041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24043 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24045 (ARM Options, -mtune): Document that this accepts the same
24046 extension options as -mcpu.
24047 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24049 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24051 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24052 permitted extensions.
24054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24056 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24057 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24058 (armv8-m.main): Add option +nodsp.
24059 * config/arm/arm-cpu-cdata.h: Regenerated.
24061 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24063 * config/arm/t-fuchsia: New file.
24064 * config.gcc (arm*-*-fuchsia*): Use it.
24066 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24068 * config/arm/t-symbian: Rewrite for new option infrastructure.
24070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24072 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24073 (MULTILIB_REQUIRED): Likewise.
24075 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24077 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24078 (MULTILIB_RESUE): Likewise.
24079 (MULTILIB_MATCHES): Likewise.
24080 (MULTLIB_REQUIRED): Likewise.
24082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24084 * config/arm/t-rtems: Rewrite for new option framework.
24086 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24088 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24089 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24090 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24091 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24092 * config/arm/t-multilib: ... here.
24093 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24094 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24095 armv7-a and armv8*-a when A-profile libraries have not been built.
24096 * config/arm/t-rmprofile: Rewrite.
24098 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24100 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24101 with a backslash. Remove the backslash after substituting unescaped
24103 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24105 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24107 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24108 the driver to pass through the default setting of -mfloat-abi.
24109 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24111 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24112 (all_feat_combs): New rule.
24113 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24115 * config/arm/t-aprofile: Rewrite.
24117 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24119 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24120 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24121 fpu is not specified by the user/command-line.
24122 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24123 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24124 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24125 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24126 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24127 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24129 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24131 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24132 * config/arm/t-arm-elf: Rewritten.
24134 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24136 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24137 have some floating-point instructions.
24138 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24139 (TARGET_MAYBE_HARD_FLOAT): New macro.
24140 * config/arm/arm-builtins.c (arm_init_builtins): Use
24141 TARGET_MAYBE_HARD_FLOAT.
24142 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24144 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24146 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24147 (configargs.h): Include it.
24148 (arm_print_hint_for_fpu_option): New function.
24149 (arm_parse_fpu_option): New function.
24150 (candidate_extension): New class.
24151 (arm_canon_for_multilib): New function.
24152 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24153 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24154 (ARCH_CANONICAL_SPECS): New macro.
24155 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24157 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24159 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24160 are set after handling multilib fragments. Set target_cpu_default2
24163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24165 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24167 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24168 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24170 (check_arch): Likewise.
24171 * config/arm/arm.c (arm_configure_build_target): Handle
24172 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24173 options in the default.
24175 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24177 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24178 when an option is an alias of another.
24179 * config/arm/parsecpu.awk (optalias): New parser token.
24180 (gen_comm_data): Mark non-alias options as such. Emit entries
24181 for extension aliases.
24182 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24183 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24184 (armv6kz, armv6zk, armv6t2): Likewise.
24185 (armv7): Make vfpv3-d16 an alias.
24186 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24188 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24189 Sort in canonical order.
24190 (armv8-a): Sort in canonical order.
24191 (armv8.1-a, armv8.2-a): Likewise.
24192 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24194 (cortex-a9): Sort in canonical order.
24195 * config/arm/arm.c (selftests.h): Include it.
24196 (arm_test_cpu_arch_data): New function.
24197 (arm_run_self_tests): New function.
24198 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24199 (targetm): Move declaration to the end of the file.
24200 * arm-cpu-cdata.h: Regenerated.
24202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24204 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24205 call to target_mode_check describing the type of option passed.
24206 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24207 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24208 arm_parse_cpu_option_name to match parameters against list of
24210 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24211 arm_arch_core_flags data structure.
24212 * config/arm/arm-cpu_cdata.h: Regenerated.
24214 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24216 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24218 (arm_print_hint_for_cpu_option): Likewise.
24219 (arm_print_hint_for_arch_option): Likewise.
24220 (arm_parse_cpu_option_name): Likewise.
24221 (arm_parse_arch_option_name): Likewise.
24222 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24223 of entries in the all_fpus list.
24224 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24225 (arm_parse_cpu_option_name): Declare.
24226 (arm_parse_arch_option_name): Declare.
24227 (arm_parse_option_features): Declare.
24228 (arm_intialize_isa): Declare.
24229 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24231 (gen_comm_data): ... here. Make definitions non-static.
24232 * config/arm/arm-cpu-data.h: Regenerated.
24233 * config/arm/arm-cpu-cdata.h: Regenerated.
24235 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24237 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24238 (cpu_arch_extension): New structure.
24239 (cpu_arch_option, arch_option, cpu_option): New structures.
24240 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24241 architecture types.
24242 (gen_data): Generate new format data tables.
24243 * config/arm/arm.c (cpu_tune): New structure.
24244 (cpu_option, processors): Delete.
24245 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24246 (arm_print_hint_for_cpu_option): ... this and ...
24247 (arm_print_hint_for_arch_option): ... this.
24248 (arm_parse_arch_cpu_name): Delete. Replace with ...
24249 (arm_parse_cpu_option_name): ... this and ...
24250 (arm_parse_arch_option_name): ... this.
24251 (arm_unrecognized_feature): Change type of target parameter to
24253 (arm_parse_arch_cpu_features): Delete. Replace with ...
24254 (arm_parse_option_features): ... this.
24255 (arm_configure_build_target): Rework to use new configuration data
24257 (arm_print_tune_info): Rework for new configuration data tables.
24258 * config/arm/arm-cpu-data.h: Regenerated.
24259 * config/arm/arm-cpu.h: Regenerated.
24261 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24263 * Makefile.in (OBJS): Move sbitmap.o from here ...
24264 (OBJS-libcommon): ... to here.
24266 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24268 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24269 (ISA_ALL_CRYPTO): New macro.
24270 (ISA_ALL_SIMD): New macro
24271 (ISA_ALL_FP): New macro.
24272 * config/arm/arm.c (fpu_bitlist): Update initializer.
24273 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24275 (arm9e): Add fpu. Add option for nofp
24276 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24277 (arm926ej-s, arm1026ej-s): Likewise.
24278 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24279 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24280 neon-fp16, neon-vfpv4, nofp and nosimd.
24281 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24282 (cortex-a8): Add fpu. Add option for nofp.
24283 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24284 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24285 (cortex-r4f): Add fpu.
24286 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24287 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24289 (cortex-r8): Likewise.
24290 (cortex-m4): Add fpu. Add option for nofp.
24291 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24292 (cortex-a17.cortex-a7): Likewise.
24293 (cortex-a32): Add fpu. Add options for crypto and nofp.
24294 (cortex-a35, cortex-a53): Likewise.
24295 (cortex-a57): Add fpu. Add option for crypto.
24296 (cortex-a72, cortex-a73): Likewise.
24297 (exynos-m1): Likewise.
24298 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24299 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24300 (cortex-m33): Add fpu. Add option for nofp.
24301 * config/arm/arm-cpu-cdata.h: Regenerated
24302 * config/arm/arm-cpu-data.h: Regenerated.
24304 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24306 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24307 (armv5te, armv5tej): Likewise.
24308 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24309 (armv7): Add options fp and vfpv3-d16.
24310 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24311 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24313 (armv7ve): Likewise.
24314 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24315 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24316 (armv8-a): Add nocrypto option.
24317 (armv8.1-a, armv8.2-a): Likewise.
24318 (armv8-m.main): add options fp, fp.dp and nofp.
24320 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24322 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24324 (armv8-a+crc): Delete.
24325 (armv8.1-a): Add options simd, crypto and nofp.
24326 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24327 (armv8.2-a+fp16): Delete.
24328 (armv8-m.main): Add option dsp.
24329 (armv8-m.main+dsp): Delete.
24330 (cortex-a8): Add fpu. Add nofp option.
24331 (cortex-a9): Add fpu. Add nofp and nosimd options.
24332 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24333 link to main cpu and architecture data structures.
24334 (gen_comm_data): Only put isa attributes from the main architecture
24336 (option): New statement for architecture and CPU entries.
24337 * arm.c (struct cpu_option): New structure.
24338 (struct processors): Add entry for options.
24339 (arm_unrecognized_feature): New function.
24340 (arm_parse_arch_cpu_name): Ignore any characters after the first
24342 (arm_parse_arch_cpu_feature): New function.
24343 (arm_configure_build_target): Separate out any CPU and architecture
24344 features and parse separately. Don't error out if -mfpu=auto is
24345 used with only an architecture string.
24346 (arm_print_asm_arch_directives): New function.
24347 (arm_file_start): Call it.
24348 * config/arm/arm-cpu-cdata.h: Regenerated.
24349 * config/arm/arm-cpu-data.h: Likewise.
24350 * config/arm/arm-tables.opt: Likewise.
24352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24354 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24355 assembler when it is not -mfpu=auto.
24357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24359 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24360 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24361 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24362 (ASM_CPU_SPEC): Rewrite.
24363 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24364 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24365 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24366 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24367 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24368 copied string is NUL-terminated. Also strip any characters prefixed
24370 (arm_rewrite_selected_arch): New function.
24371 (arm_rewrite_march): New function.
24373 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24375 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24376 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24377 (march, mcpu, mtune): Convert to string-based options.
24378 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24379 (arm_parse_arch_cpu_name): New function.
24380 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24381 identify selected architecture or CPU.
24382 (arm_option_save): New function.
24383 (TARGET_OPTION_SAVE): Redefine.
24384 (arm_option_restore): Restore string options.
24385 (arm_option_print): Print string options.
24387 2017-06-16 Martin Sebor <msebor@redhat.com>
24389 PR tree-optimization/80933
24390 PR tree-optimization/80934
24391 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24392 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24393 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24394 (gimple_fold_builtin): Call them.
24396 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24398 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24399 as unlikely; update profile.
24401 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24403 * predict.c (force_edge_cold): Handle declaring edges impossible
24406 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24408 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24410 (try_unroll_loop_completely): Fix reporting.
24412 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24414 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24416 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24419 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24420 if given a non-constant argument for an intrinsic which requires a
24423 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24425 * profile.c (compare_freqs): New function.
24426 (branch_prob): Sort edge list.
24427 (find_spanning_tree): Assume that the list is priority sorted.
24429 2017-06-16 Richard Biener <rguenther@suse.de>
24431 PR tree-optimization/81090
24432 * passes.def (pass_record_bounds): Remove.
24433 * tree-pass.h (make_pass_record_bounds): Likewise.
24434 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24435 make_pass_record_bounds): Likewise.
24436 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24437 not free niter estimates at the beginning but at the end.
24438 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24440 2017-06-16 Richard Biener <rguenther@suse.de>
24442 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24443 initializer to workaround ICE in host GCC 4.8.
24445 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24447 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24449 (clone_inlined_nodes): Update.
24451 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24453 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24454 prefetch settings, and enable prefetching by default at -O3.
24456 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24458 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24459 Set flag_prefetch_loop_arrays according to tuning data.
24461 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24463 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24464 New tune structure.
24465 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24466 [Unrelated to main purpose of the patch] Place the pointer field last
24467 to enable type checking errors when tune structure are wrongly merged.
24468 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24469 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24470 (thunderx2t99_prefetch_tune): New tune constants.
24471 (tune_params *_tunings): Update all tunings (no functional change).
24472 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24473 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24474 from tunings structures.
24476 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24479 * ubsan.c (instrument_null): Add T argument, use it instead
24480 of computing it based on IS_LHS.
24481 (instrument_object_size): Likewise.
24482 (pass_ubsan::execute): Adjust instrument_null and
24483 instrument_object_size callers to pass gimple_get_lhs or
24484 gimple_assign_rhs1 result to it. Use instrument_null instead of
24485 calling get_base_address and instrument_mem_ref. Handle
24486 aggregate call arguments for object-size sanitization.
24488 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24490 PR tree-optimization/81089
24491 * tree-vrp.c (is_masked_range_test): Validate operands of
24494 2017-06-15 Martin Sebor <msebor@redhat.com>
24497 * dumpfile.c (dump_register): Avoid calling memset to initialize
24498 a class with a default ctor.
24499 * gcc.c (struct compiler): Remove const qualification.
24500 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24501 * hash-table.h: Ditto.
24502 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24504 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24505 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24507 * params.h (struct param_info): Make struct members non-const.
24508 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24509 with default initialization.
24510 * vec.h (vec_copy_construct, vec_default_construct): New helper
24512 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24513 with vec_copy_construct.
24514 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24515 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24516 * doc/invoke.texi (-Wclass-memaccess): Document.
24518 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24520 * emit-rtl.h (is_leaf): Update comment about local
24521 register allocator.
24523 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24526 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24527 for a variable to have a section before checking if the section has a
24529 Set section to.persistent if persistent attribute is set.
24530 Warn if .persistent attribute is used on an automatic variable.
24532 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24534 PR rtl-optimization/80474
24535 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24537 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24539 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24542 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24544 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24545 (casesi): Emit bounds checking as RTL.
24546 (casesi_internal_mips16_<mode>): Remove bounds checking.
24548 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24550 * config/xtensa/xtensa.c (xtensa_option_override): Append
24551 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24552 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24553 xtensa_doloop_hooks): Define unconditionally.
24554 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24556 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24557 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24558 for it in xtensa_option_override.
24559 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24560 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24562 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24564 * doc/cppopts.texi: Document '-' special value to -MF.
24566 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24568 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24569 (cortex_a53_fconst): Likewise.
24570 (cortex_a53_fpmul): Likewise.
24571 (cortex_a53_f_load_64): Likewise.
24572 (cortex_a53_f_load_many): Likewise.
24573 (cortex_a53_advsimd_alu): Likewise.
24574 (cortex_a53_advsimd_alu_q): Likewise.
24575 (cortex_a53_advsimd_mul): Likewise.
24576 (cortex_a53_advsimd_mul_q): Likewise.
24577 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24578 Add missing fmul, r2f_cvt and fconst cases.
24580 2017-06-14 Richard Biener <rguenther@suse.de>
24582 PR middle-end/81088
24583 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24585 (fold_binary_loc): When associating do not treat pre-existing
24586 TREE_OVERFLOW on literal constants as a reason to allow
24587 TREE_OVERFLOW on associated literal constants.
24589 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24591 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24592 (MASK_FEATURES): New macro.
24593 * config/sparc/sparc.c (sparc_option_override): Remove the special
24594 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24596 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24598 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24599 a division of 0 if non-call exceptions are enabled.
24601 2017-06-14 Andrew Pinski <apinski@cavium.com>
24602 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24605 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24606 Improve vector initialization code gen for only variable case.
24608 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24610 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24612 2017-06-14 Richard Biener <rguenther@suse.de>
24614 PR tree-optimization/81083
24615 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24618 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24620 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24621 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24622 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24623 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24624 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24625 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24627 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24629 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24630 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24632 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24634 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24636 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24638 * config/rs6000/t-rtems: Don't handle SPE.
24640 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24642 * config/rs6000/t-linux: Don't handle SPE.
24644 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24646 * config/rs6000/eabispe.h: Delete file.
24648 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24650 * config/rs6000/t-spe: Delete file.
24652 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24654 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24655 (rs6000_legitimate_offset_address_p): Return false for anything in
24656 V2SImode or V2SFmode.
24658 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24660 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24661 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24662 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24664 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24665 (rs6000_legitimate_offset_address_p): Ditto.
24666 (rs6000_emit_move): Ditto.
24667 (rs6000_init_builtins): Remove V4HI_type_node.
24669 2017-06-13 Martin Liska <mliska@suse.cz>
24672 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24673 (gate_asan): Likewise.
24674 * asan.h (asan_no_sanitize_address_p): Remove the function.
24675 (sanitize_flags_p): New function.
24676 * builtins.def: Fix coding style.
24677 * common.opt: Use renamed enum value.
24678 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24679 * doc/extend.texi: Document no_sanitize attribute.
24680 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24681 to SANITIZE_UNDEFINED_NONDEFAULT.
24682 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24683 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24684 Use sanitize_flags_p.
24685 * gimplify.c (gimplify_function_tree): Likewise.
24686 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24687 * opts.c (parse_no_sanitize_attribute): New function.
24688 (common_handle_option): Use renamed enum value.
24689 * opts.h (parse_no_sanitize_attribute): Declare.
24690 * tree.c (sanitize_flags_p): New function.
24691 * tree.h: Declared here.
24692 * tsan.c: Use sanitize_flags_p.
24693 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24694 (instrument_mem_ref): Likewise.
24695 (instrument_bool_enum_load): Likewise.
24696 (do_ubsan_in_current_function): Remove the function.
24697 (pass_ubsan::execute): Use sanitize_flags_p.
24698 * ubsan.h: Remove do_ubsan_in_current_function
24699 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24700 (dump_function_to_file): Use it here.
24702 2017-06-13 Martin Jambor <mjambor@suse.cz>
24704 PR tree-optimization/80803
24705 PR tree-optimization/81063
24706 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24707 (propagate_subaccesses_across_link): Enqueue subtree whenever
24708 necessary instead of relying on the caller.
24710 2017-06-13 Martin Jambor <mjambor@suse.cz>
24712 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24713 that have a first_link.
24714 (sort_and_splice_var_accesses): Do not check first_link before
24716 (subtree_mark_written_and_enqueue): Likewise.
24717 (propagate_all_subaccesses): Likewise and do not stop at first
24718 parent with a first_link.
24720 2017-06-13 Martin Jambor <mjambor@suse.cz>
24722 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24725 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24727 * match.pd: New pattern.
24729 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24731 * tree-vrp.c (is_masked_range_test): New function.
24732 (register_edge_assert_for): Determine ranges for
24735 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24737 PR tree-optimization/67328
24738 * fold-const.c (maskable_range_p): New function.
24739 (build_range_check): Generate bittests if possible.
24741 2017-06-13 Martin Liska <mliska@suse.cz>
24743 * gimple-pretty-print.c (dump_probability): Add new argument.
24744 (dump_edge_probability): Dump both probability and count.
24745 (dump_gimple_label): Likewise.
24746 (dump_gimple_bb_header): Likewise.
24748 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24751 * config/avr/avr-devices.c: Fix indentation.
24752 * config/avr/gen-avr-mmcu-specs.c: Dito.
24754 2017-06-13 Richard Biener <rguenther@suse.de>
24756 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24757 instead get vector type from stmt_info.
24758 (vectorizable_reduction): Adjust. Remove dead code.
24760 2017-06-13 Richard Biener <rguenther@suse.de>
24762 PR middle-end/81065
24763 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24764 case of C * (x * C2 + C3).
24765 (fold_addr_of_array_ref_difference): Properly fold index difference.
24767 2017-06-12 David S. Miller <davem@davemloft.net>
24770 * config/sparc/sparc.md (return expander): Emit frame blockage if
24771 function uses alloca.
24773 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24775 * combine.c (make_field_assignment): Check len rather than the mode
24776 precision when calling force_to_mode.
24778 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24780 Support multilibs and devices that see flash in RAM address range.
24783 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24784 (avr_mcu_t) <flash_pm_offset>: New field.
24785 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24786 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24787 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24788 (AVR_TINY_PM_OFFSET): Remove macro.
24789 * config/avr/avr.opt (-mshort-calls): New option.
24790 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24791 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24792 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24793 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24794 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24795 instead of avr_arch->have_jmp_call.
24796 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24797 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24798 avr_arch->flash_pm_offset to define.
24799 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24800 new field flash_pm_offset. Add entry for avrxmega3.
24801 (avr_texinfo): Add entry for avrxmega3.
24802 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24803 attiny212, attiny214,
24804 attiny412, attiny414, attiny416, attiny417,
24805 attiny814, attiny816, attiny817,
24806 attiny1614, attiny1616, attiny1617,
24807 attiny3214, attiny3216, attiny3217.
24808 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24809 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24810 (avr_print_operand_address) [AVR_TINY]: Same.
24811 (avr_asm_init_sections) <readonly_data_section>: Only patch
24812 callback if avr_arch->flash_pm_offset = 0.
24813 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24814 for rodata if avr_arch->flash_pm_offset != 0.
24815 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24816 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24817 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24818 (m_options): Append opt_rcall.
24819 (m_dirnames): Append dir_rcall.
24820 * config/avr/t-multilib: Regenerate.
24822 * configure.ac [target=avr]: Check whether avrxmega3 default
24823 linker description file works as needed.
24824 * configure: Regenerate.
24825 * doc/avr-mmcu.texi: Regenerate.
24826 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24827 <__AVR_ARCH__>: Document avrxmega3 and 103.
24828 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24829 <__AVR_SHORT_CALLS__>: Document it.
24830 <__AVR_PM_BASE_ADDRESS__>: Document it.
24831 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24832 (AVR Variable Attributes) <progmem>: Document this is
24833 not needed for avrxmega3.
24834 (AVR Named Address Spaces) <__flash>: Dito.
24836 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24838 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24840 2017-06-12 Doug Rupp <rupp@adacore.com>
24842 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24843 Append vxworks-stdint.h to the tm_file list.
24844 * config/vxworks-stdint.h: New file.
24846 2017-06-12 Martin Liska <mliska@suse.cz>
24848 PR tree-optimization/81041
24849 * tree-profile.c (gimple_gen_ic_func_profiler):
24850 Create an extra BB in profile-generate
24851 (gimple_gen_time_profiler): Likewise.
24853 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24855 PR tree-optimization/81003
24856 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24857 (update_range_test): Use it instead of force_gimple_operand_gsi.
24859 2017-06-12 Richard Biener <rguenther@suse.de>
24861 PR tree-optimization/81053
24862 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24863 with backedge value not defined in loop. Simplify def stmt
24866 2017-06-11 Tom de Vries <tom@codesourcery.com>
24869 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24871 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24872 nvptx_cannot_force_const_mem.
24874 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24876 * opts.c (finish_options): Move test for flag_split_stack after
24877 it has been initialized.
24879 2017-06-11 Jason Merrill <jason@redhat.com>
24881 * tree.h (id_equal): New.
24882 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24883 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24884 instead of strcmp of IDENTIFIER_POINTER.
24886 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24888 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24889 (mark_all_inlined_calls_cdtor): Fix formating.
24890 (inline_transform): Rescale profile before inlining.
24892 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24894 * cgraph.h (cgraph_edge::clone): Update prototype.
24895 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24896 (cgraph_node::create_clone): Update.
24897 (cgraph_node::create_version_clone): Update.
24898 * tree-inline.c (copy_bb): Update.
24899 (expand_call_inline): Update.
24901 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24903 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24904 factored out from ...
24905 (rs6000_emit_prologue): ... here.
24907 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24909 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
24910 factored out from ...
24911 (rs6000_emit_prologue): ... here.
24913 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24915 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
24917 (handle_missing_profiles): Fix computation of tp_first_run.
24918 (counts_to_freqs): Do not touch freqs when count is 0.
24920 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24922 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
24925 2017-06-10 Tom de Vries <tom@codesourcery.com>
24927 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24928 attributes): Document signal effective target.
24930 2017-06-10 Tom de Vries <tom@codesourcery.com>
24932 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24933 Document effective target stack_size.
24935 2017-06-09 David Malcolm <dmalcolm@redhat.com>
24937 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
24938 to the edit_context if they can be auto-applied.
24940 2017-06-9 Ian Lance Taylor <iant@golang.org>
24942 * opts.c (finish_options): If -fsplit-stack, disable implicit
24943 -forder-blocks-and-partition.
24944 * doc/invoke.texi (Optimize Options): Document that when using
24945 -fsplit-stack -forder-blocks-and-partition is not implicitly
24948 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24950 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
24951 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
24952 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
24953 * builtins.def (abort, trap, unreachable): Declare cold.
24954 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
24955 * tree-core.h (ECF_COLD): New.
24956 * tree.c (set_call_expr_flags): Handle ECF_COLD.
24957 (build_common_builtin_nodes): Mark unreachable and abort as cold.
24959 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24961 * predict.c (unlikely_executed_stmt_p): Cleanup.
24963 2017-06-09 Richard Biener <rguenther@suse.de>
24965 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
24966 model if the ref is always written to.
24968 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24970 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
24972 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24974 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
24977 2017-06-09 Tom de Vries <tom@codesourcery.com>
24980 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
24981 "target cannot support label values" when encountering LABEL_REF.
24983 2017-06-09 Martin Liska <mliska@suse.cz>
24985 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
24986 (gimple_gen_ic_func_profiler): Emit direct comparison
24987 of __gcov_indirect_call_callee with NULL.
24988 (gimple_gen_time_profiler): Change probability from
24989 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
24991 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24993 * profile.c (edge_gcov_counts): Turn to pointer.
24994 (compute_branch_probabilities, compute_branch_probabilities): Update.
24995 (branch_prob): Do not clear edge_gcov_count.
24996 * profile.h (edge_gcov_counts): Turn to pointer.
24997 (edge_gcov_count): Update.
24999 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25001 * gimple.h (gimple_check_failed): Mark cold.
25003 2017-06-09 Richard Biener <rguenther@suse.de>
25005 PR tree-optimization/66623
25006 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25007 refactor check_reduction into two parts, properly computing
25008 whether we have to check reduction validity for outer loop
25011 2017-06-09 Richard Biener <rguenther@suse.de>
25013 PR tree-optimization/79483
25014 * graphite-scop-detection.c (order): New global.
25015 (get_order): Compute bb to order mapping that satisfies code
25016 generation constraints.
25017 (cmp_pbbs): New helper.
25018 (build_scops): Start domwalk at entry block, sort generated
25021 2017-06-09 Richard Biener <rguenther@suse.de>
25023 PR middle-end/81007
25024 * ipa-polymorphic-call.c
25025 (ipa_polymorphic_call_context::restrict_to_inner_class):
25026 Skip FIELD_DECLs with error_mark_node type.
25027 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25030 2017-06-09 Martin Liska <mliska@suse.cz>
25032 * predict.c (struct branch_predictor): New struct.
25033 (test_prediction_value_range): New test.
25034 (predict_c_tests): New function.
25035 * selftest-run-tests.c (selftest::run_tests): Run the function.
25036 * selftest.h: Declare new tests.
25038 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25041 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25042 gen_add3_insn did not fail.
25043 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25044 r0, construct that number in a temporary reg and add that reg to r0.
25045 If asked to put the result in r0 as well, fail.
25047 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25049 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25050 for early expansion of vec_eqv.
25052 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25054 PR middle-end/81005
25055 * ubsan.c (instrument_null): Avoid pointless code temporary.
25056 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25058 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25062 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25065 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25068 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25070 * predict.c (unlikely_executed_edge_p): Move ahead.
25071 (probably_never_executed_edge_p): Use it.
25073 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25075 PR middle-end/79988
25076 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25077 gimple_call_builtin_p call.
25079 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25081 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25082 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25083 rtl_check_failed_type2, rtl_check_failed_code1,
25084 rtl_check_failed_code2, rtl_check_failed_code_mode,
25085 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25086 rtvec_check_failed_bounds, rtl_check_failed_flag,
25087 _fatal_insn_not_found, _fatal_insn): Likewise.
25088 * tree.h (tree_contains_struct_check_failed,
25089 tree_check_failed, tree_not_check_failed,
25090 tree_class_check_failed, tree_range_check_failed,
25091 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25092 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25093 tree_operand_check_failed, omp_clause_check_failed,
25094 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25097 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25099 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25100 flag_branch_probabilities.
25101 * ipa-inline.c (edge_badness): Likewise.
25102 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25103 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25104 * predict.c (maybe_hot_frequency_p): Likewise.
25105 (probably_never_executed): Likewise.
25106 * sched-ebb.c (schedule_ebbs): Likewise.
25107 * sched-rgn.c (find_single_block_region): Likewise.
25108 * tracer.c (tail_duplicate): Likewise.
25110 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25112 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25113 longer requires x_flag_profile_use.
25115 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25117 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25118 instead of flag_reorder_blocks_and_partition.
25119 * dbxout.c (dbxout_function_end): Likewise.
25120 * dwarf2out.c (gen_subprogram_die): Likewise.
25121 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25122 * hw-doloop.c (reorg_loops): Likewise.
25123 * varasm.c (assemble_start_function,
25124 assemble_end_function): Likewise.
25125 (decide_function_section): Do not check for
25126 flag_reorder_blocks_and_partition.
25128 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25130 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25132 (chkp_get_hard_register_fake_addr_expr): Ditto.
25133 (chkp_build_addr_expr): Add check for hard reg case.
25134 (chkp_parse_array_and_component_ref): Ditto.
25135 (chkp_find_bounds_1): Ditto.
25136 (chkp_process_stmt): Don't generate bounds store for
25139 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25141 * predict.c (maybe_hot_bb_p): Do not check profile status.
25142 (maybe_hot_edge_p): Likewise.
25143 (probably_never_executed): Check for zero counts even if profile
25145 (unlikely_executed_edge_p): New function.
25146 (unlikely_executed_stmt_p): New function.
25147 (unlikely_executed_bb_p): New function.
25148 (set_even_probabilities): Use unlikely predicates.
25149 (combine_predictions_for_bb): Likewise.
25150 (predict_paths_for_bb): Likewise.
25151 (predict_paths_leading_to_edge): Likewise.
25152 (determine_unlikely_bbs): New function.
25153 (estimate_bb_frequencies): Use it.
25154 (compute_function_frequency): Use zero counts even if profile is
25156 * profile-count.h: Fix typo.
25158 2017-08-08 Julia Koval <julia.koval@intel.com>
25160 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25161 _mm512_mask_cvtsepi16_storeu_epi8,
25162 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25163 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25164 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25165 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25166 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25167 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25168 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25169 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25170 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25171 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25172 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25173 __builtin_ia32_pmovuswb256mem_mask,
25174 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25175 __builtin_ia32_pmovwb512mem_mask): New builtins.
25177 2017-08-08 Julia Koval <julia.koval@intel.com>
25179 PR target/73350,80862
25180 * config/i386/subst.md (round): Fix round pattern.
25181 * config/i386/i386.c (ix86_erase_embedded_rounding):
25182 Fix erasing rounding for the fixed pattern.
25184 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25186 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25188 2017-06-08 Martin Liska <mliska@suse.cz>
25190 PR gcov-profile/80911
25191 * gcov.c (block_info::block_info): New constructor.
25193 2017-06-07 Carl Love <cel@us.ibm.com>
25195 * config/rs6000/rs6000-c: The return type of the following
25196 built-in functions was implemented as int not long long. Fix sign
25197 of return value for the unsigned version of vec_mulo and vec_mule.
25198 vector unsigned long long vec_bperm (vector unsigned long long,
25199 vector unsigned char)
25200 vector signed long long vec_mule (vector signed int,
25202 vector unsigned long long vec_mule (vector unsigned int,
25203 vector unsigned int)
25204 vector signed long long vec_mulo (vector signed int,
25206 vector unsigned long long vec_mulo (vector unsigned int,
25207 vector unsigned int)
25208 * doc/extend.texi: Fix the documentation for the built-in
25211 2017-06-07 Carl Love <cel@us.ibm.com>
25214 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25217 2017-06-07 Carl Love <cel@us.ibm.com>
25219 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25220 support, Generate doublehv for signed int/float for BE case only.
25222 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25224 * doc/invoke.texi (mcx16): Rewrite.
25226 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25228 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25229 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25230 *mov<mode>_softfloat, and an anonymous splitter): Use
25231 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25233 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25235 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25237 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25238 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25239 (rs6000_debug_reg_global): Adjust.
25240 (rs6000_init_hard_regno_mode_ok): Adjust.
25241 (rs6000_dbx_register_number): Adjust.
25242 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25243 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25244 Remove SPE_ACC and SPEFSCR.
25245 (REG_ALLOC_ORDER): Ditto.
25246 (FRAME_POINTER_REGNUM): Change to 111.
25247 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25248 (REG_CLASS_NAMES): Ditto.
25249 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25250 (REGISTER_NAMES): Ditto.
25251 (ADDITIONAL_REG_NAMES): Ditto.
25252 (rs6000_reg_names): Ditto.
25253 * config/rs6000/rs6000.md: Renumber some register number
25256 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25258 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25260 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25261 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25263 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25264 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25265 Delete the SPE high registers.
25266 (REG_ALLOC_ORDER): Ditto.
25267 (enum reg_class): Remove SPE_HIGH_REGS.
25268 (REG_CLASS_NAMES): Ditto.
25269 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25270 (REGISTER_NAMES): Ditto.
25271 (rs6000_reg_names): Ditto.
25272 * doc/tm.texi.in: Remove SPE as example.
25273 * doc/tm.texi: Regenerate.
25275 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25277 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25278 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25279 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25280 * config/rs6000/rs6000.md (type): Remove "brinc".
25282 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25284 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25285 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25286 * config/rs6000/linuxspe.h: Delete file.
25287 * config/rs6000/rs6000.md: Don't include spe.md.
25288 * config/rs6000/spe.h: Delete file.
25289 * config/rs6000/spe.md: Delete file.
25290 * config/rs6000/t-rs6000: Remove spe.md.
25292 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25294 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25295 (reg_or_none500mem_operand): Delete.
25296 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25297 instead of reg_or_none500mem_operand.
25299 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25301 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25302 handling of SPE flags.
25303 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25305 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25307 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25309 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25310 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25311 paired_divv2sf3): Similar.
25312 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25313 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25314 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25316 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25317 Rename the paired_* instruction patterns.
25318 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25320 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25321 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25322 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25323 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25324 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25325 PAIRED_VECTOR_MODE.
25326 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25327 (spe_func_has_64bit_regs_p): Delete.
25328 (spe_expand_predicate_builtin): Delete.
25329 (spe_expand_evsel_builtin): Delete.
25330 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25331 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25332 (invalid_e500_subreg): Delete.
25333 (rs6000_legitimize_address): Always force_reg op2 as well, for
25334 paired single memory accesses.
25335 (rs6000_member_type_forces_blk): Delete.
25336 (rs6000_spe_function_arg): Delete.
25337 (rs6000_expand_unop_builtin): Delete SPE handling.
25338 (rs6000_expand_binop_builtin): Ditto.
25339 (spe_expand_stv_builtin): Delete.
25340 (bdesc_2arg_spe): Delete.
25341 (spe_expand_builtin): Delete.
25342 (spe_expand_predicate_builtin): Delete.
25343 (spe_expand_evsel_builtin): Delete.
25344 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25345 (spe_init_builtins): Delete.
25346 (spe_func_has_64bit_regs_p): Delete.
25347 (savres_routine_name): Delete "info" parameter. Adjust callers.
25348 (rs6000_emit_stack_reset): Ditto.
25349 (rs6000_dwarf_register_span): Delete.
25350 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25351 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25352 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25354 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25356 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25357 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25358 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25359 mulv2sf3, divv2sf3): Delete expanders.
25361 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25363 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25365 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25367 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25368 * config/rs6000/rs6000.c: Ditto.
25370 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25372 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25373 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25374 comparison_operator.
25376 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25378 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25379 * config/rs6000/rs6000.opt: Ditto.
25380 * config/rs6000/t-rtems: Ditto.
25382 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25384 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25385 TARGET_E500_SINGLE by 0, simplify.
25386 * config/rs6000/rs6000.c: Ditto.
25387 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25388 (spe_build_register_parallel): Delete.
25389 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25390 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25391 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25392 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25393 (E500_CONVERT): Delete.
25394 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25396 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25398 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25399 * config/rs6000/dfp.md: Ditto.
25400 (negdd2, *negdd2_fpr): Merge.
25401 (absdd2, *absdd2_fpr): Merge.
25402 (negtd2, *negtd2_fpr): Merge.
25403 (abstd2, *abstd2_fpr): Merge.
25404 * config/rs6000/e500.h: Delete file.
25405 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25406 TARGET_FPRS by 1 and simplify.
25407 * config/rs6000/rs6000-c.c: Ditto.
25408 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25409 TARGET_DF_SPE by 0.
25410 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25412 * config/rs6000/rs6000.md: Ditto.
25413 (floatdidf2, *floatdidf2_fpr): Merge.
25414 (move_from_CR_gt_bit): Delete.
25415 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25416 (E500_CR_IOR_COMPARE): Delete.
25417 (All patterns that require !TARGET_FPRS): Delete.
25418 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25420 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25422 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25424 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25426 * graphds.c (add_edge): Intitialize edge's attached data.
25427 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25428 pointer parameter. Call pointed function on each edge during
25429 graph traversing. Skip traversing the edge when the function
25431 (graphds_dfs, graphds_scc): Ditto.
25432 (for_each_edge): New parameter. Pass the new parameter to callback
25434 * graphds.h (skip_edge_callback): New function pointer type.
25435 (graphds_dfs, graphds_scc): New function pointer parameter.
25436 (graphds_edge_callback, for_each_edge): New parameter.
25438 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25440 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25441 out code checking if runtime alias check is possible to below ...
25442 Call the new function.
25443 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25444 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25446 2017-06-07 Marek Polacek <polacek@redhat.com>
25449 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25450 TYPE_OVERFLOW_WRAPS check.
25452 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25454 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25455 if versioning is required.
25456 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25457 peeling with the check for versioning.
25459 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25461 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25462 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25463 Set true to new parameter if new ssa variable is defined.
25464 (vect_gen_vector_loop_niters): Refactor. Set range information
25465 for the new vector loop bound variable.
25466 (vect_do_peeling): Ditto.
25468 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25470 * tree-affine.c (ssa.h): Include header file.
25471 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25472 has wrapping overflow behavior.
25474 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25476 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25478 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25480 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25481 (tree_to_aff_combination): ... here.
25483 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25485 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25486 reg_pressure model function.
25487 (ivopts_global_cost_for_size): Delete.
25488 (determine_set_costs, iv_ca_recount_cost): Call new model function
25489 ivopts_estimate_reg_pressure.
25491 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25493 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25494 expensive than udiv. Remove floating point cases from mod.
25496 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25498 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25499 Increase idiv cost.
25501 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25503 * config/aarch64/aarch64.md
25504 (copysignsf3): Fix mask generation.
25506 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25508 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25510 (class dump_manager): Add register_dumps method.
25511 * dumpfile.c: Include langhooks.h.
25512 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25513 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25514 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25515 (dump_manager::dump_register): Start with 512 entries instead of 32.
25516 (dump_manager::register_dumps): New method.
25517 * toplev.c (general_init): Instead of invoking register_dumps
25518 langhook, invoke register_dumps method on the dump manager.
25519 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25522 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25524 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25525 Say that # requires an associated define_split to exist, and that
25526 the define_split must be suitable for use after register allocation.
25528 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25530 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25531 (compute_outgoing_frequencies): Also initialize zero counts.
25532 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25533 around loops; preserve more of profile when nothing changes.
25535 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25537 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25539 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25540 * config/arm/arm-cpu-cdata.h: Regenerate.
25541 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25542 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25543 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25544 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25545 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25547 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25549 * config/arm/t-rmprofile: Likewise.
25550 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25552 2017-06-06 David S. Miller <davem@davemloft.net>
25555 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25556 blockage if function uses alloca.
25558 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25560 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25561 New "uid" fields to hold pretty-print IDs of group and ref.
25562 Memory references are now identified as <group_id>:<ref_id>
25563 instead of using [random] addresses.
25564 (dump_mem_details): Simplify, no functional change.
25565 (dump_mem_ref): Simplify and make output more concise.
25566 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25567 (find_or_create_group): Initialize group uid.
25568 (record_ref): Initialize ref uid. Improve debug output.
25569 (prune_group_by_reuse, should_issue_prefetch_p,)
25570 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25571 (mark_nontemporal_store, determine_loop_nest_reuse):
25572 Improve debug output.
25574 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25576 * dbgcnt.def (prefetch): New debug counter.
25577 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25578 (schedule_prefetches): Stop issueing prefetches if debug counter
25581 2017-06-06 Tom de Vries <tom@codesourcery.com>
25583 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25584 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25586 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25588 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25589 Use aarch64_reg_or_zero predicate for operand 4.
25590 (aarch64_compare_and_swap<mode> define_insn_and_split):
25591 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25592 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25594 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25596 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25597 (arm_compute_save_core_reg_mask): This.
25598 (thumb1_compute_save_reg_mask): Rename into ...
25599 (thumb1_compute_save_core_reg_mask): This.
25600 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25601 (arm_compute_frame_layout): Likewise.
25603 2017-06-06 Richard Biener <rguenther@suse.de>
25605 PR tree-optimization/80974
25606 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25607 keep or clear leaders SSA info.
25609 2017-06-06 Tom de Vries <tom@codesourcery.com>
25611 * config/nvptx/nvptx.c (split_mode_p): New function.
25612 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25614 2017-06-06 Tom de Vries <tom@codesourcery.com>
25616 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25618 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25621 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25624 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25626 * shrink-wrap.c (handle_simple_exit): Update profile.
25627 (try_shrink_wrapping): Upate profile.
25629 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25631 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25632 (tree_guess_outgoing_edge_probabilities): New.
25633 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25634 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25636 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25638 * ipa-split.c (split_function): Initialize return bb profile.
25640 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25642 * profile.c (compute_branch_probabilities): Also initialize
25643 EXIT_BLOCK profile.
25645 2017-06-06 Richard Biener <rguenther@suse.de>
25647 PR tree-optimization/80928
25648 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25649 (vect_analyze_loop_operations): Properly guard analysis for
25651 (vect_transform_loop): Likewise.
25652 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25653 (vect_model_induction_cost): Do not cost for pure SLP.
25654 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25655 of induction in inner loop vectorization.
25656 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25657 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25658 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25660 (vect_analyze_slp_cost_1): Cost induction.
25661 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25662 (vect_get_slp_vect_defs): Likewise.
25663 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25664 (vect_transform_stmt): Handle SLP reductions.
25665 * tree-vectorizer.h (vectorizable_induction): Adjust.
25667 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25669 * config/rs6000/rs6000.c (make_resolver_func): Update
25670 init_lowered_empty_function call.
25672 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25674 * doc/invoke.texi: Document the -fprofile-abs-path option.
25675 * common.opt (fprofile-abs-path): New option.
25676 * gcov-io.h (gcov_write_filename): Declare.
25677 * gcov-io.c (gcov_write_filename): New function.
25678 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25679 * profile.c (output_location): Likewise.
25681 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25683 * shring-wrap.c: Revert accidental commit.
25685 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25687 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25689 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25691 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25693 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25694 profile in callgraph edge.
25695 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25696 (apply_scale): Likewise.
25697 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25698 Also scale profile when inlining function with zero profile.
25699 (initialize_cfun): Update exit block profile even when it is zero.
25700 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25701 when profile is read.
25703 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25705 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25706 (CLONE_*): New constants to define the processors we can generate
25707 code for with the target_clone attribute.
25708 (rs6000_clone_map): New array to identify which clone processors
25709 the current program is running on.
25710 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25711 target_clone attribute.
25712 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25713 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25714 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25715 (cpu_expand_builtin): Add support for target_clone attribute.
25716 (rs6000_valid_attribute_p): Allow "default" attribute.
25717 (get_decl_name): New debug function to simplify printing the
25718 current function name in debugging statements.
25719 (rs6000_clone_priority): New functions to support the target_clone
25720 attribute, and be able to generate code to switch between ISA 2.05
25721 through ISA 3.0 (power6 through power9).
25722 (rs6000_compare_version_priority): Likewise.
25723 (rs6000_get_function_versions_dispatcher): Likewise.
25724 (make_resolver_func): Likewise.
25725 (add_condition_to_bb): Likewise.
25726 (dispatch_function_versions): Likewise.
25727 (rs6000_generate_version_dispatcher_body): Likewise.
25728 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25729 (fusion_gpr_load_p): Fix a spacing issue.
25730 * doc/extend.texi (Common Function Attributes): Document that the
25731 PowerPC supports the target_clone attribute.
25733 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25735 * config/arm/arm.h: explain F symbol found in description of ARM
25736 register allocation in its legend.
25738 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25740 * config/mips/frame-header-opt.c: Include profile-count.h.
25741 * config/riscv/riscv.c: Include profile-count.h
25743 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25745 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25747 (sm_set_flag_if_changed): Add bbs field.
25748 (execute_sm_if_changed_flag_set): Pass BBS.
25749 (execute_sm): Update.
25751 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25753 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25756 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25758 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25759 (peephole2): New peephole2 to emit the above.
25760 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25762 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25764 * config/aarch64/aarch64.c (define_peephole2 above
25765 *sub_<shift>_<mode>): New peephole.
25767 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25769 * config/i386/i386.c (make_resolver_func): Update.
25770 * Makefile.in: Add profile-count.h and profile-count.o
25771 * auto-profile.c (afdo_indirect_call): Update to new API.
25772 (afdo_set_bb_count): Update.
25773 (afdo_propagate_edge): Update.
25774 (afdo_propagate_circuit): Update.
25775 (afdo_calculate_branch_prob): Update.
25776 (afdo_annotate_cfg): Update.
25777 * basic-block.h: Include profile-count.h
25778 (struct edge_def): Turn count to profile_count.
25779 (struct basic_block_def): Likewie.
25780 (REG_BR_PROB_BASE): Move to profile-count.h
25781 (RDIV): Move to profile-count.h
25782 * bb-reorder.c (max_entry_count): Turn to profile_count.
25783 (find_traces): Update.
25784 (rotate_loop):Update.
25785 (connect_traces):Update.
25786 (sanitize_hot_paths):Update.
25787 * bt-load.c (migrate_btr_defs): Update.
25788 * cfg.c (RDIV): Remove.
25789 (init_flow): Use alloc_block.
25790 (alloc_block): Uninitialize count.
25791 (unchecked_make_edge): Uninitialize count.
25792 (check_bb_profile): Update.
25793 (dump_edge_info): Update.
25794 (dump_bb_info): Update.
25795 (update_bb_profile_for_threading): Update.
25796 (scale_bbs_frequencies_int): Update.
25797 (scale_bbs_frequencies_gcov_type): Update.
25798 (scale_bbs_frequencies_profile_count): New.
25799 * cfg.h (update_bb_profile_for_threading): Update.
25800 (scale_bbs_frequencies_profile_count): Declare.
25801 * cfgbuild.c (compute_outgoing_frequencies): Update.
25802 (find_many_sub_basic_blocks): Update.
25803 * cfgcleanup.c (try_forward_edges): Update.
25804 (try_crossjump_to_edge): Update.
25805 * cfgexpand.c (expand_gimple_tailcall): Update.
25806 (construct_exit_block): Update.
25807 * cfghooks.c (verify_flow_info): Update.
25808 (dump_bb_for_graph): Update.
25809 (split_edge): Update.
25810 (make_forwarder_block): Update.
25811 (duplicate_block): Update.
25812 (account_profile_record): Update.
25813 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25814 (get_estimated_loop_iterations): Update.
25815 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25816 (single_likely_exit): Update.
25817 * cfgloopmanip.c (scale_loop_profile): Update.
25819 (set_zero_probability): Update.
25820 (lv_adjust_loop_entry_edge): Update.
25821 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25822 (purge_dead_edges): Update.
25823 (rtl_account_profile_record): Update.
25824 * cgraph.c (cgraph_node::create): Uninitialize count.
25825 (symbol_table::create_edge): Uninitialize count.
25826 (cgraph_update_edges_for_call_stmt_node): Update.
25827 (cgraph_edge::dump_edge_flags): Update.
25828 (cgraph_node::dump): Update.
25829 (cgraph_edge::maybe_hot_p): Update.
25830 * cgraph.h: Include profile-count.h
25831 (create_clone), create_edge, create_indirect_edge): Update.
25832 (cgraph_node): Turn count to profile_count.
25833 (cgraph_edge0: Likewise.
25834 (make_speculative, clone): Update.
25835 (create_edge): Update.
25836 (init_lowered_empty_function): Update.
25837 * cgraphclones.c (cgraph_edge::clone): Update.
25838 (duplicate_thunk_for_node): Update.
25839 (cgraph_node::create_clone): Update.
25840 * cgraphunit.c (cgraph_node::analyze): Update.
25841 (cgraph_node::expand_thunk): Update.
25842 * final.c (dump_basic_block_info): Update.
25843 * gimple-streamer-in.c (input_bb): Update.
25844 * gimple-streamer-out.c (output_bb): Update.
25845 * graphite.c (print_global_statistics): Update.
25846 (print_graphite_scop_statistics): Update.
25847 * hsa-brig.c: Include basic-block.h.
25848 * hsa-dump.c: Include basic-block.h.
25849 * hsa-gen.c (T sum_slice): Update.
25850 (convert_switch_statements):Update.
25851 * hsa-regalloc.c: Include basic-block.h.
25852 * ipa-chkp.c (chkp_produce_thunks): Update.
25853 * ipa-cp.c (struct caller_statistics): Update.
25854 (init_caller_stats): Update.
25855 (gather_caller_stats): Update.
25856 (ipcp_cloning_candidate_p): Update.
25857 (good_cloning_opportunity_p): Update.
25858 (get_info_about_necessary_edges): Update.
25859 (dump_profile_updates): Update.
25860 (update_profiling_info): Update.
25861 (update_specialized_profile): Update.
25862 (perhaps_add_new_callers): Update.
25863 (decide_about_value): Update.
25864 (ipa_cp_c_finalize): Update.
25865 * ipa-devirt.c (struct odr_type_warn_count): Update.
25866 (struct decl_warn_count): Update.
25867 (struct final_warning_record): Update.
25868 (possible_polymorphic_call_targets): Update.
25869 (ipa_devirt): Update.
25870 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25871 * ipa-icf.c (sem_function::merge): Update.
25872 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25873 * ipa-inline.c (compute_uninlined_call_time): Update.
25874 (compute_inlined_call_time): Update.
25875 (want_inline_small_function_p): Update.
25876 (want_inline_self_recursive_call_p): Update.
25877 (edge_badness): Update.
25878 (lookup_recursive_calls): Update.
25879 (recursive_inlining): Update.
25880 (inline_small_functions): Update.
25881 (dump_overall_stats): Update.
25882 (dump_inline_stats): Update.
25883 * ipa-profile.c (ipa_profile_generate_summary): Update.
25884 (ipa_propagate_frequency): Update.
25885 (ipa_profile): Update.
25886 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25887 * ipa-utils.c (ipa_merge_profiles): Update.
25888 * loop-doloop.c (doloop_modify): Update.
25889 * loop-unroll.c (report_unroll): Update.
25890 (unroll_loop_runtime_iterations): Update.
25891 * lto-cgraph.c (lto_output_edge): Update.
25892 (lto_output_node): Update.
25893 (input_node): Update.
25894 (input_edge): Update.
25895 (merge_profile_summaries): Update.
25896 * lto-streamer-in.c (input_cfg): Update.
25897 * lto-streamer-out.c (output_cfg): Update.
25898 * mcf.c (create_fixup_graph): Update.
25899 (adjust_cfg_counts): Update.
25900 (sum_edge_counts): Update.
25901 * modulo-sched.c (sms_schedule): Update.
25902 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25903 * predict.c (maybe_hot_count_p): Update.
25904 (probably_never_executed): Update.
25905 (dump_prediction): Update.
25906 (combine_predictions_for_bb): Update.
25907 (propagate_freq): Update.
25908 (handle_missing_profiles): Update.
25909 (counts_to_freqs): Update.
25910 (rebuild_frequencies): Update.
25911 (force_edge_cold): Update.
25912 * predict.h: Include profile-count.h
25913 (maybe_hot_count_p, counts_to_freqs): UPdate.
25914 * print-rtl-function.c: Do not include cfg.h
25915 * print-rtl.c: Include basic-block.h
25916 * profile-count.c: New file.
25917 * profile-count.h: New file.
25918 * profile.c (is_edge_inconsistent): Update.
25919 (correct_negative_edge_counts): Update.
25920 (is_inconsistent): Update.
25921 (set_bb_counts): Update.
25922 (read_profile_edge_counts): Update.
25923 (compute_frequency_overlap): Update.
25924 (compute_branch_probabilities): Update; Initialize and deinitialize
25926 (branch_prob): Update.
25927 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
25928 (edge_gcov_count): New.
25929 (bb_gcov_count): New.
25930 * shrink-wrap.c (try_shrink_wrapping): Update.
25931 * tracer.c (better_p): Update.
25932 * trans-mem.c (expand_transaction): Update.
25933 (ipa_tm_insert_irr_call): Update.
25934 (ipa_tm_insert_gettmclone_call): Update.
25935 * tree-call-cdce.c: Update.
25936 * tree-cfg.c (gimple_duplicate_sese_region): Update.
25937 (gimple_duplicate_sese_tail): Update.
25938 (gimple_account_profile_record): Update.
25939 (execute_fixup_cfg): Update.
25940 * tree-inline.c (copy_bb): Update.
25941 (copy_edges_for_bb): Update.
25942 (initialize_cfun): Update.
25943 (freqs_to_counts): Update.
25944 (copy_cfg_body): Update.
25945 (expand_call_inline): Update.
25946 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25947 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
25948 (try_unroll_loop_completely): Update.
25949 (try_peel_loop): Update.
25950 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25951 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
25952 * tree-ssa-loop-split.c (connect_loops): Update.
25953 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
25954 * tree-ssa-reassoc.c (branch_fixup): Update.
25955 * tree-ssa-tail-merge.c (replace_block_by): Update.
25956 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
25957 (compute_path_counts): Update.
25958 (update_profile): Update.
25959 (recompute_probabilities): Update.
25960 (update_joiner_offpath_counts): Update.
25961 (estimated_freqs_path): Update.
25962 (freqs_to_counts_path): Update.
25963 (clear_counts_path): Update.
25964 (ssa_fix_duplicate_block_edges): Update.
25965 (duplicate_thread_path): Update.
25966 * tree-switch-conversion.c (case_bit_test_cmp): Update.
25967 (struct switch_conv_info): Update.
25968 * tree-tailcall.c (decrease_profile): Update.
25969 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25970 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25971 * value-prof.c (check_counter): Update.
25972 (gimple_divmod_fixed_value): Update.
25973 (gimple_mod_pow2): Update.
25974 (gimple_mod_subtract): Update.
25975 (gimple_ic_transform): Update.
25976 (gimple_stringop_fixed_value): Update.
25977 * value-prof.h (gimple_ic): Update.
25979 2017-06-02 Carl Love <cel@us.ibm.com>
25981 * config/rs6000/rs6000-c: Add support for built-in functions
25982 vector double vec_doublee (vector signed int);
25983 vector double vec_doublee (vector unsigned int);
25984 vector double vec_doublee (vector float);
25985 vector double vec_doubleh (vector signed int);
25986 vector double vec_doubleh (vector unsigned int);
25987 vector double vec_doubleh (vector float);
25988 vector double vec_doublel (vector signed int);
25989 vector double vec_doublel (vector unsigned int);
25990 vector double vec_doublel (vector float);
25991 vector double vec_doubleo (vector signed int);
25992 vector double vec_doubleo (vector unsigned int);
25993 vector double vec_doubleo (vector float);.
25994 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
25995 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
25997 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
25998 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
25999 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26001 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26002 vec_doublel, vec_doubleh.
26003 * doc/extend.texi: Update the built-in documentation file for the
26004 new built-in functions.
26006 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26009 * ipa-inline-analysis.c (free_growth_caches): Set
26010 edge_removal_hook_holder to NULL after removing it.
26012 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26014 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26015 comparision with zero.
26017 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26018 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26019 for early expansion of vec_min and vec_max builtins.
26020 (builtin_function_type): Add min/max unsigned variants to those
26021 identified as having unsigned arguments.
26023 2017-06-02 Olivier Hainque <hainque@adacore.com>
26025 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26027 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26029 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26030 Use VALL_F16 iterator rather than VALL.
26032 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26034 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26035 Emit CBNZ inside loop when doing a strong exchange and comparing
26036 against zero. Generate the CC flags after the loop.
26038 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26040 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26041 (dl_section_ref): New.
26042 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26043 On AIX, append an expression to subtract the size of the
26044 section length to dl_section_ref.
26046 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26048 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26049 for early expansion of vector absolute builtins.
26051 2017-06-02 Richard Biener <rguenther@suse.de>
26053 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26054 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26056 2017-06-02 Richard Biener <rguenther@suse.de>
26058 PR tree-optimization/80948
26059 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26060 stmt order as well.
26062 2017-06-02 Richard Biener <rguenther@suse.de>
26064 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26066 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26067 for inductions as relevant.
26069 2017-06-02 Richard Biener <rguenther@suse.de>
26071 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26072 (vectorizable_induction): ... this. Remove dead code.
26074 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26076 * builtins. (expand_builtin_alloca): Remove second parameter and
26077 infer its value from the first parameter instead.
26078 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26080 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26082 PR rtl-optimization/80903
26083 * loop-doloop.c (add_test): Unshare sequence.
26085 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26087 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26089 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26091 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26093 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26094 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26095 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26096 (xlogue_layout::get_stub_rtx): Make static.
26097 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26098 (xlogue_layout::compute_stub_managed_regs): Rename to...
26099 (xlogue_layout::count_stub_managed_regs): ...this.
26100 (xlogue_layout::is_stub_managed_reg): New function.
26101 (xlogue_layout::m_stub_names): Rename to...
26102 (xlogue_layout::s_stub_names): ...this, make static.
26103 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26104 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26105 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26106 xlogue_layout::s_stub_names): Instantiate statics.
26107 (stub_managed_regs): Remove.
26108 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26109 (disable_call_ms2sysv_xlogues): Rename to...
26110 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26111 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26113 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26114 change after reload_completed.
26115 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26117 (ix86_expand_prologue): Likewise.
26118 (ix86_expand_epilogue): Likewise.
26119 (ix86_expand_split_stack_prologue): Likewise.
26120 (ix86_compute_frame_layout): Remove frame parameter ...
26121 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26122 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26124 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26125 (ix86_frame): Move from here ...
26126 * config/i386/i386.h (ix86_frame): ... to here.
26127 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26128 complete ix86_frame data structure instead. Remove some_ld_name.
26130 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26132 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26133 symbols that hold a DECL_VALUE_EXPR.
26135 2017-06-01 Martin Jambor <mjambor@suse.cz>
26137 PR tree-optimization/80898
26138 * tree-sra.c (process_subtree_disqualification): Removed.
26139 (disqualify_candidate): Do not acll
26140 process_subtree_disqualification.
26141 (subtree_mark_written_and_enqueue): New function.
26142 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26143 RHS has been disqualified and re-queue LHS if necessary. Apart
26144 from that, ignore disqualified RHS.
26146 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26148 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26149 address fetch for z10 or later.
26151 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26153 * config/arc/arc.md (tst_movb): Add guard when splitting.
26155 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26157 * config/arc/arc.c (arc_can_eliminate): Test against
26158 arc_frame_pointer_needed.
26160 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26162 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26163 to prevent store reordering.
26164 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26165 (type): Add block type.
26166 (stack_tie): Define special instruction to be used in
26169 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26171 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26172 constraint. It is not valid for the pattern.
26173 (noncommutative_binary_comparison): Likewise.
26175 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26177 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26180 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26182 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26183 be used by the reg-alloc.
26185 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26187 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26188 reg-alloc when having mul64 or mul32x16 instructions.
26189 (mulsidi3): Likewise.
26190 (umulsidi3): Likewise.
26191 (mulsi32x16): New pattern.
26192 (mulsi64): Likewise.
26193 (mulsidi64): Likewise.
26194 (umulsidi64): Likewise.
26195 (MUL32x16_REG): Define.
26196 (mul64_600): Use MUL32x16_REG.
26197 (mac64_600): Likewise.
26198 (umul64_600): Likewise.
26199 (umac64_600): Likewise.
26201 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26203 * config/arc/arc.md (mulsi3_700): Make it commutative.
26205 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26207 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26209 (*sign_extendsidi2_insn): Likewise for movstosw.
26211 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26213 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26214 the type of the input discriminant value. Convert the
26215 discriminant value of signedness vary.
26217 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26219 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26220 Add to -Wall section.
26222 2017-06-01 Richard Biener <rguenther@suse.de>
26224 PR middle-end/66313
26225 * fold-const.c (fold_plusminus_mult_expr): If the factored
26226 factor may be zero use a wrapping type for the inner operation.
26227 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26228 and handle moved defs.
26229 (process_assignment): Properly guard the unary op case. Return a
26230 tri-state indicating that moving the stmt before the call may allow
26231 to continue. Pass through to_move.
26232 (find_tail_calls): Handle moving unrelated defs before
26235 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26238 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26239 splitter result in the canonical way.
26241 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26243 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26244 also for 32bit target. Update insn attributes.
26245 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26247 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26249 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26250 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26251 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26252 (_mm_maskz_min_ss): New intrinsics.
26254 2017-05-31 Martin Liska <mliska@suse.cz>
26256 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26257 Change comment style to one we normally use.
26258 (vectorizable_reduction): Likewise.
26259 (vectorizable_induction): Likewise.
26260 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26261 (vectorizable_call): Likewise.
26262 (vectorizable_simd_clone_call): Likewise.
26263 (vectorizable_conversion): Likewise.
26264 (vectorizable_assignment): Likewise.
26265 (vectorizable_shift): Likewise.
26266 (vectorizable_operation): Likewise.
26267 (vectorizable_store): Likewise.
26268 (vectorizable_load): Likewise.
26269 * tree-vectorizer.h: Likewise.
26271 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26273 * passes.c (emergency_dump_function): New.
26274 * tree-pass.h (emergency_dump_function): Declare.
26275 * plugin.c (plugins_internal_error_function): Remove.
26276 * plugin.h (plugins_internal_error_function): Remove declaration.
26277 * toplev.c (internal_error_function): New static function. Use it...
26278 (general_init): ...here.
26280 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26282 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26283 (arc_rtx_costs): Add costs for new patterns.
26284 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26285 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26287 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26289 * tree-ssa-strlen.c (get_next_strinfo): New function.
26290 (get_stridx_plus_constant): Use it.
26291 (zero_length_string): Likewise.
26292 (adjust_related_strinfos): Likewise.
26293 (adjust_last_stmt): Likewise.
26295 2017-05-31 Richard Biener <rguenther@suse.de>
26298 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26299 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26301 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26303 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26304 loop_vinfo argument and use of dependence distance vectors.
26305 Check instead whether the two references differ only in their
26306 initial value and assume that they have the same alignment if the
26307 difference is a multiple of the vector alignment.
26308 (vect_analyze_data_refs_alignment): Update call accordingly.
26310 2017-05-31 Martin Liska <mliska@suse.cz>
26313 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26315 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26317 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26318 (create_intersect_range_checks): Move from ...
26319 * tree-data-ref.c (create_intersect_range_checks_index)
26320 (create_intersect_range_checks): ... to here.
26321 (create_runtime_alias_checks): New function factored from ...
26322 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26323 here. Call above function.
26324 * tree-data-ref.h (create_runtime_alias_checks): New function.
26326 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26328 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26329 segment length for dr_b and compute it in wide_int.
26331 2017-05-31 Richard Biener <rguenther@suse.de>
26333 PR tree-optimization/80906
26334 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26335 and pass through iv_map.
26336 (copy_bb_and_scalar_dependences): Adjust.
26337 (translate_pending_phi_nodes): Likewise.
26338 (copy_loop_close_phi_args): Handle code-generating IVs instead
26341 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26343 * diagnostic-color.c (color_dict): Add "type-diff".
26344 (parse_gcc_colors): Update comment.
26345 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26346 -fdiagnostics-show-template-tree and -fno-elide-type.
26347 (GCC_COLORS): Add type-diff to example.
26349 (-fdiagnostics-show-template-tree): New.
26350 (-fno-elide-type): New.
26351 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26352 the pp_format_decoder callback. Call any m_format_postprocessor's
26354 (pretty_printer::pretty_printer): Initialize
26355 m_format_postprocessor.
26356 (pretty_printer::~pretty_printer): Delete any
26357 m_format_postprocessor.
26358 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26359 (class format_postprocessor): New class.
26360 (struct pretty_printer::format_decoder): Document the new parameters.
26361 (struct pretty_printer::m_format_postprocessor): New field.
26362 * tree-diagnostic.c (default_tree_printer): Update for new
26363 bool and const char ** params.
26364 * tree-diagnostic.h (default_tree_printer): Likewise.
26366 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26368 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26369 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26370 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26371 rs6000_gen_cell_microcode code.
26372 (rs6000_final_prescan_insn): Delete.
26373 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26374 "warn-cell-microcode" entries.
26375 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26376 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26377 throughout. Change cc_reg_not_micro_cr0_operand to
26378 cc_reg_not_cr0_operand throughout.
26379 (*extendhi<mode>2_noload): Delete.
26380 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26381 (mwarn-cell-microcode): Delete.
26382 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26383 -mgen-cell-microcode and -mwarn-cell-microcode.
26385 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26388 * config/i386/constraints.md (Yd): New constraint.
26390 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26391 and (?Yd, r) alternatives. Update insn attributes.
26392 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26393 and (?*Yd, r) alternatives. Update insn attributes.
26394 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26396 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26398 * gimplify.c (gimplify_modify_expr): Don't create a
26399 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26402 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26404 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26406 2017-05-30 Richard Biener <rguenther@suse.de>
26408 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26409 and reduc_def fields.
26410 (STMT_VINFO_REDUC_TYPE): New define.
26411 (STMT_VINFO_REDUC_DEF): Likewise.
26412 (vect_force_simple_reduction): Adjust prototype.
26413 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26414 (vect_is_simple_reduction): Remove check_reduction argument.
26415 (vect_force_simple_reduction): Adjust and set
26416 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26417 (vectorizable_reduction): Do not re-do reduction analysis
26418 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26419 * tree-parloops.c (gather_scalar_reductions): Adjust.
26421 2017-05-30 Richard Biener <rguenther@suse.de>
26423 PR middle-end/80901
26424 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26427 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26429 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26430 Introduce unknown_misalignment parameter and remove vf.
26431 (vect_peeling_hash_get_lowest_cost):
26432 Pass unknown_misalignment parameter.
26433 (vect_enhance_data_refs_alignment):
26434 Fix unsupportable data ref treatment.
26436 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26438 * tree-vect-data-refs.c (vect_get_data_access_cost):
26439 Workaround for SLP handling.
26440 (vect_enhance_data_refs_alignment):
26441 Compute costs for doing no peeling at all, compare to the best
26442 peeling costs so far and avoid peeling if cheaper.
26444 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26446 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26447 Return peeling info and set costs to zero for unlimited cost
26449 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26450 with unknown misalignment. Compute and costs for unknown
26451 misalignment, compare them to the costs for known misalignment
26452 and choose the cheapest for peeling.
26454 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26456 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26457 (vect_get_peeling_costs_all_drs): Create function.
26458 (vect_peeling_hash_get_lowest_cost):
26459 Use vect_get_peeling_costs_all_drs.
26460 (vect_peeling_supportable): Create function.
26461 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26463 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26465 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26466 DR_HAS_NEGATIVE_STEP.
26467 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26468 (vect_enhance_data_refs_alignment): Use.
26469 (vect_duplicate_ssa_name_ptr_info): Use.
26470 * tree-vectorizer.h (dr_misalignment): Use.
26471 (known_alignment_for_access_p): Use.
26473 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26476 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26478 (has_section_name): New function.
26480 2017-05-30 Martin Liska <mliska@suse.cz>
26483 * auto-profile.c (get_function_decl_from_block): Fix
26486 2017-05-30 Richard Biener <rguenther@suse.de>
26488 PR middle-end/80876
26489 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26491 2017-05-30 Martin Liska <mliska@suse.cz>
26493 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26494 * dumpfile.h (struct dump_file_info): Remove ctors.
26496 2017-05-30 Martin Liska <mliska@suse.cz>
26498 * predict.def: Fix GNU coding style.
26500 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26502 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26503 Mark 'to' argument with ATTRIBUTE_UNUSED.
26505 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26507 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26508 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26509 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26512 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26514 * doc/install.texi (Options specification): Restore entry of
26515 --enable-sjlj-exceptions.
26517 2017-05-27 Michael Eager <eager@eagercon.com>
26520 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26522 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26524 * config/microblaze/microblaze.h
26525 (FIXED_REGISTERS): Update in macro.
26526 (CALL_USED_REGISTERS): Update in macro.
26528 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26530 * doc/install.texi: Add links to macOS binary distributions.
26532 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26536 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26538 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26540 2017-05-26 Martin Liska <mliska@suse.cz>
26542 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26544 2017-05-26 Martin Liska <mliska@suse.cz>
26546 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26547 always leading ';; '.
26548 (dump_bb_info): Likewise.
26549 (brief_dump_cfg): Likewise.
26550 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26551 * dumpfile.c: Remove usage of TDF_VERBOSE.
26552 * dumpfile.h (enum dump_kind): Likewise.
26553 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26554 * print-tree.c (debug_verbose): Remove.
26555 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26556 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26557 * tree-diagnostic.c (default_tree_printer): Replace
26558 TDF_DIAGNOSTIC with TDF_SLIM.
26560 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26562 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26563 in parameter loop, rather than loop_vinfo.
26564 (create_intersect_range_checks): Ditto.
26565 (vect_create_cond_for_alias_checks): Update call to above functions.
26567 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26569 PR tree-optimization/80815
26570 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26571 for merging runtime alias checks. Handle negative DR_STEPs.
26573 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26575 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26577 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26578 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26579 out code pruning runtime alias checks.
26580 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26581 factored out from above.
26582 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26584 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26586 (prune_runtime_alias_test_list): New decalaration.
26588 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26590 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26591 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26592 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26593 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26594 (operator==, comp_dr_with_seg_len_pair): Ditto.
26595 (vect_prune_runtime_alias_test_list): Ditto.
26597 2017-05-26 Martin Liska <mliska@suse.cz>
26600 * params.def: Bound partial-inlining-entry-probability param.
26602 2017-05-26 Marek Polacek <polacek@redhat.com>
26605 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26608 2017-05-26 Richard Biener <rguenther@suse.de>
26610 PR tree-optimization/80842
26611 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26614 2017-05-26 Richard Biener <rguenther@suse.de>
26616 PR tree-optimization/80844
26617 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26619 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26621 * doc/md.texi (Machine Constraints): Update x86 family
26622 machine constraints section to match 'config/i386/constraints.md'.
26624 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26626 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26628 2017-05-25 Nathan Sidwell <nathan@acm.org>
26630 * doc/invoke.texi (--enable-languages): Update documentation.
26632 2017-05-25 Martin Liska <mliska@suse.cz>
26634 * dumpfile.c: Add TDF_FOLDING.
26635 * dumpfile.h (enum dump_kind): Likewise.
26636 * genmatch.c (dt_simplify::gen_1): Use it.
26638 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26640 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26642 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26644 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26645 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26647 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26649 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26650 * match.pd (X == C): Rewrite it here.
26651 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26652 with_certain_nonzero_bits2): New predicates.
26653 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26655 2017-05-24 Nathan Sidwell <nathan@acm.org>
26657 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26660 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26663 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26665 * config/powerpcspe: New port. Files are copied from the rs6000
26666 port, with "rs6000" in filenames replaced by "powerpcspe".
26668 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26670 PR rtl-optimization/80754
26671 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26673 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26675 * config/sparc/sparc.md (length): Return the correct value for -mflat
26676 sibcalls to match output_sibcall.
26678 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26682 * config/rs6000/rs6000.c (struct machine_function): Add new field
26684 (rs6000_get_separate_components): Init that field, use it.
26685 (rs6000_components_for_bb): Use the field.
26687 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26689 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26691 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26693 PR middle-end/80823
26694 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26696 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26699 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26700 address against address_operand predicate.
26701 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26703 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26705 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26706 parameters passed indirectly.
26708 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26710 * config/i386/i386.md (*movdi_internal): Remove SSE4
26711 alternative 18 (?r, *v). Update insn attributes.
26712 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26713 Update insn attributes.
26714 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26715 Update insn attributes.
26716 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26717 alternative 1 (r, v). Remove isa attribute.
26718 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26719 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26720 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26722 2017-05-23 Tom de Vries <tom@codesourcery.com>
26724 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26727 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26729 * cgraphunit.c (symbol_table::process_new_functions): Update.
26730 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26731 (inline_generate_summary): Rename to ...
26732 (ipa_fn_summary_generate): ... this one.
26733 (inline_read_summary): Rename to ...
26734 (ipa_fn_summary_read): ... this one.
26735 (inline_write_summary): Rename to ...
26736 (ipa_fn_summary_write): ... this one.
26737 (inline_free_summary): Rename to ...
26738 (ipa_free_fn_summary): ... this one.
26739 (pass_data_local_fn_summary, pass_local_fn_summary,
26740 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26741 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26742 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26743 make_pass_ipa_fn_summary): New.
26744 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26745 inline_write_summary, inline_free_summary): Remove.
26746 (ipa_free_fn_summary) : New.
26747 * ipa-inline.c (ipa_inline): Update.
26748 (pass_ipa_inline): Do not generate summaries.
26749 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26751 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26752 and add pass_ipa_fn_summary.
26753 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26755 (make_pass_inline_parameters): Remove.
26757 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26759 * omp-low.c (struct omp_context): Remove "default_kind" member.
26762 * omp-offload.c (execute_oacc_device_lower): Remove the
26763 parallelism dimensions function attributes for unparallelized
26764 OpenACC kernels constructs.
26766 2017-05-23 Martin Liska <mliska@suse.cz>
26768 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26770 (cgraph_edge::make_speculative): Likewise.
26771 (cgraph_edge::resolve_speculation): Likewise.
26772 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26773 (cgraph_node::dump): Likewise.
26774 * cgraph.h: Likewise.
26775 * cgraphunit.c (analyze_functions): Likewise.
26776 (symbol_table::compile): Likewise.
26777 * ipa-cp.c (print_all_lattices): Likewise.
26778 (determine_versionability): Likewise.
26779 (initialize_node_lattices): Likewise.
26780 (ipcp_verify_propagated_values): Likewise.
26781 (estimate_local_effects): Likewise.
26782 (update_profiling_info): Likewise.
26783 (create_specialized_node): Likewise.
26784 (perhaps_add_new_callers): Likewise.
26785 (decide_about_value): Likewise.
26786 (decide_whether_version_node): Likewise.
26787 (identify_dead_nodes): Likewise.
26788 (ipcp_store_bits_results): Likewise.
26789 * ipa-devirt.c (dump_targets): Likewise.
26790 (ipa_devirt): Likewise.
26791 * ipa-icf.c (sem_item::dump): Likewise.
26792 (sem_function::equals): Likewise.
26793 (sem_variable::equals): Likewise.
26794 (sem_item_optimizer::read_section): Likewise.
26795 (sem_item_optimizer::execute): Likewise.
26796 (congruence_class::dump): Likewise.
26797 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26798 (dump_inline_summary): Likewise.
26799 (estimate_node_size_and_time): Likewise.
26800 (inline_analyze_function): Likewise.
26801 * ipa-inline-transform.c (inline_call): Likewise.
26802 * ipa-inline.c (report_inline_failed_reason): Likewise.
26803 (want_early_inline_function_p): Likewise.
26804 (edge_badness): Likewise.
26805 (update_edge_key): Likewise.
26806 (inline_small_functions): Likewise.
26807 * ipa-profile.c (ipa_profile): Likewise.
26808 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26809 (ipa_make_edge_direct_to_target): Likewise.
26810 (remove_described_reference): Likewise.
26811 (ipa_impossible_devirt_target): Likewise.
26812 (propagate_controlled_uses): Likewise.
26813 (ipa_print_node_params): Likewise.
26814 (ipcp_transform_function): Likewise.
26815 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26816 (propagate_pure_const): Likewise.
26817 * ipa-reference.c (generate_summary): Likewise.
26818 (read_write_all_from_decl): Likewise.
26819 (propagate): Likewise.
26820 (ipa_reference_read_optimization_summary): Likewise.
26821 * ipa-utils.c (ipa_merge_profiles): Likewise.
26822 * ipa.c (walk_polymorphic_call_targets): Likewise.
26823 (symbol_table::remove_unreachable_nodes): Likewise.
26824 (ipa_single_use): Likewise.
26825 * passes.c (execute_todo): Likewise.
26826 * predict.c (drop_profile): Likewise.
26827 * symtab.c (symtab_node::get_dump_name): New function.
26828 (symtab_node::dump_name): Likewise.
26829 (symtab_node::dump_asm_name): Likewise.
26830 (symtab_node::dump_references): Likewise.
26831 (symtab_node::dump_referring): Likewise.
26832 (symtab_node::dump_base): Likewise.
26833 (symtab_node::debug_symtab): Likewise.
26834 * tree-sra.c (convert_callers_for_node): Likewise.
26835 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26836 * value-prof.c (init_node_map): Likewise.
26838 2017-05-23 Martin Liska <mliska@suse.cz>
26840 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26841 and symtab_node::debug_symtab to symbol_table::debug.
26842 * cgraphunit.c (analyze_functions): Use the renamed function.
26843 (symbol_table::compile): Likewise.
26844 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26845 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26846 * passes.c (execute_todo): Likewise.
26847 * symtab.c (symbol_table::dump): New function.
26848 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26850 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26852 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26853 that nonconst implies exec.
26855 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26857 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26858 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26859 (inline_edge_summary_vec): Turn into ...
26860 (ipa_call_summaries): ... this one.
26861 (redirect_to_unreachable, edge_set_predicate,
26862 evaluate_properties_for_edge, inline_summary_alloc,
26863 reset_ipa_call_summary, reset_inline_summary,
26864 inline_summary_t::duplicate): Update.
26865 (inline_edge_duplication_hook): Turn to ...
26866 (ipa_call_summary_t::duplicate): ... this one.
26867 (inline_edge_removal_hook): Turn to ...
26868 (ipa_call_summary_t::remove): ... this one.
26869 (dump_inline_edge_summary): Turn to ...
26870 (dump_ipa_call_summary): ... this one.
26871 (estimate_function_body_sizes): Update.
26872 (inline_update_callee_summaries): Update.
26873 (remap_edge_change_prob): Update.
26874 (remap_edge_summaries): Update.
26875 (inline_merge_summary): Update.
26876 (do_estimate_edge_time): Update.
26877 (inline_generate_summary): Update.
26878 (inline_read_section): Update.
26879 (inline_read_summary): Update.
26880 (inline_free_summary): Update.
26881 * ipa-inline.c (can_inline_edge_p): Update.
26882 (compute_inlined_call_time): Update.
26883 (want_inline_small_function_p): Update.
26884 (edge_badness): Update.
26885 (early_inliner): Update.
26886 * ipa-inline.h (inline_edge_summary): Turn to ...
26887 (ipa_call_summary): ... this one.
26888 (ipa_call_summary_t): New class.
26889 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26890 (ipa_call_summaries): New.
26891 (inline_edge_summary): Remove.
26892 (estimate_edge_growth): Update.
26893 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26894 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26895 * ipa-split.c (execute_split_functions): Update.
26896 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26898 2017-05-23 Tom de Vries <tom@codesourcery.com>
26900 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26901 attributes): Document rdrand effective target.
26903 2017-05-23 Tom de Vries <tom@codesourcery.com>
26905 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26906 attributes): Sort alphabetically.
26908 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
26910 * config/avr/genmultilib.awk: Use gsub instead of gensub.
26912 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26915 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
26916 V2DF/V2DI splat into two separate patterns, one that handles
26917 registers, and the other that only handles memory. Drop support
26918 for splatting from a GPR on ISA 2.07 and then splitting the
26919 splat into direct move and splat.
26920 (vsx_splat_<mode>_reg): Likewise.
26921 (vsx_splat_<mode>_mem): Likewise.
26923 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
26925 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
26927 2017-05-22 Jakub Jelinek <jakub@redhat.com>
26929 PR middle-end/80809
26930 * omp-low.c (finish_taskreg_remap): New function.
26931 (finish_taskreg_scan): If unit size of ctx->record_type
26932 is non-constant, unshare the size expression and replace
26933 decls in it with possible outer var refs.
26935 PR middle-end/80809
26936 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
26937 GOVD_SHARED rather than GOVD_PRIVATE with it.
26938 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
26939 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
26941 PR middle-end/80853
26942 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
26943 as last argument to build_outer_var_ref for pointer bases of array
26944 section reductions.
26946 2017-05-19 Martin Sebor <msebor@redhat.com>
26948 * print-tree.c (print_node): Print DECL_READ_P flag.
26950 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26952 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
26953 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
26954 * cgraph.c: Likewise.
26955 * cgraphunit.c: Likewise.
26956 * gengtype.c: Likewise.
26957 * ipa-cp.c: Likewise.
26958 * ipa-devirt.c: Likewise.
26959 * ipa-icf.c: Likewise.
26960 * ipa-predicate.c: Likewise.
26961 * ipa-profile.c: Likewise.
26962 * ipa-prop.c: Likewise.
26963 * ipa-split.c: Likewise.
26965 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
26966 edge_predicate_pool, dump_inline_hints,
26967 inline_summary::account_size_time, redirect_to_unreachable,
26968 edge_set_predicate, set_hint_predicate,
26969 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
26970 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
26971 inline_summary_t::remove, remap_hint_predicate_after_duplication,
26972 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
26973 ipa_call_summary_t::remove, initialize_growth_caches,
26974 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
26975 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
26976 mark_modified, unmodified_parm_1, unmodified_parm,
26977 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
26978 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
26979 compute_bb_predicates, will_be_nonconstant_expr_predicate,
26980 will_be_nonconstant_predicate, record_modified_bb_info,
26981 get_minimal_bb, record_modified, param_change_prob,
26982 phi_result_unknown_predicate, predicate_for_phi_result,
26983 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
26984 estimate_function_body_sizes, compute_inline_parameters,
26985 compute_inline_parameters_for_curren, pass_data_inline_parameters,
26986 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
26987 inline_update_callee_summaries, remap_edge_change_prob,
26988 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
26989 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
26990 inline_analyze_function, inline_summary_t::insert,
26991 inline_generate_summary, read_ipa_call_summary, inline_read_section,
26992 inline_read_summary, write_ipa_call_summary, inline_write_summary,
26993 inline_free_summary): Move to ipa-fnsummary.h
26994 (predicate_t): Remove.
26995 * ipa-fnsummary.c: New file.
26996 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
26997 (enum inline_hints_vals, inline_hints, agg_position_info,
26998 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
26999 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27000 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27001 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27002 inline_read_summary, inline_write_summary, inline_free_summary,
27003 inline_analyze_function, initialize_inline_failed,
27004 inline_merge_summary, inline_update_overall_summary,
27005 compute_inline_parameters): Move to ipa-fnsummary.h
27006 * ipa-fnsummary.h: New file.
27007 * ipa-inline-transform.h: Include ipa-inline.h.
27008 * ipa-inline.c: LIkewise.
27010 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27012 * ipa-inline.c (edge_badness): Use inlined_time instead of
27013 inline_summaries->get.
27015 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27017 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27019 2017-05-22 Nathan Sidwell <nathan@acm.org>
27021 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27022 (fdump-lang): Document 'raw' option.
27023 * dumpfile.h (TDI_tu): Delete.
27024 * dumpfile.c (dump_files): Remove translation-unit.
27025 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27027 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27029 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27030 command option from $(AWK) call.
27031 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27032 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27033 [FORMAT]: Remove handling of variable.
27034 * config/avr/t-multilib: Regenerate.
27036 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27038 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27040 (dump_inline_summary): Do not print self_time.
27041 (estimate_function_body_sizes): Do not set self_time.
27042 (compute_inline_parameters): Likewise.
27043 (inline_read_section, inline_write_summary): Do not stream self_time.
27044 * ipa-inline.h (inline_summary): Drop self_time.
27046 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27048 * ipa-inline-analysis.c (account_size_time): Rename to ...
27049 (inline_summary::account_size_time): ... this one.
27050 (reset_ipa_call_summary): Turn to ...
27051 (ipa_call_summary::reset): ... this one.
27052 (reset_inline_summary): Turn to ...
27053 (inline_summary::reset): ... this one.
27054 (inline_summary_t::remove): Update.
27055 (inline_summary_t::duplicate): Update.
27056 (ipa_call_summary_t::remove): Update.
27057 (dump_inline_summary): Update.
27058 (estimate_function_body_sizes): Update.
27059 (compute_inline_parameters): Update.
27060 (estimate_node_size_and_time): Update.
27061 (inline_merge_summary): Update.
27062 (inline_update_overall_summary): Update.
27063 (inline_read_section): Update.
27064 (inline_write_summary): Update.
27065 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27066 add account_size_time and reset member functions.
27067 (ipa_call_summary): Add reset function.
27068 * ipa-predicate.h (predicate::operator &): Constify.
27070 2017-05-22 Richard Biener <rguenther@suse.de>
27072 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27074 2017-05-19 Jason Merrill <jason@redhat.com>
27076 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27078 2017-05-19 Marek Polacek <polacek@redhat.com>
27081 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27082 TYPE_OVERFLOW_WRAPS checks.
27084 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27086 * tree-core.h (enum omp_clause_default_kind): Add
27087 "OMP_CLAUSE_DEFAULT_PRESENT".
27088 * tree-pretty-print.c (dump_omp_clause): Handle it.
27089 * gimplify.c (enum gimplify_omp_var_data): Add
27090 "GOVD_MAP_FORCE_PRESENT".
27091 (gimplify_adjust_omp_clauses_1): Map it to
27092 "GOMP_MAP_FORCE_PRESENT".
27093 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27095 * gimplify.c (oacc_default_clause): Clarify.
27097 2017-05-19 Nathan Sidwell <nathan@acm.org>
27099 LANG_HOOK_REGISTER_DUMPS
27100 * toplev.c (general_init): Call register dump lang hook.
27101 * doc/invoke.texi: Document -fdump-lang option family.
27102 * dumpfile.c (dump_files): Remove class dump here.
27103 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27104 * dumpfile.h (tree_dump_index): Remove TDI_class.
27105 * langhooks-def.h (lhd_register_dumps): Declare.
27106 (LANG_HOOKS_REGISTER_DUMPS): Define.
27107 (LANG_HOOKS_INITIALIZER): Add it.
27108 * langhooks.c (lhd_register_dumps): Define.
27109 * langhooks.h (struct lang_hooks): Add register_dumps.
27111 2017-05-19 Nathan Sidwell <nathan@acm.org>
27113 * context.h (context::set_passes): New.
27114 * context.c (context::context): Do not create pass manager.
27115 * toplev.c (general_init): Create pass manager here.
27117 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27119 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27120 use this splitter if two add or or instructions would also work for
27121 the constant we want to generate.
27123 2017-05-19 Richard Biener <rguenther@suse.de>
27126 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27127 predicate evaluation.
27129 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27131 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27133 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27136 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27138 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27139 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27140 (inline_edge_summary_vec): Turn into ...
27141 (ipa_call_summaries): ... this one.
27142 (redirect_to_unreachable, edge_set_predicate,
27143 evaluate_properties_for_edge, inline_summary_alloc,
27144 reset_ipa_call_summary, reset_inline_summary,
27145 inline_summary_t::duplicate): Update.
27146 (inline_edge_duplication_hook): Turn to ...
27147 (ipa_call_summary_t::duplicate): ... this one.
27148 (inline_edge_removal_hook): Turn to ...
27149 (ipa_call_summary_t::remove): ... this one.
27150 (dump_inline_edge_summary): Turn to ...
27151 (dump_ipa_call_summary): ... this one.
27152 (estimate_function_body_sizes): Update.
27153 (inline_update_callee_summaries): Update.
27154 (remap_edge_change_prob): Update.
27155 (remap_edge_summaries): Update.
27156 (inline_merge_summary): Update.
27157 (do_estimate_edge_time): Update.
27158 (inline_generate_summary): Update.
27159 (inline_read_section): Update.
27160 (inline_read_summary): Update.
27161 (inline_free_summary): Update.
27162 * ipa-inline.c (can_inline_edge_p): Update.
27163 (compute_inlined_call_time): Update.
27164 (want_inline_small_function_p): Update.
27165 (edge_badness): Update.
27166 (early_inliner): Update.
27167 * ipa-inline.h (inline_edge_summary): Turn to ...
27168 (ipa_call_summary): ... this one.
27169 (ipa_call_summary_t): New class.
27170 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27171 (ipa_call_summaries): New.
27172 (inline_edge_summary): Remove.
27173 (estimate_edge_growth): Update.
27174 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27175 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27176 * ipa-split.c (execute_split_functions): Update.
27177 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27179 2017-05-19 Richard Biener <rguenther@suse.de>
27181 PR middle-end/80764
27182 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27184 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27186 * config/rs6000/rs6000.c (struct machine_function): Add field
27187 fpr_is_wrapped_separately.
27188 (rs6000_get_separate_components): Use 64 components. Handle the
27189 new FPR components.
27190 (rs6000_components_for_bb): Handle the FPR components.
27191 (rs6000_emit_prologue_components): Handle the FPR components.
27192 (rs6000_emit_epilogue_components): Handle the FPR components.
27193 (rs6000_set_handled_components): Handle the FPR components.
27194 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27195 that are already separately shrink-wrapped.
27196 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27197 that are already separately shrink-wrapped.
27199 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27202 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27205 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27206 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27207 cases where the register allocator uses a move and an offsettable
27208 memory operation to/from a FPR register on ISA 2.06/2.07.
27209 (define_peephole2 for Altivec d-form store): Likewise.
27211 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27214 * config/i386/mmx.md (*mov<mode>_internal): Enable
27215 alternatives 11, 12, 13 and 14 also for 32bit targets.
27216 Remove alternatives 15, 16, 17 and 18.
27217 * config/i386/sse.md (vec_concatv2di): Change
27218 alternative (!x, *y) to (x, ?!*Yn).
27220 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27222 * dumpfile.h (enum dump_kind): Remove stray comma.
27224 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27226 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27227 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27228 predicate::num_conditions
27229 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27230 (CHANGED): turn into predicate::changed.
27231 (agg_position_info): Move to ipa-predicate.h
27232 (add_condition, predicate::add_clause, predicate::operator &=,
27233 predicate::or_with, predicate::evaluate, predicate::probability,
27234 dump_condition, dump_clause, predicate::dump,
27235 predicate::remap_after_duplication, predicate::remap_after_inlining,
27236 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27237 (evaluate_conditions_for_known_args): Update.
27238 (set_cond_stmt_execution_predicate): Update.
27239 * ipa-inline.h: Include ipa-predicate.h
27240 (condition, inline_param_summary, conditions, agg_position_info,
27241 predicate): Move to ipa-predicate.h
27242 * ipa-predicate.c: New file.
27243 * ipa-predicate.h: New file.
27245 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27247 * final.c (leaf_function_p): Check we are not in a sequence.
27249 2017-05-18 Martin Liska <mliska@suse.cz>
27251 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27252 * dumpfile.c (dump_register): Use new enum dump_kind.
27253 (get_dump_file_name): Likewise.
27254 (dump_enable_all): Likewise.
27255 (dump_switch_p_1): Likewise.
27256 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27257 * dumpfile.h (enum dump_kind): New enum type.
27258 (struct dump_file_info): Create constructor and
27259 format fields and comments.
27260 * passes.c (pass_manager::register_one_dump_file):
27262 * statistics.c (statistics_early_init): Likewise.
27263 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27264 TDF_TREE with TDF_SLIM.
27265 (gather_memory_references_ref): Likewise.
27267 2017-05-18 Martin Liska <mliska@suse.cz>
27269 * vec.h (struct vnull): Use it.
27271 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27273 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27274 (true_predicate, false_predicate, true_predicate_p,
27275 false_predicate_p): Remove.
27276 (single_cond_predicate, not_inlined_predicate): Turn to member function
27278 (add_condition): Update.
27279 (add_clause): Turn to...
27280 (predicate::add_clause): ... this one; update; allow passing NULL
27282 (and_predicates): Turn to ...
27283 (predicate::operator &=): ... this one.
27284 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27285 (or_predicates): Turn to ...
27286 (predicate::or_with): ... this one.
27287 (evaluate_predicate): Turn to ...
27288 (predicate::evaluate): ... this one.
27289 (predicate_probability): Turn to ...
27290 (predicate::probability): ... this one.
27291 (dump_condition): Update.
27292 (dump_predicate): Turn to ...
27293 (predicate::dump): ... this one.
27294 (account_size_time): Update.
27295 (edge_set_predicate): Update.
27296 (set_hint_predicate): UPdate.
27297 (evaluate_conditions_for_known_args): Update.
27298 (evaluate_properties_for_edge): Update.
27299 (remap_predicate_after_duplication): Turn to...
27300 (predicate::remap_after_duplication): ... this one.
27301 (remap_hint_predicate_after_duplication): Update.
27302 (inline_summary_t::duplicate): UPdate.
27303 (dump_inline_edge_summary): Update.
27304 (dump_inline_summary): Update.
27305 (set_cond_stmt_execution_predicate): Update.
27306 (set_switch_stmt_execution_predicate): Update.
27307 (compute_bb_predicates): Update.
27308 (will_be_nonconstant_expr_predicate): Update.
27309 (will_be_nonconstant_predicate): Update.
27310 (phi_result_unknown_predicate): Update.
27311 (predicate_for_phi_result): Update.
27312 (array_index_predicate): Update.
27313 (estimate_function_body_sizes): Update.
27314 (estimate_node_size_and_time): Update.
27315 (estimate_ipcp_clone_size_and_time): Update.
27316 (remap_predicate): Rename to ...
27317 (predicate::remap_after_inlining): ... this one.
27318 (remap_hint_predicate): Update.
27319 (inline_merge_summary): Update.
27320 (inline_update_overall_summary): Update.
27321 (estimate_size_after_inlining): Update.
27322 (read_predicate): Rename to ...
27323 (predicate::stream_in): ... this one.
27324 (read_inline_edge_summary): Update.
27325 (write_predicate): Rename to ...
27326 (predicate::stream_out): ... this one.
27327 (write_inline_edge_summary): Update.
27328 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27329 (clause_t): Turn to uint32_t
27330 (predicate): Turn to class; implement constructor and operators
27332 (size_time_entry): Update.
27333 (inline_summary): Update.
27334 (inline_edge_summary): Update.
27336 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27338 * fold-const.c (fold_binary_loc): Move transformation...
27339 * match.pd (C - X CMP X): ... here.
27341 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27343 * config/sparc/sparc.c (sparc_option_override): Set function
27344 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27345 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27347 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27349 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27351 2017-05-18 Marek Polacek <polacek@redhat.com>
27354 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27355 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27356 gimple_assign_load_p.
27358 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27360 PR middle-end/80692
27361 * real.c (do_compare): Give decimal_do_compare preference over
27362 comparing just the signs.
27364 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27366 * doc/md.texi (Canonicalization of Instructions): Describe the
27367 canonical form of instructions that inherently set a condition
27370 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27372 PR middle-end/80775
27373 * tree-cfg.c: Move deletion of unreachable case statements to after
27374 the merging of consecutive case labels.
27376 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27378 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27379 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27380 restoring of callee-saved registers.
27382 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27384 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27385 * config/visium/visium.c (single_set_and_flags): Likewise.
27386 * config/visium/visium.md (Substitutions): Likewise.
27388 2017-05-17 Martin Liska <mliska@suse.cz>
27390 * cfg.c: Introduce dump_flags_t type and
27391 use it instead of int type.
27393 * cfghooks.c: Likewise.
27394 * cfghooks.h (struct cfg_hooks): Likewise.
27395 * cfgrtl.c: Likewise.
27396 * cfgrtl.h: Likewise.
27397 * cgraph.c (cgraph_node::get_body): Likewise.
27398 * coretypes.h: Likewise.
27399 * domwalk.c: Likewise.
27400 * domwalk.h: Likewise.
27401 * dumpfile.c (struct dump_option_value_info): Likewise.
27402 (dump_enable_all): Likewise.
27403 (dump_switch_p_1): Likewise.
27404 (opt_info_switch_p): Likewise.
27405 * dumpfile.h (enum tree_dump_index): Likewise.
27406 (struct dump_file_info): Likewise.
27407 * genemit.c: Likewise.
27408 * generic-match-head.c: Likewise.
27409 * gengtype.c (open_base_files): Likewise.
27410 * gimple-pretty-print.c: Likewise.
27411 * gimple-pretty-print.h: Likewise.
27412 * graph.c (print_graph_cfg): Likewise.
27413 * graphite-scop-detection.c (dot_all_sese): Likewise.
27414 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27415 * loop-unroll.c (report_unroll): Likewise.
27416 * passes.c (pass_manager::register_one_dump_file): Likewise.
27417 * print-tree.c: Likewise.
27418 * statistics.c: Likewise.
27419 * tree-cfg.c: Likewise.
27420 * tree-cfg.h: Likewise.
27421 * tree-dfa.c: Likewise.
27422 * tree-dfa.h: Likewise.
27423 * tree-dump.c (dump_function): Likewise.
27424 * tree-dump.h (struct dump_info): Likewise.
27425 * tree-pretty-print.c: Likewise.
27426 * tree-pretty-print.h: Likewise.
27427 * tree-ssa-live.c: Likewise.
27428 * tree-ssa-live.h: Likewise.
27429 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27430 * tree-vect-loop.c: Likewise.
27431 * tree-vect-slp.c: Likewise.
27433 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27434 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27436 PR tree-optimization/80457
27437 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27438 of all arguments to a statement as scalar_to_vec operations.
27439 (vectorizable_call): Adjust call to vect_model_simple_cost for
27441 (vectorizable_conversion): Likewise.
27442 (vectorizable_assignment): Likewise.
27443 (vectorizable_shift): Likewise.
27444 (vectorizable_operation): Likewise.
27445 (vectorizable_comparison): Likewise.
27446 (vect_is_simple_cond): Record the def types for operands.
27447 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27448 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27449 for statement argument count.
27451 2017-05-16 Carl Love <cel@us.ibm.com>
27453 * config/rs6000/rs6000-c: Add support for built-in functions
27454 vector unsigned long long vec_bperm (vector unsigned long long,
27455 vector unsigned char)
27456 vector signed long long vec_mule (vector signed int,
27458 vector unsigned long long vec_mule (vector unsigned int,
27459 vector unsigned int)
27460 vector signed long long vec_mulo (vector signed int,
27462 vector unsigned long long vec_mulo (vector unsigned int,
27463 vector unsigned int)
27464 vector signed char vec_sldw (vector signed char,
27465 vector signed char,
27467 vector unsigned char vec_sldw (vector unsigned char,
27468 vector unsigned char,
27470 vector signed short vec_sldw (vector signed short,
27471 vector signed short,
27473 vector unsigned short vec_sldw (vector unsigned short,
27474 vector unsigned short,
27476 vector signed int vec_sldw (vector signed int,
27479 vector unsigned int vec_sldw (vector unsigned int,
27480 vector unsigned int,
27482 vector signed long long vec_sldw (vector signed long long,
27483 vector signed long long,
27485 vector unsigned long long vec_sldw (vector unsigned long long,
27486 vector unsigned long long,
27488 * config/rs6000/rs6000-c: Add support for built-in functions
27489 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27490 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27491 * doc/extend.texi: Update the built-in documentation for the
27492 new built-in functions.
27494 2017-05-16 Marek Polacek <polacek@redhat.com>
27498 * tree.c (save_expr): Don't fold the expression.
27500 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27502 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27503 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27504 and (?*y,m). Update insn attributes.
27506 2017-05-16 Martin Liska <mliska@suse.cz>
27508 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27509 flags argument of print_gimple_stmt, print_gimple_expr,
27510 print_generic_stmt and print_generic_expr.
27511 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27512 * coretypes.h: Likewise.
27513 * except.c (dump_eh_tree): Likewise.
27514 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27515 * gimple-pretty-print.h: Likewise.
27516 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27517 (backprop::push_to_worklist): Likewise.
27518 (backprop::pop_from_worklist): Likewise.
27519 (backprop::process_use): Likewise.
27520 (backprop::intersect_uses): Likewise.
27521 (note_replacement): Likewise.
27522 * gimple-ssa-store-merging.c
27523 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27524 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27525 (pass_store_merging::execute): Likewise.
27526 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27527 (ssa_base_cand_dump_callback): Likewise.
27528 (dump_incr_vec): Likewise.
27529 (replace_refs): Likewise.
27530 (replace_mult_candidate): Likewise.
27531 (create_add_on_incoming_edge): Likewise.
27532 (create_phi_basis): Likewise.
27533 (insert_initializers): Likewise.
27534 (all_phi_incrs_profitable): Likewise.
27535 (introduce_cast_before_cand): Likewise.
27536 (replace_one_candidate): Likewise.
27537 * gimplify.c (gimplify_expr): Likewise.
27538 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27539 (set_rename): Likewise.
27540 (rename_uses): Likewise.
27541 (copy_loop_phi_nodes): Likewise.
27542 (add_close_phis_to_merge_points): Likewise.
27543 (copy_loop_close_phi_args): Likewise.
27544 (copy_cond_phi_args): Likewise.
27545 (graphite_copy_stmts_from_block): Likewise.
27546 (translate_pending_phi_nodes): Likewise.
27547 * graphite-poly.c (print_pdr): Likewise.
27548 (dump_gbb_cases): Likewise.
27549 (dump_gbb_conditions): Likewise.
27550 (print_scop_params): Likewise.
27551 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27552 (build_cross_bb_scalars_use): Likewise.
27553 (gather_bbs::before_dom_children): Likewise.
27554 * hsa-dump.c (dump_hsa_immed): Likewise.
27555 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27556 (get_replacement_map): Likewise.
27557 * ipa-inline-analysis.c (dump_condition): Likewise.
27558 (estimate_function_body_sizes): Likewise.
27559 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27560 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27561 * ipa-prop.c (ipa_dump_param): Likewise.
27562 (ipa_print_node_jump_functions_for_edge): Likewise.
27563 (ipa_modify_call_arguments): Likewise.
27564 (ipa_modify_expr): Likewise.
27565 (ipa_dump_param_adjustments): Likewise.
27566 (ipa_dump_agg_replacement_values): Likewise.
27567 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27568 * ipa-pure-const.c (check_stmt): Likewise.
27569 (pass_nothrow::execute): Likewise.
27570 * ipa-split.c (execute_split_functions): Likewise.
27571 * omp-offload.c (dump_oacc_loop_part): Likewise.
27572 (dump_oacc_loop): Likewise.
27573 * trans-mem.c (tm_log_emit): Likewise.
27574 (tm_memopt_accumulate_memops): Likewise.
27575 (dump_tm_memopt_set): Likewise.
27576 (dump_tm_memopt_transform): Likewise.
27577 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27578 (print_loop): Likewise.
27579 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27580 (chkp_gather_checks_info): Likewise.
27581 (chkp_get_check_result): Likewise.
27582 (chkp_remove_check_if_pass): Likewise.
27583 (chkp_use_outer_bounds_if_possible): Likewise.
27584 (chkp_reduce_bounds_lifetime): Likewise.
27585 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27586 (chkp_mark_completed_bounds): Likewise.
27587 (chkp_register_incomplete_bounds): Likewise.
27588 (chkp_mark_invalid_bounds): Likewise.
27589 (chkp_maybe_copy_and_register_bounds): Likewise.
27590 (chkp_build_returned_bound): Likewise.
27591 (chkp_get_bound_for_parm): Likewise.
27592 (chkp_build_bndldx): Likewise.
27593 (chkp_get_bounds_by_definition): Likewise.
27594 (chkp_generate_extern_var_bounds): Likewise.
27595 (chkp_get_bounds_for_decl_addr): Likewise.
27596 * tree-chrec.c (chrec_apply): Likewise.
27597 * tree-data-ref.c (dump_data_reference): Likewise.
27598 (dump_subscript): Likewise.
27599 (dump_data_dependence_relation): Likewise.
27600 (analyze_overlapping_iterations): Likewise.
27601 * tree-inline.c (expand_call_inline): Likewise.
27602 (tree_function_versioning): Likewise.
27603 * tree-into-ssa.c (dump_defs_stack): Likewise.
27604 (dump_currdefs): Likewise.
27605 (dump_names_replaced_by): Likewise.
27606 (dump_update_ssa): Likewise.
27607 (update_ssa): Likewise.
27608 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27609 * tree-parloops.c (build_new_reduction): Likewise.
27610 (try_create_reduction_list): Likewise.
27611 (ref_conflicts_with_region): Likewise.
27612 (oacc_entry_exit_ok_1): Likewise.
27613 (oacc_entry_exit_single_gang): Likewise.
27614 * tree-pretty-print.h: Likewise.
27615 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27616 (get_scalar_evolution): Likewise.
27617 (add_to_evolution): Likewise.
27618 (get_loop_exit_condition): Likewise.
27619 (analyze_evolution_in_loop): Likewise.
27620 (analyze_initial_condition): Likewise.
27621 (analyze_scalar_evolution): Likewise.
27622 (instantiate_scev): Likewise.
27623 (number_of_latch_executions): Likewise.
27624 (gather_chrec_stats): Likewise.
27625 (final_value_replacement_loop): Likewise.
27626 (scev_const_prop): Likewise.
27627 * tree-sra.c (dump_access): Likewise.
27628 (disqualify_candidate): Likewise.
27629 (create_access): Likewise.
27630 (reject): Likewise.
27631 (maybe_add_sra_candidate): Likewise.
27632 (create_access_replacement): Likewise.
27633 (analyze_access_subtree): Likewise.
27634 (analyze_all_variable_accesses): Likewise.
27635 (sra_modify_assign): Likewise.
27636 (initialize_constant_pool_replacements): Likewise.
27637 (find_param_candidates): Likewise.
27638 (decide_one_param_reduction): Likewise.
27639 (replace_removed_params_ssa_names): Likewise.
27640 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27641 * tree-ssa-copy.c (dump_copy_of): Likewise.
27642 (copy_prop_visit_cond_stmt): Likewise.
27643 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27644 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27645 (record_equivalences_from_stmt): Likewise.
27646 * tree-ssa-dse.c (compute_trims): Likewise.
27647 (delete_dead_call): Likewise.
27648 (delete_dead_assignment): Likewise.
27649 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27650 (forward_propagate_into_cond): Likewise.
27651 (pass_forwprop::execute): Likewise.
27652 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27653 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27655 (move_computations_worker): Likewise.
27656 (execute_sm): Likewise.
27657 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27658 (remove_exits_and_undefined_stmts): Likewise.
27659 (remove_redundant_iv_tests): Likewise.
27660 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27661 (adjust_iv_update_pos): Likewise.
27662 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27663 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27664 (value_replacement): Likewise.
27665 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27666 * tree-ssa-pre.c (print_pre_expr): Likewise.
27667 (get_representative_for): Likewise.
27668 (create_expression_by_pieces): Likewise.
27669 (insert_into_preds_of_block): Likewise.
27670 (eliminate_insert): Likewise.
27671 (eliminate_dom_walker::before_dom_children): Likewise.
27672 (eliminate): Likewise.
27673 (remove_dead_inserted_code): Likewise.
27674 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27675 * tree-ssa-reassoc.c (get_rank): Likewise.
27676 (eliminate_duplicate_pair): Likewise.
27677 (eliminate_plus_minus_pair): Likewise.
27678 (eliminate_not_pairs): Likewise.
27679 (undistribute_ops_list): Likewise.
27680 (eliminate_redundant_comparison): Likewise.
27681 (update_range_test): Likewise.
27682 (optimize_range_tests_var_bound): Likewise.
27683 (optimize_vec_cond_expr): Likewise.
27684 (rewrite_expr_tree): Likewise.
27685 (rewrite_expr_tree_parallel): Likewise.
27686 (linearize_expr): Likewise.
27687 (break_up_subtract): Likewise.
27688 (linearize_expr_tree): Likewise.
27689 (attempt_builtin_powi): Likewise.
27690 (attempt_builtin_copysign): Likewise.
27691 (transform_stmt_to_copy): Likewise.
27692 (transform_stmt_to_multiply): Likewise.
27693 (dump_ops_vector): Likewise.
27694 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27695 (print_scc): Likewise.
27696 (set_ssa_val_to): Likewise.
27697 (visit_reference_op_store): Likewise.
27698 (visit_use): Likewise.
27699 (sccvn_dom_walker::before_dom_children): Likewise.
27700 (run_scc_vn): Likewise.
27701 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27703 (expr_hash_elt::print): Likewise.
27704 (const_and_copies::pop_to_marker): Likewise.
27705 (const_and_copies::record_const_or_copy_raw): Likewise.
27706 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27707 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27708 (dump_predicates): Likewise.
27709 (find_uninit_use): Likewise.
27710 (warn_uninitialized_phi): Likewise.
27711 (pass_late_warn_uninitialized::execute): Likewise.
27712 * tree-ssa.c (verify_vssa): Likewise.
27713 (verify_ssa): Likewise.
27714 (maybe_optimize_var): Likewise.
27715 * tree-vrp.c (dump_value_range): Likewise.
27716 (dump_all_value_ranges): Likewise.
27717 (dump_asserts_for): Likewise.
27718 (register_edge_assert_for_2): Likewise.
27719 (vrp_visit_cond_stmt): Likewise.
27720 (vrp_visit_switch_stmt): Likewise.
27721 (vrp_visit_stmt): Likewise.
27722 (vrp_visit_phi_node): Likewise.
27723 (simplify_cond_using_ranges_1): Likewise.
27724 (fold_predicate_in): Likewise.
27725 (evrp_dom_walker::before_dom_children): Likewise.
27726 (evrp_dom_walker::push_value_range): Likewise.
27727 (evrp_dom_walker::pop_value_range): Likewise.
27728 (execute_early_vrp): Likewise.
27730 2017-05-16 Richard Biener <rguenther@suse.de>
27732 * dwarf2out.c (loc_list_from_tree_1): Do not create
27733 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27735 2017-05-16 Richard Biener <rguenther@suse.de>
27737 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27739 (note_variable_value_in_expr): If we resolved the decl ref
27740 do not push to the stack.
27742 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27744 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27745 operations in fast-math mode.
27746 (vaddq_f16): Likewise.
27747 (vmul_f16): Likewise.
27748 (vmulq_f16): Likewise.
27749 (vsub_f16): Likewise.
27750 (vsubq_f16): Likewise.
27751 * config/arm/neon.md (add<mode>3): New.
27753 (fma:<VH:mode>3): New. Also remove outdated comment.
27756 2017-05-16 Martin Liska <mliska@suse.cz>
27760 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27761 (odr_types_equivalent_p): Likewise.
27763 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27765 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27767 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27770 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27771 non-interunit SSE move alternatives with '?'.
27772 (zero-extendsidi peephole2): New peephole to skip intermediate
27773 general register in SSE zero-extend sequence.
27775 2017-05-15 Jeff Law <law@redhat.com>
27777 * reorg.c (relax_delay_slots): Create a new variable to hold
27778 the temporary target rather than clobbering TARGET_LABEL.
27780 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27781 missing argument to extract_bit_field call.
27782 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27784 2017-05-15 Martin Liska <mliska@suse.cz>
27787 * gcc.c (process_command): Do not allow empty argument of -o option.
27789 2017-05-15 Renlin Li <renlin.li@arm.com>
27791 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27792 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27793 * config/aarch64/constraints.md (Usf): Add long call check.
27794 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27795 (call_value): Likewise.
27796 (sibcall): Likewise.
27797 (sibcall_value): Likewise.
27799 (call_value_insn): New.
27800 (sibcall_insn): Update rtx pattern.
27801 (sibcall_value_insn): Likewise.
27802 (call_internal): Remove.
27803 (call_value_internal): Likewise.
27804 (sibcall_internal): Likewise.
27805 (sibcall_value_internal): Likewise.
27806 (call_reg): Likewise.
27807 (call_symbol): Likewise.
27808 (call_value_reg): Likewise.
27809 (call_value_symbol): Likewise.
27811 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27814 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27816 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27818 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27819 compatible with CCGOCmode and with CCZmode.
27821 2017-05-14 Martin Sebor <msebor@redhat.com>
27823 PR middle-end/77671
27824 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27825 (gimple_fold_builtin_snprintf): Same.
27826 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27827 (gimple_fold_builtin_snprintf): Same.
27828 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27829 of character types.
27830 (is_call_safe): New function.
27831 (try_substitute_return_value): Call it.
27832 (try_simplify_call): New function.
27833 (pass_sprintf_length::handle_gimple_call): Call it.
27835 2017-05-14 Martin Sebor <msebor@redhat.com>
27837 PR middle-end/80669
27838 * builtins.c (expand_builtin_stpncpy): Simplify.
27840 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27842 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27843 * config/i386/i386.h
27844 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27845 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27846 (struct machine_function): Add new members call_ms2sysv,
27847 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27848 (struct machine_frame_state): New fields sp_realigned and
27849 sp_realigned_offset.
27850 * config/i386/i386.c
27851 (enum xlogue_stub): New enum.
27852 (enum xlogue_stub_sets): New enum.
27853 (class xlogue_layout): New class.
27854 (struct ix86_frame): New fields stack_realign_allocate_offset,
27855 stack_realign_offset and outlined_save_offset. Modify comments to
27856 detail stack layout when using out-of-line stubs.
27857 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27858 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27859 -mcall-ms2sysv-xlogues.
27860 (stub_managed_regs): New static variable.
27861 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27862 registers managed by out-of-line stub.
27863 (disable_call_ms2sysv_xlogues): New function.
27864 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27865 m->call_ms2sysv when appropriate and compute frame layout for
27867 (sp_valid_at, fp_valid_at): New inline functions.
27868 (choose_basereg): New function.
27869 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27871 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27872 Use align parameter of choose_baseaddr to generated aligned SSE movs
27874 (pro_epilogue_adjust_stack): Modify to track
27875 machine_frame_state::sp_realigned.
27876 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27877 (ix86_nsaved_sseregs): Likewise.
27878 (ix86_emit_save_regs): Likewise.
27879 (ix86_emit_save_regs_using_mov): Likewise.
27880 (ix86_emit_save_sse_regs_using_mov): Likewise.
27881 (get_scratch_register_on_entry): Likewise.
27882 (gen_frame_set): New function.
27883 (gen_frame_load): Likewise.
27884 (gen_frame_store): Likewise.
27885 (emit_outlined_ms2sysv_save): Likewise.
27886 (emit_outlined_ms2sysv_restore): Likewise.
27887 (ix86_expand_prologue): Modify stack re-alignment code and call
27888 emit_outlined_ms2sysv_save when appropriate.
27889 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27890 parameter rtx_insn *insn, which allows the function to be used to only
27891 generate the notes.
27892 (ix86_expand_epilogue): Modify validity checks of frame and stack
27893 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27894 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27895 * config/i386/predicates.md
27896 (save_multiple): New predicate.
27897 (restore_multiple): Likewise.
27898 * config/i386/sse.md
27899 (save_multiple<mode>): New pattern.
27900 (save_multiple_realign<mode>): Likewise.
27901 (restore_multiple<mode>): Likewise.
27902 (restore_multiple_and_return<mode>): Likewise.
27903 (restore_multiple_leave_return<mode>): Likewise.
27904 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27906 2017-05-14 Julia Koval <julia.koval@intel.com>
27908 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
27909 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
27910 (__builtin_ia32_xsetbv): New builtins.
27911 * config/i386/i386.c (ix86_expand_special_args_builtin):
27913 (ix86_expand_builtin): Special expand for new intrinsics.
27914 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
27915 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
27916 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
27918 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27920 * cfganal.c (inverted_post_order_compute): Change argument type
27922 * cfganal.h (inverted_post_order_compute): Adjust prototype.
27923 * df-core.c (rest_of_handle_df_initialize): Adjust.
27924 (rest_of_handle_df_finish): Likewise.
27925 (df_analyze_1): Likewise.
27926 (df_analyze): Likewise.
27927 (loop_inverted_post_order_compute): Change argument to be a vec *.
27928 (df_analyze_loop): Adjust.
27929 (df_get_n_blocks): Likewise.
27930 (df_get_postorder): Likewise.
27931 * df.h (struct df_d): Change field to be a vec.
27932 * lcm.c (compute_laterin): Adjust.
27933 (compute_available): Likewise.
27934 * lra-lives.c (lra_create_live_ranges_1): Likewise.
27935 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
27936 * tree-ssa-pre.c (compute_antic): Likewise.
27938 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27940 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
27941 (depth_first_search::depth_first_search): Change structure init
27942 function to this constructor.
27943 (depth_first_search::add_bb): Rename function to this member.
27944 (depth_first_search::execute): Likewise.
27945 (flow_dfs_compute_reverse_finish): Adjust.
27947 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27949 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
27950 (longest_simple_path): Likewise.
27951 * shrink-wrap.c (spread_components): Likewise.
27952 (disqualify_problematic_components): Likewise.
27953 (emit_common_heads_for_components): Likewise.
27954 (emit_common_tails_for_components): Likewise.
27955 (insert_prologue_epilogue_for_components): Likewise.
27957 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27959 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
27962 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27964 * df-core.c (df_set_blocks): Start using auto_bitmap.
27965 (df_compact_blocks): Likewise.
27966 * df-problems.c (df_rd_confluence_n): Likewise.
27967 * df-scan.c (df_insn_rescan_all): Likewise.
27968 (df_process_deferred_rescans): Likewise.
27969 (df_update_entry_block_defs): Likewise.
27970 (df_update_exit_block_uses): Likewise.
27971 (df_entry_block_bitmap_verify): Likewise.
27972 (df_exit_block_bitmap_verify): Likewise.
27973 (df_scan_verify): Likewise.
27974 * lra-constraints.c (lra_constraints): Likewise.
27975 (undo_optional_reloads): Likewise.
27976 (lra_undo_inheritance): Likewise.
27977 * lra-remat.c (calculate_gen_cands): Likewise.
27978 (do_remat): Likewise.
27979 * lra-spills.c (assign_spill_hard_regs): Likewise.
27980 (spill_pseudos): Likewise.
27981 * tree-ssa-pre.c (bitmap_set_and): Likewise.
27982 (bitmap_set_subtract_values): Likewise.
27984 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27986 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
27987 management with auto_bitmap.
27988 (fix_inter_tick): Likewise.
27989 (fix_recovery_deps): Likewise.
27990 * ira.c (add_store_equivs): Likewise.
27991 (find_moveable_pseudos): Likewise.
27992 (split_live_ranges_for_shrink_wrap): Likewise.
27993 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
27994 (rtx_reuse_manager::seen_def_p): Likewise.
27995 (rtx_reuse_manager::set_seen_def): Likewise.
27996 * print-rtl.h (class rtx_reuse_manager): Likewise.
27998 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28000 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28002 (migrate_btr_def): Likewise.
28003 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28004 * df-core.c (loop_post_order_compute): Likewise.
28005 (loop_inverted_post_order_compute): Likewise.
28006 * hsa-common.h: Likewise.
28007 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28008 * init-regs.c (initialize_uninitialized_regs): Likewise.
28009 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28010 (inline_small_functions): Likewise.
28011 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28012 * ira.c (combine_and_move_insns): Likewise.
28013 (build_insn_chain): Likewise.
28014 * loop-invariant.c (find_invariants): Likewise.
28015 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28016 * predict.c (tree_predict_by_opcode): Likewise.
28017 (predict_paths_leading_to): Likewise.
28018 (predict_paths_leading_to_edge): Likewise.
28019 (estimate_loops_at_level): Likewise.
28020 (estimate_loops): Likewise.
28021 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28022 (spread_components): Likewise.
28023 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28024 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28025 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28026 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28027 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28028 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28029 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28030 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28031 (mark_threaded_blocks): Likewise.
28032 (thread_through_all_blocks): Likewise.
28033 * tree-ssa.c (verify_ssa): Likewise.
28034 (execute_update_addresses_taken): Likewise.
28035 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28037 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28039 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28041 (post_order_compute): Likewise.
28042 (inverted_post_order_compute): Likewise.
28043 (pre_and_rev_post_order_compute_fn): Likewise.
28045 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28047 * genrecog.c (int_set::int_set): Explicitly construct our
28048 auto_vec base class.
28049 * vec.h (auto_vec::auto_vec): New constructor.
28051 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28053 * bitmap.h (class auto_bitmap): New constructor taking
28054 bitmap_obstack * argument.
28056 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28058 * bitmap.h (class auto_bitmap): Change type of m_bits to
28059 bitmap_head, and adjust ctor / dtor and member operators.
28061 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28063 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28064 when returned register mode doesn't match original mode.
28066 2017-05-12 Jeff Law <law@redhat.com>
28067 Jakub Jelinek <jakub@redhat.com>
28069 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28070 we look for cc setter after the compare-elim changes.
28071 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28072 within the vector to match what compare-elim now expects.
28073 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28074 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28076 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28077 after the compare-elim changes.
28078 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28079 the vector to match what compare-elim now expects.
28080 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28081 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28082 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28083 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28084 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28086 * config/visium/visium.c (single_set_and_flags): Fix where
28087 we look for cc setter after the compare-elim changes.
28088 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28089 with the vector to match what compare-elim now expects.
28090 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28091 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28092 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28093 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28094 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28095 (neg<mode>2_insn_set_overflow): Likewise.
28097 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28099 PR middle-end/79794
28100 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28101 maybe_expand_insn call, set ops[0].target. If still set after call,
28102 set alt_rtl. Add extra arg to recursive calls.
28103 (extract_bit_field): Add alt_rtl argument. Pass to
28105 * expmed.h (extract_bit_field): Fix prototype.
28106 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28107 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28108 to extract_bit_field_calls.
28109 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28110 Pass alt_rtl to extract_bit_field calls.
28111 * calls.c (store_unaligned_arguments_into_psuedos)
28112 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28113 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28115 * optabs.h (struct expand_operand): Add target bitfield.
28117 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28119 * compare-elim.c (try_eliminate_compare): Canonicalize
28120 operation with embedded compare to
28121 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28122 (set (reg) (operation)].
28124 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28126 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28129 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28130 cost of adding a carry flag for ADC instruction.
28131 [case MINUS]: Ignore the cost of subtracting a carry flag
28132 for SBB instruction.
28134 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28136 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28138 * config/rs6000/bmiintrin.h: New file.
28139 * config/rs6000/bmi2intrin.h: New file.
28140 * config/rs6000/x86intrin.h: New file.
28142 2017-05-12 Jeff Law <law@redhat.com>
28144 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28147 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28149 PR middle-end/80707
28150 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28152 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28154 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28155 early expansion of vector divide builtins.
28156 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28157 builtins identified as having unsigned arguments.
28159 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28161 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28162 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28163 expansion of vector logical operations (and, andc, or, xor,
28166 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28168 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28169 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28171 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28173 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28174 early GIMPLE expansion of vector multiplies.
28176 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28178 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28179 TARGET_HAVE_MOVT conditional.
28180 (movt splitter): Likewise.
28182 2017-05-12 Richard Biener <rguenther@suse.de>
28184 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28185 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28186 Fold all stmts not inplace.
28188 2017-05-12 Richard Biener <rguenther@suse.de>
28190 PR tree-optimization/80713
28191 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28192 inserted_exprs bit for not removed stmts.
28194 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28196 PR middle-end/69921
28197 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28198 parallelized" attribute for parallelized OpenACC kernels.
28199 * omp-offload.c (execute_oacc_device_lower): Use it.
28201 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28202 Set "oacc kernels" attribute.
28203 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28204 parameter. Adjust all users.
28205 (oacc_fn_attrib_kernels_p): Remove function.
28206 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28207 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28208 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28209 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28210 assert "oacc kernels" attribute is set.
28212 2017-05-11 Carl Love <cel@us.ibm.com>
28214 * config/rs6000/rs6000-c: Add support for built-in functions
28215 vector unsigned char vec_popcnt (vector signed char)
28216 vector unsigned char vec_popcnt (vector unsigned char)
28217 vector unsigned short vec_popcnt (vector signed short)
28218 vector unsigned short vec_popcnt (vector unsigned short)
28219 vector unsigned int vec_popcnt (vector signed int)
28220 vector unsigned int vec_popcnt (vector unsigned int)
28221 vector unsigned long long vec_popcnt (vector signed long long)
28222 vector unsigned long long vec_popcnt (vector unsigned long long)
28223 vector signed long long vec_slo (vector signed long long,
28224 vector signed char)
28225 vector signed long long vec_slo (vector signed long long,
28226 vector unsigned char)
28227 vector unsigned long long vec_slo (vector unsigned long long,
28228 vector signed char)
28229 vector unsigned long long vec_slo (vector unsigned long long,
28230 vector unsigned char)
28231 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28232 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28233 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28234 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28235 * doc/extend.texi: Update the built-in documentation file for the
28236 new built-in functions.
28238 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28240 * attribs.h (sorted_attr_string): Move machine independent
28241 functions for target clone support from the i386 port to common
28242 code. Rename ix86_function_versions to common_function_versions.
28243 Rename make_name to make_unique_name.
28244 (common_function_versions): Likewise.
28245 (make_unique_name): Likewise.
28246 (make_dispatcher_decl): Likewise.
28247 (is_function_default_version): Likewise.
28248 * attribs.c (attr_strcmp): Likewise.
28249 (sorted_attr_string): Likewise.
28250 (common_function_versions): Likewise.
28251 (make_unique_name): Likewise.
28252 (make_dispatcher_decl): Likewise.
28253 (is_function_default_version): Likewise.
28254 * config/i386/i386.c (attr_strcmp): Likewise.
28255 (sorted_attr_string): Likewise.
28256 (ix86_function_versions): Likewise.
28257 (make_name): Likewise.
28258 (make_dispatcher_decl): Likewise.
28259 (is_function_default_version): Likewise.
28260 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28262 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28265 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28266 Account for direct move costs for vec_construct of integer
28269 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28272 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28273 (UNSPEC_STX_ATOMIC): Ditto.
28274 (loaddi_via_sse): New insn.
28275 (storedi_via_sse): Ditto.
28276 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28277 Update corresponding peephole2 patterns.
28278 (atomic_storedi_fpu): Ditto.
28280 2017-05-11 Julia Koval <julia.koval@intel.com>
28282 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28283 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28285 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28286 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28287 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28289 2017-05-11 Nathan Sidwell <nathan@acm.org>
28291 * graphite-poly.c: Include dumpfile.h.
28293 * dumpfle.h (dump_function): Declare here ...
28294 * tree-dump.h (dump_function): ... not here.
28295 * dumpfile.c: #include tree-cfg.h.
28296 (dump_function): Move here from ...
28297 * tree-dump.c (dump_function): ... here.
28298 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28299 * graphite-poly.c: Don't include tree-dump.h.
28300 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28301 * print-tree.c: Likewise.
28302 * stor-layout.c: Likewise.
28303 * tree-nested.c: Likewise.
28305 * dumpfile.c (dump_start): Use TDF_FLAGS.
28306 (dump_enable_all): Fix TDF_KIND check thinko.
28308 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28310 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28311 array entries to represent two legal parameterizations of the
28312 overloaded __builtin_cmpb function, as represented by the
28313 P6_OV_BUILTIN_CMPB constant.
28314 (altivec_resolve_overloaded_builtin): Add special case handling
28315 for the __builtin_cmpb function, as represented by the
28316 P6_OV_BUILTIN_CMPB constant.
28317 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28318 (BU_P6_64BIT_2): New macro.
28319 (BU_P6_OVERLOAD_2): New macro
28320 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28321 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28322 (CMPB): Add overload support to represent both 32-bit and 64-bit
28323 compare-bytes function.
28324 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28325 support for TARGET_CMPB.
28326 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28327 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28328 documentation of the __builtin_cmpb overloaded built-in function.
28330 2017-05-11 Richard Biener <rguenther@suse.de>
28332 PR tree-optimization/80705
28333 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28334 bases are not vectorizable.
28336 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28338 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28339 when counting register pressure.
28341 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28343 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28344 (iv_ca_more_deps): Renamed to ...
28345 (iv_ca_compare_deps): ... this.
28346 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28348 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28350 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28352 (determine_group_iv_costs): ... here.
28353 (find_inv_vars_cb): Record inv var if it's not recorded before.
28355 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28357 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28358 (get_shiftadd_cost): Ditto.
28360 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28362 * tree-ssa-address.c: Include header file.
28363 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28365 (add_to_parts): Refactor.
28366 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28367 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28370 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28372 PR tree-optimization/53090
28373 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28375 (extract_cond_operands): Detect condition with IV on both sides
28376 and return COMP_IV_EXPR_2.
28377 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28378 (rewrite_use_compare): Simplify by removing call to function
28379 extract_cond_operands.
28381 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28383 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28384 (extract_cond_operands): Detect condition comparing against non-
28385 invariant bound and return appropriate enum value.
28386 (find_interesting_uses_cond): Update use of extract_cond_operands.
28387 Handle its return value accordingly.
28388 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28390 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28392 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28393 nonlinear iv_use computation in loop invariant sensitive way.
28395 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28397 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28398 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28400 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28402 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28403 (dump_cand): Support iv_cand.inv_exprs.
28404 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28406 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28409 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28411 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28413 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28414 as local function. Include necessary header files.
28415 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28417 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28419 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28421 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28423 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28424 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28425 RSHIFT_EXPR and BIT_NOT_EXPR.
28427 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28429 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28430 (adjust_setup_cost): New parameter supporting round up adjustment.
28431 (struct address_cost_data): Delete.
28432 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28433 (split_address_cost, ptr_difference_cost): Delete.
28434 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28435 (struct ainc_cost_data): New struct.
28436 (get_address_cost_ainc): New function.
28437 (get_address_cost, get_computation_cost): Reimplement.
28438 (determine_group_iv_cost_address): Record inv_expr for all uses of
28440 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28441 (iv_ca_has_deps): Reimplemented to ...
28442 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28444 (iv_ca_extend): Call iv_ca_more_deps.
28446 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28448 * tree-ssa-address.c (struct mem_address): Move to header file.
28449 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28450 * tree-ssa-address.h (struct mem_address): Move from C file.
28451 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28453 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28455 * tree-affine.h (aff_combination_type): New interface.
28456 (aff_combination_zero_p): Remove static.
28457 (aff_combination_const_p): New interface.
28458 (aff_combination_singleton_var_p): New interfaces.
28460 2017-05-11 Richard Biener <rguenther@suse.de>
28462 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28463 Skip unreachable blocks and destinations.
28464 (eliminate): Move stmt removal and fixup ...
28465 (fini_eliminate): ... here. Skip inserted exprs.
28466 (pass_pre::execute): Move fini_pre after fini_eliminate.
28467 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28468 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28469 PRE to get rid of dead code that has invalid SSA form and
28470 split critical edges again.
28472 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28474 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28476 2017-05-11 Richard Biener <rguenther@suse.de>
28478 * passes.c (execute_function_todo): Verify loops if they are
28479 said to be up-to-date.
28480 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28481 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28483 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28486 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28487 handle calling assemble_external ourself.
28490 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28491 modes with zero size. Enhance comment.
28493 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28495 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28496 built-ins for vec_xl and vec_xst with short and char pointer
28499 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28501 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28502 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28503 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28504 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28505 (_mm_maskz_min_round_ss): New intrinsics.
28506 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28507 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28508 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28509 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28510 (__builtin_ia32_minss_mask_round): New builtins.
28511 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28512 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28513 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28515 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28516 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28518 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28521 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28523 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28524 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28525 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28526 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28527 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28528 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28529 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28530 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28531 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28532 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28533 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28534 (__builtin_ia32_mulss_mask_round): New builtins.
28535 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28536 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28537 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28539 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28540 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28542 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28545 2017-05-10 Julia Koval <julia.koval@intel.com>
28547 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28548 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28549 (_mm256_setr_m128i): New intrinsics.
28551 2017-05-10 Julia Koval <julia.koval@intel.com>
28553 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28554 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28555 (_mm_maskz_rcp14_ss): New intrinsics.
28556 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28557 (__builtin_ia32_rcp14ss_mask): New builtins.
28558 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28560 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28562 PR tree-optimization/51513
28563 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28564 (assert_unreachable_fallthru_edge_p): Use it.
28565 (group_case_labels_stmt): Likewise.
28566 * tree-cfg.h: Prototype it.
28567 * stmt.c: Include cfghooks.h and tree-cfg.h.
28568 (emit_case_dispatch_table) <gap_label>: New local variable.
28569 Use it to fill dispatch table gaps.
28570 Test for default_label before updating probabilities.
28571 (expand_case) <default_label>: Remove unneeded initialization.
28572 Test for unreachable default case statement and remove its edge.
28573 Set default_label accordingly.
28574 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28576 2017-05-10 Carl Love <cel@us.ibm.com>
28578 * config/rs6000/rs6000-c: Add support for built-in functions
28579 vector signed char vec_neg (vector signed char)
28580 vector signed short int vec_neg (vector short int)
28581 vector signed int vec_neg (vector signed int)
28582 vector signed long long vec_neg (vector signed long long)
28583 vector float vec_neg (vector float)
28584 vector double vec_neg (vector double)
28585 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28587 * config/rs6000/altivec.h: Add define for vec_neg
28588 * doc/extend.texi: Update the built-in documentation for the
28589 new built-in functions.
28591 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28593 PR tree-optimization/77644
28594 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28596 2017-05-10 Nathan Sidwell <nathan@acm.org>
28598 * dumpfile.h (TDI_lang_all): New.
28599 (TDF_KIND): New. Renumber others
28600 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28602 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28604 (get_dump_file_name): Adjust suffix generation.
28605 (dump_enable_all): Use TDF_KIND.
28606 * doc/invoke.texi (-fdump-lang-all): Document.
28608 * dumpfile.h: Tabify.
28610 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28613 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28614 Move member access before delete.
28616 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28618 * tree-inline.c (expand_call_inline): Split block at stmt
28621 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28624 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28625 are now unused after splitting mov{sf,sd}_hardfloat.
28626 (f32_lr2): Likewise.
28627 (f32_lm): Likewise.
28628 (f32_lm2): Likewise.
28629 (f32_li): Likewise.
28630 (f32_li2): Likewise.
28631 (f32_lv): Likewise.
28632 (f32_sr): Likewise.
28633 (f32_sr2): Likewise.
28634 (f32_sm): Likewise.
28635 (f32_sm2): Likewise.
28636 (f32_si): Likewise.
28637 (f32_si2): Likewise.
28638 (f32_sv): Likewise.
28639 (f32_dm): Likewise.
28640 (f32_vsx): Likewise.
28641 (f32_av): Likewise.
28642 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28643 For movsf, order stores so the VSX stores occur before the GPR
28644 store which encourages the register allocator to use a traditional
28645 FPR instead of a GPR. For movsd, order the stores so that the GPR
28646 store comes before the VSX stores to allow the power6 to work.
28647 This is due to the power6 not having a 32-bit integer store
28648 instruction from a FPR.
28649 (movsf_hardfloat): Likewise.
28650 (movsd_hardfloat): Likewise.
28652 2017-05-09 Martin Sebor <msebor@redhat.com>
28654 PR translation/80280
28655 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28658 PR translation/80280
28659 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28660 data member added in r247778.
28661 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28663 2017-05-09 Nathan Sidwell <nathan@acm.org>
28665 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28667 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28670 2017-05-09 Marek Polacek <polacek@redhat.com>
28672 * doc/invoke.texi: Fix typo.
28674 2017-05-09 Richard Biener <rguenther@suse.de>
28676 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28677 (vrp_val_is_min): Likewise.
28678 (set_value_range_to_value): Likewise.
28679 (set_value_range_to_nonnegative): Likewise.
28680 (gimple_assign_nonzero_p): Likewise.
28681 (gimple_stmt_nonzero_p): Likewise.
28682 (vrp_int_const_binop): Likewise. Remove unreachable case.
28683 (adjust_range_with_scev): Adjust comments.
28684 (compare_range_with_value): Likewise.
28685 (extract_range_from_phi_node): Likewise.
28686 (test_for_singularity): Likewise.
28688 2017-05-09 Richard Biener <rguenther@suse.de>
28690 * tree-vrp.c (get_single_symbol): Add assert that we don't
28691 get overflowed constants as invariant part.
28692 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28693 checks. Use wi::cmp instead of recursing for integer constants.
28694 (compare_values): Just ignore whether we assumed undefined
28695 overflow instead of failing the compare.
28696 (extract_range_for_var_from_comparison_expr): Add comment before the
28697 TREE_NO_WARNING sets.
28698 (test_for_singularity): Likewise.
28699 (extract_range_from_comparison): Do not disable optimization
28700 when we assumed undefined overflow.
28701 (extract_range_basic): Remove init of unused var.
28703 2017-05-09 Richard Biener <rguenther@suse.de>
28705 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28706 (extract_range_from_multiplicative_op_1): Adjust.
28707 (extract_range_from_binary_expr_1): Use int_const_binop.
28709 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28712 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28713 rs6000_store_data_bypass_p in seven define_bypass directives and
28714 in several comments.
28715 * config/rs6000/rs6000-protos.h: Add prototype for
28716 rs6000_store_data_bypass_p function.
28717 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28718 function implements slightly different (rs6000-specific) semantics
28719 than store_data_bypass_p, returning false rather than aborting
28720 with assertion error when arguments do not satisfy the
28721 requirements of store data bypass.
28722 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28723 rs6000_store_data_bypass_p.
28725 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28727 * config/xtensa/xtensa-protos.h
28728 (xtensa_initial_elimination_offset): New declaration.
28729 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28730 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28731 macro definition, add case for FRAME_POINTER_REGNUM when
28732 FRAME_GROWS_DOWNWARD.
28733 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28734 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28735 xtensa_initial_elimination_offset.
28737 2017-05-08 Nathan Sidwell <nathan@acm.org>
28739 * doc/invoke.texi: Alphabetize -fdump options.
28741 2017-05-08 Martin Sebor <msebor@redhat.com>
28743 PR translation/80280
28744 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28746 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28748 * target.def (compute_frame_layout): New optional target hook.
28749 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28750 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28751 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28753 * reload1.c (verify_initial_elim_offsets): Likewise.
28754 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28755 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28756 (arm_get_frame_offsets): Split up into this ...
28757 (arm_compute_frame_layout): ... and this function.
28759 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28761 * config/aarch64/constraints.md (Usa): New constraint.
28762 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28764 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28766 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28767 with_multilib_list after it has been checked.
28769 2017-05-08 Richard Biener <rguenther@suse.de>
28771 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28772 (bitmap_set_subtract_values): Likewise.
28774 2017-05-08 Richard Biener <rguenther@suse.de>
28776 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28777 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28779 (gimple_stmt_nonzero_warnv_p): Rename to ...
28780 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28782 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28783 (extract_range_basic): Adjust, do not disable propagation on
28784 strict overflow sensitive simplification.
28785 (vrp_visit_cond_stmt): Likewise.
28787 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28789 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28790 body size unconditionally.
28792 2017-05-07 Jeff Law <law@redhat.com>
28795 2017-05-06 Jeff Law <law@redhat.com>
28796 PR tree-optimization/78496
28797 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28800 PR tree-optimization/78496
28801 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28802 (simplify_stmt_using_ranges): Call it.
28803 (vrp_dom_walker::before_dom_children): Extract equivalences
28804 from an ASSERT_EXPR with an equality comparison against a
28807 2017-05-06 Jeff Law <law@redhat.com>
28809 PR tree-optimization/78496
28810 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28813 PR tree-optimization/78496
28814 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28815 (simplify_stmt_using_ranges): Call it.
28816 (vrp_dom_walker::before_dom_children): Extract equivalences
28817 from an ASSERT_EXPR with an equality comparison against a
28820 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28822 * lra-constraints.c (lra_copy_reg_equiv): New function.
28823 (split_reg): Use it to copy equivalence information from the
28824 original register to the spill register.
28826 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28828 PR rtl-optimization/75964
28829 * simplify-rtx.c (simplify_const_relational_operation): Remove
28830 invalid handling of comparisons of integer ABS.
28832 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28834 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28835 initialize to zero.
28836 (init_regs): Remove declaration.
28837 (function_arg_advance_32): Initialize error_p as boolean variable.
28839 2017-05-05 Nathan Sidwell <nathan@acm.org>
28841 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28842 lines. Use for (;;).
28844 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28846 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28847 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28848 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28849 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28850 VF=2 that require versioning.
28852 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28854 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28857 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28859 * diagnostic.h (diagnostic_override_option_index): Convert from
28860 macro to inline function.
28862 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28864 * diagnostic.c (last_module_changed_p): New function.
28865 (set_last_module): New function.
28866 (diagnostic_report_current_module): Convert macro usage to
28867 the above functions.
28868 * diagnostic.h (diagnostic_context::last_module): Strengthen
28869 from const line_map * to const line_map_ordinary *.
28870 (diagnostic_last_module_changed): Delete macro.
28871 (diagnostic_set_last_module): Delete macro.
28873 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28875 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28876 with diagnostic_report_diagnostic.
28877 (diagnostic_n_impl_richloc): Likewise.
28878 * diagnostic.h (report_diagnostic): Delete macro.
28879 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28880 with diagnostic_report_diagnostic.
28881 * substring-locations.c (format_warning_va): Likewise.
28883 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28885 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28886 save/restor of format_spec. Move option-printing code to...
28887 (print_option_information): ...this new function, and
28888 reimplement by simply printing to the pretty_printer,
28889 rather than appending to the format string.
28891 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28893 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28894 handling logic into...
28895 (update_effective_level_from_pragmas): ...this new function.
28897 2017-05-04 Andrew Waterman <andrew@sifive.com>
28899 * config/riscv/riscv.opt (mstrict-align): New option.
28900 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28901 (SLOW_UNALIGNED_ACCESS): Define.
28902 (riscv_slow_unaligned_access): Declare.
28903 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28905 (riscv_slow_unaligned_access): New variable.
28906 (rocket_tune_info): Set slow_unaligned_access to true.
28907 (optimize_size_tune_info): Set slow_unaligned_access to false.
28908 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
28909 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
28910 (riscv_option_override): Set riscv_slow_unaligned_access.
28911 * doc/invoke.texi: Add -mstrict-align to RISC-V.
28913 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
28915 * config/riscv/riscv.md: Unify indentation.
28917 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28922 * config/rs6000/rs6000.md (u code attribute): Add FIX and
28924 (extendsi<mode>2): Add support for doing sign extension via
28925 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
28926 don't have ISA 3.0 instructions.
28927 (extendsi<mode>2 splitter): Likewise.
28928 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
28929 generate the normal insns since SImode can now go in vector
28930 registers. Disallow the special UNSPECs needed for previous
28931 machines to hide SImode being used. Add new insns
28932 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
28933 (fix_trunc<mode>si2_stfiwx): Likewise.
28934 (fix_trunc<mode>si2_internal): Likewise.
28935 (fixuns_trunc<mode>si2): Likewise.
28936 (fixuns_trunc<mode>si2_stfiwx): Likewise.
28937 (fctiw<u>z_<mode>_smallint): Likewise.
28938 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
28939 of floating point to 32-bit integer from doing a direct move to
28940 the GPR registers to do a store.
28941 (fctiwz_<mode>): Break long line.
28943 2017-05-05 Bin Cheng <bin.cheng@arm.com>
28945 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
28946 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
28947 (addr_list, addr_offset_valid_p): New.
28948 (split_address_groups): Check offset validity with above function.
28949 (gt-tree-ssa-loop-ivopts.h): Include header file.
28951 2017-05-05 Nathan Sidwell <nathan@acm.org>
28953 * config.gcc (arm*-*-*): Add missing 'fi'.
28955 2017-05-05 Steve Ellcey <sellcey@cavium.com>
28957 * doc/invoke.texi (-fopt-info): Explicitly say order of options
28958 included in -fopt-info does not matter.
28959 * doc/optinfo.texi (-fopt-info): Fix description of default
28960 behavour. Explicitly say order of options included in -fopt-info
28963 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28965 * config.gcc: Allow combinations of aprofile and rmprofile values for
28966 --with-multilib-list.
28967 * config/arm/t-multilib: New file.
28968 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
28969 variables. Remove setting of ISA and floating-point ABI in
28970 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
28971 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
28972 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
28973 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
28975 * config/arm/t-rmprofile: Likewise except for the matches changes.
28976 * doc/install.texi (--with-multilib-list): Document the combination of
28977 aprofile and rmprofile values and warn about pitfalls in doing that.
28979 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28981 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
28982 (movdi_aarch64): Likewise.
28984 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28986 PR tree-optimization/80632
28987 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
28989 (build_arrays): Initialize it for virtual phis.
28990 (fix_phi_nodes): Use it for virtual phis.
28992 PR tree-optimization/80558
28993 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
28994 [x, y] op z into [x op, y op z] for op & or | if conditions
28997 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
28998 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29001 * config/arm/arm.md (use_literal_pool): Remove.
29002 (64-bit immediate split): No longer takes cost into consideration
29003 if arm_disable_literal_pool is enabled.
29004 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29005 used when arm_disable_literal_pool is enabled.
29006 (arm_max_const_double_inline_cost): Remove use of
29007 arm_disable_literal_pool.
29008 (push_minipool_fix): Add assert.
29009 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29010 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29011 (no_literal_pool_sf_immediate): New.
29013 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29015 PR tree-optimization/80613
29016 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29017 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29019 2017-05-05 Richard Biener <rguenther@suse.de>
29021 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29023 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29025 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29026 of this flag from insn conditions due to removal from r247495.
29028 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29030 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29032 (arm_early_store_addr_dep_ptr): Likewise.
29033 * config/arm/aarch-common-protos.h
29034 (arm_early_load_addr_dep_ptr): Add prototype.
29035 (arm_early_store_addr_dep_ptr): Likewise.
29036 * config/arm/cortex-a53.md: Add new bypasses.
29038 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29040 * tree.c (next_type_uid): Change type to unsigned.
29041 (type_hash_canon): Decrement back next_type_uid if
29042 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29043 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29046 2017-05-04 Martin Sebor <msebor@redhat.com>
29048 * builtins.c: Fix a trivial typo in a comment.
29050 PR middle-end/79234
29051 * builtins.c (check_sizes): Adjust to handle reading past the end.
29052 Avoid printing excessive upper bound of ranges. Use %E to print
29053 tree nodes instead of converting them to %wu.
29054 (expand_builtin_memchr): New function.
29055 (compute_dest_size): Rename...
29056 (compute_objsize): ...to this.
29057 (expand_builtin_memcpy): Adjust.
29058 (expand_builtin_mempcpy): Adjust.
29059 (expand_builtin_strcat): Adjust.
29060 (expand_builtin_strcpy): Adjust.
29061 (check_strncat_sizes): Adjust.
29062 (expand_builtin_strncat): Adjust.
29063 (expand_builtin_strncpy): Adjust and simplify.
29064 (expand_builtin_memset): Adjust.
29065 (expand_builtin_bzero): Adjust.
29066 (expand_builtin_memcmp): Adjust.
29067 (expand_builtin): Handle memcmp.
29068 (maybe_emit_chk_warning): Check strncat just once.
29070 2017-05-04 Martin Sebor <msebor@redhat.com>
29072 PR preprocessor/79214
29073 PR middle-end/79222
29074 PR middle-end/79223
29075 * builtins.c (check_sizes): Add inlining context and issue
29076 warnings even when -Wno-system-headers is set.
29077 (check_strncat_sizes): Same.
29078 (expand_builtin_strncat): Same.
29079 (expand_builtin_memmove): New function.
29080 (expand_builtin_stpncpy): Same.
29081 (expand_builtin): Handle memmove and stpncpy.
29083 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29085 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29086 which is not used any more.
29088 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29090 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29092 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29094 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29095 (cortexa53_tunings): Likewise.
29096 (cortexa57_tunings): Likewise.
29097 (cortexa72_tunings): Likewise.
29098 (cortexa73_tunings): Likewise.
29100 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29102 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29103 Set loop alignment to 8.
29105 2017-05-04 Martin Sebor <msebor@redhat.com>
29107 PR translation/80280
29108 * builtins.c (expand_builtin_object_size): Add missing quoting to
29109 %D and like directives.
29110 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29111 (hsa_type_for_tree_type): Same.
29112 (verify_function_arguments): Same.
29113 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29114 * varasm.c (get_section): Same.
29117 2017-05-04 Martin Sebor <msebor@redhat.com>
29119 PR translation/80280
29120 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29122 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29124 * config/aarch64/aarch64.c (generic_addrcost_table):
29125 Change HI/TI mode setting.
29127 2017-05-04 Martin Jambor <mjambor@suse.cz>
29129 PR tree-optimization/80622
29130 * tree-sra.c (comes_initialized_p): New function.
29131 (build_accesses_from_assign): Only set write lazily when
29132 comes_initialized_p is false.
29133 (analyze_access_subtree): Use comes_initialized_p.
29134 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29135 instead of testing for PARM_DECL.
29137 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29139 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29140 constraint on operand 0 to allow more general addressing modes.
29141 Adjust output template.
29142 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29144 * config/aarch64/aarch64-protos.h
29145 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29146 * config/aarch64/constraints.md (Dp): New address constraint.
29147 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29150 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29152 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29153 update use of estimate_ipcp_clone_size_and_time.
29154 (estimate_local_effects): Update use of
29155 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29156 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29157 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29158 Return nonspecialized time.
29160 2017-05-04 Richard Biener <rguenther@suse.de>
29162 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29163 for the last VUSE which def dominates the PHI. Directly call
29165 (get_continuation_for_phi_1): Remove.
29167 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29169 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29170 to explain the use of truncating division. Cap the number of
29171 iterations to the maximum given by nb_iterations_upper_bound,
29174 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29176 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29177 * configure: Regenerate.
29178 * config.in: Regenerate.
29179 * config/i386/driver-mingw32.c: new file.
29180 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29181 * config.host: Link driver-mingw32.o on MinGW host.
29182 * doc/install.texi: Document new --enable-mingw-wildcard configure
29185 2017-05-04 Marek Polacek <polacek@redhat.com>
29187 PR tree-optimization/80612
29188 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29190 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29191 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29193 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29194 (movt splitter): Likewise.
29195 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29196 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29197 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29198 block for Thumb-1 with MOVT.
29199 (thumb2_legitimate_address_p): Move code block ...
29200 (can_avoid_literal_pool_for_label_p): ... into this new function.
29201 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29203 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29204 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29205 "M-profile targets with the MOVT instruction".
29207 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29209 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29210 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29211 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29213 2017-05-04 Martin Liska <mliska@suse.cz>
29215 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29216 variable cond_code.
29218 2017-05-04 Richard Biener <rguenther@suse.de>
29220 * tree.c (array_at_struct_end_p): Handle arrays at struct
29221 end with flexarrays more conservatively. Refactor and treat
29222 arrays of arrays or aggregates more strict. Fix
29223 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29224 * tree.c (array_at_struct_end_p): Adjust prototype.
29225 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29226 * gimple-fold.c (get_range_strlen): Likewise.
29227 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29229 2017-05-04 Richard Biener <rguenther@suse.de>
29231 PR tree-optimization/31130
29232 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29234 (supports_overflow_infinity): Likewise.
29235 (is_negative_overflow_infinity): Likewise.
29236 (is_positive_overflow_infinity): Likewise.
29237 (is_overflow_infinity): Likewise.
29238 (stmt_overflow_infinity): Likewise.
29239 (overflow_infinity_range_p): Likewise.
29240 (usable_range_p): Remove as always returning true.
29241 (make_overflow_infinity): Remove.
29242 (negative_overflow_infinity): Likewise.
29243 (positive_overflow_infinity): Likewise.
29244 (avoid_overflow_infinity): Likewise.
29245 (set_value_range): Adjust accordingly.
29246 (set_value_range_to_nonnegative): Likewise, remove now unused
29247 overflow_infinity arg.
29248 (vrp_operand_equal_p): Adjust.
29249 (update_value_range): Likewise.
29250 (range_int_cst_singleton_p): Likewise.
29251 (operand_less_p): Likewise.
29252 (compare_values_warnv): Likewise.
29253 (extract_range_for_var_from_comparison_expr): Likewise.
29254 (vrp_int_const_binop): Likewise.
29255 (zero_nonzero_bits_from_vr): Likewise.
29256 (extract_range_from_multiplicative_op_1): Likewise.
29257 (extract_range_from_binary_expr_1): Likewise.
29258 (extract_range_from_unary_expr): Likewise.
29259 (extract_range_from_comparison): Likewise.
29260 (extract_range_basic): Likewise.
29261 (adjust_range_with_scev): Likewise.
29262 (compare_ranges): Likewise.
29263 (compare_range_with_value): Likewise.
29264 (dump_value_range): Likewise.
29265 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29267 (simplify_cond_using_ranges): Adjust.
29269 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29271 * brig-builtins.def: Added a builtin for class_f64.
29272 * builtin-types.def: Added a builtin type needed by class_f64.
29274 2017-05-03 Jason Merrill <jason@redhat.com>
29276 * timevar.def: Add TV_CONSTEXPR.
29278 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29280 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29282 2017-05-03 Martin Jambor <mjambor@suse.cz>
29284 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29285 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29286 * ipa-cp.c (ipcp_propagate_stage): Do not call
29287 ipa_update_after_lto_read.
29288 * ipa-inline.c (ipa_inline): Likewise.
29290 2017-05-03 Martin Jambor <mjambor@suse.cz>
29292 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29293 tag. Added a default constructor and a destructor.
29294 (ipa_edge_args_sum_t): New class;
29295 (ipa_edge_args_sum): Declare.
29296 (ipa_edge_args_vector): Remove declaration.
29297 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29298 (ipa_free_edge_args_substructures): Remove declaration.
29299 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29300 (ipa_edge_args_info_available_for_edge_p): Likewise.
29301 * ipa-prop.c (ipa_edge_args_vector): Removed.
29302 (edge_removal_hook_holder): Likewise.
29303 (edge_duplication_hook_holder): Likewise.
29304 (ipa_edge_args_sum): New variable.
29305 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29306 ipa_edge_args_vector.
29307 (ipa_free_edge_args_substructures): Likewise.
29308 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29309 ipa_edge_args_vector.
29310 (ipa_edge_removal_hook): Turned into method
29311 ipa_edge_args_sum_t::remove.
29312 (ipa_edge_duplication_hook): Turned into method
29313 ipa_edge_args_sum_t::duplicate.
29314 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29315 registering edge hooks.
29316 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29317 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29318 ipa_edge_args_sum instead of ipa_edge_args_vector.
29319 * ipa-profile.c (ipa_profile): Likewise.
29321 2017-05-03 Martin Jambor <mjambor@suse.cz>
29323 * symbol-summary.h (function_summary): New method exists.
29324 (function_summary::symtab_removal): Deallocate through release.
29325 (call_summary): New class.
29326 (gt_ggc_mx): New overload.
29327 (gt_pch_nx): Likewise.
29328 (gt_pch_nx): Likewise.
29330 2017-05-03 Jeff Law <law@redhat.com>
29332 PR tree-optimization/78496
29333 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29334 from simplify_cond_using_ranges. Split off code to walk
29335 backwards through casts into ...
29336 (simplify_cond_using_ranges_2): New function.
29337 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29338 (execute_vrp): After identifying jump threads, call
29339 simplify_cond_using_ranges_2.
29341 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29344 * ipa-inline.h (inline_summary): Add ctor.
29345 (create_ggc): Do not use ggc_cleared_alloc.
29347 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29348 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29350 * gcc.c (handle_braces): Support escaping in switch matching
29352 * doc/invoke.texi (Spec Files): Document it.
29353 Remove superfluous @code markup in items.
29355 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29357 * diagnostic-show-locus.c (struct column_range): New struct.
29358 (get_affected_columns): New function.
29359 (get_printed_columns): New function.
29360 (struct correction): New struct.
29361 (correction::ensure_capacity): New function.
29362 (correction::ensure_terminated): New function.
29363 (struct line_corrections): New struct.
29364 (line_corrections::~line_corrections): New dtor.
29365 (line_corrections::add_hint): New function.
29366 (layout::print_trailing_fixits): Reimplement in terms of the new
29368 (selftest::test_overlapped_fixit_printing): New function.
29369 (selftest::diagnostic_show_locus_c_tests): Call it.
29371 2017-05-03 Nathan Sidwell <nathan@acm.org>
29373 Canonicalize canonical type hashing
29374 * tree.h (type_hash_canon_hash): Declare.
29375 * tree.c (type_hash_list, attribute_hash_list): Move into
29376 type_hash_canon_hash.
29377 (build_type_attribute_qual_variant): Break out hash code calc into
29378 type_hash_canon_hash.
29379 (type_hash_canon_hash): New. Generic type hash computation.
29380 (build_range_type_1, build_array_type_1, build_function_type,
29381 build_method_type_directly, build_offset_type, build_complex_type,
29382 make_vector_type): Call it.
29384 2017-05-03 Richard Biener <rguenther@suse.de>
29386 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29387 When all DRs have unknown misaligned do not always peel
29388 when there is a store but apply the same costing model as if
29389 there were only loads.
29391 2017-05-03 Richard Biener <rguenther@suse.de>
29394 PR tree-optimization/80492
29395 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29396 compare_base_decls returning dont-know properly.
29398 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29400 * config/arm/iterators.md (CCSI): New mode iterator.
29401 (arch): New mode attribute.
29402 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29403 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29404 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29405 code iterator for success result mode.
29406 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29407 the corresponding new insn generators.
29409 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29412 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29413 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29415 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29417 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29418 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29419 (DDR_A): Wrap DDR argument in brackets.
29420 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29421 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29422 (DDR_REVERSED_P): Likewise.
29424 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29426 PR tree-optimization/79472
29427 * tree-switch-conversion.c (struct switch_conv_info): Add
29428 contiguous_range and default_case_nonstandard fields.
29429 (collect_switch_conv_info): Compute contiguous_range and
29430 default_case_nonstandard fields, don't clear final_bb if
29431 contiguous_range and only the default case doesn't have the required
29433 (check_all_empty_except_final): Set default_case_nonstandard instead
29434 of failing if contiguous_range and the default case doesn't have empty
29436 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29437 and only the default case doesn't have the required constants. Skip
29439 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29440 if default_case_nonstandard.
29441 (build_constructors): Build constant 1 just once. Assert that default
29442 values aren't inserted in between cases if contiguous_range. Skip
29444 (build_arrays): Skip virtual phis.
29445 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29446 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29447 Handle virtual phis.
29448 (gen_inbound_check): Handle default_case_nonstandard case.
29449 (process_switch): Adjust check_final_bb caller. Call
29450 gather_default_values with the first non-default case instead of
29451 default case if default_case_nonstandard.
29453 2017-05-02 Nathan Sidwell <nathan@acm.org>
29455 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29456 check. Fix formatting.
29458 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29460 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29461 errors when comparing specialized and unspecialized times.
29463 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29465 * diagnostic-show-locus.c
29466 (layout::should_print_annotation_line_p): Make private.
29467 (layout::print_annotation_line): Make private.
29468 (layout::annotation_line_showed_range_p): Make private.
29469 (layout::show_ruler): Make private.
29470 (layout::print_source_line): Make private. Pass in line and
29471 line_width, rather than calling location_get_source_line. Drop
29473 (layout::print_leading_fixits): New method.
29474 (layout::print_any_fixits): Rename to...
29475 (layout::print_trailing_fixits): ...this, and make private.
29476 Don't print newline fixits.
29477 (diagnostic_show_locus): Move logic for printing one row into...
29478 (layout::print_line): ...this new function. Move the
29479 location_get_source_line call and error-handling from
29480 print_source_line to here. Call print_leading_fixits, and rename
29481 print_any_fixits to print_trailing_fixits.
29482 (selftest::test_fixit_insert_containing_newline): Update now that
29483 newlines are partially supported.
29484 (selftest::test_fixit_insert_containing_newline_2): New test.
29485 (selftest::test_fixit_replace_containing_newline): Update comments.
29486 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29487 * edit-context.c (class added_line): New class.
29488 (class edited_line): Describe newline handling in comment.
29489 (edited_line::actually_edited_p): New method.
29490 (edited_line::print_content): Delete redundant decl.
29491 (edited_line::m_predecessors): New field.
29492 (edited_file::print_content): Call edited_line::print_content.
29493 (edited_file::print_diff): Update to support newlines.
29494 (edited_file::print_diff_hunk): Likewise.
29495 (edited_file::print_run_of_changed_lines): New function.
29496 (edited_file::print_diff_line): Convert to...
29497 (print_diff_line): ...this.
29498 (edited_file::get_effective_line_count): New function.
29499 (edited_line::edited_line): Initialize new field m_predecessors.
29500 (edited_line::~edited_line): Clean up m_predecessors.
29501 (edited_line::apply_fixit): Handle newlines.
29502 (edited_line::get_effective_line_count): New function.
29503 (edited_line::print_content): New function.
29504 (edited_line::print_diff_lines): New function.
29505 (selftest::test_applying_fixits_insert_containing_newline): New
29507 (selftest::test_applying_fixits_replace_containing_newline): New
29509 (selftest::insert_line): New function.
29510 (selftest::test_applying_fixits_multiple_lines): Add example of
29512 (selftest::edit_context_c_tests): Call the new tests.
29514 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29516 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29517 parameter cand. Update dump information.
29518 (get_computation_cost): Update uses.
29520 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29522 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29523 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29524 (get_computation_at, rewrite_use_address): Update use of
29525 get_computation_aff.
29527 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29529 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29530 (get_computation): Delete.
29531 (get_computation_cost): Implement like get_computation_cost_at.
29532 Use get_computation_at.
29533 (get_computation_cost_at): Delete.
29534 (rewrite_use_nonlinear_expr): Use get_computation_at.
29535 (rewrite_use_compare, remove_unused_ivs): Ditto.
29537 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29539 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29541 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29543 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29544 (ivopts_global_cost_for_size): Rename parameter and update uses.
29545 (iv_ca_recount_cost): Update uses.
29546 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29547 candidates seperately in n_invs and n_cands.
29548 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29550 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29552 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29553 (find_inv_vars_cb): New.
29554 (find_depends): Renamed to ...
29555 (find_inv_vars): ... this.
29556 (add_candidate_1, force_var_cost): Call find_inv_vars.
29557 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29559 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29561 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29562 inv_vars. Add inv_exprs.
29563 (struct iv_cand): Rename depends_on to inv_vars.
29564 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29565 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29566 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29567 (dump_cand): Dump inv_vars.
29568 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29569 (record_invariant, find_depends, add_candidate_1): Ditto.
29570 (set_group_iv_cost, force_var_cost): Ditto.
29571 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29572 (get_computation_cost_at, get_computation_cost): Ditto.
29573 (determine_group_iv_cost_generic): Ditto.
29574 (determine_group_iv_cost_address): Ditto.
29575 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29576 (determine_group_iv_costs): Ditto.
29577 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29578 (iv_ca_set_remove_invariants): Renamed to ...
29579 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29580 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29581 (iv_ca_set_add_invariants): Renamed to ...
29582 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29583 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29584 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29585 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29586 (create_new_ivs): Remove useless dump.
29588 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29590 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29592 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29593 (iv_ca_set_no_cp, create_new_iv): Ditto.
29595 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29597 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29599 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29601 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29602 function tree_check2.
29604 2017-05-02 Martin Liska <mliska@suse.cz>
29606 * doc/gcov.texi: Add missing preposition.
29607 * gcov.c (function_info::function_info): Properly fill up
29608 all member variables.
29610 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29612 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29614 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29616 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29618 2017-05-02 Martin Liska <mliska@suse.cz>
29621 * lto-streamer-in.c (lto_read_tree_1): Remove
29622 LTO_STREAMER_DEBUG.
29623 * lto-streamer.c (struct tree_hash_entry): Likewise.
29624 (struct tree_entry_hasher): Likewise.
29625 (tree_entry_hasher::hash): Likewise.
29626 (tree_entry_hasher::equal): Likewise.
29627 (lto_streamer_init): Likewise.
29628 (lto_orig_address_map): Likewise.
29629 (lto_orig_address_get): Likewise.
29630 (lto_orig_address_remove): Likewise.
29631 * lto-streamer.h: Likewise.
29632 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29633 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29635 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29637 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29638 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29639 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29640 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29641 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29642 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29643 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29644 (mm_maskz_sub_ss): New intrinsics.
29645 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29646 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29647 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29648 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29649 (__builtin_ia32_subss_mask_round): New builtins.
29650 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29651 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29652 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29654 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29655 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29657 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29660 2017-05-02 Martin Jambor <mjambor@suse.cz>
29662 PR tree-optimization/78687
29663 * tree-sra.c (access): New field parent.
29664 (process_subtree_disqualification): New function.
29665 (disqualify_candidate): Call it.
29666 (build_accesses_from_assign): Reset write flag if creating an
29668 (build_access_subtree): Fill in parent field and also prpagate
29669 down grp_write flag.
29670 (create_artificial_child_access): New parameter set_grp_write, set
29671 grp_write to its value.
29672 (propagate_subaccesses_across_link): Also propagate grp_write flag
29674 (propagate_all_subaccesses): Push the closest parent back to work
29675 queue if add_access_to_work_queue returned true.
29677 2017-05-02 Richard Biener <rguenther@suse.de>
29679 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29680 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29681 -fstrict-overflow documentation.
29682 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29683 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29684 flag_strict_overflow.
29685 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29686 * lto-opts.c (lto_write_options): Do not stream it.
29687 * lto-wrapper.c (merge_and_complain): Do not handle it.
29688 * opts.c (default_options_table): Do not set -fstrict-overflow.
29689 (finish_options): Likewise do not clear it when sanitizing.
29690 * simplify-rtx.c (simplify_const_relational_operation): Do not
29691 test flag_strict_overflow.
29693 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29695 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29696 using enabled attribute.
29697 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29698 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29699 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29700 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29701 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29702 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29703 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29704 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29705 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29706 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29708 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29710 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29712 2017-05-02 Richard Biener <rguenther@suse.de>
29714 PR tree-optimization/80591
29716 2017-04-10 Richard Biener <rguenther@suse.de>
29718 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29721 2017-05-02 Richard Biener <rguenther@suse.de>
29723 PR tree-optimization/80549
29724 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29725 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29726 headers if they do not have a preheader.
29728 2017-05-02 Martin Liska <mliska@suse.cz>
29731 * common.opt: Fix typo.
29732 * doc/invoke.texi: Likewise.
29734 2017-05-01 Jan Beulich <jbeulich@suse.com>
29736 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29737 swapping, add (x,x,m,x,n) alternative.
29739 2017-05-01 Nathan Sidwell <nathan@acm.org>
29741 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29742 unnecessary unadjusted_alignment check.
29744 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29747 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29749 * gimplify.c (gimplify_cilk_detach): New function.
29750 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29751 * tree-core.h: Document EXPR_CILK_SPAWN.
29752 * tree.h (EXPR_CILK_SPAWN): Define.
29754 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29756 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29757 to use new fixit_hint representation, using the "replace" logic.
29758 (get_line_span_for_fixit_hint): Likewise.
29759 (layout::print_any_fixits): Likewise.
29760 (selftest::test_one_liner_many_fixits): Rename to...
29761 (selftest::test_one_liner_many_fixits_1): ...this, and update
29762 comment and expected output to reflect that the multiple fix-it
29763 hints are now consolidated into one insertion.
29764 (selftest::test_one_liner_many_fixits_2): New test.
29765 (selftest::test_diagnostic_show_locus_one_liner): Update for
29767 (selftest::test_fixit_consolidation): Update for fix-it API
29769 * diagnostic.c (print_parseable_fixits): Likewise.
29770 * edit-context.c (edited_line::m_line_events): Convert from
29771 auto_vec <line_event *> to auto_vec <line_event>.
29772 (class line_event): Convert from abstract base class to a concrete
29773 class, taking over the role of replace_event.
29774 (class insert_event): Delete.
29775 (class replace_event): Rename to class line_event. Convert to
29777 (edit_context::add_fixits): Reimplement.
29778 (edit_context::apply_insert): Delete.
29779 (edit_context::apply_replace): Rename to...
29780 (edit_context::apply_fixit): ...this. Convert to half-open range.
29781 (edited_file::apply_insert): Delete.
29782 (edited_file::apply_replace): Rename to...
29783 (edited_file::apply_fixit): ...this.
29784 (edited_line::~edited_line): Drop deletion of events.
29785 (edited_line::apply_insert): Delete.
29786 (edited_line::apply_replace): Rename to...
29787 (edited_line::apply_fixit): ...this. Convert to half-open range.
29788 Update for change to type of m_line_events.
29789 * edit-context.h (edit_context::apply_insert): Delete.
29790 (edit_context::apply_replace): Rename to...
29791 (edit_context::apply_fixit): ...this.
29793 2017-05-01 Martin Sebor <msebor@redhat.com>
29795 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29798 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29801 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29802 __get_cpuid_max returns 0.
29803 (__get_cpuid_count): Ditto.
29805 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29807 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29808 replacement expression is another instance of one of its arguments.
29810 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29813 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29814 check for stack push/pop autoinc.
29815 * config/i386/i386.c (ix86_agi_dependent): Return false
29816 if the only reason why modified_in_p returned true is that
29817 addr is SP based and set_insn is a push or pop.
29819 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29821 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29824 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29827 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29828 (account_size_time): Use two predicates - exec_pred and
29830 (evaluate_conditions_for_known_args): Compute both clause and
29832 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29833 (inline_summary_t::duplicate): Update.
29834 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29836 (compute_inline_parameters): Likewise.
29837 (estimate_edge_size_and_time): Update caluclation of time.
29838 (estimate_node_size_and_time): Compute both time and nonspecialized
29840 (estimate_ipcp_clone_size_and_time): Update.
29841 (inline_merge_summary): Update.
29842 (do_estimate_edge_time): Update.
29843 (do_estimate_edge_size): Update.
29844 (do_estimate_edge_hints): Update.
29845 (inline_read_section, inline_write_summary): Stream both new predicates.
29846 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29848 (compute_inlined_call_time): Cleanup.
29849 (big_speedup_p): Update.
29850 (edge_badness): Update.
29851 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29852 (size_time_entry): Replace predicate by exec_predicate and
29853 nonconst_predicate.
29854 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29855 (estimate_edge_time): Return also nonspec_time.
29856 (reset_edge_growth_cache): Update.
29858 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29860 PR rtl-optimization/80491
29861 * ifcvt.c (noce_process_if_block): When looking for x setter
29862 with missing else_bb, don't check only the insn right before
29863 cond_earliest, but look for the last insn that x is modified in
29864 within the same bb.
29866 PR rtl-optimization/80491
29867 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29869 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29871 PR tree-optimization/80487
29872 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29874 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29876 PR tree-optimization/79697
29877 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29878 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29879 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29881 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29882 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29884 2017-04-28 Martin Sebor <msebor@redhat.com>
29886 PR tree-optimization/80523
29887 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29888 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29890 (maybe_warn, format_directive, parse_directive): Use new functions.
29891 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29893 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29895 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29897 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29899 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29900 target_header_dir): Set correctly.
29901 * configure: Regenerated.
29902 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29903 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29904 instead of SYSTEM_HEADER_DIR.
29906 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
29908 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
29909 (estimate_local_effects): Likewise.
29910 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
29911 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
29912 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
29913 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
29914 do_estimate_edge_time, estimate_edge_time): Likewise.
29915 * ipa-inline-analysis.c (estimate_node_size_and_time,
29916 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
29917 (estimate_time_after_inlining): Remove.
29919 2017-04-28 Martin Liska <mliska@suse.cz>
29921 * doc/gcov.texi: Enhance documentation of gcov.
29923 2017-04-28 Martin Liska <mliska@suse.cz>
29925 * doc/gcov.texi: Sort options in alphabetic order.
29926 * doc/gcov-dump.texi: Likewise.
29927 * doc/gcov-tool.texi: Likewise.
29928 * gcov.c (print_usage): Likewise.
29929 * gcov-dump.c (print_usage): Likewise.
29930 * gcov-tool.c (print_merge_usage_message): Likewise.
29931 (print_rewrite_usage_message): Likewise.
29932 (print_overlap_usage_message): Likewise.
29934 2017-04-28 Martin Liska <mliska@suse.cz>
29936 PR gcov-profile/53915
29937 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
29939 2017-04-28 Martin Liska <mliska@suse.cz>
29941 PR gcov-profile/79891
29942 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
29943 is marked by compiler as living on a line.
29944 (get_cycles_count): Remove usage of the union.
29945 (output_intermediate_file): Likewise.
29946 (find_source): Fix GNU coding style.
29947 (accumulate_line_counts): Remove old non-all block mode.
29948 (output_lines): Remove usage of the union.
29949 * profile.c (output_location): Include all BBs, even if
29950 belonging to a same line (and file) as a previous BB.
29952 2017-04-28 Martin Liska <mliska@suse.cz>
29954 * gcov.c (process_args): Handle new argument 'w'.
29955 (read_graph_file): Assign ID to BBs.
29956 (output_branch_count): Display BB # if verbose flag is set.
29957 (output_lines): Likewise for arcs.
29958 (print_usage): Add '--verbose' option help.
29959 * doc/gcov.texi: Document --verbose (-w) option.
29961 2017-04-28 Martin Liska <mliska@suse.cz>
29963 * gcov.c (struct block_location_info): New struct.
29964 (process_file): Fill up the new structure.
29965 (read_graph_file): Replace usage of encoding by the newly added
29967 (add_line_counts): Likewise.
29968 (accumulate_line_counts): Remove usage of the union.
29969 (function_info::function_info): New function.
29970 (function_info::~function_info): Likewise.
29971 (process_file): Call delete instead of release_function.
29972 (release_function): Release the function.
29973 (release_structures): Call delete instead of release_function.
29974 (solve_flow_graph): Replace usage of num_blocks.
29975 (find_exception_blocks): Likewise.
29976 (output_lines): Fix GNU coding style.
29978 2017-04-28 Martin Liska <mliska@suse.cz>
29981 * coverage.c (coverage_remove_note_file): New function.
29982 * coverage.h: Declare the function.
29983 * toplev.c (finalize): Clean if an error has been seen.
29985 2017-04-28 Martin Liska <mliska@suse.cz>
29987 PR gcov-profile/80031
29988 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
29989 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
29990 * gcov.c (read_graph_file): Read just number of blocks.
29991 * profile.c (branch_prob): Do not stream 0 flags per a basic
29994 2017-04-28 Martin Liska <mliska@suse.cz>
29996 * gcov-dump.c (tag_*): Add new argument to declarations.
29997 (dump_gcov_file): Likewise.
29998 (tag_blocks): Add and use new argument depth.
29999 (tag_arcs): Likewise.
30000 (tag_lines): Likewise.
30001 (tag_counters): Likewise.
30002 (tag_summary): Likewise.
30003 (dump_working_sets): Use depth to do a proper indentation.
30005 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30008 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30009 * symtab.c (symtab_node::debug_symtab): Move definition here.
30011 2017-04-28 Richard Biener <rguenther@suse.de>
30013 * lto-streamer.h (LTO_major_version): Bump to 7.
30015 2017-04-28 Richard Biener <rguenther@suse.de>
30017 * tree-vrp.c (assert_info): New struct.
30018 (add_assert_info): New helper.
30019 (register_edge_assert_for_2): Refactor to add asserts to a vector
30021 (register_edge_assert_for_1): Likewise.
30022 (register_edge_assert_for): Likewise.
30023 (finish_register_edge_assert_for): New helper actually registering
30024 asserts where live on edge.
30025 (find_conditional_asserts): Adjust.
30026 (find_switch_asserts): Likewise.
30027 (evrp_dom_walker::try_find_new_range): Generalize.
30028 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30030 2017-04-27 Marek Polacek <polacek@redhat.com>
30033 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30034 arg10 and arg11 to itype.
30036 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30038 * doc/extend.texi (Object Size Checking): Improve grammar.
30040 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30043 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30044 that the logic for permitting reciprocal estimates matches that
30047 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30050 * tree.c (type_cache_hasher::equal): Only compare
30051 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30052 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30053 non-aggregate element types.
30054 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30055 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30057 2017-04-27 Richard Biener <rguenther@suse.de>
30059 PR middle-end/80533
30060 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30061 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30062 keeping a reference to a trailing array.
30064 2017-04-27 Richard Biener <rguenther@suse.de>
30066 PR middle-end/80539
30067 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30068 being in loop-closed SSA form conservatively.
30069 (chrec_fold_multiply_poly_poly): Likewise.
30071 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30073 PR middle-end/79665
30074 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30075 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30077 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30080 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30081 (aarch64_function_arg_alignment): Return unsigned int again, but still
30082 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30083 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30084 Don't emit -Wpsabi note.
30085 (aarch64_function_arg_boundary): Likewise.
30086 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30089 2017-04-26 Nathan Sidwell <nathan@acm.org>
30091 * tree.h (crc32_unsigned_n): Declare.
30092 (crc32_unsigned, crc32_unsigned): Make inline.
30093 * tree.c (crc32_unsigned_bits): Replace with ...
30094 (crc32_unsigned_n): ... this.
30095 (crc32_unsigned, crc32_byte): Remove.
30096 (crc32_string): Remove unnecessary braces.
30098 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30100 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30101 * ipa-inline-analysis.c (MAX_TIME): Remove.
30102 (account_size_time): Use sreal for time.
30103 (dump_inline_summary): Update.
30104 (estimate_function_body_sizes): Update.
30105 (estimate_edge_size_and_time): Update.
30106 (estimate_calls_size_and_time): Update.
30107 (estimate_node_size_and_time): Update.
30108 (inline_merge_summary): Update.
30109 (inline_update_overall_summary): Update.
30110 (estimate_time_after_inlining): Update.
30111 (inline_read_section): Update.
30112 (inline_write_summary): Update.
30113 * ipa-inline.c (compute_uninlined_call_time): Update.
30114 (compute_inlined_call_time): Update.
30115 (recursive_inlining): Update.
30116 (inline_small_functions): Update.
30117 (dump_overall_stats): Update.
30118 * ipa-inline.h: Include sreal.h.
30119 (size_time_entry): Turn time to sreal.
30120 (inline_summary): Turn self_time nad time to sreal.
30122 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30124 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30126 (sreal::stream_out, sreal::stream_in): New.
30127 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30129 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30131 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30134 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30137 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30139 (dimode_scalar_chain::compute_convert_gain): Ditto.
30140 (dimode_scalar_chain::make_vector_copies): Ditto.
30141 (dimode_scalar_chain::convert_reg): Ditto.
30142 (dimode_scalar_chain::convert_insn): Ditto.
30143 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30144 (VI248_AVX512BW_1): New mode iterator.
30145 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30146 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30149 2017-04-25 Martin Sebor <msebor@redhat.com>
30151 PR tree-optimization/80497
30152 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30153 constants are representable in HOST_WIDE_INT.
30154 (parse_directive): Ditto.
30156 2017-04-25 Martin Sebor <msebor@redhat.com>
30159 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30160 (new_zero_array): Adjust signature.
30161 (dom_info::dom_init): Used unsigned rather that size_t.
30162 (dom_info::dom_info): Same.
30164 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30165 Jakub Jelinek <jakub@redhat.com>
30168 * config/arm/arm.c: Include gimple.h.
30169 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30170 returns negative, increment ncrn only if it returned positive.
30171 (arm_needs_doubleword_align): Return int instead of bool,
30172 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30173 members, but if there is any such non-FIELD_DECL
30174 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30175 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30176 returns negative, increment nregs only if it returned positive.
30177 (arm_setup_incoming_varargs): Likewise.
30178 (arm_function_arg_boundary): Emit -Wpsabi note if
30179 arm_needs_doubleword_align returns negative, return
30180 DOUBLEWORD_ALIGNMENT only if it returned positive.
30182 2017-04-25 Marek Polacek <polacek@redhat.com>
30185 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30186 first argument to type.
30188 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30191 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30192 type checks to test for compatibility instead of equality.
30194 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30195 Jakub Jelinek <jakub@redhat.com>
30198 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30200 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30201 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30202 the alignment computation, but return their maximum in warn_alignment.
30203 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30204 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30206 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30207 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30210 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30212 * config/arc/simdext.md (dmpyh): Fix typo.
30214 2017-04-25 Richard Biener <rguenther@suse.de>
30216 PR tree-optimization/80492
30217 * alias.c (compare_base_decls): Handle registers with asm
30218 specification conservatively.
30219 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30220 compare_base_decls returning dont-know properly.
30222 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30224 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30225 (legitimate_offset_address_p): New function.
30226 (arc_legitimate_address_p): Use above function.
30228 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30230 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30232 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30234 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30235 ACCH registers whenever they are available.
30237 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30239 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30240 double regs fix when not used.
30242 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30244 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30246 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30247 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30249 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30251 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30252 when emitting short ADD instructions.
30254 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30256 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30258 (cmpsi_cc_c_insn): Likewise.
30259 (cbranchsi4_scratch): Compute proper instruction length using
30260 compact_hreg_operand.
30261 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30263 2017-04-25 Richard Biener <rguenther@suse.de>
30265 PR middle-end/80509
30266 * passes.c (pass_manager::pass_manager): Initialize
30267 m_name_to_pass_map.
30269 2017-04-25 Richard Biener <rguenther@suse.de>
30271 PR tree-optimization/79201
30272 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30274 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30277 * config/s390/vector.md: Split MEM->GPR vector moves for
30278 non-s_operand addresses.
30280 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30283 * config/s390/predicates.md (reload_const_wide_int_operand): New
30285 * config/s390/s390.md ("movti"): Remove d/P alternative.
30286 ("movti_bigconst"): New pattern definition.
30288 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30291 * s390-protos.h (s390_expand_cs_hqi): Removed.
30292 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30293 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30294 modes as well as CCZ1mode and CCZmode.
30295 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30296 signature of s390_emit_compare_and_swap.
30297 (s390_expand_cs_hqi): Likewise, make static.
30298 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30299 compare-and-swap, in some cases.
30300 (s390_expand_cs): Wrapper function.
30301 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30303 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30304 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30305 patterns for small and large integers. Forbid symref memory operands.
30306 Move expander to s390.c. Require cc register.
30307 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30308 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30309 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30310 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30311 symref memory operands. Remove CC mode and call s390_match_ccmode
30313 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30315 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30317 * config/s390/s390.md (define_peephole2): New peephole to help
30318 combining the load-and-test pattern with volatile memory.
30320 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30322 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30323 with CCZmode for TARGET_Z196.
30325 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30327 PR rtl-optimization/80501
30328 * combine.c (make_compound_operation_int): Set subreg_code to SET
30329 even for AND with mask of the sign bit of mode.
30331 PR rtl-optimization/80500
30332 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30333 sum's initial value.
30335 2017-04-25 Julian Brown <julian@codesourcery.com>
30336 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30338 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30340 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30342 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30344 2017-04-25 Julian Brown <julian@codesourcery.com>
30345 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30347 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30348 (thunderx2t99_sha): New Reservation.
30350 2017-04-25 Julian Brown <julian@codesourcery.com>
30351 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30353 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30354 type for 1-element load.
30356 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30358 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30360 2017-04-24 Martin Jambor <mjambor@suse.cz>
30362 PR tree-optimization/80293
30363 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30364 char arrays not totally scalarizable if it is false.
30365 (analyze_all_variable_accesses): Pass correct value in the new
30366 parameter. Add a statistics counter.
30368 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30370 PR middle-end/79931
30371 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30373 2017-04-24 Richard Biener <rguenther@suse.de>
30375 PR tree-optimization/80494
30376 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30377 out for complex types.
30379 2017-04-24 Richard Biener <rguenther@suse.de>
30381 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30382 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30383 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30384 (DFS): Adjust and never fail.
30385 (sccvn_dom_walker::fail): Remove.
30386 (sccvn_dom_walker::before_dom_children): Adjust.
30387 (run_scc_vn): Likewise and never fail.
30388 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30389 (pass_fre::execute): Likewise.
30391 2017-04-24 Richard Biener <rguenther@suse.de>
30393 PR tree-optimization/79725
30394 * tree-ssa-sink.c (statement_sink_location): Return whether
30395 failure reason was zero uses. Move that check later.
30396 (sink_code_in_bb): Deal with zero uses by removing the stmt
30399 2017-04-24 Richard Biener <rguenther@suse.de>
30402 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30403 pointer-based references.
30405 2017-04-24 Richard Biener <rguenther@suse.de>
30408 * pass_manager.h (pass_manager::operator new): Remove.
30409 (pass_manager::operator delete): Likewise.
30410 * passes.c (pass_manager::operator new): Remove.
30411 (pass_manager::operator delete): Likewise.
30412 (pass_manager::pass_manager): Zero individual pass members.
30414 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30417 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30418 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30419 Check "XEXP (src, 1)" operand here.
30420 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30421 Check "XEXP (src, 1)" operand here.
30422 (dimode_scalar_chain::make_vector_copies): Detect count register
30423 of a shift instruction. Zero extend count register from QImode
30424 to DImode to satisfy vector shift pattern count operand predicate.
30425 Substitute vector shift count operand with a DImode copy.
30426 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30429 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30431 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30432 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30433 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30434 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30435 (UNSPEC_NOREX_MEM): Remove definition.
30437 2017-04-21 Richard Biener <rguenther@suse.de>
30439 PR tree-optimization/79547
30440 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30441 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30442 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30443 without any constraints.
30445 2017-04-21 Richard Biener <rguenther@suse.de>
30447 PR tree-optimization/78847
30448 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30450 2017-04-21 Richard Biener <rguenther@suse.de>
30452 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30453 (build_distinct_type_copy): Likewise.
30454 (build_variant_type_copy): Likewise.
30455 * tree.c (build_qualified_type): Pass down mem-stat info.
30456 (build_distinct_type_copy): Likewise.
30457 (build_variant_type_copy): Likewise.
30459 2017-04-21 Richard Biener <rguenther@suse.de>
30461 PR tree-optimization/80237
30462 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30464 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30465 for a simplified result.
30467 2016-04-21 Richard Biener <rguenther@suse.de>
30469 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30470 sth as strict as a simple_iv but a chrec without symbols and an
30471 operand defined in the loop we are peeling (and not some subloop).
30472 (propagate_constants_for_unrolling): Propagate all constants.
30474 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30477 * config/i386/i386.c (print_reg): Remove assert for disalowed
30478 regno values, call output_operand_lossage instead.
30480 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30483 * config/i386/constraints.md (Yc): New register constraint.
30484 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30485 Use Yc constraint for alternative 2 of operand 0. Remove
30486 preferred_for_speed attribute.
30488 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30490 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30491 lastprivate clauses in SIMT case.
30493 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30495 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30497 2017-04-20 Richard Biener <rguenther@suse.de>
30499 PR tree-optimization/57796
30500 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30501 as N scalar stores.
30502 (vect_model_load_cost): Cost gathers as N scalar loads.
30504 2017-04-20 Richard Biener <rguenther@suse.de>
30506 * ggc-page.c (ggc_allocated_p): Rename to ...
30507 (safe_lookup_page_table_entry): ... this and return the lookup
30509 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30511 2017-04-20 Richard Biener <rguenther@suse.de>
30513 PR tree-optimization/80453
30514 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30515 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30516 from the conditions.
30517 (vn_phi_eq): Pass them down.
30518 (vn_phi_lookup): Record them.
30519 (vn_phi_insert): Likewise.
30521 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30523 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30524 uninitialized variable warning to avoid buffer overrun.
30526 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30529 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30530 is suppressed for '{ 0 }' in C.
30532 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30534 * BASE-VER: Set to 8.0.0.
30536 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30538 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30539 priority .init_array and .fini_array section with SECTION_NOTYPE
30542 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30544 PR middle-end/80423
30545 * tree.h (build_array_type): Add typeless_storage default argument.
30546 * tree.c (type_cache_hasher::equal): Also compare
30547 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30548 (build_array_type): Add typeless_storage argument, set
30549 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30551 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30552 (build_array_type): Likewise. Add typeless_storage argument.
30554 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30555 Jakub Jelinek <jakub@redhat.com>
30557 PR tree-optimization/80426
30558 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30559 operation on symbolic operands, also compute the overflow for the
30560 invariant part when the operation degenerates into a negation.
30562 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30565 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30566 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30569 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30571 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30574 * config/avr/avr.c (tree.h): Include it.
30575 (cgraph.h): Include it.
30576 (avr_encode_section_info): Don't warn for uninitialized progmem
30577 variable if it's just an alias.
30579 2017-04-19 Richard Biener <rguenther@suse.de>
30582 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30583 when needed by AutoPGO.
30585 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30588 * doc/lto.texi: Remove an extra 'that'.
30590 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30592 PR rtl-optimization/80429
30593 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30594 are only used in debug insns.
30596 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30597 Vladimir Makarov <vmakarov@redhat.com>
30599 * config/sparc/predicates.md (input_operand): Add comment. Return
30600 true for any memory operand when LRA is in progress.
30601 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30603 2017-04-18 Jeff Law <law@redhat.com>
30606 * mips.md ({return,simple_return}_internal): Do not overwrite
30609 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30611 PR tree-optimization/80443
30612 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30613 instead of adding 1, subtract -1 and similarly instead of subtracting
30616 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30618 PR rtl-optimization/80357
30619 * haifa-sched.c (tmp_bitmap): New variable.
30620 (model_recompute): Handle duplicate use records.
30621 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30622 (free_global_sched_pressure_data): Free it.
30624 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30627 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30628 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30629 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30630 instead of SYSTEM_HEADER_DIR.
30632 2017-04-18 Jeff Law <law@redhat.com>
30634 PR middle-end/80422
30635 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30636 predecessors after walking up the insn chain.
30638 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30641 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30642 sizetype type into debug info.
30644 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30647 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30648 unneeded test for TARGET_UPPER_REGS_SF.
30649 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30651 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30654 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30655 instead of gsi_after_labels.
30657 2017-04-18 Jeff Law <law@redhat.com>
30659 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30663 2017-04-13 Jeff Law <law@redhat.com>
30664 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30665 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30667 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30670 * config/avr/avr.c (intl.h): Include it.
30671 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30673 2017-04-18 Martin Liska <mliska@suse.cz>
30675 PR gcov-profile/78783
30676 * gcov-tool.c (gcov_output_files): Validate that destination
30677 file is either removed by the tool or by a user.
30679 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30680 Guy Benyei <guybe@mellanox.com>
30682 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30683 block, and do not negate it, the stored id is already negative.
30685 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30687 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30689 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30692 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30693 masks of options that should be turned off if the VSX vector
30694 options are turned off.
30695 (OTHER_P8_VECTOR_MASKS): Likewise.
30696 (OTHER_VSX_VECTOR_MASKS): Likewise.
30697 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30698 rs6000_disable_incompatible_switches to validate no type switches
30700 (rs6000_incompatible_switch): New function to disallow turning on
30701 other vector options if -mno-vsx, -mno-power8-vector, or
30702 -mno-power9-vector are specified.
30704 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30706 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30708 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30710 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30711 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30712 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30713 (ARG_POINTER_CFA_OFFSET): Likewise.
30715 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30717 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30718 conditions to take advantage of various optimizations.
30720 2017-04-13 Jeff Law <law@redhat.com>
30722 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30723 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30724 (zero_extendsidi2_dext): Likewise.
30726 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30729 * fold-const.c (fold_ternary_loc): Revert
30730 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30733 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30735 PR rtl-optimization/80343
30736 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30737 new scratch pseudo.
30739 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30742 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30743 to ubsan_encode_value.
30745 2017-04-13 Jeff Law <law@redhat.com>
30747 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30748 appearing in DEBUG_INSNs.
30750 2017-04-13 Martin Liska <mliska@suse.cz>
30752 PR gcov-profile/80413
30753 * gcov-io.c (gcov_write_string): Copy to buffer just when
30754 allocated size is greater than zero.
30756 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30759 * dwarf2out.c (decls_for_scope): Ignore declarations of
30760 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30762 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30765 * ipa-visibility.c (non_local_p): Fix typos.
30766 (localize_node): When localizing symbol in same comdat group,
30767 dissolve the group only when we know external symbols are going
30769 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30771 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30773 PR tree-optimization/79390
30774 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30775 order does not result in usable sequence, retry with reversed operand
30781 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30782 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30783 op0 instead of fold_convert_loc (loc, type, arg0).
30785 2017-04-12 Jeff Law <law@redhat.com>
30787 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30788 has a delay slot in the generated code.
30790 * config/cris/cris.md (cris_preferred_reload_class): Return
30791 GENNONACR_REGS rather than GENERAL_REGS.
30793 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30796 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30797 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30798 signedness of the result type.
30800 2017-04-12 Richard Biener <rguenther@suse.de>
30801 Jeff Law <law@redhat.com>
30803 PR tree-optimization/80359
30804 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30805 trim stores to TARGET_MEM_REFs.
30807 2017-04-12 Richard Biener <rguenther@suse.de>
30809 PR tree-optimization/79390
30810 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30811 threading case even more.
30813 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30816 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30817 for quad_address_p for TImode, instead of just not indexed_address.
30819 2017-04-12 Richard Biener <rguenther@suse.de>
30820 Bernd Edlinger <bernd.edlinger@hotmail.de>
30822 PR middle-end/79671
30823 * alias.c (component_uses_parent_alias_set_from): Handle
30824 TYPE_TYPELESS_STORAGE.
30825 (get_alias_set): Likewise.
30826 * tree-core.h (tree_type_common): Add typeless_storage flag.
30827 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30828 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30829 for types containing members with TYPE_TYPELESS_STORAGE.
30830 (place_field): Likewise.
30831 (layout_type): Likewise for ARRAY_TYPE.
30832 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30833 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30834 TYPE_TYPELESS_STORAGE.
30835 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30837 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30840 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30841 first argument to type.
30843 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30847 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30848 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30849 (rs6000_expand_binop_builtin): Likewise.
30850 (rs6000_expand_ternop_builtin): Likewise; also add missing
30851 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30852 vshasigma built-ins.
30853 * doc/extend.texi: Document that vec_xxpermdi's third argument
30854 must be a constant.
30856 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30858 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30859 Use shift_const cost parameter when calculating gain of STV shifts.
30861 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30863 PR rtl-optimization/70478
30864 * lra-constraints.c (process_alt_operands): Check memory for
30865 disfavoring memory insn operand.
30867 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30869 PR middle-end/80100
30870 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30871 left shift in unsigned HOST_WIDE_INT type.
30873 PR rtl-optimization/80385
30874 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30875 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30878 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30879 if they have any depend clauses.
30881 2017-04-11 Martin Liska <mliska@suse.cz>
30884 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30885 * ipa-split.c (split_function): Create a local comdat symbol
30886 if caller is in a comdat group.
30888 2017-04-11 Martin Liska <mliska@suse.cz>
30891 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30894 2017-04-11 Martin Sebor <msebor@redhat.com>
30896 PR middle-end/80364
30897 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30898 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30900 (directive::set_width, directive::set_precision, format_character):
30902 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30905 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
30908 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
30909 conflict, set target->arch_name instead of target->cpu_name.
30911 2017-04-11 Richard Biener <rguenther@suse.de>
30913 PR tree-optimization/80374
30914 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
30915 build_zero_cst, remove fold_convertible_p check again.
30917 2017-04-11 Martin Liska <mliska@suse.cz>
30920 * ubsan.c (instrument_object_size): Do not instrument register
30923 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30926 * config/i386/i386-builtin-types.def
30927 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
30928 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
30929 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
30930 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
30931 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
30932 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
30933 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
30934 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
30935 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
30936 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
30937 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
30938 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
30939 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
30940 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
30941 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
30942 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
30943 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
30944 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
30945 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
30946 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
30947 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
30948 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
30949 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
30950 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
30951 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
30952 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
30953 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
30954 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
30955 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
30956 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
30957 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
30958 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
30959 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
30960 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
30961 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
30962 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
30963 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
30964 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
30965 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
30966 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
30967 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
30968 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
30969 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
30970 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
30971 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
30972 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
30974 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
30975 flag to second_arg_count, handle 4 argument function type _COUNT
30976 aliases, handle second_arg_count on second argument rather than last.
30978 2017-04-10 Jeff Law <law@redhat.com>
30980 PR tree-optimization/80374
30981 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
30982 record anything if we can not convert integer_zero_node to the
30985 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30988 * config/rs6000/rs6000.c (rs6000_option_override_internal):
30989 Enhance special handling given to the TARGET_P9_MINMAX option in
30990 relation to certain other options.
30992 2017-04-10 Bin Cheng <bin.cheng@arm.com>
30994 PR tree-optimization/80153
30995 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
30996 remove POINTER_PLUS_EXPR's base part directly, rather than through
30999 2017-04-10 Richard Biener <rguenther@suse.de>
31000 Bin Cheng <bin.cheng@arm.com>
31002 PR tree-optimization/80153
31003 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31004 the first element of pointer type aff_tree. Build result expr in
31006 (add_elt_to_tree): Convert to type unconditionally. Remove other
31007 fold_convert calls.
31008 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31009 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31011 2017-04-10 Richard Biener <rguenther@suse.de>
31013 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31016 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31018 PR rtl-optimization/70478
31019 * lra-constraints.c (curr_small_class_check): New.
31020 (update_and_check_small_class_inputs): New.
31021 (process_alt_operands): Update curr_small_class_check. Disfavor
31022 alternative insn memory operands. Check available regs for small
31025 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31028 * config/mips/mips.opt (-mvirt): Update description.
31029 * doc/invoke.texi (-mvirt): Likewise.
31031 2017-04-10 Richard Biener <rguenther@suse.de>
31033 PR middle-end/80362
31034 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31035 looking for NEGATE_EXPR in -A / -B to A / B folding.
31037 2017-04-10 Martin Liska <mliska@suse.cz>
31039 PR gcov-profile/80224
31040 * gcov.c (print_usage): Fix usage string.
31041 (get_gcov_intermediate_filename): Remove.
31042 (output_gcov_file): Use both for normal and intermediate format.
31043 (generate_results): Do not initialize special file for
31044 intermediate format.
31046 2017-04-10 Richard Biener <rguenther@suse.de>
31048 PR tree-optimization/80304
31049 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31052 2017-04-10 Nathan Sidwell <nathan@acm.org>
31055 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31056 (rs6000_init_builtins): Use it.
31058 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31060 * config/arm/arm.md (<mrc>): Add mode to SET source.
31061 (<mrrc>): Likewise.
31063 2017-04-10 Richard Biener <rguenther@suse.de>
31065 PR middle-end/80344
31066 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31068 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31071 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31072 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31073 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31074 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31075 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31076 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31077 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31078 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31079 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31080 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31081 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31082 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31083 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31084 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31085 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31086 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31087 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31088 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31089 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31090 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31091 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31092 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31093 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31095 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31097 PR rtl-optimization/70478
31098 * lra-constraints.c: Reverse the last patch.
31100 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31102 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31103 Add comment for WCHAR_T.
31105 2017-04-08 Martin Liska <mliska@suse.cz>
31108 2017-04-07 Martin Liska <mliska@suse.cz>
31111 * ipa-split.c (split_function): Add function part to a same comdat
31114 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31117 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31119 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31121 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31122 (vec_pairing): ...to this.
31123 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31124 (rs6000_sched_init): Adjust for name change.
31125 (struct rs6000_sched_context): Likewise.
31126 (rs6000_init_sched_context): Likewise.
31127 (rs6000_set_sched_context): Likewise.
31129 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31135 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31137 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31138 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31139 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31141 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31143 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31145 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31147 PR rtl-optimization/70703
31148 * ira-color.c (update_conflict_hard_regno_costs): Use
31149 int64_t instead of HOST_WIDE_INT.
31151 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31153 PR rtl-optimization/70478
31154 * lra-constraints.c (process_alt_operands): Disfavor alternative
31155 insn memory operands.
31157 2017-04-07 Jeff Law <law@redhat.com>
31159 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31160 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31162 2017-04-07 Martin Liska <mliska@suse.cz>
31165 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31166 Show error message instead of an ICE.
31168 2017-04-07 Martin Liska <mliska@suse.cz>
31171 * ipa-split.c (split_function): Add function part to a same comdat
31174 2017-04-07 Richard Biener <rguenther@suse.de>
31176 PR middle-end/80341
31177 * tree.c (get_unwidened): Also handle ! for_type case for
31179 * convert.c (do_narrow): Split out from ...
31180 (convert_to_integer_1): ... here. Do not pass final truncation
31181 type to get_unwidened for TRUNC_DIV_EXPR.
31183 2017-04-07 Richard Biener <rguenther@suse.de>
31185 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31187 (aff_combination_const): Adjust.
31188 (aff_combination_scale): Likewise.
31189 (aff_combination_add_elt): Likewise.
31190 (aff_combination_add_cst): Likewise.
31191 (aff_combination_convert): Likewise.
31192 (add_elt_to_tree): Likewise. Remove unused argument.
31193 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31195 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31197 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31199 * config/arm/arm.c (arm_default_short_enums): Use
31200 ARM_DEFAULT_SHORT_ENUMS.
31201 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31203 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31206 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31207 members with redundant out-of-class redeclaration.
31209 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31212 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31213 * config/i386/i386.md (*zero_extendsidi2):
31214 Add (?*x,*x) and (?*v,*v) alternatives.
31216 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31219 * config/i386/i386.c (ix86_expand_builtin)
31220 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31221 mode from insn data. Convert operands to insn operand mode.
31222 Copy operands that don't satisfy insn predicate to a register.
31224 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31226 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31229 2017-04-06 Richard Biener <rguenther@suse.de>
31231 PR tree-optimization/80334
31232 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31233 preserve alignment of accesses.
31235 2017-04-06 Richard Biener <rguenther@suse.de>
31237 PR tree-optimization/80262
31238 * tree-sra.c (build_ref_for_offset): Preserve address-space
31240 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31241 Drop useless address-space information on MEM_REF offsets.
31243 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31245 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31247 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31249 PR rtl-optimization/70703
31250 * ira-color.c (update_conflict_hard_regno_costs): Use
31251 HOST_WIDE_INT instead of long.
31253 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31256 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31257 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31259 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31260 for x86_64 target. Handle -m3dnowa option.
31262 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31264 PR rtl-optimization/70703
31265 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31266 (update_conflict_hard_regno_costs): Use long instead of unsigned
31267 arithmetic for cost calculation.
31269 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31270 Bernd Edlinger <bernd.edlinger@hotmail.de>
31273 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31276 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31279 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31280 ptr_mode with Pmode throughout.
31281 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31282 into probe_stack_range and use DImode.
31284 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31287 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31288 call_eh_return is true.
31290 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31292 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31293 Initialize last_match_fntype_index.
31295 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31298 * tree-nvr.c: Include internal-fn.h.
31299 (pass_return_slot::execute): Ignore internal calls without
31302 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31303 Richard Biener <rguenther@suse.de>
31306 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31307 captures used multiple times, except for the last use.
31308 * generic-match-head.c: Include gimplify.h.
31310 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31312 PR tree-optimization/79390
31313 * target.h (struct noce_if_info): Declare.
31314 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31315 * target.def (noce_conversion_profitable_p): New target hook.
31316 * ifcvt.h (struct noce_if_info): New type, moved from ...
31317 * ifcvt.c (struct noce_if_info): ... here.
31318 (noce_conversion_profitable_p): Renamed to ...
31319 (default_noce_conversion_profitable_p): ... this. No longer
31321 (noce_try_store_flag_constants, noce_try_addcc,
31322 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31323 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31324 instead of noce_conversion_profitable_p.
31325 * config/i386/i386.c: Include ifcvt.h.
31326 (ix86_option_override_internal): Don't override
31327 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31328 (ix86_noce_conversion_profitable_p): New function.
31329 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31330 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31331 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31332 * doc/tm.texi: Regenerated.
31334 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31336 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31339 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31342 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31343 instructions for small multiply cores.
31345 2017-04-04 Jeff Law <law@redhat.com>
31347 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31349 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31350 that are not set by the loop over the elements.
31352 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31355 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31356 int mode, convert_modes it to mode as unsigned, otherwise use
31357 lowpart_subreg to mode rather than SImode.
31358 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31359 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31360 Use DImode instead of SImode for the shift count operand.
31361 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31364 2017-04-04 Richard Biener <rguenther@suse.de>
31366 PR middle-end/80281
31367 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31368 arithmetic done for the negate or the plus. Simplify.
31369 (A - (-B) -> A + B): Likewise.
31370 * fold-const.c (split_tree): Make sure to not negate pointers.
31372 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31374 PR rtl-optimization/60818
31375 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31376 a compare of comparisons with the thing compared if this results
31377 in a different machine mode.
31379 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31381 * alias.c (base_alias_check): Fix typo in comment.
31382 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31383 * cgraphunit.c (symbol_table::compile): Likewise.
31384 * collect2.c (maybe_run_lto_and_relink): Likewise.
31385 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31386 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31387 * config/avr/avr.c (avr_map_op_t): Likewise.
31388 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31389 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31390 * config/epiphany/epiphany.md (movcc): Likewise.
31391 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31392 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31394 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31395 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31396 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31397 * config/sh/sh.c (sh_rtx_costs): Likewise.
31398 * fold-const.c (fold_truth_andor): Likewise.
31399 * genautomata.c (collapse_flag): Likewise.
31400 * gengtype.h (struct type::u::s): Likewise.
31401 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31402 * input.c (FORMAT_AMOUNT): Likewise.
31403 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31404 (known_aggs_to_agg_replacement_list): Likewise.
31405 * ipa-inline-analysis.c: Likewise.
31406 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31407 * ipa-polymorphic-call.c
31408 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31409 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31410 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31412 * modulo-sched.c (apply_reg_moves): Likewise.
31413 * omp-expand.c (build_omp_regions_1): Likewise.
31414 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31415 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31416 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31417 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31418 * value-prof.c: Likewise.
31419 * var-tracking.c (val_reset): Likewise.
31421 2017-04-03 Richard Biener <rguenther@suse.de>
31423 PR tree-optimization/80275
31424 * fold-const.c (split_address_to_core_and_offset): Handle
31427 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31429 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31430 descriptors is at least equal to that of functions.
31432 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31434 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31436 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31439 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31440 (mov<IMOD4:mode>): New expander.
31441 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31443 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31445 PR rtl-optimization/79405
31446 * fwprop.c (propagations_left): New variable.
31447 (forward_propagate_into): Decrement it.
31448 (fwprop_init): Initialize it.
31449 (fw_prop): If the variable has reached zero, stop propagating.
31450 (fwprop_addr): Ditto.
31452 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31455 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31456 a FUNCTION_DECL, pass it as decl instead of origin to
31459 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31461 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31464 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31467 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31468 TARGET_VSX_SMALL_INTEGER.
31470 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31472 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31473 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31475 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31477 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31478 extraction from odd-numbered MSA register.
31480 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31482 PR middle-end/80173
31483 * expmed.c (store_bit_field_1): Don't attempt to create
31484 a word subreg out of hard registers wider than word if they
31485 have HARD_REGNO_NREGS of 1 for their mode.
31487 PR middle-end/80163
31488 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31489 conversions to integer types wider than word and pointer.
31492 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31493 (rtx_equal_for_cselib_p): Pass 0 to it.
31494 * cselib.c (cselib_hasher::equal): Likewise.
31495 (rtx_equal_for_cselib_1): Add depth argument. If depth
31496 is 128, don't look up VALUE locs and punt. Increment
31497 depth in recursive calls when walking VALUE locs.
31499 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31501 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31502 (make_gcov_file_name): Use the canonical path name for generating
31504 (read_line): Fix handling of files with ascii null bytes.
31506 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31508 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31509 to initialise a vector register instead
31510 of using a const_int.
31512 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31514 PR translation/80189
31515 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31516 diagnostic messages.
31518 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31521 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31522 (dfp_diex_<mode>): Update mode of operand 1.
31523 * doc/extend.texi (dxex, dxexq): Document change to return type.
31524 (diex, diexq): Document change to argument type.
31526 2017-03-30 Martin Jambor <mjambor@suse.cz>
31529 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31530 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31531 it reflects the signature changes performed at the callee side.
31532 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31533 to cgraph_build_function_type_skip_args.
31534 (build_function_decl_skip_args): Adjust call to the above function.
31536 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31539 * config/i386/sse.md
31540 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31541 register as dest whenever it is a MEM not rtx_equal_p to the
31542 corresponding dup operand, and when forcing into reg move the
31543 reg into the memory afterwards.
31544 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31545 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31546 for the force_reg mode.
31547 (avx512vl_vextractf128<mode>): Use register as dest either
31548 always when a MEM, or when it is a MEM not rtx_equal_p to the
31549 corresponding dup operand, or even not when it is a CONST_VECTOR
31550 depending on the mode and lo vs. hi.
31551 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31553 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31554 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31555 Likewise. Require that operands[2] is even.
31556 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31557 Remove extraneous parens. Require that operands[2] is a multiple
31559 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31560 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31561 disallow memory then.
31563 2017-03-30 Richard Biener <rguenther@suse.de>
31565 PR tree-optimization/77498
31566 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31567 to non-constants over backedges.
31569 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31571 PR rtl-optimization/80233
31572 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31573 as last_combined_insn. Do not test for BARRIER_P separately.
31575 2017-03-29 Andreas Schwab <schwab@suse.de>
31578 * calls.c (prepare_call_address): Convert funexp to Pmode before
31579 copying to temp reg.
31581 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31583 PR tree-optimization/80158
31584 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31585 Handle possible future case of more than one alternate
31587 (replace_rhs_if_not_dup): Likewise.
31588 (replace_one_candidate): Likewise.
31590 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31592 PR rtl-optimization/80193
31593 * ira.c (ira): Do not check allocation for LRA.
31595 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31597 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31598 (nvptx_output_simt_exit): Declare.
31599 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31600 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31601 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31602 (nvptx_declare_function_name): ...here. Emit declaration of local
31603 memory space buffer for omp_simt_enter insn.
31604 (nvptx_output_unisimt_switch): New.
31605 (nvptx_output_softstack_switch): New.
31606 (nvptx_output_simt_enter): New.
31607 (nvptx_output_simt_exit): New.
31608 * config/nvptx/nvptx.h (struct machine_function): New fields
31609 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31610 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31611 (UNSPECV_SIMT_EXIT): Ditto.
31612 (omp_simt_enter_insn): New insn.
31613 (omp_simt_enter): New expansion.
31614 (omp_simt_exit): New insn.
31615 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31617 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31618 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31619 (expand_GOMP_SIMT_EXIT): New.
31620 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31621 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31622 (GOMP_SIMT_EXIT): Ditto.
31623 * target-insns.def (omp_simt_enter): New insn.
31624 (omp_simt_exit): Ditto.
31625 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31627 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31628 (lower_rec_input_clauses): Likewise.
31629 (lower_lastprivate_clauses): Handle SIMT privatization.
31631 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31632 (ompdevlow_adjust_simt_enter): New.
31633 (find_simtpriv_var_op): New.
31634 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31635 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31637 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31638 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31639 (copy_decl_for_dup_finish): Ditto.
31641 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31643 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31646 * config/i386/i386.c (ix86_option_override_internal): Always
31647 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31649 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31651 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31653 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31655 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31656 mark new edge's irreducible flag accordign to it.
31657 (vect_do_peeling): Check loop preheader edge's irreducible flag
31658 and pass it to function slpeel_add_loop_guard.
31660 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31662 PR tree-optimization/80218
31663 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31664 Update block frequencies and counts.
31666 2017-03-28 Richard Biener <rguenther@suse.de>
31668 PR tree-optimization/78644
31669 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31670 of a simplification result we may not use it at all.
31672 2017-03-28 Richard Biener <rguenther@suse.de>
31675 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31676 without arguments, generate default definition of a SSA name.
31678 2017-03-28 Richard Biener <rguenther@suse.de>
31680 PR middle-end/80222
31681 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31682 TYPE_REF_CAN_ALIAS_ALL references.
31683 * fold-const.c (fold_indirect_ref_1): Likewise.
31685 2017-03-28 Martin Liska <mliska@suse.cz>
31688 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31689 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31691 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31692 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31694 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31695 (EXTRA_SPECS): Define.
31696 (SUBTARGET_EXTRA_SPECS): Likewise.
31697 (SUBTARGET_CPP_SPEC): Likewise.
31698 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31699 SUBTARGET_EXTRA_SPECS.
31700 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31702 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31704 * config/arc/simdext.md (vst64_insn): Update pattern.
31705 (vld32wh_insn): Likewise.
31706 (vld32wl_insn): Likewise.
31707 (vld64_insn): Likewise.
31708 (vld32_insn): Likewise.
31710 2017-03-28 Marek Polacek <polacek@redhat.com>
31713 * fold-const.c (fold_comparison): Use protected_set_expr_location
31714 instead of SET_EXPR_LOCATION.
31716 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31718 * tree.c (add_expr): Avoid name lookup warning.
31720 2017-03-27 Jeff Law <law@redhat.com>
31722 PR tree-optimization/80216
31723 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31724 function name. Limit recursion depth.
31725 (record_temporary_equivalences): Corresponding changes.
31727 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31729 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31732 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31735 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31737 * cfgcleanup.c (reg_note_cfa_p): New array.
31738 (insns_have_identical_cfa_notes): New function.
31739 (old_insns_match_p): Don't cross-jump in between /f
31740 and non-/f instructions. If both i1 and i2 are frame related,
31741 verify all CFA notes, their order and content.
31743 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31746 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31747 HImode and SImode with zero extend to DImode to one insn.
31748 (bswap<mode>2_extenddi): Likewise.
31749 (bswapsi2_extenddi): Likewise.
31750 (bswaphi2_extendsi): Likewise.
31751 (bswaphi2): Combine bswap HImode and SImode into one insn.
31752 Separate memory insns from swapping register.
31753 (bswapsi2): Likewise.
31754 (bswap<mode>2): Likewise.
31755 (bswaphi2_internal): Delete, no longer used.
31756 (bswapsi2_internal): Likewise.
31757 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31758 store, and gpr<-gpr swap insns.
31759 (bswap<mode>2_store): Likewise.
31760 (bswaphi2_reg): Register only splitter, combine with the splitter.
31761 (bswaphi2 splitter): Likewise.
31762 (bswapsi2_reg): Likewise.
31763 (bswapsi2 splitter): Likewise.
31764 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31765 the insns into load, store, and register/register insns.
31766 (bswapdi2_ldbrx): Likewise.
31767 (bswapdi2_load): Likewise.
31768 (bswapdi2_store): Likewise.
31769 (bswapdi2_reg): Likewise.
31771 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31773 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31774 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31776 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31779 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31781 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31782 special handling for target option conflicts between dform
31783 options (-mpower9-dform, -mpower9-dform-vector,
31784 -mpower9-dform-scalar) and -mno-direct-move.
31786 2017-03-27 Richard Biener <rguenther@suse.de>
31788 PR tree-optimization/80181
31789 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31791 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31793 * config/arc/predicates.md (move_double_src_operand): Replace the
31794 call to move_double_src_operand with a call to address_operand.
31796 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31798 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31799 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31800 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31802 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31804 * config/arc/predicates.md (long_immediate_loadstore_operand):
31805 Consider scaled addresses cases.
31807 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31809 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31810 restored when in interrupt.
31811 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31812 doesn't have delay slot.
31814 2017-03-27 Richard Biener <rguenther@suse.de>
31817 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31818 inlined thunk clones.
31820 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31823 * asan.c (instrument_derefs): Copy over last operand from
31824 original COMPONENT_REF to the new COMPONENT_REF with
31825 DECL_BIT_FIELD_REPRESENTATIVE.
31826 * ubsan.c (instrument_object_size): Likewise.
31828 2017-03-27 Richard Biener <rguenther@suse.de>
31830 PR tree-optimization/80170
31831 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31832 sure DR/SCEV didnt fold in constants we do not see when looking
31833 at the reference base alignment.
31835 2017-03-27 Richard Biener <rguenther@suse.de>
31837 PR middle-end/80171
31838 * gimple-fold.c (fold_ctor_reference): Properly guard against
31839 NULL return value from canonicalize_constructor_val.
31841 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31844 * config/i386/i386.c (ix86_expand_builtin)
31845 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31846 flags reg setting and flags reg using instructions.
31847 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31848 clobbering instructions to zero extend op2.
31850 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31852 * doc/install.texi (Configuration) <--with-aix-soname>:
31853 Update link to AIX ld.
31855 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31857 PR rtl-optimization/80160
31858 PR rtl-optimization/80159
31859 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31860 reg_alternate_class into account.
31862 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31865 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31866 to consider in curr_insn_transform.
31868 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31870 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31871 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31872 and emit_mode_inner.
31874 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31876 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31877 argument to the overloaded builtin variants. Use the new flag to
31878 deprecate certain builtin variants.
31879 * config/s390/s390-builtin-types.def: Add new builtin types.
31880 * config/s390/s390-builtins.h: Support new flags field for
31881 overloaded builtins.
31882 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31883 (s390_macro_to_expand): Enable vector float data type.
31884 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31885 builtins by incrementing the __VEC__ version number.
31886 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31888 (s390_resolve_overloaded_builtin): Emit error messages depending
31889 on the builtin flags.
31890 * config/s390/s390.c (s390_expand_builtin): Support additional
31891 flags argument. Change error message to match the messages
31892 emitted in s390-c.c.
31893 * config/s390/s390.md: New UNSPEC_* constants.
31894 (op_type): Add new instruction types.
31895 * config/s390/vecintrin.h: Add new builtins and test data class
31897 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31898 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31899 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31900 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31901 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31902 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31904 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31905 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31906 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
31907 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
31909 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
31910 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
31911 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
31912 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
31914 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
31915 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
31916 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
31917 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
31918 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
31919 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
31920 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
31922 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
31923 ("vec_scatter_element<V_HW_4:mode>_DI")
31924 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
31925 ("vec_fpint<mode>", "vflls")
31926 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
31927 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
31928 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
31929 ("*vec_cmphe<mode>_cc"): ... these.
31931 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
31932 mode constant instead of magic value.
31934 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31936 * config/s390/s390.c (s390_expand_vec_compare): Support other
31937 vector floating point modes than just V2DF.
31938 (s390_expand_vcond): Likewise.
31939 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
31940 (s390_cannot_change_mode_class): Prevent mode changes between TF
31941 and V1TF in vector registers.
31942 * config/s390/s390.md (DF, SF): New mode attributes.
31943 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
31944 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
31945 SFmode support for VRs.
31946 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
31948 (VFT, VF_HW): New mode iterators.
31949 (vw, sdx): New mode attributes.
31950 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
31951 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
31952 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
31953 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
31954 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
31955 also the new vector floating point modes. Renaming to ...
31957 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
31958 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
31959 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
31960 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
31961 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
31962 ("vec_unordered<mode>"): ... these.
31964 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
31965 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
31966 ("*vec_extendv2df"): New insn definitions.
31968 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31970 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
31971 ("mulditi3_2", "*muldi3_sign"): New patterns.
31972 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
31973 rename the pattern definition.
31975 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31977 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
31979 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
31981 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31983 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
31984 instruction if possible.
31985 * config/s390/vector.md (vec_halfnumelts): New mode
31987 ("*vec_vllezlf<mode>"): New pattern.
31989 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31991 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
31992 ("popcountv4si2", "popcountv2di2"): Rename to ...
31993 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
31994 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
31996 ("popcount<mode>2_vxe"): New pattern.
31998 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32000 * common/config/s390/s390-common.c (processor_flags_table): Add
32002 * config.gcc: Add arch12.
32003 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32004 Default to arch12 for unknown CPU model numbers.
32005 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32006 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32007 PROCESSOR_max sanity check.
32008 * config/s390/s390-opts.h (enum processor_type): Add
32010 * config/s390/s390.c (processor_table): Add arch12.
32011 (s390_expand_builtin): Add check for B_VXE flag.
32012 (s390_issue_rate): Add PROCESSOR_ARCH12.
32013 (s390_get_sched_attrmask): Likewise.
32014 (s390_get_unit_mask): Likewise.
32015 (s390_sched_score): Enable z13 scheduling for arch12.
32016 (s390_sched_reorder): Likewise.
32017 (s390_sched_variable_issue): Likewise.
32018 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32020 (s390_tune_attr): Use z13 scheduling also for arch12.
32021 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32022 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32023 (TARGET_VXE_P): New macros.
32024 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32025 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32026 * config/s390/s390.opt: Add arch12 as processor_type.
32028 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32030 * config/s390/s390.md
32031 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32032 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32033 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32035 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32036 Rename expanders to ...
32038 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32039 ("fixuns_truncdddi2_emu"): ... these.
32041 ("fixuns_trunc<mode>si2_emu"): New expander.
32043 ("*fixuns_truncdfdi2_z13"): Rename to ...
32044 ("*fixuns_truncdfdi2_vx"): ... this.
32046 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32048 * config/s390/2964.md: Remove the single element vector compare
32049 instructions which are no longer used.
32050 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32052 (s390_canonicalize_comparison): Remove handling of DFmode
32054 (s390_expand_vec_compare_scalar): Remove function.
32055 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32056 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32058 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32060 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32062 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32064 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32065 will anyway by matched by mov<mode>_64dfp.
32067 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32069 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32070 vlef/vstef. Add missing operand to vleif.
32072 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32074 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32075 pair for all vector types with 64 bit elements.
32076 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32077 * config/s390/vector.md (V_HW_64): ... here.
32078 (V_128_NOSINGLE): New mode iterator.
32079 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32080 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32081 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32082 ("*vec_load_pairv2di"): Change to ...
32083 ("*vec_load_pair<mode>"): ... this one.
32085 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32087 * config/s390/constraints.md: Add comments.
32088 (jKK): Reject element sizes > 8 bytes.
32089 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32091 * config/s390/s390.md: Add the s_operand checks formerly in
32092 s390_split_ok_p to various splitters where they are still
32094 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32095 for 128 bit vectors. Plus two splitters.
32097 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32099 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32102 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32105 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32106 error if the boundary argument is not constant.
32108 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32110 PR rtl-optimization/80112
32111 * loop-doloop.c (doloop_condition_get): Don't check condition
32112 if cmp isn't SET with IF_THEN_ELSE src.
32114 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32116 PR tree-optimization/80158
32117 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32118 replacing a candidate statement, also replace it for the
32119 candidate's alternate interpretation.
32120 (replace_rhs_if_not_dup): Likewise.
32121 (replace_one_candidate): Likewise.
32123 2017-03-24 Richard Biener <rguenther@suse.de>
32125 PR tree-optimization/80167
32126 * graphite-isl-ast-to-gimple.c
32127 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32129 (translate_isl_ast_to_gimple::get_rename): Likewise.
32131 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32133 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32134 handling of certain combinations of target options, including the
32135 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32136 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32138 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32141 * config/arm/arm.md (*load_multiple): Add reload_completed to
32142 matching condition.
32144 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32145 Richard Biener <rguenth@suse.de>
32147 PR tree-optimization/79908
32148 PR tree-optimization/80136
32149 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32150 been cast away, gimplify_and_add suffices.
32152 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32154 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32156 2017-03-23 Richard Biener <rguenther@suse.de>
32158 PR tree-optimization/80032
32159 * gimplify.c (gimple_push_cleanup): Forced unconditional
32160 cleanups still have to go to the conditional_cleanups
32163 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32165 PR tree-optimization/80072
32166 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32168 (next_operand_entry_id): Change type to unsigned int.
32169 (sort_by_operand_rank): Make sure to return the right return value
32170 even if unsigned fields are bigger than INT_MAX.
32171 (struct oecount): Change cnt and id type to unsigned int.
32172 (oecount_hasher::equal): Formatting fix.
32173 (oecount_cmp): Make sure to return the right return value
32174 even if unsigned fields are bigger than INT_MAX.
32175 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32178 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32179 TREE_READONLY on result if writing it more than once.
32182 * doc/invoke.texi (-fsanitize=thread): Document that with
32183 -fnon-call-exceptions atomics are not able to throw
32187 * tsan.c: Include tree-eh.h.
32188 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32189 maybe_clean_or_replace_eh_stmt where needed.
32190 (instrument_memory_accesses): Add cfg_changed argument.
32191 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32192 if it returned true.
32193 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32195 PR rtl-optimization/63191
32196 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32197 wrapper function, moved the whole old content into ...
32198 (ix86_delegitimize_address_1): ... this. New inline function.
32199 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32200 true as last argument instead of ix86_delegitimize_address.
32202 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32204 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32205 cortexa57_branch_cost.
32207 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32209 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32211 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32214 * doc/md.texi (Constraints): Document wA constraint.
32215 * config/rs6000/constraints.md (wA): New.
32216 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32217 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32218 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32219 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32221 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32224 * gimplify.c (is_oacc_declared): New function.
32225 (oacc_default_clause): Use it to set default flags for acc declared
32226 variables inside parallel regions.
32227 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32228 declared variables.
32229 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32230 declare attribute to any decl as necessary.
32232 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32235 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32236 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32237 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32238 (arm_arch_lpae): This.
32239 * config/arm/arm.c (arm_arch7ve): Rename into ...
32240 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32241 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32244 2017-03-22 Martin Liska <mliska@suse.cz>
32247 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32248 error message instead of an ICE.
32250 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32252 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32254 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32256 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32258 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32261 2017-03-21 Martin Sebor <msebor@redhat.com>
32263 * doc/extend.texi: Use "cannot" instead of "can't."
32264 * doc/hostconfig.texi: Same.
32265 * doc/install.texi: Same.
32266 * doc/invoke.texi: Same.
32267 * doc/loop.texi: Same.
32268 * doc/md.texi: Same.
32269 * doc/objc.texi: Same.
32270 * doc/rtl.texi: Same.
32271 * doc/tm.texi: Same.
32272 * doc/tm.texi.in: Same.
32273 * doc/trouble.texi: Same.
32275 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32278 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32279 (collect_checksum_attributes): Set it.
32280 (die_checksum_ordered): Use it.
32282 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32284 PR tree-optimization/79908
32285 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32286 change: For a VA_ARG whose LHS has been cast away, use
32287 force_gimple_operand to construct the side effects.
32289 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32291 PR translation/80001
32292 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32293 more amenable to translation.
32294 (oacc_loop_auto_partitions): Likewise.
32296 2017-03-21 Marek Polacek <polacek@redhat.com>
32297 Martin Sebor <msebor@redhat.com>
32299 PR tree-optimization/80109
32300 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32301 on INTEGRAL_TYPE_P.
32303 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32304 Segher Boessenkool <segher@kernel.crashing.org>
32307 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32308 check reg_used_between_p between insn and one of succ or succ2
32309 depending on if succ is artificial insn not inserted into insn
32312 2017-03-21 Martin Liska <mliska@suse.cz>
32314 PR gcov-profile/80081
32315 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32316 * doc/gcc.texi: Include gcov-dump stuff.
32317 * doc/gcov-dump.texi: New file.
32319 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32321 PR rtl-optimization/79150
32322 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32323 conditional jump, if the jump is the last insn of the loop.
32325 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32326 Richard Biener <rguenth@suse.de>
32328 PR tree-optimization/79908
32329 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32330 been cast away, use force_gimple_operand to construct the side
32333 2017-03-21 Martin Liska <mliska@suse.cz>
32335 PR libfortran/79956
32336 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32339 2017-03-21 Brad Spengler <spender@grsecurity.net>
32342 * plugin.c (htab_hash_plugin): New function.
32343 (add_new_plugin): Use it and adjust.
32344 (parse_plugin_arg_opt): Adjust.
32345 (init_one_plugin): Likewise.
32347 2017-03-21 Richard Biener <rguenther@suse.de>
32349 PR tree-optimization/80032
32350 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32351 if set force the cleanup to happen unconditionally.
32352 (gimplify_target_expr): Push inserted clobbers with force_uncond
32353 to avoid them being removed by control-dependent DCE.
32355 2017-03-21 Richard Biener <rguenther@suse.de>
32357 PR tree-optimization/80122
32358 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32359 va_arg_pack_len when the inlined call stmt requires pack
32361 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32363 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32366 * tsan.c (instrument_builtin_call): If the memory model argument
32367 is not a constant, assume it is valid.
32370 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32373 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32375 PR rtl-optimization/79910
32376 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32377 if its dest is used by an insn before I2 (other than the combined
32378 insns themselves, which are properly handled already).
32380 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32383 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32385 * combine.c (record_used_regs): New static function.
32386 (try_combine): Handle situations where there is an additional
32387 instruction between I2 and I3 which needs to have a LOG_LINK
32391 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32393 * combine.c (try_combine): Delete redundant i1 test. Call
32394 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32396 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32399 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32400 alternatives 13/14.
32402 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32404 PR tree-optimization/80054
32405 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32406 the optimization if a PHI or any of its arguments is not dominated
32407 by the candidate's basis. Use gphi* rather than gimple* as
32409 (replace_profitable_candidates): Clean up a gimple* variable that
32410 should be a gphi* variable.
32412 2017-03-20 Martin Sebor <msebor@redhat.com>
32415 * doc/extend.texi (attribute constructor): Document present limitation.
32417 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32420 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32421 __POWER9_VECTOR__ #ifdef control, change template definition to
32422 use Power9-specific built-in function.
32423 (vec_any_eq): Likewise.
32424 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32425 to control outcomes from this test.
32426 (vector_ae_<mode>p): For VEC_F modes, likewise.
32428 2017-03-20 Ian Lance Taylor <iant@google.com>
32430 * config/i386/i386.c (ix86_function_regparm): Save an extra
32431 register for -fsplit-stack with DECL_STATIC_CHAIN.
32433 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32436 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32437 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32439 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32441 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32443 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32446 2017-03-20 Marek Polacek <polacek@redhat.com>
32449 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32451 2017-03-20 Richard Biener <rguenther@suse.de>
32453 PR tree-optimization/80113
32454 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32455 allocate extra SSA name for PHI def.
32456 (add_close_phis_to_outer_loops): Likewise.
32457 (add_close_phis_to_merge_points): Likewise.
32458 (copy_loop_close_phi_args): Likewise.
32459 (copy_cond_phi_nodes): Likewise.
32461 2017-03-20 Martin Liska <mliska@suse.cz>
32463 PR middle-end/79753
32464 * tree-chkp.c (chkp_build_returned_bound): Do not build
32465 returned bounds for a LHS that's not a BOUNDED_P type.
32467 2017-03-20 Martin Liska <mliska@suse.cz>
32471 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32472 COMPLEX_CST and VECTOR_CST.
32474 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32477 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32478 target operand. A new splitter adds the clobber statement in case
32479 the target operand is dead anyway.
32481 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32483 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32484 to age-old versions of binutils and glibc.
32486 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32488 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32490 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32492 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32494 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32496 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32497 requirement for binutils 2.13.
32499 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32501 * combine.c (try_combine): Delete redundant i1 test. Call
32502 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32504 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32506 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32507 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32509 <riscv64-*-elf>: Re-arrange section
32510 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32511 <riscv32-*-linux>: Likewise.
32512 <riscv64-*-elf>: Likewise
32513 <riscv64-*-linux>: Likewise.
32515 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32518 * aarch64.opt(verbose-cost-dump): Fix typo.
32520 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32523 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32524 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32526 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32528 * reload.c (find_reloads): When reloading a nonoffsettable address,
32529 use RELOAD_OTHER for it and its address reloads.
32531 PR rtl-optimization/79910
32532 * combine.c (record_used_regs): New static function.
32533 (try_combine): Handle situations where there is an additional
32534 instruction between I2 and I3 which needs to have a LOG_LINK
32537 2017-03-17 Jeff Law <law@redhat.com>
32539 PR tree-optimization/71437
32540 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32541 conditional in the hash table first.
32542 (vrp_dom_walker::before_dom_children): Extract condition from
32543 ASSERT_EXPR. Record condition, its inverion and any implied
32544 conditions as well.
32546 2017-03-17 Marek Polacek <polacek@redhat.com>
32547 Markus Trippelsdorf <markus@trippelsdorf.de>
32549 PR tree-optimization/80079
32550 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32553 2017-03-17 Richard Biener <rguenther@suse.de>
32555 PR middle-end/80075
32556 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32557 Properly verify the LHS before the RHS possibly claims to be
32559 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32562 2017-03-17 Martin Jambor <mjambor@suse.cz>
32564 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32565 (List of -O2 options): Likewise.
32566 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32569 2017-03-17 Tom de Vries <tom@codesourcery.com>
32571 * gcov-dump.c (print_usage): Print bug_report_url.
32573 2017-03-17 Richard Biener <rguenther@suse.de>
32575 PR middle-end/80050
32576 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32577 (parser::peek): Likewise.
32579 2017-03-17 Richard Biener <rguenther@suse.de>
32581 PR tree-optimization/80048
32582 * sese.c (free_sese_info): Properly release rename_map and
32583 copied_bb_map elements.
32585 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32587 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32588 Add linked-list forward and backlinks. Insert on
32589 construction, remove on destruction.
32590 (class pass_store_merging): Add m_stores_head field.
32591 (pass_store_merging::terminate_and_process_all_chains):
32592 Iterate over m_stores_head list.
32593 (pass_store_merging::terminate_all_aliasing_chains):
32595 (pass_store_merging::execute): Check for debug stmts first.
32596 Push new chains onto the m_stores_head stack.
32598 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32601 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32602 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32603 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32605 2017-03-16 Jeff Law <law@redhat.com>
32607 PR tree-optimization/71437
32608 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32609 member function. Implementation moved into after_dom_children
32610 member function and into the threader's thread_outgoing_edges
32612 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32613 some code into new thread_outgoing_edges.
32614 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32615 definition. Simplify marker handling (do it here). Assume we always
32616 have the available expression and the const/copies tables.
32617 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32619 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32620 * tree-vrp.c (equiv_stack): No longer file scoped.
32621 (vrp_dom_walker): New class.
32622 (vrp_dom_walker::before_dom_children): New member function.
32623 (vrp_dom_walker::after_dom_children): Likewise.
32624 (identify_jump_threads): Setup domwalker. Use it rather than
32625 walking edges in a random order by hand. Simplify setup/finalization.
32626 (finalize_jump_threads): Remove.
32627 (vrp_finalize): Do not call identify_jump_threads here.
32628 (execute_vrp): Do it here instead and call thread_through_all_blocks
32631 PR tree-optimization/71437
32632 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32634 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32636 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32637 (dom_opt_dom_walker::thread_across_edge): Remove
32638 handle_dominating_asserts argument. All callers changed.
32639 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32640 changes. Remove calls to lhs_of_dominating_assert. Other
32641 uses of handle_dominating_asserts turn into unconditional code
32642 (simplify_control_stmt_condition_1): Likewise.
32643 (simplify_control_stmt_condition): Likewise.
32644 (thread_through_normal_block, thread_across_edge): Likewise.
32645 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32646 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32647 object if it is not an SSA_NAME.
32648 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32649 before calling into the VRP specific simplifiers.
32650 (identify_jump_threads): Remove handle_dominating_asserts
32653 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32656 * tree-diagnostic.c (default_tree_printer): No longer static.
32657 * tree-diagnostic.h (default_tree_printer): New prototype.
32659 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32661 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32662 Change ins into fmov.
32664 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32666 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32667 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32668 Use h_con constraint for operand 1.
32669 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32670 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32672 2017-03-15 Jeff Law <law@redhat.com>
32674 PR tree-optimization/71437
32675 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32676 (record_temporary_equivalences): Use it.
32678 PR tree-optimization/71437
32679 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32680 tree-ssa-scopedtables.
32681 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32682 (record_conditions, record_cond, vuse_eq): Likewise.
32683 (record_edge_info): Adjust to API tweak of record_conditions.
32684 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32685 (record_temporary_equivalences, optimize_stmt): Likewise.
32686 (eliminate_redundant_computations): Likewise.
32687 (record_equivalences_from_stmt): Likewise.
32688 * tree-ssa-scopedtables.c: Include options.h and params.h.
32689 (vuse_eq): New function, moved from tree-ssa-dom.c
32690 (build_and_record_new_cond): Likewise.
32691 (record_conditions): Likewise. Accept vector of conditions rather
32692 than edge_equivalence structure for first argument.
32693 for the first argument.
32694 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32695 from tree-ssa-dom.c.
32696 (avail_exprs_stack::record_cond): Likewise.
32697 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32698 from tree-ssa-dom.c.
32699 (avail_exprs_stack): Add new member functions lookup_avail_expr
32701 (record_conditions): Declare.
32703 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32706 * lra-constraints.c (process_alt_operands): Increase reject for
32707 reloading an input/output operand.
32709 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32712 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32713 insns to convert from signed/unsigned char/short to IEEE 128-bit
32715 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32717 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32720 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32721 subreg of inner mode for values already in registers.
32723 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32725 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32726 iteration reg is used after the loop.
32728 2017-03-14 Martin Sebor <msebor@redhat.com>
32730 PR tree-optimization/79800
32731 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32732 precision in negative-positive range.
32733 (format_floating): Call non-const overload with adjusted precision.
32735 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32738 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32741 2017-03-14 Martin Sebor <msebor@redhat.com>
32743 PR middle-end/80020
32744 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32745 * builtins.def (aligned_alloc): Use it.
32748 * Makefile.in (GTFILES): Add calls.c.
32749 * calls.c: Include "gt-calls.h".
32751 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32753 PR rtl-optimization/79728
32754 * regs.h (struct target_regs): New field
32755 x_contains_allocatable_regs_of_mode.
32756 (contains_allocatable_regs_of_mode): New macro.
32757 * reginfo.c (init_reg_sets_1): Initialize it, and change
32758 contains_reg_of_mode so it includes global regs as well.
32759 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32760 rather than contains_regs_of_mode.
32762 2017-03-14 Martin Liska <mliska@suse.cz>
32764 * doc/invoke.texi: Document options that can't be combined with
32765 -fcheck-pointer-bounds.
32767 2017-03-14 Martin Liska <mliska@suse.cz>
32769 PR middle-end/79831
32770 * doc/invoke.texi (-Wchkp): Document the option.
32772 2017-03-14 Martin Liska <mliska@suse.cz>
32774 * Makefile.in: Install gcov-dump.
32776 2017-03-14 Martin Liska <mliska@suse.cz>
32778 * multiple_target.c (expand_target_clones): Bail out for
32779 an invalid attribute.
32781 2017-03-14 Richard Biener <rguenther@suse.de>
32783 * alias.c (struct alias_set_entry): Pack properly.
32784 * cfgloop.h (struct loop): Likewise.
32785 * cse.c (struct set): Likewise.
32786 * ipa-utils.c (struct searchc_env): Likewise.
32787 * loop-invariant.c (struct invariant): Likewise.
32788 * lra-remat.c (struct cand): Likewise.
32789 * recog.c (struct change_t): Likewise.
32790 * rtl.h (struct address_info): Likewise.
32791 * symbol-summary.h (function_summary): Likewise.
32792 * tree-loop-distribution.c (struct partition): Likewise.
32793 * tree-object-size.c (struct object_size_info): Likewise.
32794 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32795 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32796 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32797 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32798 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32799 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32800 (struct _stmt_vec_info): Likewise.
32802 2017-03-14 Martin Liska <mliska@suse.cz>
32805 * multiple_target.c (create_dispatcher_calls): Check that
32806 a target can create a function dispatcher.
32808 2017-03-14 Martin Liska <mliska@suse.cz>
32811 * multiple_target.c (expand_target_clones): Drop local.local
32812 flag for default implementation.
32814 2017-03-14 Richard Biener <rguenther@suse.de>
32816 PR tree-optimization/80030
32817 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32819 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32821 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32822 gcc_fallthrough() instead of __attribute__((fallthrough));
32824 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32826 * doc/gcc.texi: Remove "up" link to (DIR).
32827 * doc/gccint.texi: Ditto.
32829 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32831 * doc/install.texi (Specific) <avr>: Remove reference to
32834 2017-03-13 Jeff Law <law@redhat.com>
32836 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32837 attribute rather than comments.
32839 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32840 match_scratch operand is highest.
32842 2017-03-13 Martin Liska <mliska@suse.cz>
32844 PR middle-end/78339
32845 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32846 is a CHKP clone, use original declaration.
32848 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32850 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32851 (arc_conditional_register_usage): Use a different allocation order
32852 when optimizing for size.
32853 * common/config/arc/arc-common.c (arc_option_optimization_table):
32854 Section anchors default on when optimizing for size.
32856 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32858 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32860 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32862 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32863 * config/arc/arc.md (cpu_facility): Add cd variant.
32864 (*movqi_insn): Add code density variant.
32865 (*movhi_insn): Likewise.
32866 (*movqi_insn): Likewise.
32867 (*addsi3_mixed): Likewise.
32868 (subsi3_insn): Likewise.
32870 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32872 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32874 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32876 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32877 expressions with MINUS and UNARY ops.
32879 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32882 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32884 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32885 between vec_select and vector argument.
32886 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32887 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32888 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32889 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32890 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32891 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32893 2017-03-13 Richard Biener <rguenther@suse.de>
32896 * params.def (vect-max-peeling-for-alignment): Fix typo.
32898 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32900 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32901 issue that only occurred with binutils below 2.18.
32903 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32905 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32906 refer to binutils 2.11/2.12 minimum.
32908 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32910 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
32911 ftp.kernel.org and simplify binutils requirement.
32913 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
32915 * doc/invoke.texi (Warning Options): Fix spelling of link-time
32917 (Optimize Options): Ditto. Also remove redundancy.
32919 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32921 PR translation/79848
32922 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
32924 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
32925 to G_ to avoid double translation.
32927 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32929 PR translation/79923
32930 * auto-profile.c (get_combined_location): Convert leading
32931 character of diagnostics to lower case and remove trailing period.
32932 (read_profile): Likewise for various diagnostics.
32933 * config/arm/arm.c (arm_option_override): Remove trailing period
32934 from various diagnostics.
32935 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
32936 (msp430_expand_delay_cycles): Likewise.
32938 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32941 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
32942 full command-line argument, rather than just "str".
32943 (aarch64_validate_march): Likewise.
32944 (aarch64_validate_mtune): Likewise.
32946 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
32948 PR rtl-optimization/78911
32949 * lra-assigns.c (must_not_spill_p): New function.
32950 (spill_for): Use it.
32952 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32954 PR tree-optimization/79981
32955 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
32956 ATOMIC_COMPARE_EXCHANGE ifn result.
32957 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
32958 IFN_ATOMIC_COMPARE_EXCHANGE.
32960 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32963 * opts.c (parse_sanitizer_options): Add missing question mark to
32964 "did you mean" message.
32966 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32968 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
32970 (VMULEUH_UNS): Likewise.
32971 (VMULOUB_UNS): Likewise.
32972 (VMULOUH_UNS): Likewise.
32973 * config/rs6000/rs6000.c (builtin_function_type): Remove
32974 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
32976 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32979 * read-rtl-function.c (function_reader::read_rtx_operand): Update
32980 x with result of extra_parsing_for_operand_code_0.
32981 (function_reader::extra_parsing_for_operand_code_0): Convert
32982 return type from void to rtx, returning x. When reading
32983 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
32984 larger size containing struct block_symbol.
32986 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
32988 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
32989 -mfloat128-hardware without -m64.
32991 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
32994 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
32995 entries to the case statement that marks unsigned arguments to
32996 overloaded functions.
32998 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33000 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33001 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33003 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33006 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33007 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33009 2017-03-10 Martin Liska <mliska@suse.cz>
33013 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33014 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33017 2017-03-10 Olivier Hainque <hainque@adacore.com>
33019 * tree-switch-conversion (array_value_type): Start by resetting
33020 candidate type to it's main variant.
33022 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33024 PR rtl-optimization/79909
33025 * combine.c (try_combine): Use simplify_replace_rtx on individual
33026 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33027 of the whole CALL_INSN_FUNCTION_USAGE.
33029 PR tree-optimization/79972
33030 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33031 get_range_info on SSA_NAMEs. Formatting fixes.
33033 2017-03-10 Richard Biener <rguenther@suse.de>
33034 Jakub Jelinek <jakub@redhat.com>
33036 PR tree-optimization/77975
33037 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33038 edge to be constant.
33039 (get_val_for): For constant x return it. Formatting fix.
33040 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33041 would use the same bases as the current one.
33043 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33045 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33046 instead of vec_select for V1TImode.
33047 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33049 (VSX_LE_128): Add V1TI to this mode iterator.
33050 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33051 (*vsx_le_perm_store_<mode>): Likewise.
33052 (pre-reload splitter for VSX stores): Likewise.
33053 (post-reload splitter for VSX stores): Likewise.
33054 (*vsx_xxpermdi2_le_<mode>): Likewise.
33055 (*vsx_lxvd2x2_le_<mode>): Likewise.
33056 (*vsx_stxvd2x2_le_<mode>): Likewise.
33058 2017-03-09 Michael Eager <eager@eagercon.com>
33060 Correct failures with --enable-checking=yes,rtl.
33062 * config/microblaze/microblaze.c (microblaze_expand_shift):
33063 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33064 test for const0_rtx.
33065 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33066 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33068 2017-03-09 Richard Biener <rguenther@suse.de>
33070 PR tree-optimization/79977
33071 * graphite-scop-detection.c (scop_detection::merge_sese):
33072 Handle the case of extra exits to blocks dominating the entry.
33074 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33076 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33079 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33081 PR rtl-optimization/79949
33082 * lra-constraints.c (process_alt_operands): Check memory when
33083 trying to predict a cycle. Print about the overall increase.
33085 2017-03-09 Richard Biener <rguenther@suse.de>
33087 PR middle-end/79971
33088 * gimple-expr.c (useless_type_conversion_p): Preserve
33089 TYPE_SATURATING for fixed-point types.
33091 2017-03-09 Richard Biener <rguenther@suse.de>
33094 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33095 alignment of BLKmode params.
33097 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33100 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33101 (VALL_NO_V2Q): Likewise.
33103 * config/aarch64/aarch64-simd.md
33104 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33106 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33107 VALL_NO_V2Q mode iterator.
33108 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33110 2017-03-09 Martin Liska <mliska@suse.cz>
33112 PR tree-optimization/79631
33113 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33114 tree_int_cst_sign_bit just for INTEGER constants.
33116 2017-03-09 Martin Liska <mliska@suse.cz>
33120 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33123 2017-03-09 Marek Polacek <polacek@redhat.com>
33126 * tree.c (inchash::add_expr): Handle TREE_VEC.
33128 2017-03-09 Martin Liska <mliska@suse.cz>
33131 (chkp_narrow_size_and_offset): New function.
33132 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33133 (void chkp_parse_bit_field_ref): New function.
33134 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33135 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33137 2017-03-09 Martin Liska <mliska@suse.cz>
33140 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33141 (chkp_find_bounds_1): Remove gcc_unreachable.
33143 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33146 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33147 BUILT_IN_SYNC*, determine the access type from the size suffix and
33148 always build a MEM_REF with that type. Handle forgotten
33149 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33152 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33153 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33154 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33155 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33156 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33157 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33158 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33159 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33160 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33161 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33162 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33163 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33164 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33165 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33166 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33167 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33168 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33169 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33170 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33171 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33172 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33173 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33174 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33175 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33176 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33177 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33178 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33179 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33180 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33181 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33182 definitions outside of __OPTIMIZE__ guarded section.
33185 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33186 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33187 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33188 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33191 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33193 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33194 ("vfenez<mode>"): Add missing constraints.
33196 2017-03-08 Martin Sebor <msebor@redhat.com>
33199 * config/nds32/nds32.c (nds32_option_override):
33200 Fix misspelled diagnostic.
33202 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33205 * gimplify.c (gimplify_omp_for): Replace index var in outer
33206 taskloop statement with an artificial variable and add
33207 OMP_CLAUSE_PRIVATE clause for it.
33209 2017-03-08 Richard Biener <rguenther@suse.de>
33211 PR tree-optimization/79955
33212 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33213 for accesses that are completely outside of the variable.
33215 2017-03-08 Andrew Haley <aph@redhat.com>
33217 PR tree-optimization/79943
33218 * tree-ssa-loop-split.c (compute_new_first_bound): When
33219 calculating the new upper bound, (END-BEG) should be added, not
33222 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33224 * config/avr/avr.md (setmemhi): Make sure match_dup
33225 operand number comes before match_scratch.
33227 2017-03-08 Richard Biener <rguenther@suse.de>
33229 PR tree-optimization/79920
33230 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33231 with ncopies == 1 to ...
33232 (vect_transform_slp_perm_load): ... here. Properly compute
33233 all element loads by iterating VF times over the group. Do
33234 not handle ncopies (computed in a broken way) in
33235 vect_create_mask_and_perm.
33237 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33240 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33241 is a uniform vector, use uniform_vector_p return value instead of
33242 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33244 2017-03-07 Marek Polacek <polacek@redhat.com>
33246 PR middle-end/79809
33247 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33248 (alloca_call_type): Likewise.
33250 2017-03-07 Martin Liska <mliska@suse.cz>
33252 * gcov.c (process_args): Put comment to correct location.
33254 2017-03-07 Martin Liska <mliska@suse.cz>
33256 PR middle-end/68270
33257 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33258 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33259 (chkp_narrow_bounds_for_field): Likewise.
33260 (chkp_parse_array_and_component_ref): Pass one more argument to
33263 2017-03-07 Richard Biener <rguenther@suse.de>
33265 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33268 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33270 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33271 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33273 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33276 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33277 to end of description.
33278 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33280 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33282 PR rtl-optimization/79901
33283 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33285 (*avx512f_<code><mode>3<mask_name>): ... this.
33286 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33287 iterator instead of VI8_AVX2_AVX512BW.
33289 PR rtl-optimization/79901
33290 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33291 min/max expander, expand it using expand_vec_cond_expr.
33294 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33297 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33300 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33301 to void * for PCH reasons.
33302 * dwarf2out.c (output_loc_operands, output_die): Cast
33303 v.val_vec.array to unsigned char *.
33305 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33308 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33311 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33313 PR rtl-optimization/79571
33314 * lra-constraints.c (process_alt_operands): Calculate static
33315 reject and subtract it from overall when only addresses will be
33318 2017-03-06 Julia Koval <julia.koval@intel.com>
33321 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33322 incoming stack boundary to 128 for 64-bit targets.
33324 2017-03-06 Richard Biener <rguenther@suse.de>
33326 PR tree-optimization/79894
33327 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33328 to NULL after folding it.
33330 2017-03-06 Richard Biener <rguenther@suse.de>
33332 PR tree-optimization/79824
33333 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33334 check disabling peeling for gaps.
33336 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33338 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33339 attributes): Document gettimeofday.
33341 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33343 * config/s390/s390.c (s390_option_override_internal): Set
33344 PARAM_MIN_VECT_LOOP_BOUND
33346 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33348 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33349 * config/s390/s390.md: Likewise.
33351 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33354 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33355 (<avx2_avx512>_perm<mode>): Rename to ...
33356 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33358 (<avx512>_perm<mode>_mask): Rename to ...
33359 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33361 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33362 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33363 instead of VI8F_256_512.
33364 (avx512f_perm<mode>): New define_expand.
33365 (avx512f_perm<mode>_mask): Likewise.
33366 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33367 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33369 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33371 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33372 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33374 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33376 2017-03-06 Martin Liska <mliska@suse.cz>
33379 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33380 when having a SSA NAME w/o VAR_DECL assigned to it.
33382 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33384 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33385 msa_dpsub_<su>_d): Fix MODE for vec_select.
33387 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33389 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33391 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33393 2017-03-06 Richard Biener <rguenther@suse.de>
33395 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33396 * plugin.c (register_plugin_info): Likewise.
33397 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33399 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33401 * config/i386/sse.md (sse_storehps, sse_storelps,
33402 avx_<castmode><avxsizesuffix>_<castmode>,
33403 avx512f_<castmode><avxsizesuffix>_<castmode>,
33404 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33405 in condition that at least one operand is not a MEM.
33407 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33409 PR middle-end/79805
33410 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33411 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33413 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33414 gimple_call_nothrow_p flag based on whether original builtin can throw.
33415 If it can, emit following stmts on the fallthrough edge.
33416 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33417 don't create new bb if inserting just debug stmts on the edge, try to
33418 insert them on the fallthru bb or just reset debug stmts.
33420 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33423 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33424 restore recog_data (including the operand rtxes inside it) around
33425 the call to get_insn_template.
33427 2017-03-03 Martin Sebor <msebor@redhat.com>
33429 PR tree-optimization/79699
33430 * context.c (context::~context): Free MPFR caches to avoid
33431 a memory leak on program exit.
33433 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33435 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33436 Use wide_int::ulow () instead of .elt (0).
33438 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33440 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33441 (*pushxf): Limit oF constraint to 32bit targets and add oC
33442 constraint for 64bit targets.
33443 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33444 (*pushdf): Change rmF constraint to rmC.
33446 2017-03-03 Martin Liska <mliska@suse.cz>
33448 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33449 Remove unused variable.
33451 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33454 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33455 is a memory operand, increase num_memory.
33456 (ix86_expand_args_builtin): Likewise.
33458 2017-03-03 Jan Hubicka <jh@suse.cz>
33461 * ipa-devirt.c (maybe_record_node): Properly handle
33462 __cxa_pure_virtual visibility.
33464 2017-03-03 Martin Liska <mliska@suse.cz>
33466 PR tree-optimization/79803
33467 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33469 (pass_loop_prefetch::execute): Disabled optimization if an
33470 assumption about L1 cache size is not met.
33472 2017-03-03 Martin Liska <mliska@suse.cz>
33474 PR rtl-optimization/79574
33475 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33476 (hash_scan_set): Likewise.
33477 (dump_hash_table): Likewise.
33478 (hoist_code): Likewise.
33480 2017-03-03 Richard Biener <rguenther@suse.de>
33482 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33483 in place of uhigh ().
33484 (fixed_convert_from_real): Likewise.
33486 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33489 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33491 2017-03-03 Richard Biener <rguenther@suse.de>
33493 PR middle-end/79818
33494 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33495 TYPE_OVERFLOW_UNDEFINED check.
33497 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33499 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33501 (vector_ae_<mode>_p): Likewise.
33502 (vector_nez_<mode>_p): Likewise.
33503 (vector_ne_v2di_p): Likewise.
33504 (vector_ae_v2di_p): Likewise.
33505 (vector_ne_<mode>_p): Likewise.
33506 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33508 (vsx_tsqrt<mode>2_fe): Likewise.
33510 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33513 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33515 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33517 PR rtl-optimization/79780
33518 * cprop.c (one_cprop_pass): When second and further conditional trap
33519 in a single basic block is turned into an unconditional trap, turn it
33520 into a deleted note to avoid RTL verification failures.
33522 2017-03-02 Richard Biener <rguenther@suse.de>
33524 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33526 2017-03-02 Richard Biener <rguenther@suse.de>
33528 PR tree-optimization/79345
33530 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33531 param and abort the walk, returning -1 if it is hit.
33532 (walk_aliased_vdefs): Take a limit param and pass it on.
33533 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33534 defaulting to 0 and return a signed int.
33535 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33536 (check_defs): New helper.
33537 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33538 about uninitialized memory.
33539 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33540 bogus uninitialized warning.
33541 (fixed_convert_from_real): Likewise.
33543 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33545 PR tree-optimization/66768
33546 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33547 iv_use if base object can't be determined.
33549 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33551 PR tree-optimization/79345
33552 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33553 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33554 (get_pattern_stats): Initialize it.
33555 * genemit.c (gen_expand): Verify match_scratch numbers come after
33556 match_operand/match_dup numbers.
33557 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33558 match_scratch numbers.
33559 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33561 * config/s390/s390.md (trunctdsd2): Likewise.
33563 2017-03-02 Richard Biener <rguenther@suse.de>
33565 * wide-int.h (wide_int_storage::operator=): Implement in terms
33568 2017-03-02 Richard Biener <rguenther@suse.de>
33570 PR tree-optimization/79777
33571 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33572 the to insert expression to sth existing.
33574 2017-03-01 Martin Sebor <msebor@redhat.com>
33576 PR middle-end/79692
33577 * gimple-ssa-sprintf.c
33578 (directive::known_width_and_precision): New function.
33579 (format_integer): Use it.
33580 (get_mpfr_format_length): Consider the full range of precision
33581 when computing %g output with the # flag. Set the likely byte
33582 count to 3 rather than 1 when precision is indeterminate.
33583 (format_floating): Correct the lower bound of precision.
33585 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33587 * doc/invoke.texi: Document default code model for 64-bit Linux.
33589 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33592 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33593 udiv rather than div since input pattern is unsigned.
33595 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33597 * config/i386/i386.c (print_reg): Warn for values of
33598 unsupported size in integer register.
33600 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33603 * config/rs6000/predicates.md (current_file_function_operand): Do
33604 not allow self calls to be local if the function is replaceable.
33606 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33609 * config/rs6000/altivec.h (vec_ctz and others): Change the
33610 preprocessor macro that controls conditional compilation from
33611 _ARCH_PWR9 to __POWER9_VECTOR__.
33612 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33613 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33614 control (instead of _ARCH_PWR9 control) so that template
33615 definition uses power9-specific function.
33616 (vec_any_eq): Likewise.
33617 (vec_all_ne): Change macro definition to use a power9-specific
33618 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33619 _ARCH_PWR9 control).
33620 (vec_any_eq) Likewise.
33621 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33622 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33623 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33624 support for xvcmpnedp instruction.
33625 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33626 macro expansion so that Power9 implementation of vec_all_ne does
33627 not use the AltiVec predicate framework.
33628 (VCMPNEH_P): Likewise.
33629 (VCMPNEW_P): Likewise.
33630 (VCMPNED_P): Likewise.
33631 (VCMPNEFP_P): Likewise.
33632 (VCMPNEDP_P): Likewise.
33633 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33634 implementation of vec_any_eq to not use AltiVec predicate
33636 (VCMPAEH_P): Likewise.
33637 (VCMPAEW_P): Likewise.
33638 (VCMPAED_P): Likewise.
33639 (VCMPAEFP_P): Likewise.
33640 (VCMPAEDP_P): Likewise.
33641 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33642 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33643 not use the AltiVec predicate framework.
33644 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33645 of vec_any_eq to not use AltiVec predicate framework.
33646 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33647 support for predefined __POWER9_VECTOR__ macro to indicate that
33648 Power9 instruction selection is enabled.
33649 (altivec_overloaded_builtins): Remove extraneous
33650 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33651 function argument types RS6000_BTI_bool_V16QI and
33652 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33653 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33654 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33655 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33656 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33657 Power9 for implementations of vec_cmpne. Change the signature for
33658 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33659 (representing vec_all_ne) to remove the previously described first
33660 argument of type RS6000_BTI_INTSI, as this was an artifact of
33661 reliance on the AltiVec predicate framework, which is no longer
33662 used in the implementation of these functions. Add
33663 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33664 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33665 since, unlike the AltiVec predicate framework implementation, we
33666 do not share function descriptors between vec_alle and vec_anyeq.
33667 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33668 set of modes that receive special treatment even when
33669 TARGET_P9_VECTOR is true. The special treatment emits code that
33670 does not depend on Power9 instructions.
33671 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33672 define_expand to not rely on AltiVec predicate framework.
33673 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33675 (vector_ne_v2di_p): Change this define_expand to not rely on
33676 AltiVec predicate framework.
33677 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33679 (vector_ne_<mode>_p): Change this define_expand to not rely on
33680 AltiVec predicate framework.
33681 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33683 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33684 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33685 define_insn pattern.
33686 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33687 define_insn pattern because the xvcmpne<VSs>. instruction is not
33689 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33690 instruction is not supported.
33692 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33694 * config/nvptx/nvptx.c: Include intl.h.
33696 2017-03-01 Martin Jambor <mjambor@suse.cz>
33699 * ipa-prop.h (ipa_bits): Removed field known.
33700 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33701 to pointers. Adjusted their comments to warn about their sharing.
33702 (ipcp_transformation_summary): Change bits to a vector of pointers.
33703 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33704 (ipa_get_ipa_bits_for_value): Declare.
33705 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33706 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33707 (ipa_bits_hash_table): Likewise.
33708 (ipa_vr_ggc_hash_traits): Likewise.
33709 (ipa_vr_hash_table): Likewise.
33710 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33711 being pointers and vr_known being removed.
33712 (ipa_set_jf_unknown): Likewise.
33713 (ipa_get_ipa_bits_for_value): New function.
33714 (ipa_set_jfunc_bits): Likewise.
33715 (ipa_get_value_range): New overloaded functions.
33716 (ipa_set_jfunc_vr): Likewise.
33717 (ipa_compute_jump_functions_for_edge): Use the above functions to
33718 construct bits and vr parts of jump functions.
33719 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33720 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33722 (ipcp_grow_transformations_if_necessary): Also allocate
33723 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33725 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33726 them. Fix too long lines.
33727 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33728 vr_known being removed.
33729 (ipa_read_jump_function): Use new setter functions to construct bits
33730 and vr parts of jump functions or set them to NULL.
33731 (write_ipcp_transformation_info): Adjust for bits being pointers.
33732 (read_ipcp_transformation_info): Likewise.
33733 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33735 Include gt-ipa-prop.h.
33736 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33738 (ipcp_store_bits_results): Likewise.
33739 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33740 Do not write to existing jump functions but use a temporary instead.
33742 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33745 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33746 attempt to use its first operand as BIT_FIELD_REF base.
33748 2017-03-01 Richard Biener <rguenther@suse.de>
33750 PR middle-end/79721
33751 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33752 interpolating formula in wrapping arithmetic.
33753 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33755 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33757 PR tree-optimization/79734
33758 * tree-vect-generic.c (expand_vector_condition): Optimize
33759 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33760 Handle VEC_COND_EXPR where comparison has different inner width from
33761 type's inner width.
33763 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33765 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33766 markup, and similar issues. Remove @opindex entries for things
33767 that aren't options. Add missing -mmpy-option entries.
33769 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33771 PR tree-optimization/79737
33772 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33773 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33774 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33775 instead of byte_size. Formatting fix.
33776 (shift_bytes_in_array_right): Formatting fix.
33778 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33781 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33782 condition on optimize for the leaf function test.
33784 2017-02-28 Martin Liska <mliska@suse.cz>
33787 * read-rtl-function.c (function_reader::handle_unknown_directive):
33788 Bail out when one uses -flto.
33790 2017-02-28 Martin Liska <mliska@suse.cz>
33792 * common.opt: Replace space with tabular for options of <number>
33794 * config/i386/i386.opt: Show <number> value for
33795 -mlarge-data-threshold.
33796 * opts.c (print_filtered_help): Do not display number in hexadecimal
33799 2017-02-28 Martin Liska <mliska@suse.cz>
33801 * common.opt: Fix --help=option -Q for options which are of
33804 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33806 * config/i386/i386.c (print_reg): Error out for values
33807 of 8-bit size in invalid integer register.
33809 2017-02-28 Martin Sebor <msebor@redhat.com>
33811 PR tree-optimization/79691
33812 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33814 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33817 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33818 gcc_unreachable with output_operand_lossage.
33820 2017-02-28 Richard Biener <rguenther@suse.de>
33822 PR tree-optimization/79740
33823 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33825 (visit_nary_op): Insert the nary into the hashtable if we
33826 pattern-matched sth.
33827 * tree-ssa-pre.c (eliminate_insert): Robustify.
33829 2017-02-28 Richard Biener <rguenther@suse.de>
33831 PR middle-end/79731
33832 * fold-const.c (decode_field_reference): Reject out-of-bound
33835 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33837 * config/i386/i386.c: Include intl.h.
33838 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33839 instead of just cond ? "..." : "...".
33840 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33841 * coverage.c (read_counts_file): Likewise.
33842 * omp-offload.c: Include intl.h.
33843 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33844 of just cond ? "..." : "...".
33845 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33846 of just cond ? "..." : "...".
33848 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33851 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33853 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33854 'tune for' CPU name.
33855 * config/arm/arm-cpu-data.h: Regenerated.
33857 2017-02-28 Richard Biener <rguenther@suse.de>
33859 PR tree-optimization/79732
33860 * tree-inline.c (expand_call_inline): Do not shadow var.
33862 2017-02-28 Richard Biener <rguenther@suse.de>
33864 PR tree-optimization/79723
33865 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33866 address-space properly.
33868 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33870 * doc/optinfo.texi (Optimization groups): Fix option used for
33872 * doc/invoke.texi (-fopt-info): Document "omp".
33873 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33874 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33875 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33876 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33877 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33879 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33881 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33882 OPTGROUP_OMP with "omp".
33884 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33887 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33888 for arithmetic shift of unsigned V2DI.
33890 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33892 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33893 arc/linux.h headers.
33894 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33895 (LINK_SPEC): Likewise.
33896 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33897 (EXTRA_SPECS): Likewise.
33898 (STARTFILE_SPEC): Likewise.
33899 (ENDFILE_SPEC): Likewise.
33900 (LIB_SPEC): Likewise.
33901 (TARGET_SDATA_DEFAULT): Likewise.
33902 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33903 (MULTILIB_DEFAULTS): Likewise.
33904 (DWARF2_UNWIND_INFO): Likewise.
33905 * config/arc/big.h: New file.
33906 * config/arc/elf.h: Likewise.
33907 * config/arc/linux.h: Likewise.
33908 * config/arc/t-uClibc: Remove.
33910 2017-02-27 Bin Cheng <bin.cheng@arm.com>
33912 PR tree-optimization/77536
33913 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
33914 (tree_transform_and_unroll_loop): Use above function to compute the
33915 estimated niter of unrolled loop and use it when scaling profile.
33916 Also use count info rather than frequency if it's non-zero.
33917 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
33918 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
33919 (vect_transform_loop): Call above function.
33921 2017-02-27 Richard Biener <rguenther@suse.de>
33923 PR tree-optimization/45397
33924 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
33925 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
33926 (visit_nary_op): Add pattern matching for CSEing sign-changed
33927 or truncated operations with wider ones.
33929 2017-02-27 Richard Biener <rguenther@suse.de>
33931 PR tree-optimization/79690
33932 * tree-vect-stmts.c (vectorizable_store): Use vector type
33933 built from the DR with address-space.
33935 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
33937 * doc/invoke.texi (Optimize Options): Refine the description
33938 of asan-use-after-return.
33940 2017-02-25 Alan Modra <amodra@gmail.com>
33942 PR rtl-optimization/79584
33943 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
33944 base, not ad->base_term, the reg within base. Remove assertion
33945 that ad->base == ad->base_term. Replace gen_int_mode using
33946 bogus mode with const0_rtx.
33948 2017-02-25 Jakub Jelinek <jakub@redhat.com>
33950 PR middle-end/79396
33951 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
33952 FMA_EXPR like tcc_binary or tcc_unary.
33954 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
33957 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
33959 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
33960 (output_loc_operands): Handle DW_OP_call_ref and
33961 DW_OP_GNU_variable_value.
33962 (struct variable_value_struct): New type.
33963 (struct variable_value_hasher): Likewise.
33964 (variable_value_hash): New variable.
33965 (string_types): Remove.
33966 (copy_loc_descr): New function.
33967 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
33968 (prepend_loc_descr_to_each): New function.
33969 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
33970 instead of add_loc_descr_to_each if the first argument is single
33971 location list and the second has multiple.
33972 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
33973 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
33974 when looking for variable value which doesn't have other location info.
33975 (loc_list_from_tree): Formatting fix.
33976 (gen_array_type_die): Simplify DW_AT_string_length handling.
33977 (adjust_string_types): Remove.
33978 (gen_subprogram_die): Don't call adjust_string_types nor test/set
33979 string_types. Call resolve_variable_values.
33980 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
33981 (resolve_addr_in_expr): Likewise. Add A argument.
33982 (copy_deref_exprloc): Remove deref argument. Adjust for the
33983 original expression being DW_OP_GNU_variable_value with optionally
33984 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
33985 optionally after it.
33986 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
33987 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
33988 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
33989 (variable_value_hasher::hash, variable_value_hasher::equal): New
33991 (resolve_variable_value_in_expr, resolve_variable_value,
33992 resolve_variable_values, note_variable_value_in_expr,
33993 note_variable_value): New functions.
33994 (dwarf2out_early_finish): Call note_variable_value on all toplevel
33997 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34000 * opts.h (handle_generated_option): Add GENERATED_P argument.
34001 * opts-common.c (handle_option): Adjust function comment.
34002 (handle_generated_option): Add GENERATED_P argument, pass it to
34004 (control_warning_option): Pass false to handle_generated_option
34006 * opts.c (maybe_default_option): Pass true to handle_generated_option
34008 * optc-gen.awk: Likewise.
34010 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34012 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34013 a REG, look at the REG it is a SUBREG of.
34014 (splitter for cmpeqsi_t): Ditto.
34016 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34018 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34019 the special USEs with the pattern of the insn, not the insn itself.
34021 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34024 * doc/invoke.texi: Document -mload-store-pairs.
34026 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34027 Sandra Loosemore <sandra@codesourcery.com>
34029 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34030 argument isn't a CONST_INT.
34031 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34032 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34033 (nios2_validate_compare): Bypass alternate compare logic if *op2
34034 is not a CONST_INT.
34035 (ldstwm_operation_p): Return false if first_base is not a REG or
34036 if first_offset is not a CONST_INT.
34038 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34040 * config/cris/cris.md: Use correct operand in a define_peephole2.
34042 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34044 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34046 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34048 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34049 this_insn if it is an INSN or JUMP_INSN.
34050 (force_offsettable): Look at base, not at addr.
34051 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34052 on things that aren't necessarily CONST_INTs.
34054 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34056 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34057 -mfpmath=sse is the default also for x86-32 targets with SSE2
34058 instruction set when @option{-ffast-math} is enabled
34060 2017-02-24 Jeff Law <law@redhat.com>
34062 PR rtl-optimizatoin/79286
34063 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34066 2017-02-24 Richard Biener <rguenther@suse.de>
34068 PR tree-optimization/79389
34069 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34072 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34074 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34075 function comment to reflect reality.
34076 (loop_exits_before_overflow): Fix typo in function description.
34078 2017-02-24 Richard Biener <rguenther@suse.de>
34080 PR tree-optimization/79389
34081 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34082 properly that a threading opportunity exists. Detect conditional
34083 copy/constant propagation opportunities.
34085 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34087 * config/visium/visium.md (type): Add trap.
34088 (b): New mode attribute.
34089 (*btst): Rename into...
34090 (*btst<mode>): ...this and adjust.
34091 (*cbranchsi4_btst_insn): Rename into...
34092 (*cbranch<mode>4_btst_insn): ...this and adjust.
34093 (trap): New define_insn.
34095 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34097 PR tree-optimization/79389
34098 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34099 (noce_reversed_cond_code): New function.
34100 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34101 reversed_comparison_code. Formatting fix.
34102 (noce_try_store_flag): Test rev_cond != NULL in addition to
34103 reversed_comparison_code.
34104 (noce_try_store_flag_constants): Likewise.
34105 (noce_try_store_flag_mask): Likewise.
34106 (noce_try_addcc): Use rev_cond if non-NULL instead of
34107 reversed_comparison_code.
34108 (noce_try_cmove_arith): Likewise. Formatting fixes.
34109 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34110 (noce_find_if_block): Initialize rev_cond.
34111 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34112 instead of false as last argument never attempt to reverse it
34115 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34117 PR tree-optimization/79663
34118 * tree-predcom.c (combine_chains): Process refs in reverse order
34119 only for ZERO length chains, and add explaining comment.
34121 2017-02-23 Jeff Law <law@redhat.com>
34123 PR tree-optimization/79578
34124 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34125 in call to operand_equal_p.
34127 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34130 * config/i386/cpuid.h: Fix another undefined behavior.
34132 2017-02-23 Richard Biener <rguenther@suse.de>
34134 PR tree-optimization/79683
34135 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34136 vector types for data-refs.
34138 2017-02-23 Martin Liska <mliska@suse.cz>
34140 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34142 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34144 PR middle-end/79665
34145 * internal-fn.c (get_range_pos_neg): Moved to ...
34146 * tree.c (get_range_pos_neg): ... here. No longer static.
34147 * tree.h (get_range_pos_neg): New prototype.
34148 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34149 are known to be in between 0 and signed maximum inclusive, try to
34150 expand both unsigned and signed divmod and use the cheaper one from
34153 2017-02-22 Jeff Law <law@redhat.com>
34155 PR tree-optimization/79578
34156 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34157 to compare base operands.
34159 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34162 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34163 gpc_reg_operand instead of fpr_reg_operand.
34165 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34167 * config/mips/mips.c (mips_return_in_memory): Force FP
34168 vector types to be returned in memory for o32 ABI.
34170 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34172 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34173 instead of DW_TAG_member for static data member declarations and don't
34174 set no_linkage_name for static inline data members.
34175 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34178 2017-02-22 Martin Liska <mliska@suse.cz>
34180 * doc/invoke.texi: Replace inequality signs with square brackets
34183 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34186 * lra-constraints.c (simplify_operand_subreg): Handle
34187 WORD_REGISTER_OPERATIONS targets.
34189 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34192 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34193 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34194 elimination by swapping fld*.
34196 2017-02-22 Richard Biener <rguenther@suse.de>
34198 PR tree-optimization/79673
34199 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34200 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34201 irrelevant address-space qualifiers and avoiding a
34202 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34204 2017-02-22 Richard Biener <rguenther@suse.de>
34206 PR tree-optimization/79666
34207 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34208 to not symbolically negate if that may introduce undefined
34211 2017-02-22 Martin Liska <mliska@suse.cz>
34214 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34215 * data-streamer-out.c (streamer_write_gcov_count_stream):
34217 * value-prof.c (stream_out_histogram_value): Make assert more
34218 precise based on type of counter.
34220 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34223 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34224 Use nonimmediate_operand instead of memory_operand for operand 1.
34225 (float-extend standard_x87sse_constant_load splitter): Ditto.
34227 2017-02-21 Jeff Law <law@redhat.com>
34229 PR tree-optimization/79621
34230 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34231 blocks with edges to themselves.
34233 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34236 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34237 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34238 Use gimple_call_builtin_p.
34241 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34242 on temporarily removed DEBUG_INSNs.
34244 PR tree-optimization/79649
34245 * tree-loop-distribution.c (classify_partition): Give up on
34246 non-generic address space loads/stores.
34248 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34250 * doc/loop.texi (Loop manipulation): Remove nonexistent
34251 tree_ssa_loop_version from the documentation.
34252 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34254 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34257 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34258 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34259 * config/rs6000/rs6000.c: Include except.h.
34260 (rs6000_expand_split_stack_prologue): Call
34261 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34263 2017-02-21 Martin Jambor <mjambor@suse.cz>
34266 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34267 have been analyzed.
34269 2017-02-21 Martin Jambor <mjambor@suse.cz>
34271 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34272 for backward compatibility only.
34273 * doc/invoke.texi (Option Summary): Remove all references to
34274 -fipa-cp-alignment.
34276 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34280 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34282 * lra-constraints.c (curr_insn_transform): Handle
34283 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34285 2017-02-21 Martin Liska <mliska@suse.cz>
34287 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34289 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34291 PR translation/79638
34292 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34294 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34297 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34298 (arm_function_ok_for_sibcall): Return false for an indirect call by
34299 descriptor if all the argument registers are used.
34300 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34301 alignment of the function.
34303 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34305 PR tree-optimization/61441
34306 * simplify-rtx.c (simplify_const_unary_operation): For
34307 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34308 the sNaN unmodified.
34310 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34312 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34313 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34314 instead of SYSTEM_HEADER_DIR.
34316 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34317 Martin Liška <mliska@suse.cz>
34319 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34320 Fix typos and grammar, use active voice, and clarify.
34322 2017-02-20 Marek Polacek <polacek@redhat.com>
34324 PR middle-end/79537
34325 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34328 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34330 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34333 * config/i386/i386.c (ix86_expand_builtin): Handle
34334 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34335 ix86_builtins_isa[fcode].isa as a requirement of those
34336 flags and any other flag in the bitmask.
34337 (ix86_init_mmx_sse_builtins): Use 0 instead of
34338 ~OPTION_MASK_ISA_64BIT as mask.
34339 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34340 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34341 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34342 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34344 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34347 * lra-constraints.c (split_reg): Check requested split mode
34348 is supported by the register.
34350 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34352 * lra-constraints.c (simplify_operand_subreg): Remove early
34355 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34358 * lra-constraints.c (curr_insn_transform): Tighten condition
34359 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34361 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34364 * lra-constraints.c (curr_insn_transform): Handle
34365 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34367 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34370 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34372 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34374 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34377 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34380 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34382 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34385 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34387 * final.c (last_columnnum, override_columnnum): New variables.
34388 (final_start_function): Set last_columnnum, pass it to begin_prologue
34389 hook and pass 0 to dwarf2out_begin_prologue.
34390 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34391 to source_line debug hook.
34392 (notice_source_line): Compute last_columnnum and for debug_column_info
34393 return true on column changes.
34394 * debug.h (struct gcc_debug_hooks): Add column argument to
34395 source_line and begin_prologue hooks.
34396 (debug_nothing_int_charstar_int_bool): Remove prototype.
34397 (debug_nothing_int_int_charstar,
34398 debug_nothing_int_int_charstar_int_bool): New prototypes.
34399 (dwarf2out_begin_prologue): Add column argument.
34400 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34401 begin_prologue hooks.
34402 (debug_nothing_int_charstar_int_bool): Remove.
34403 (debug_nothing_int_int_charstar,
34404 debug_nothing_int_int_charstar_int_bool): New functions.
34405 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34406 through to dwarf2out_source_line.
34407 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34408 (dwarf2out_source_line): Add column argument, emit it if requested.
34409 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34411 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34412 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34413 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34414 through to dwarf2out_begin_prologue.
34415 (vmsdbgout_source_line): Add column argument, pass it through to
34416 dwarf2out_source_line.
34417 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34418 dbxout_source_line caller.
34419 (dbxout_source_line): Add column argument.
34421 * common.opt (gno-column-info, gcolumn-info): New options.
34422 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34423 (check_die): Also test for multiple DW_AT_decl_column attributes.
34424 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34425 DW_AT_decl_column if requested.
34426 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34428 (gen_variable_die): Likewise.
34429 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34430 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34433 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34434 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34435 (ix86_handle_option): Handle OPT_m3dnowa.
34436 * doc/invoke.texi (-m3dnowa): Document.
34437 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34438 -m3dnowa instead of -m3dnow -march=athlon.
34441 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34442 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34444 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34447 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34448 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34449 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34450 generator for vsx_xxpermdi_<mode>_be.
34451 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34452 force big-endian semantics.
34453 (vsx_xxpermdi_<mode>_be): New define_expand with same
34454 implementation as previous version of vsx_xxpermdi_<mode>.
34456 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34458 PR tree-optimization/79327
34459 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34460 variable, its initialization and use.
34462 2017-02-17 Julia Koval <julia.koval@intel.com>
34464 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34465 (OPTION_MASK_ISA_PKU_UNSET): New.
34466 (ix86_handle_option): Handle -mrdpid.
34467 * config/i386/cpuid.h (bit_RDPID): New.
34468 * config/i386/driver-i386.c (host_detect_local_cpu):
34469 Detect RDPID feature.
34470 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34471 * config/i386/i386-c.c (ix86_target_macros_internal):
34473 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34474 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34475 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34476 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34477 * config/i386/i386.md (define_insn "rdpid"): New.
34478 * config/i386/i386.opt Add -mrdpid.
34479 * config/i386/immintrin.h (_rdpid_u32): New.
34481 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34483 PR rtl-optimization/79541
34484 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34485 instead of transforming it into USE.
34487 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34489 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34490 If HONOR_SNANS (SFmode) force the input to a register.
34491 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34492 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34493 an frsp or similar insn.
34495 2017-02-17 Martin Liska <mliska@suse.cz>
34497 PR rtl-optimization/79577
34498 * params.def (selsched-max-sched-times): Increase minimum to 1.
34500 2017-02-17 Martin Liska <mliska@suse.cz>
34502 PR rtl-optimization/79574
34503 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34505 2017-02-17 Martin Liska <mliska@suse.cz>
34507 PR tree-optimization/79529
34508 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34509 ssa_defined_default_def_p to handle cases which are implicitly
34511 * tree-ssa.c (ssa_defined_default_def_p): New function.
34512 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34513 which are implicitly defined.
34514 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34516 2017-02-17 Richard Biener <rguenther@suse.de>
34518 PR middle-end/79576
34519 * params.def (max-ssa-name-query-depth): Limit to 10.
34521 2017-02-17 Richard Biener <rguenther@suse.de>
34523 PR tree-optimization/79552
34524 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34527 2017-02-17 Richard Biener <rguenther@suse.de>
34530 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34532 2017-02-17 Marek Polacek <polacek@redhat.com>
34534 PR middle-end/79536
34535 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34536 (fold_negate_expr): New wrapper.
34538 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34540 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34541 Correct terminology and de-emphasize pre-standard behavior.
34543 2017-02-16 Alan Modra <amodra@gmail.com>
34545 PR rtl-optimization/79286
34546 * ira.c (def_dominates_uses): New function.
34547 (update_equiv_regs): Don't create an equivalence for insns that
34548 may trap where the register def does not dominate the use.
34550 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34552 PR rtl-optimization/78127
34553 * lra.c (lra): Call lra_eliminate before finish the loop after
34556 2017-02-16 Richard Biener <rguenther@suse.de>
34558 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34560 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34561 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34562 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34563 (isl_val_int_from_wi): New function.
34564 (extract_affine_gmp): Rename to ...
34565 (extract_affine_wi): ... this, take a widest_int.
34566 (extract_affine_int): Just wrap extract_affine_wi.
34567 (add_param_constraints): Use isl_val_int_from_wi.
34568 (add_loop_constraints): Likewise, and extract_affine_wi.
34570 2017-02-15 Jeff Law <law@redhat.com>
34572 PR middle-end/79521
34573 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34574 ira_init_register_move_cost_if_necessary.
34576 2017-02-15 Martin Sebor <msebor@redhat.com>
34578 PR middle-end/32003
34579 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34580 removed in a prior commit.
34582 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34584 PR tree-optimization/79347
34585 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34586 counters during peeling.
34588 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34590 * Makefile.in (site.exp): Remove "set ISLVER".
34592 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34595 * real.c (real_from_integer): Call real_convert even for decimal.
34597 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34600 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34602 2017-02-14 Andrew Pinski <apinski@cavium.com>
34604 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34605 cores and change the partno/implementer to be correct.
34606 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34607 the 'B" as the implementer.
34608 * config/aarch64/aarch64-tune.md: Regenerate.
34610 2017-02-14 Carl Love <cel@us.ibm.com>
34612 * config/rs6000/rs6000.c: Add case statement entry to make the
34613 xvcvuxdsp built-in argument unsigned.
34614 * config/rs6000/vsx.md: Fix the source and return operand types so they
34615 match the instruction definitions from the ISA document. Fix typo
34616 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34619 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34622 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34623 member early_clobber_alts.
34624 * lra-lives.c (reg_early_clobber_p): New.
34625 (process_bb_lives): Use it.
34626 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34627 (debug_operand_data): Initialize early_clobber_alts.
34628 (setup_operand_alternative): Set up early_clobber_alts.
34629 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34630 alternatives to new_insn_reg.
34631 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34633 (lra_update_insn_regno_info): Pass the new arg.
34635 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34637 PR middle-end/79505
34638 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34639 (new_oacc_loop_raw): Don't clear already cleared fields.
34642 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34643 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34644 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34646 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34649 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34651 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34654 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34655 the extra instruction to the right place to store 128-bit constant
34658 2017-02-14 Martin Sebor <msebor@redhat.com>
34660 PR middle-end/79448
34661 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34662 warning for strings of unknown length.
34664 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34666 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34668 2017-02-14 Jeff Law <law@redhat.com>
34671 * ira-costs.c (scan_one_insn): Initialize register move costs
34672 for pseudos seen in USE/CLOBBER insns.
34674 PR tree-optimization/79095
34675 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34676 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34677 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34678 if the operands are known to be not equal, then the resulting range
34680 (intersect_ranges): If the new range is ~[0,0] and the old range is
34681 wide, then prefer ~[0,0].
34682 * tree-vrp.c (overflow_comparison_p_1): New function.
34683 (overflow_comparison_p): New function.
34684 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34685 if NAME is used in an overflow test.
34686 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34687 overflow check that can be expressed as an equality test, then adjust
34688 ops to be that equality test.
34690 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34692 * config/s390/s390-builtin-types.def: Remove flags argument.
34693 * config/s390/s390.c (s390_init_builtins): Likewise.
34695 2017-02-14 Martin Liska <mliska@suse.cz>
34697 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34698 vector. Fix trailing white spaces.
34700 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34702 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34705 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34707 PR rtl-optimization/68664
34708 * config/arm/arm.c (arm_sched_can_speculate_insn):
34709 New function. Declare prototype.
34710 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34712 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34714 PR rtl-optimization/68664
34715 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34717 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34719 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34721 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34722 max skip bytes for function, loop and jump.
34724 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34726 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34727 ABS_EXPR for gimple dump.
34729 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34732 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34734 PR tree-optimization/79408
34735 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34736 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34737 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34738 also if rhs1 is INTEGER_CST.
34740 2017-02-14 Richard Biener <rguenther@suse.de>
34742 PR middle-end/79432
34743 * tree-into-ssa.c (insert_phi_nodes): When the function can
34744 have abnormal edges rewrite SSA names with broken use-def
34745 dominance out of SSA and register them for PHI insertion.
34747 2017-02-13 Martin Sebor <msebor@redhat.com>
34749 PR middle-end/79496
34750 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34751 clearing info.nowrite flag when snprintf size argument is a range.
34753 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34755 * cprop.c (cprop_jump): Add missing space in string literal.
34756 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34757 (get_constraint_for_component_ref): Likewise.
34758 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34759 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34760 * lra-constraints.c (process_alt_operands): Likewise.
34761 * ipa-inline.c (inline_small_functions): Likewise.
34762 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34763 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34764 * trans-mem.c (diagnose_tm_1_op): Likewise.
34765 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34766 (grid_parallel_clauses_gridifiable): Likewise.
34768 * config/nvptx/mkoffload.c (process): Add space in between
34771 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34772 "MOD4_SSE_REGS" and "ALL_REGS".
34774 * spellcheck.c (test_data): Add , in between "foo" and "food".
34776 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34779 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34780 boundary crossing check and subsequent code generation agree.
34782 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34784 * config/aarch64/aarch64.c (has_memory_op): Delete.
34785 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34788 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34790 PR rtl-optimization/79388
34791 PR rtl-optimization/79450
34792 * combine.c (distribute_notes): When removing TEM_INSN for which
34793 corresponding dest has last value recorded, invalidate that last
34796 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34798 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34799 of explicit '@'. Add missing assembly comment marker on branch costs
34802 2017-02-13 Nathan Sidwell <nathan@acm.org>
34804 * gengtype-lex.l (<in_struct>): Add '/'.
34806 2017-02-13 Martin Liska <mliska@suse.cz>
34809 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34811 2017-02-13 Richard Biener <rguenther@suse.de>
34813 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34815 * configure: Re-generate.
34816 * config.in: Likewise.
34817 * graphite-dependences.c: Simplify as if
34818 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34819 * graphite-isl-ast-to-gimple.c: Likewise.
34820 * graphite-optimize-isl.c: Likewise.
34821 * graphite-poly.c: Likewise.
34822 * graphite-sese-to-poly.c: Likewise.
34823 * graphite.h: Likewise.
34824 * toplev.c: Include isl/version.h and use isl_version () for
34825 printing the ISL version.
34826 * doc/install.texi: Update ISL requirement.
34828 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34830 * doc/standards.texi (Standards): Update reference to
34833 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34835 * doc/extend.texi (Named Address Spaces): sourceware.org now
34837 * doc/install.texi (Binaries): Ditto.
34840 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34842 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34843 standard terminology "stringize"/"stringizing" throughout.
34844 * doc/cppinternals.texi: Likewise.
34846 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34848 * doc/extend.texi: Fix some spelling mistakes and typos.
34849 * doc/invoke.texi: Likewise.
34851 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34854 * params.def (inline-min-speedup) Change from 10 to 8.
34856 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34858 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34861 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34864 * ipa-inline-analysis.c (get_minimal_bb): New function.
34865 (record_modified): Use it.
34866 (remap_edge_change_prob): Handle also ancestor functions.
34868 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34870 * doc/contrib.texi (Contributors): Remove broken link into
34871 the Mauve CVS repository.
34873 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34875 PR middle-end/79454
34876 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34877 result computation whenever lhs doesn't have vector mode, not
34878 just when it has BLKmode.
34880 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34882 * doc/makefile.texi (profiledbootstrap): Refer to the
34883 installation instructions only in textual form.
34885 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34888 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34890 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34892 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34893 (Specific): Update mingw-w64 reference.
34895 (Specific): Remove broken link to Renesas RX processor.
34897 2017-02-10 Richard Biener <rguenther@suse.de>
34899 * toplev.c (process_options): Do not mention obsolete graphite
34900 options when printing sorry message about missing graphite support.
34901 Mention -floop-nest-optimize.
34903 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34905 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34906 (vtst_p16): Likewise.
34907 (vtstq_p8): Likewise.
34908 (vtstq_p16): Likewise.
34910 (vtstq_p64): Likewise.
34911 * config/arm/arm_neon.h (vgetq_lane_p64): New.
34912 (vset_lane_p64): New.
34913 (vsetq_lane_p64): New.
34915 2017-02-10 Jakub Jelinek <jakub@redhat.com>
34917 PR tree-optimization/79411
34918 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
34919 stmt operands are SSA_NAMEs used in abnormal phis.
34920 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
34923 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34926 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
34929 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34931 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
34933 2017-02-09 Jakub Jelinek <jakub@redhat.com>
34935 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
34939 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
34940 not arbitrary TREE_CONSTANT.
34943 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
34944 "omp declare target link" attribute unless is_global_var.
34945 * omp-offload.c (find_link_var_op): Likewise.
34947 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34948 Chung-Lin Tang <cltang@codesourcery.com>
34950 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
34952 (gimplify_adjust_omp_clauses): Don't delete TILE.
34953 (gimplify_omp_for): Deal with TILE.
34954 * internal-fn.c (expand_GOACC_TILE): New function.
34955 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
34957 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
34958 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
34960 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
34961 avoid DIV for outermost collapse var.
34962 (expand_oacc_for): Insert tile element loop as needed. Adjust.
34963 Remove out of date comments, fix whitespace.
34964 * omp-general.c (omp_extract_for_data): Deal with tiling.
34965 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
34966 adjust OLF_DIM_BASE value.
34967 (struct omp_for_data): Add tiling field.
34968 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
34969 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
34970 for auto loops. Remove default auto determining, moved to
34971 oacc_loop_fixed_partitions.
34972 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
34973 stmts, add e_mask field.
34974 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
34975 (oacc_thread_numbers): Use oacc_dim_call.
34976 (oacc_xform_tile): New.
34977 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
34978 (finish_oacc_loop): Adjust for ifns vector.
34979 (oacc_loop_discover_walk): Append loop abstraction sites to list,
34980 add case for GOACC_TILE fns.
34981 (oacc_loop_xform_loop): Delete.
34982 (oacc_loop_process): Iterate over call list directly, and add
34983 handling for GOACC_TILE fns.
34984 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
34986 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
34987 vector partitioning to outer loops. Assign 2 partitions to loops
34988 when available. Add TILE handling.
34989 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
34990 (execite_oacc_device_lower): Process GOACC_TILE fns,
34991 ignore unknown specs.
34992 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
34993 * tree.c (omp_clause_num_ops): Adjust TILE ops.
34994 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
34996 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
34998 * configure.ac (ACX_BUGURL): Update.
34999 * configure: Regenerate.
35001 2017-02-09 Richard Biener <rguenther@suse.de>
35003 PR tree-optimization/69823
35004 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35005 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35007 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35009 * config/arc/arc-c.def: Add __NPS400__ definition.
35010 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35011 (TARGET_NPS400): Define.
35013 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35015 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35017 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35018 pointer, arch_info.
35019 (arc_cpu_types): Fill the arch_info field with a pointer into the
35020 arc_arch_types table.
35021 (arc_selected_cpu): Declare.
35022 * config/arc/arc.c (arc_selected_cpu): Make global.
35023 (arc_selected_arch): Delete.
35024 (arc_base_cpu): Delete.
35025 (arc_override_options): Remove references to deleted variables,
35026 update access to arch information.
35027 (ARC_OPT): Update access to arch information.
35028 (ARC_OPTX): Likewise.
35029 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35030 (TARGET_ARC600): Update access to arch information.
35031 (TARGET_ARC601): Likewise.
35032 (TARGET_ARC700): Likewise.
35033 (TARGET_EM): Likewise.
35034 (TARGET_HS): Likewise.
35035 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35038 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35041 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35042 condition/operands for integer GE/LE/GEU/LEU operations.
35044 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35046 PR translation/79397
35047 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35050 2017-02-08 Martin Jambor <mjambor@suse.cz>
35053 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35054 whether allocation happened.
35055 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35056 nothing was allocated.
35058 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35060 PR tree-optimization/79408
35061 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35062 constant, but SSA_NAME with a known integer range, use the minimum
35063 of that range instead of op1 to determine if modulo can be replaced
35064 with its first operand.
35066 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35068 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35070 2017-02-08 Richard Biener <rguenther@suse.de>
35072 PR tree-optimization/71824
35073 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35074 Check all loops contained in the merged region.
35076 2017-02-07 Andrew Pinski <apinski@cavium.com>
35078 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35080 2017-02-07 Andrew Pinski <apinski@cavium.com>
35082 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35083 (thunderxt88): Likewise.
35084 (thunderxt81): Disable LSE and change v8.1 to v8.
35085 (thunderxt83): Likewise.
35087 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35088 Richard Biener <rguenther@suse.de>
35090 PR middle-end/79399
35091 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35092 type from int to size_t.
35093 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35095 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35097 PR rtl-optimization/79386
35098 * cprop.c (bypass_conditional_jumps): Initialize
35099 bypass_last_basic_block already before splitting bbs after
35100 unconditional traps...
35101 (bypass_conditional_jumps): ... rather than here.
35104 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35105 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35106 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35107 fix -masm=intel patterns.
35109 2017-02-07 Richard Biener <rguenther@suse.de>
35111 PR tree-optimization/79256
35112 PR middle-end/79278
35113 * builtins.c (get_object_alignment_2): Use min_align_of_type
35114 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35115 and ADJUST_FIELD_ALIGN.
35117 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35119 * doc/tm.texi: Regenerate.
35120 * stor-layout.c (layout_decl): Adjust.
35121 (update_alignment_for_field): Likewise.
35122 (place_field): Likewise.
35123 (min_align_of_type): Likewise.
35124 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35125 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35126 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35127 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35128 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35129 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35130 * config/i386/i386.c (x86_field_alignment): Likewise.
35131 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35132 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35133 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35134 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35135 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35136 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35140 2017-01-30 Richard Biener <rguenther@suse.de>
35142 PR tree-optimization/79256
35143 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35144 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35147 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35149 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35150 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35151 builtins to SImode and emit a zero-extend, if necessary.
35153 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35155 * docs/invoke.texi (RISC-V Options): Alphabetize.
35157 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35159 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35162 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35164 * config/riscv/riscv.c: New file.
35165 * common/config/riscv/riscv-common.c: Likewise.
35166 * config.gcc: Likewise.
35167 * config/riscv/constraints.md: Likewise.
35168 * config/riscv/elf.h: Likewise.
35169 * config/riscv/generic.md: Likewise.
35170 * config/riscv/linux.h: Likewise.
35171 * config/riscv/multilib-generator: Likewise.
35172 * config/riscv/peephole.md: Likewise.
35173 * config/riscv/pic.md: Likewise.
35174 * config/riscv/predicates.md: Likewise.
35175 * config/riscv/riscv-builtins.c: Likewise.
35176 * config/riscv/riscv-c.c: Likewise.
35177 * config/riscv/riscv-ftypes.def: Likewise.
35178 * config/riscv/riscv-modes.def: Likewise.
35179 * config/riscv/riscv-opts.h: Likewise.
35180 * config/riscv/riscv-protos.h: Likewise.
35181 * config/riscv/riscv.h: Likewise.
35182 * config/riscv/riscv.md: Likewise.
35183 * config/riscv/riscv.opt: Likewise.
35184 * config/riscv/sync.md: Likewise.
35185 * config/riscv/t-elf-multilib: Likewise.
35186 * config/riscv/t-linux: Likewise.
35187 * config/riscv/t-linux-multilib: Likewise.
35188 * config/riscv/t-riscv: Likewise.
35189 * configure.ac: Likewise.
35190 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35191 Waterman as RISC-V maintainers.
35192 * doc/install.texi: Add RISC-V entries.
35193 * doc/invoke.texi: Add RISC-V options section.
35194 * doc/md.texi: Add RISC-V constraints section.
35195 * configure: Regenerated.
35197 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35200 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35201 false values to be constant vectors with all 0 or all 1 bits set.
35202 (vcondu<mode><mode>): Likewise.
35203 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35205 (fpmask_comparison_operator): Update comment.
35206 (vecint_comparison_operator): New predicate.
35207 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35208 vector conditionals when the true and false values are constant
35209 vectors with all 0 bits or all 1 bits set.
35211 2017-02-06 Martin Sebor <msebor@redhat.com>
35213 PR tree-optimization/79376
35214 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35216 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35218 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35219 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35220 to simplify split condition.
35222 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35224 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35225 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35228 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35230 PR rtl-optimization/68664
35231 * target.def (can_speculate_insn): New hook.
35232 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35233 * doc/tm.texi: Regenerate.
35234 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35235 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35236 (rs6000_sched_can_speculate_insn): New function.
35238 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35240 PR tree-optimization/79284
35241 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35242 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35243 vectorizable_mask_load_store, vectorizable_operation,
35244 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35245 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35246 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35247 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35248 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35249 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35250 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35251 is_gimple_assign (stmt). Replace another such test with
35252 is_gimple_assign (stmt).
35254 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35257 * config/avr/avr.c (rtl-iter.h): Include it.
35258 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35259 (avr_legitimate_combined_insn): ...and implementation.
35261 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35263 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35264 * config/s390/s390.c (s390_const_operand_ok)
35265 (s390_canonicalize_comparison, s390_extract_part)
35266 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35267 (s390_contiguous_bitmask_p, s390_rtx_costs)
35268 (legitimize_pic_address): Likewise.
35269 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35270 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35271 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35272 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35273 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35275 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35277 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35278 REGNO($0) == REGNO($1).
35280 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35282 * config/s390/linux.h(SIZE_TYPE): Add comment.
35284 2017-02-06 Julian Brown <julian@codesourcery.com>
35285 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35286 Virendra Pathak <virendra.pathak@broadcom.com>
35288 * config/aarch64/aarch64-cores.def: Change the scheduler
35290 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35291 * config/aarch64/thunderx2t99.md: New file.
35293 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35295 * doc/standards.texi (Go Language): Update link to language
35298 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35300 * tree-eh.c (lower_resx): Sanitize profile.
35301 (cleanup_empty_eh_move_lp): Likewise.
35303 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35306 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35308 * cfgloopmanip.h (loop_version): Update prototype.
35309 * modulo-sched.c (sms_schedule): Update call of loop_version.
35310 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35311 * tree-parloops.c (gen_parallel_loop): Likewise.
35312 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35313 * tree-ssa-loop-split.c (split_loop): Likewise.
35314 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35315 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35317 2017-02-05 Martin Liska <mliska@suse.cz>
35320 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35322 (print_operand_address): Initialize a struct to zero.
35324 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35326 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35327 garbage collector only in textual form.
35329 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35331 * doc/extend.texi (x86 specific memory model extensions for
35332 transactional memory): Simplify a phrase.
35334 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35337 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35338 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35339 (atomic_storedi_1): Likewise.
35341 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35343 PR tree-optimization/79338
35344 * tree-parloops.c (gather_scalar_reductions): Don't call
35345 vect_analyze_loop_form for loop->inner before destroying loop's
35348 2017-02-03 Martin Sebor <msebor@redhat.com>
35350 PR tree-optimization/79327
35351 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35352 when precision has resulted in leading zeros.
35353 (format_integer): Adjust the likely counter to assume an unknown
35354 argument that may be zero is non-zero.
35356 2017-02-03 Jason Merrill <jason@redhat.com>
35359 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35360 avoid copying non-taken branch.
35362 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35364 PR tree-optimization/79340
35365 * tree-vect-loop.c (vectorizable_reduction): Release
35366 vec_defs elements after safe_splicing them into other vectors.
35369 PR tree-optimization/79327
35370 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35371 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35373 (format_integer): Use wide_int_to_tree instead of build_int_cst
35374 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35375 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35376 of shortest and longest sequence.
35378 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35380 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35381 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35383 2017-02-03 Walter Lee <walt@tilera.com>
35386 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35387 after initial stackframe link reg save.
35388 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35390 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35393 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35394 wu for stxssp alternative.
35396 2017-02-03 Martin Sebor <msebor@redhat.com>
35398 PR tree-optimization/79352
35399 * gimple-fold.c (get_range_strlen): Add argument.
35400 (get_range_strlen): Change return type to bool.
35401 (get_maxval_strlen): Pass in a dummy argument.
35402 * gimple-fold.h (get_range_strlen): Change return type to bool.
35403 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35404 * tree.h (array_at_struct_end_p): Add argument.
35405 * tree.c (array_at_struct_end_p): Handle it.
35407 2017-02-03 Martin Liska <mliska@suse.cz>
35410 * multiple_target.c (create_dispatcher_calls): Redirect edge
35411 from a caller of a dispatcher.
35412 (expand_target_clones): Make the clones local.
35413 (ipa_target_clone): Do both target clones and resolvers.
35414 (ipa_dispatcher_calls): Remove the pass.
35415 (pass_dispatcher_calls::gate): Likewise.
35416 (make_pass_dispatcher_calls): Likewise.
35417 * passes.def (pass_target_clone): Put as very first IPA early
35420 2017-02-03 Martin Liska <mliska@suse.cz>
35422 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35423 in case of a function with ifunc attribute.
35425 2017-02-03 Martin Liska <mliska@suse.cz>
35427 * cgraph.c (cgraph_node::dump): Dump function version info.
35428 * symtab.c (symtab_node::dump_base): Add missing new line.
35430 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35432 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35433 (ifcombine_ifandif): Use it.
35435 2017-02-03 Martin Liska <mliska@suse.cz>
35437 * doc/invoke.texi: Document default value for
35438 use-after-scope-direct-emission-threshold.
35440 2017-02-03 Martin Liska <mliska@suse.cz>
35442 PR tree-optimization/79339
35443 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35444 (format_floating): Likewise.
35446 2017-02-03 Martin Liska <mliska@suse.cz>
35449 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35451 (ipa_node_params_t::remove): Likewise.
35452 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35453 initialization from removed ipa_node_params_t::insert.
35454 (ipa_node_params::~ipa_node_params): Move from removed
35455 ipa_node_params_t::release.
35456 * symbol-summary.h (symbol_summary::m_released): New member.
35457 Do not release a summary twice. Do not allow to call finalizer
35458 for types of a summary that live in GGC memory.
35460 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35462 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35465 2017-02-02 Martin Sebor <msebor@redhat.com>
35467 PR middle-end/79275
35468 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35469 (format_string): Tighten up the range of output for non-constant
35470 strings and correct the expected range for wide non-constant strings.
35472 2017-02-02 Martin Sebor <msebor@redhat.com>
35474 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35476 PR middle-end/32003
35477 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35479 (-fdump-tree-@var): Add to index and document how to come up
35480 with pass-specific option and dump file names.
35481 (-fdump-passes): Clarify where to look for output.
35483 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35485 PR middle-end/77445
35486 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35487 statistics of the analyzed path; allow threading for speed when
35488 any of BBs along the path are optimized for speed.
35490 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35492 PR middle-end/78468
35493 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35494 settings of the virtual registers.
35497 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35499 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35500 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35503 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35505 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35506 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35508 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35510 * config/s390/s390.md: Add missing comments with the expanded
35512 * config/s390/vector.md: Likewise.
35513 * config/s390/vx-builtins.md: Likewise.
35515 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35518 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35519 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35520 conditions on a single line.
35522 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35524 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35525 __S390_VX__ to __VX__.
35527 2017-02-01 Andrew Pinski <apinski@cavium.com>
35529 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35530 stmt_info to record_stmt_cost.
35531 (vect_get_known_peeling_cost): Pass stmt_info if known to
35533 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35534 cpu_vector_cost field into
35535 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35536 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35537 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35538 splitting of scalar_stmt_cost and vec_stmt_cost.
35539 (thunderx_vector_cost): Likewise.
35540 (cortexa57_vector_cost): LIkewise.
35541 (exynosm1_vector_cost): Likewise.
35542 (xgene1_vector_cost): Likewise.
35543 (thunderx2t99_vector_cost): Improve after the splitting of the two
35545 (aarch64_builtin_vectorization_cost): Update for the splitting of
35546 scalar_stmt_cost and vec_stmt_cost.
35548 2017-02-01 Torvald Riegel <triegel@redhat.com>
35549 Richard Henderson <rth@redhat.com>
35551 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35552 conditional on existance of a fast atomic load.
35553 * optabs-query.c (can_atomic_load_p): New function.
35554 * optabs-query.h (can_atomic_load_p): Declare it.
35555 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35556 no fast atomic load is available for the particular size of access.
35557 (expand_atomic_compare_and_swap): Likewise.
35558 (expand_atomic_load): Likewise.
35559 (expand_atomic_store): Likewise.
35560 (expand_atomic_fetch_op): Likewise.
35561 * testsuite/lib/target-supports.exp
35562 (check_effective_target_sync_int_128): Remove x86 because it provides
35563 no fast atomic load.
35564 (check_effective_target_sync_int_128_runtime): Likewise.
35566 2017-02-01 Richard Biener <rguenther@suse.de>
35568 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35569 (graphite_transform_loops): Provide opt-info for optimized nests.
35570 * tree-parloop.c (parallelize_loops): Provide opt-info for
35571 parallelized loops.
35573 2017-02-01 Richard Biener <rguenther@suse.de>
35575 PR middle-end/79315
35576 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35577 was not set before.
35579 2017-02-01 Richard Biener <rguenther@suse.de>
35581 PR tree-optimization/71824
35582 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35583 Verify the loops are valid in the merged SESE region.
35584 (scop_detection::can_represent_loop_1): Check analyzing the
35585 evolution of the number of iterations in the region succeeds.
35587 2017-01-31 Ian Lance Taylor <iant@golang.org>
35589 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35590 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35592 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35594 PR preprocessor/79210
35595 * input.c (get_substring_ranges_for_loc): Replace line_width
35596 assertion with error-handling.
35598 2017-01-31 Richard Biener <rguenther@suse.de>
35600 PR tree-optimization/77318
35601 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35602 (create_pw_aff_from_tree): Take loop parameter.
35603 (add_condition_to_pbb): Pass loop of the condition to
35604 create_pw_aff_from_tree.
35606 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35608 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35609 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35611 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35615 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35617 (convert_int_to_float128): Likewise.
35618 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35619 (convert_int_to_float128): Likewise.
35620 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35621 (UNSPEC_IEEE128_CONVERT): Likewise.
35622 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35623 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35624 Use local variables for IBM extended format.
35625 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35626 (fix_trunc<mode>si2_fprs): Likewise.
35627 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35628 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35629 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35630 to know that we can now have integers of all sizes in vector
35632 (fix<uns>_<mode>di2_hw): Likewise.
35633 (float<uns>_<mode>si2_hw): Likewise.
35634 (fix_<mode>si2_hw): Likewise.
35635 (fixuns_<mode>si2_hw): Likewise.
35636 (float<uns>_<mode>di2_hw): Likewise.
35637 (float_<mode>di2_hw): Likewise.
35638 (float_<mode>si2_hw): Likewise.
35639 (floatuns_<mode>di2_hw): Likewise.
35640 (floatuns_<mode>si2_hw): Likewise.
35641 (xscvqp<su>wz_<mode>): Delete, no longer used.
35642 (xscvqp<su>dz_<mode>): Likewise.
35643 (xscv<su>dqp_<mode>): Likewise.
35644 (ieee128_mfvsrd_64bit): Likewise.
35645 (ieee128_mfvsrd_32bit): Likewise.
35646 (ieee128_mfvsrwz): Likewise.
35647 (ieee128_mtvsrw): Likewise.
35648 (ieee128_mtvsrd_64bit): Likewise.
35649 (ieee128_mtvsrd_32bit): Likewise.
35651 2017-01-31 Martin Liska <mliska@suse.cz>
35654 * ipa-prop.c (ipa_free_all_node_params): Call release method
35655 instead of ~sumbol_summary to not to trigger double times
35658 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35660 PR tree-optimization/71691
35661 * bitmap.h (class auto_bitmap): New.
35662 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35663 is_maybe_undefined instead of ssa_undefined_value_p.
35665 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35667 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35668 __S390_ARCH_LEVEL__ to __ARCH__.
35670 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35672 PR tree-optimization/79267
35673 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35674 if should_remove_lhs_p is true.
35676 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35679 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35680 (add_alignment_attribute): New.
35681 (base_type_die): Add alignment attribute.
35682 (subrange_type_die): Likewise.
35683 (modified_type_die): Likewise.
35684 (gen_array_type_die): Likewise.
35685 (gen_descr_array_type_die: Likewise.
35686 (gen_enumeration_type_die): Likewise.
35687 (gen_subprogram_die): Likewise.
35688 (gen_variable_die): Likewise.
35689 (gen_field_die): Likewise.
35690 (gen_ptr_to_mbr_type_die): Likewise.
35691 (gen_struct_or_union_type_die): Likewise.
35692 (gen_subroutine_type_die): Likewise.
35693 (gen_typedef_die): Likewise.
35694 (base_type_cmp): Compare alignment attribute.
35696 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35699 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35700 (setb_unsigned) New pattern for setb with CCUNS.
35701 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35702 subfc./subfe sequence to avoid overflow problems. Generate a
35703 shorter sequence with cmpld/setb for power9.
35704 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35705 for generating subfc. instruction.
35706 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35707 now uses this instruction.
35709 2017-01-30 Ian Lance Taylor <iant@google.com>
35712 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35713 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35715 2017-01-30 Martin Sebor <msebor@redhat.com>
35717 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35718 Move constant to the right of a relational operator.
35719 (get_mpfr_format_length, format_character, format_string): Ditto.
35720 (should_warn_p, maybe_warn): Same.
35722 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35724 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35727 * asan.c (get_translation_unit_decl): Remove function.
35728 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35730 2017-01-30 Martin Liska <mliska@suse.cz>
35732 PR gcov-profile/79259
35733 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35734 -fprofile-generate.
35736 2017-01-30 Martin Liska <mliska@suse.cz>
35739 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35740 Initialize variables with NULL value.
35742 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35745 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35747 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35749 2017-01-30 Richard Biener <rguenther@suse.de>
35751 PR tree-optimization/79276
35752 * tree-vrp.c (process_assert_insertions): Properly adjust common
35753 when removing a duplicate.
35755 2017-01-30 Richard Biener <rguenther@suse.de>
35757 PR tree-optimization/79256
35758 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35759 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35761 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35763 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35766 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35767 ("*r<noxa>sbg_<mode>_sll_bitmask")
35768 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35769 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35770 Use contiguous_bitmask_nowrap_operand.
35772 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35775 * config/rs6000/altivec.h (vec_xl): Revise #define.
35776 (vec_xst): Likewise.
35778 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35780 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35782 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35784 PR rtl-optimization/79194
35785 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35786 traps before call to bypass_conditional_jumps.
35788 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35790 PR tree-optimization/71374
35791 * lra-constraints.c (check_conflict_input_operands): New.
35792 (match_reload): Use it.
35794 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35797 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35798 account to calculate conflict_set.
35800 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35802 PR rtl-optimization/78559
35803 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35804 other_insn in combine.
35806 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35808 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35809 uint16_type_node for BT_UINT16.
35811 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35813 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35814 "RTL Tests" to menu.
35815 (GIMPLE Tests): New node.
35816 (RTL Tests): New node.
35818 2017-01-27 Richard Biener <rguenther@suse.de>
35820 PR tree-optimization/79245
35821 * tree-loop-distribution.c (distribute_loop): Apply cost
35822 modeling also to detected patterns.
35824 2017-01-27 Richard Biener <rguenther@suse.de>
35826 PR tree-optimization/71433
35827 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35828 (compare_assert_loc): New function.
35829 (process_assert_insertions): Sort and optimize assert locations
35830 to remove duplicates and push down identical assertions on
35831 edges to their destination block.
35833 2017-01-27 Richard Biener <rguenther@suse.de>
35835 PR tree-optimization/79244
35836 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35837 out SSA names even if abnormal.
35839 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35841 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35842 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35843 instead of MPFR_RNDN.
35845 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35848 * arm.c (arm_option_override): Don't call build_target_option_node
35849 until after doing all option overrides.
35850 (arm_valid_target_attribute_tree): Likewise.
35852 2017-01-27 Martin Liska <mliska@suse.cz>
35854 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35855 for {cd}tors and C++ {cd}tors.
35857 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35859 * config/s390/s390.md ("*setmem_long_and")
35860 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35862 2017-01-26 Martin Sebor <msebor@redhat.com>
35864 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35867 2017-01-26 Martin Sebor <msebor@redhat.com>
35869 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35870 HAVE_DFmode before using XFmode or DFmode.
35871 (parse_directive): Avoid using the z length modifier to avoid
35872 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35874 PR middle-end/78703
35875 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35876 to accept adjustment as an array.
35877 (get_int_range): New function.
35878 (struct directive): Make width and prec arrays.
35879 (directive::set_width, directive::set_precision): Call get_int_range.
35880 (format_integer, format_floating): Handle width and precision ranges.
35881 (format_string, parse_directive): Same.
35883 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35886 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35887 comdat_type_p set, just clone them, but keep the children in the
35891 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35892 which have direct callers with -fvar-tracking-assignments enabled
35894 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35895 inside of type units.
35897 2017-01-26 Martin Sebor <msebor@redhat.com>
35899 PR middle-end/78703
35900 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35902 (struct format_result): Replace number_chars, number_chars_min,
35903 and number_chars_max with a single member of struct result_range.
35905 (format_result::operator+=): Adjust.
35906 (struct fmtresult): Remove bounded. Handle likely and unlikely
35908 (fmtresult::adjust_for_width_or_precision): New function.
35909 (fmtresult:type_max_digits): New function.
35910 (bytes_remaining): Handle likely and unlikely counters.
35911 (min_bytes_remaining): Remove.
35912 (format_percent): Simplify.
35913 (format_integer, format_floating): Set likely and unlikely counters.
35914 (get_string_length, format_character, format_string): Same.
35915 (format_plain, should_warn_p): New function.
35916 (maybe_warn): Call should_warn_p. Update diagnostic messages
35917 and handle those for all directives, including plain strings.
35918 (format_directive): Handle likely and unlikely counters.
35919 Remove unnecessary quoting from diagnostics. Add an informational
35921 (add_bytes): Remove.
35922 (pass_sprintf_length::compute_format_length): Simplify.
35923 (try_substitute_return_value): Handle likely and unlikely counters.
35925 2017-01-26 Carl Love <cel@us.ibm.com>
35927 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
35928 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
35930 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
35933 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
35934 endianess for subregs into account.
35935 * lra-constraints.c (lra_constraints): Do risky transformations
35936 always on the first iteration.
35937 * lra-lives.c (check_pseudos_live_through_calls): Add arg
35938 last_call_used_reg_set.
35939 (process_bb_lives): Define and use last_call_used_reg_set.
35940 * lra.c (lra): Always continue after lra_constraints on the first
35943 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
35945 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
35947 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
35949 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35951 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
35952 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
35953 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
35954 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
35955 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
35956 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
35957 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
35958 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
35959 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
35961 2017-01-26 Marek Polacek <polacek@redhat.com>
35964 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
35965 for the third operand.
35967 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35969 PR middle-end/79236
35970 * omp-low.c (struct omp_context): Add simt_stmt field.
35971 (scan_omp_for): Return omp_context *.
35972 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
35973 context to the _simt_ SIMD stmt.
35974 (lower_omp_for): For combined SIMD with sibling _simt_
35975 SIMD, make sure to use the same decls in _looptemp_
35976 clauses as in the sibling.
35978 2017-01-26 David Sherwood <david.sherwood@arm.com>
35980 PR middle-end/79212
35981 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
35984 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35987 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
35988 emit fld b; fld a; if possible.
35990 * brig-builtins.def: Update copyright years.
35991 * config/arm/arm_acle_builtins.def: Update copyright years.
35993 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
35996 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
35997 constraint instead of o for the stxsd instruction.
35999 2017-01-25 Carl Love <cel@us.ibm.com>
36001 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36002 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36004 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36006 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36008 2017-01-25 Richard Biener <rguenther@suse.de>
36010 PR tree-optimization/69264
36011 * target.def (vector_alignment_reachable): Improve documentation.
36012 * doc/tm.texi: Regenerate.
36013 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36015 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36016 earlier changes with respect to TYPE_USER_ALIGN.
36017 (vector_alignment_reachable_p): Likewise. Improve dumping.
36019 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36022 * config/arm/arm.md (xordi3): Force constant operand into a register
36025 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36027 * doc/invoke.texi (-fstore-merging): Correct default optimization
36028 levels at which it is enabled.
36029 (-O): Move -fstore-merging from list to...
36032 2017-01-25 Richard Biener <rguenther@suse.de>
36035 * omp-expand.c: Include debug.h.
36036 (expand_omp_taskreg): Make sure to generate early debug before
36037 outlining anything from a function.
36038 (expand_omp_target): Likewise.
36039 (grid_expand_target_grid_body): Likewise.
36041 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36044 * asan.c (get_translation_unit_decl): New function.
36045 (asan_add_global): Extract modules file name from globals
36046 TRANSLATION_UNIT_DECL name.
36048 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36051 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36052 for long calls with APCS frame and VFP.
36054 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36056 * cfg.c (original_copy_tables_initialized_p): New function.
36057 * cfg.h (original_copy_tables_initialized_p): New decl.
36058 * cfgrtl.c (relink_block_chain): Guard the call to
36059 free_original_copy_tables with a call to
36060 original_copy_tables_initialized_p.
36061 * cgraph.h (symtab_node::native_rtl_p): New decl.
36062 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36063 (symtab_node::needed_p): Don't assert for early assembly output
36064 for __RTL functions.
36065 (cgraph_node::finalize_function): Set "force_output" for __RTL
36067 (cgraph_node::analyze): Bail out early for __RTL functions.
36068 (analyze_functions): Update assertion to support __RTL functions.
36069 (cgraph_node::expand): Bail out early for __RTL functions.
36070 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36072 * function.h (struct function): Update comment for field
36074 * gimple-expr.c: Include "tree-pass.h".
36075 (gimple_has_body_p): Return false for __RTL functions.
36076 * Makefile.in (OBJS): Add run-rtl-passes.o.
36077 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36079 (gcc::pass_manager::get_clean_slate): New accessor.
36080 * passes.c: Include "insn-addr.h".
36081 (should_skip_pass_p): Add logging. Update logic for running
36082 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36083 property-provider override so it is only done for gimple passes.
36085 (skip_pass): New function.
36086 (execute_one_pass): Call skip_pass when skipping passes.
36087 * read-md.c (md_reader::read_char): Support filtering
36088 the input to a subset of line numbers.
36089 (md_reader::md_reader): Initialize fields
36090 m_first_line and m_last_line.
36091 (md_reader::read_file_fragment): New function.
36092 * read-md.h (md_reader::read_file_fragment): New decl.
36093 (md_reader::m_first_line): New field.
36094 (md_reader::m_last_line): New field.
36095 * read-rtl-function.c (function_reader::create_function): Only
36096 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36097 curr_properties. Set DECL_INITIAL to a dummy block.
36098 (read_rtl_function_body_from_file_range): New function.
36099 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36101 * run-rtl-passes.c: New file.
36102 * run-rtl-passes.h: New file.
36104 2017-01-24 Jeff Law <law@redhat.com>
36106 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36109 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36111 PR tree-optimization/79159
36112 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36113 (record_nonwrapping_iv): Improve boundary using above function if no
36114 value range information.
36116 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36117 Martin Jambor <mjambor@suse.cz>
36119 * brig-builtins.def: New file.
36120 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36121 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36122 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36123 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36124 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36125 * builtin-types.def (BT_INT8): New.
36126 (BT_INT16): Likewise.
36127 (BT_UINT8): Likewise.
36128 (BT_UINT16): Likewise.
36129 (BT_FN_ULONG): Likewise.
36130 (BT_FN_UINT_INT): Likewise.
36131 (BT_FN_UINT_ULONG): Likewise.
36132 (BT_FN_UINT_LONG): Likewise.
36133 (BT_FN_UINT_PTR): Likewise.
36134 (BT_FN_ULONG_PTR): Likewise.
36135 (BT_FN_INT8_FLOAT): Likewise.
36136 (BT_FN_INT16_FLOAT): Likewise.
36137 (BT_FN_UINT32_FLOAT): Likewise.
36138 (BT_FN_UINT16_FLOAT): Likewise.
36139 (BT_FN_UINT8_FLOAT): Likewise.
36140 (BT_FN_UINT64_FLOAT): Likewise.
36141 (BT_FN_UINT16_UINT32): Likewise.
36142 (BT_FN_UINT32_UINT16): Likewise.
36143 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36144 (BT_FN_INT_PTR_INT): Likewise.
36145 (BT_FN_UINT_PTR_UINT): Likewise.
36146 (BT_FN_LONG_PTR_LONG): Likewise.
36147 (BT_FN_ULONG_PTR_ULONG): Likewise.
36148 (BT_FN_VOID_UINT64_UINT64): Likewise.
36149 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36150 (BT_FN_INT8_INT8_INT8): Likewise.
36151 (BT_FN_INT16_INT16_INT16): Likewise.
36152 (BT_FN_INT_INT_INT): Likewise.
36153 (BT_FN_UINT_FLOAT_UINT): Likewise.
36154 (BT_FN_FLOAT_UINT_UINT): Likewise.
36155 (BT_FN_ULONG_UINT_UINT): Likewise.
36156 (BT_FN_ULONG_UINT_PTR): Likewise.
36157 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36158 (BT_FN_UINT_UINT_UINT): Likewise.
36159 (BT_FN_VOID_UINT_PTR): Likewise.
36160 (BT_FN_UINT_UINT_PTR: Likewise.
36161 (BT_FN_UINT32_UINT64_PTR): Likewise.
36162 (BT_FN_INT_INT_UINT_UINT): Likewise.
36163 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36164 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36165 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36166 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36167 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36168 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36169 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36170 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36171 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36172 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36173 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36174 * doc/frontends.texi: List BRIG FE.
36175 * doc/install.texi (Testing): Add BRIG tesring requirements.
36176 * doc/invoke.texi (Overall Options): Mention BRIG.
36177 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36179 2017-01-24 Richard Biener <rguenther@suse.de>
36181 PR translation/79208
36182 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36184 2017-01-24 Martin Jambor <mjambor@suse.cz>
36187 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36188 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36189 and known_contexts.
36191 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36193 PR middle-end/79123
36194 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36195 casts from signed to unsigned really don't have a range.
36197 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36199 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36200 GMP_RNDx for compatiblity.
36202 2017-01-24 Martin Liska <mliska@suse.cz>
36205 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36206 that would prevent us to call alloca with -1 as argument.
36208 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36210 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36211 Avoid -Wformat-security warning.
36213 2017-01-23 Andrew Pinski <apinski@cavium.com>
36215 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36218 2017-01-23 Martin Sebor <msebor@redhat.com>
36220 PR middle-end/78703
36221 * gimple-ssa-sprintf.c (warn_level): New global.
36222 (format_integer): Use it here and throughout the rest of the file.
36223 Use the same switch to compute sign as base.
36224 (maybe_warn): New function.
36225 (format_directive): Factor out warnings into maybe_warn.
36226 Add debugging output. Use warn_level.
36227 (add_bytes): Use warn_level.
36228 (pass_sprintf_length::compute_format_length): Add debugging output.
36229 (try_substitute_return_value): Same.
36230 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36232 PR middle-end/78703
36233 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36234 (struct fmtresult, format_integer, format_floating): Adjust.
36235 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36236 (get_string_length, format_string,format_directive): Same.
36237 (pass_sprintf_length::compute_format_length): Same.
36238 (try_substitute_return_value): Simplify slightly.
36240 PR middle-end/78703
36241 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36242 (fmtresult::operator+=): Outlined.
36243 (struct fmtresult): Add ctors.
36244 (struct conversion_spec): Rename...
36245 (struct directive): ...to this. Add and remove data members.
36246 (directive::set_width, directive::set_precision): New functions.
36247 (format_percent): Use fmtresult ctor.
36248 (get_width_and_precision): Remove.
36249 (format_integer): Make naming changes. Avoid computing width and
36251 (format_floating): Same. Adjust indentation.
36252 (format_character, format_none): New functions.
36253 (format_string): Moved character handling to format_character.
36254 (format_directive): Remove arguments, change return type.
36255 (parse_directive): New function.
36256 (pass_sprintf_length::compute_format_length): Move directive
36257 parsing to parse_directive.
36259 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36261 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36262 (assign_assembler_name_if_needed): ... this.
36263 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36264 (assign_assembler_name_if_needed): ... this.
36265 (free_lang_data_in_cgraph): Adjust callers.
36266 * cgraphunit.c (cgraph_node::analyze): Likewise.
36267 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36270 2017-01-23 Richard Biener <rguenther@suse.de>
36272 PR tree-optimization/79088
36273 PR tree-optimization/79188
36274 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36275 resetting loop bounds after last path deletion. Reset loop
36276 bounds of the target loop, make code match the comments.
36277 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36278 Make sure loops need no fixups.
36280 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36282 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36283 exponent support with double type for first argument.
36284 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36285 type returned by __builtin_vec_extract_sig,
36286 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36287 functions from "vector int" to "vector unsigned int" or from
36288 "vector long long int" to "vector unsigned long long int".
36289 Changed type returned by __builtin_vec_extract_exp,
36290 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36291 functions from "vector int" to "vector unsigned int" or from
36292 "vector long long int" to "vector unsigned long long int".
36293 Changed return type of __builtin_vec_test_data_class,
36294 __builtin_vec_test_data_class_sp, and
36295 __builtin_vec_test_data_class_dp from "vector int" to
36296 "vector bool int" or from "vector long long int" to "vector bool
36297 long long int" and changed second argument type from "unsigned
36298 int" to "int". Added new overloaded function forms "vector float
36299 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36300 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36301 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36302 double, vector unsigned long long int)" and "vector double
36303 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36304 long int)". Changed return type of
36305 __builtin_scalar_test_data_class and
36306 __builtin_scalar_test_data_class_sp and
36307 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36308 int" and changed second argument from "unsigned int" to "int".
36309 Changed type returned by __builtin_scalar_test_neg,
36310 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36311 from "int" to "bool int". Added new overloaded function form
36312 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36313 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36314 exponent double-precision with floating point first argument.
36315 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36316 documentation of scalar_test_data_class, scalar_test_neg,
36317 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36318 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36319 vec_test_data_class built-in functions to reflect refinements in
36320 their type signatures.
36322 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36324 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36326 (aarch64_elf_asm_destructor): Likewise.
36328 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36330 PR rtl-optimization/78634
36331 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36332 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36333 * ifcvt.c (noce_try_cmove): Add missing cost check.
36335 PR rtl-optimization/71724
36336 * combine.c (if_then_else_cond): Look for situations where it is
36337 beneficial to undo the work of one of the recursive calls.
36339 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36341 PR tree-optimization/70754
36342 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36343 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36344 combined stmt before it if not NULL.
36345 (combine_chains): Process refs reversely and compute dominance point
36348 2017-01-23 Martin Liska <mliska@suse.cz>
36350 PR tree-optimization/79196
36351 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36352 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36354 (strlen_optimize_stmt): Call the renamed function.
36356 2017-01-23 Michael Matz <matz@suse.de>
36358 PR tree-optimization/78384
36359 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36361 2017-01-23 Richard Biener <rguenther@suse.de>
36363 PR tree-optimization/79186
36364 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36365 both incoming edges before moving an assert.
36367 2017-01-23 Martin Jambor <mjambor@suse.cz>
36369 * ipa-prop.c (load_from_param_1): Removed.
36370 (load_from_unmodified_param): Bits from load_from_param_1 put back
36372 (load_from_param): Removed.
36373 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36374 with stmt. Reverted back to use of load_from_unmodified_param.
36376 2017-01-23 Martin Jambor <mjambor@suse.cz>
36379 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36380 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36381 field a pointer to garbage collected vector, mark lattices and
36382 ipcp_orig_node with GTY((skip)).
36383 (ipa_get_param_count): Adjust to descriptors being a pointer.
36384 (ipa_get_param): Likewise.
36385 (ipa_get_type): Likewise.
36386 (ipa_get_param_move_cost): Likewise.
36387 (ipa_set_param_used): Likewise.
36388 (ipa_get_controlled_uses): Likewise.
36389 (ipa_set_controlled_uses): Likewise.
36390 (ipa_is_param_used): Likewise.
36391 (ipa_node_params_t): Move into garbage collector. New methods insert
36393 (ipa_node_params_sum): Annotate wth GTY(()).
36394 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36396 (ipa_load_from_parm_agg): Adjust declaration.
36397 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36398 * ipa-profile.c (ipa_profile): Likewise.
36399 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36400 (ipa_populate_param_decls): Make descriptors parameter garbage
36402 (ipa_dump_param): Adjust to descriptors being a pointer.
36403 (ipa_alloc_node_params): Likewise.
36404 (ipa_initialize_node_params): Likewise.
36405 (load_from_param_1): Make descriptors parameter garbage collected.
36406 (load_from_unmodified_param): Likewise.
36407 (load_from_param): Likewise.
36408 (ipa_load_from_parm_agg): Likewise.
36409 (ipa_node_params::~ipa_node_params): Removed.
36410 (ipa_free_all_node_params): Remove call to delete operator.
36411 (ipa_node_params_t::insert): New.
36412 (ipa_node_params_t::remove): Likewise.
36413 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36414 copy known_csts and known_contexts vectors.
36415 (ipa_read_node_info): Adjust to descriptors being a pointer.
36416 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36418 (ipcp_transform_function): Make descriptors variable garbage
36421 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36423 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36424 * config/i386/avx512dqintrin.h: Ditto.
36425 * config/i386/avx512fintrin.h: Ditto.
36426 * config/i386/i386.c: Handle new builtins.
36427 * config/i386/i386-builtin.def: Add new builtins.
36428 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36429 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36431 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36432 Martin Liska <mliska@suse.cz>
36434 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36435 * asan.c (asan_expand_poison_ifn): Support stores and use
36436 appropriate ASAN report function.
36437 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36438 * internal-fn.def (ASAN_POISON_USE): Declare.
36439 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36440 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36441 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36442 ASAN_POISON calls w/o LHS.
36443 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36444 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36445 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36446 * gimplify.c (asan_poison_variables): Add attribute
36447 use_after_scope_memory to variables that really needs to live
36449 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36450 having the attribute.
36452 2017-01-23 Martin Liska <mliska@suse.cz>
36454 * asan.c (create_asan_shadow_var): New function.
36455 (asan_expand_poison_ifn): Likewise.
36456 * asan.h (asan_expand_poison_ifn): New declaration.
36457 * internal-fn.c (expand_ASAN_POISON): Likewise.
36458 * internal-fn.def (ASAN_POISON): New builtin.
36459 * sanopt.c (pass_sanopt::execute): Expand
36460 asan_expand_poison_ifn.
36461 * tree-inline.c (copy_decl_for_dup_finish): Make function
36463 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36464 * tree-ssa.c (is_asan_mark_p): New function.
36465 (execute_update_addresses_taken): Rewrite local variables
36466 (identified just by use-after-scope as addressable) into SSA.
36468 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36470 * doc/install.texi (Specific): opensource.apple.com uses https
36471 now. Remove trailing slash.
36473 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36475 * README.Portability: Remove note on an Irix compatibility issue.
36477 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36479 * gcov.c (INCLUDE_ALGORITHM): Define.
36480 (INCLUDE_VECTOR): Define.
36481 No longer include <vector> and <algorithm> directly.
36483 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36485 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36487 * doc/invoke.texi (Code Gen Options): Ditto.
36489 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36492 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36494 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36496 rtl-optimization/79125
36497 * cprop.c (local_cprop_pass): Handle cases where we make an
36498 unconditional trap.
36500 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36504 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36505 read from, for big endian.
36507 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36509 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36510 register pauth builtins for LP64 only.
36512 2017-01-20 Marek Polacek <polacek@redhat.com>
36515 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36518 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36520 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36522 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36523 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36524 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36526 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36529 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36530 in big-endian mode when they are not a single duplicated value.
36532 2017-01-20 Richard Biener <rguenther@suse.de>
36534 * BASE-VER: Bump to 7.0.1.
36536 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36538 * omp-low.c (omplow_simd_context): New struct. Use it...
36539 (lower_rec_simd_input_clauses): ...here and...
36540 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36541 references to idx, lane, max_vf, is_simt.
36543 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36545 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36548 2017-01-20 Martin Jambor <mjambor@suse.cz>
36550 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36551 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36553 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36554 (GTFILES): Rename hsa.c to hsa-common.c.
36555 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36556 * hsa-dump.c: Likewise.
36557 * hsa-gen.c: Likewise.
36558 * hsa-regalloc.c: Likewise.
36559 * ipa-hsa.c: Likewise.
36560 * omp-expand.c: Likewise.
36561 * omp-low.c: Likewise.
36562 * toplev.c: Likewise.
36564 2017-01-20 Marek Polacek <polacek@redhat.com>
36567 * doc/invoke.texi: Document -Wduplicated-branches.
36568 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36569 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36570 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36571 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36572 return 0 only when not OEP_LEXICOGRAPHIC.
36573 (fold_build_cleanup_point_expr): Use the expression
36574 location when building CLEANUP_POINT_EXPR.
36575 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36576 * tree.c (add_expr): Handle error_mark_node.
36578 2017-01-20 Martin Liska <mliska@suse.cz>
36581 * tree-profile.c (init_ic_make_global_vars): Do not call
36583 (gimple_init_gcov_profiler): Likewise.
36585 2017-01-20 Martin Liska <mliska@suse.cz>
36588 * cgraph.h (maybe_create_reference): Remove argument and
36590 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36592 * ipa-cp.c (create_specialized_node): Likewise.
36593 * symtab.c (symtab_node::maybe_create_reference): Handle
36594 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36596 2017-01-20 Martin Liska <mliska@suse.cz>
36598 * read-rtl-function.c (function_reader::create_function): Use
36599 build_decl instread of build_decl_stat.
36601 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36603 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36604 * config/i386/avx512dqintrin.h: Ditto.
36605 * config/i386/avx512fintrin.h: Ditto.
36606 * config/i386/i386-builtin-types.def: Add new types.
36607 * config/i386/i386.c: Handle new types.
36608 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36609 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36610 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36611 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36612 (__builtin_ia32_kshiftridi): New.
36613 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36615 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36619 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36620 define to rs6000_init_stack_protect_guard.
36621 (rs6000_init_stack_protect_guard): New function.
36623 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36624 Yunqiang Su <yunqiang.su@imgtec.com>
36626 * config.gcc (supported_defaults): Add madd4.
36627 (with_madd4): Add validation.
36628 (all_defaults): Add madd4.
36629 * config/mips/mips.opt (mmadd4): New option.
36630 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36632 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36634 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36635 (ISA_HAS_FUSED_MADD4): Likewise.
36636 * doc/invoke.texi (-mmadd4): Document the new option.
36637 * doc/install.texi (--with-madd4): Document the new option.
36639 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36641 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36642 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36643 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36644 (aarch64_init_pauth_hint_builtins): New.
36645 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36646 (aarch64_expand_builtin): Expand new builtins.
36648 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36650 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36651 * combine-stack-adj.c (no_unhandled_cfa): Handle
36652 REG_CFA_TOGGLE_RA_MANGLE.
36653 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36654 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36655 info for return address signing.
36656 (aarch64_expand_epilogue): Likewise.
36658 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36660 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36661 * config/aarch64/aarch64-protos.h
36662 (aarch64_return_address_signing_enabled): New declaration.
36663 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36665 (aarch64_expand_prologue): Sign return address before it's pushed onto
36667 (aarch64_expand_epilogue): Authenticate return address fetched from
36669 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36670 (aarch64_attributes): New function attributes for "sign-return-address".
36671 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36672 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36673 ("*do_return"): Generate combined instructions according to key index.
36674 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36675 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36677 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36678 * config/aarch64/aarch64.opt (msign-return-address=): New.
36679 * doc/extend.texi (AArch64 Function Attributes): Documents
36680 "sign-return-address=".
36681 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36683 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36685 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36686 overall option summary.
36688 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36690 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36691 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36692 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36693 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36695 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36697 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36698 -mpower9-minmax by default for -mcpu=power9.
36699 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36700 128-bit floating point.
36702 2017-01-20 Alan Modra <amodra@gmail.com>
36704 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36705 optimizing for size.
36707 2017-01-20 Alan Modra <amodra@gmail.com>
36710 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36711 for strcmp and strncmp from corresponding builtin decl.
36713 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36715 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36716 instead of i386/rtems-64.h.
36717 * config/i386/rtems-64.h: Remove.
36719 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36723 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36725 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36727 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36729 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36730 Change int to HOST_WIDE_INT.
36731 * config/aarch64/aarch64-protos.h
36732 (aarch64_simd_gen_const_vector_dup): Likewise.
36733 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36735 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36737 * langhooks-def.h (lhd_type_for_size): New decl.
36738 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36739 * langhooks.c (lhd_type_for_size): New function, taken from
36742 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36744 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36745 define_bypass for CR latency.
36746 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36747 (power9-alu2): Add define_bypass for CR latency.
36749 (power9-mul): Update insn latency.
36750 (power9-mul-compare): Update insn latency, bypass latency and remove
36753 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36755 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36757 * config/aarch64/aarch64.md
36758 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36759 aarch64_nopcrelative_literal_loads.
36760 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36762 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36764 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36765 TARGET_LOONGSON_3A.
36766 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36768 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36771 * config.gcc (supported_defaults): Add lxc1-sxc1.
36772 (with_lxc1_sxc1): Add validation.
36773 (all_defaults): Add lxc1-sxc1.
36774 * config/mips/mips.opt (mlxc1-sxc1): New option.
36775 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36777 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36778 __mips_no_lxc1_sxc1.
36779 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36780 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36781 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36783 2017-01-19 Richard Biener <rguenther@suse.de>
36785 PR tree-optimization/72488
36786 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36787 sure to restore SSA info.
36788 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36790 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36792 PR rtl-optimization/79121
36793 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36794 of the inner type when shifting an extended value.
36796 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36799 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36800 interposable aliases.
36802 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36805 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36806 Use the evmergelohi instruction.
36807 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36808 (mov_sitf_e500_subreg8_2_be): Likewise.
36809 (mov_sitf_e500_subreg12_2_le): Likewise.
36810 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36811 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36812 (mov_sitf_e500_subreg8_2_le): Likewise.
36813 (mov_sitf_e500_subreg12_2_be): Likewise.
36815 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36817 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36818 attribute from vecsimple to vecperm.
36819 (altivec_vbpermq2): Likewise.
36821 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36824 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36826 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36827 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36828 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36829 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36830 case where N arg is SIZE_MAX.
36831 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36832 (cmpstrsi): Add pattern.
36834 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36836 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36837 __builtin_vec_revb builtins.
36838 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36839 built-in functions to support generation of the ISA 3.0 XXBR<x>
36840 vector byte reverse instructions.
36841 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36842 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36843 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36844 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36845 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36846 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36847 (P9V_BUILTIN_VEC_REVB): Likewise.
36848 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36849 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36850 (p9_xxbrq_v16qi): Likewise.
36851 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36852 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36853 (p9_xxbrh_v8hi): Likewise.
36854 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36855 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36856 vec_revb built-in functions.
36858 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36860 PR rtl-optimization/78952
36861 * config/i386/i386.md (any_extract): New code iterator.
36862 (*insvqi_2): Use any_extract for source operand.
36863 (*insvqi_3): Use any_shiftrt for source operand.
36865 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36867 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36869 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36871 2017-01-18 Matthias Klose <doko@ubuntu.com>
36873 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36875 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36877 * config/rs6000/altivec.h (vec_bperm): Change #define.
36878 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36879 (altivec_vbpermq2): New define_insn.
36880 (altivec_vbpermd): Likewise.
36881 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36882 function interface.
36883 (VBPERMD): Likewise.
36884 (VBPERM): New polymorphic function interface.
36885 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36886 Add entries for P9V_BUILTIN_VEC_VBPERM.
36887 * doc/extend.texi: Add interfaces for vec_bperm.
36889 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36891 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36892 first letter of error messages.
36893 (s390_resolve_overloaded_builtin): Likewise.
36894 * config/s390/s390.c (s390_expand_builtin): Likewise.
36895 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36896 (s390_valid_target_attribute_inner_p): Likewise.
36897 * config/s390/s390.md ("tabort"): Likewise.
36899 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36901 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36902 (ISA_AVOID_DIV_HILO): New macro.
36903 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36904 (ISA_HAS_DDIV): Likewise.
36906 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36908 * doc/invoke.texi (fabi-version): Correct number of occurrences.
36910 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36912 * doc/invoke.texi (fabi-version): Spelling fix.
36914 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36917 * doc/invoke.texi (fabi-version): Mention mangling fix for
36920 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36923 * doc/invoke.texi (fabi-version): Document discriminator mangling.
36925 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
36928 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
36929 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
36931 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
36933 (stack_protect_test): Ditto.
36934 * config/rs6000/rs6000.opt (mstack-protector-guard=,
36935 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
36937 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
36938 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
36939 -mstack-protector-guard-offset=.
36940 (RS/6000 and PowerPC Options): Ditto.
36942 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36944 * config/i386/i386.h (MASK_CLASS_P): New define.
36945 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
36946 there are no registers from different register sets also when
36947 mask registers are used. Update function comment.
36948 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
36949 to (*k/*r) and (*k/*km) alternatives.
36951 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
36953 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
36954 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
36955 (EH_RETURN_HANDLER_RTX): New define.
36956 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
36957 Force frame pointer in EH return functions.
36958 (aarch64_expand_epilogue): Add barrier for eh_return.
36959 (aarch64_final_eh_return_addr): Remove.
36960 (aarch64_eh_return_handler_rtx): New function.
36961 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
36963 (aarch64_eh_return_handler_rtx): New prototype.
36965 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36967 * config/rs6000/altivec.h (vec_rlmi): New #define.
36968 (vec_vrlnm): Likewise.
36969 (vec_rlnm): Likewise.
36970 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
36971 (UNSPEC_VRLNM): Likewise.
36972 (VIlong): New mode iterator.
36973 (altivec_vrl<VI_char>mi): New define_insn.
36974 (altivec_vrl<VI_char>nm): Likewise.
36975 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
36977 (VRLDNM): Likewise.
36978 (RLNM): New polymorphic function entry.
36979 (VRLWMI): New monomorphic function entry.
36980 (VRLDMI): Likewise.
36981 (RLMI): New polymorphic function entry.
36982 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
36983 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
36984 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
36987 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36990 * dwarf2out.c (field_byte_offset): Restore the
36991 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
36992 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
36993 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
36996 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36999 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37001 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37004 * dwarf2out.c (add_data_member_location_attribute): For constant
37005 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37006 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37007 DW_AT_byte_size attributes.
37009 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37011 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37012 after forcing to constant memory when the code model is medium.
37014 2017-01-17 Julia Koval <julia.koval@intel.com>
37017 * config/i386/avx512fintrin.h
37018 (_mm512_i32gather_ps): Change __addr type to void const*.
37019 (_mm512_mask_i32gather_ps): Ditto.
37020 (_mm512_i32gather_pd): Ditto.
37021 (_mm512_mask_i32gather_pd): Ditto.
37022 (_mm512_i64gather_ps): Ditto.
37023 (_mm512_mask_i64gather_ps): Ditto.
37024 (_mm512_i64gather_pd): Ditto.
37025 (_mm512_mask_i64gather_pd): Ditto.
37026 (_mm512_i32gather_epi32): Ditto.
37027 (_mm512_mask_i32gather_epi32): Ditto.
37028 (_mm512_i32gather_epi64): Ditto.
37029 (_mm512_mask_i32gather_epi64): Ditto.
37030 (_mm512_i64gather_epi32): Ditto.
37031 (_mm512_mask_i64gather_epi32): Ditto.
37032 (_mm512_i64gather_epi64): Ditto.
37033 (_mm512_mask_i64gather_epi64): Ditto.
37034 (_mm512_i32scatter_ps): Change __addr type to void*.
37035 (_mm512_mask_i32scatter_ps): Ditto.
37036 (_mm512_i32scatter_pd): Ditto.
37037 (_mm512_mask_i32scatter_pd): Ditto.
37038 (_mm512_i64scatter_ps): Ditto.
37039 (_mm512_mask_i64scatter_ps): Ditto.
37040 (_mm512_i64scatter_pd): Ditto.
37041 (_mm512_mask_i64scatter_pd): Ditto.
37042 (_mm512_i32scatter_epi32): Ditto.
37043 (_mm512_mask_i32scatter_epi32): Ditto.
37044 (_mm512_i32scatter_epi64): Ditto.
37045 (_mm512_mask_i32scatter_epi64): Ditto.
37046 (_mm512_i64scatter_epi32): Ditto.
37047 (_mm512_mask_i64scatter_epi32): Ditto.
37048 (_mm512_i64scatter_epi64): Ditto.
37049 (_mm512_mask_i64scatter_epi64): Ditto.
37050 * config/i386/avx512pfintrin.h
37051 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37052 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37053 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37054 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37055 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37056 (_mm512_prefetch_i32scatter_ps): Ditto.
37057 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37058 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37059 (_mm512_prefetch_i64scatter_pd): Ditto.
37060 (_mm512_prefetch_i64scatter_ps): Ditto.
37061 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37062 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37063 * config/i386/avx512vlintrin.h
37064 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37065 (_mm_mmask_i32gather_ps): Ditto.
37066 (_mm256_mmask_i32gather_pd): Ditto.
37067 (_mm_mmask_i32gather_pd): Ditto.
37068 (_mm256_mmask_i64gather_ps): Ditto.
37069 (_mm_mmask_i64gather_ps): Ditto.
37070 (_mm256_mmask_i64gather_pd): Ditto.
37071 (_mm_mmask_i64gather_pd): Ditto.
37072 (_mm256_mmask_i32gather_epi32): Ditto.
37073 (_mm_mmask_i32gather_epi32): Ditto.
37074 (_mm256_mmask_i32gather_epi64): Ditto.
37075 (_mm_mmask_i32gather_epi64): Ditto.
37076 (_mm256_mmask_i64gather_epi32): Ditto.
37077 (_mm_mmask_i64gather_epi32): Ditto.
37078 (_mm256_mmask_i64gather_epi64): Ditto.
37079 (_mm_mmask_i64gather_epi64): Ditto.
37080 (_mm256_i32scatter_ps): Change __addr type to void*.
37081 (_mm256_mask_i32scatter_ps): Ditto.
37082 (_mm_i32scatter_ps): Ditto.
37083 (_mm_mask_i32scatter_ps): Ditto.
37084 (_mm256_i32scatter_pd): Ditto.
37085 (_mm256_mask_i32scatter_pd): Ditto.
37086 (_mm_i32scatter_pd): Ditto.
37087 (_mm_mask_i32scatter_pd): Ditto.
37088 (_mm256_i64scatter_ps): Ditto.
37089 (_mm256_mask_i64scatter_ps): Ditto.
37090 (_mm_i64scatter_ps): Ditto.
37091 (_mm_mask_i64scatter_ps): Ditto.
37092 (_mm256_i64scatter_pd): Ditto.
37093 (_mm256_mask_i64scatter_pd): Ditto.
37094 (_mm_i64scatter_pd): Ditto.
37095 (_mm_mask_i64scatter_pd): Ditto.
37096 (_mm256_i32scatter_epi32): Ditto.
37097 (_mm256_mask_i32scatter_epi32): Ditto.
37098 (_mm_i32scatter_epi32): Ditto.
37099 (_mm_mask_i32scatter_epi32): Ditto.
37100 (_mm256_i32scatter_epi64): Ditto.
37101 (_mm256_mask_i32scatter_epi64): Ditto.
37102 (_mm_i32scatter_epi64): Ditto.
37103 (_mm_mask_i32scatter_epi64): Ditto.
37104 (_mm256_i64scatter_epi32): Ditto.
37105 (_mm256_mask_i64scatter_epi32): Ditto.
37106 (_mm_i64scatter_epi32): Ditto.
37107 (_mm_mask_i64scatter_epi32): Ditto.
37108 (_mm256_i64scatter_epi64): Ditto.
37109 (_mm256_mask_i64scatter_epi64): Ditto.
37110 (_mm_i64scatter_epi64): Ditto.
37111 (_mm_mask_i64scatter_epi64): Ditto.
37112 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37113 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37114 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37115 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37116 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37117 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37118 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37119 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37120 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37121 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37122 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37123 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37124 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37125 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37126 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37127 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37128 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37129 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37130 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37131 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37132 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37133 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37134 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37135 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37136 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37137 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37138 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37139 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37140 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37141 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37142 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37143 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37144 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37145 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37146 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37147 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37148 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37149 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37150 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37151 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37152 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37153 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37154 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37155 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37156 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37157 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37158 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37159 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37160 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37161 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37162 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37163 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37164 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37165 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37166 definitions accordingly.
37168 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37169 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37172 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37175 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37178 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37179 hard regs for inner regno.
37181 2017-01-17 Martin Liska <mliska@suse.cz>
37184 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37185 assumption and add comment.
37187 2017-01-17 Nathan Sidwell <nathan@acm.org>
37189 * ipa-visibility.c (localize_node): New function, broken out of ...
37190 (function_and_variable_visibility): ... here. Call it.
37192 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37194 PR middle-end/77445
37195 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37196 correctly set frequency of oudgoing edge.
37197 (duplicate_thread_path): Fix profile updating.
37199 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37202 * configure.ac: Add GCC_BASE_VER.
37203 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37204 version from BASE-VER file.
37205 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37206 (gcc.o): Depend on $(BASEVER).
37207 * common.opt (dumpfullversion): New option.
37208 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37209 * doc/invoke.texi: Document -dumpfullversion.
37210 * doc/install.texi: Document --with-gcc-major-version-only.
37211 * configure: Regenerated.
37213 2017-01-17 Richard Biener <rguenther@suse.de>
37215 PR tree-optimization/71433
37216 * tree-vrp.c (register_new_assert_for): Merge same asserts
37217 on all incoming edges.
37218 (process_assert_insertions_for): Handle insertions at the
37221 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37223 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37224 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37226 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37229 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37232 2017-01-17 Alan Modra <amodra@gmail.com>
37235 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37236 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37237 symbolic stack limit when pic.
37239 2017-01-16 Martin Sebor <msebor@redhat.com>
37241 PR tree-optimization/78608
37242 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37244 2017-01-16 Jeff Law <law@redhat.com>
37247 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37248 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37249 for several include directories that may be relative to sysroot.
37250 * config/i386/x-mingw32 (gplus_includedir): Define.
37251 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37252 (native_system_includedir): Likewise.
37253 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37254 override if TARGET_SYSTEM_ROOT is defined.
37255 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37257 PR tree-optimization/79090
37258 PR tree-optimization/33562
37259 PR tree-optimization/61912
37260 PR tree-optimization/77485
37261 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37262 and computed trims into the dump file.
37264 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37266 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37268 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37271 * gimplify.c (gimplify_init_constructor): If want_value and
37272 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37276 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37277 sequence. Formatting fixes.
37278 (doloop_optimize): Formatting fixes.
37281 * gcc.c (debug_level_greater_than_spec_func): New function.
37282 (static_spec_functions): Add debug-level-gt spec function.
37283 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37285 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37286 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37287 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37288 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37289 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37290 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37292 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37294 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37295 QImode fixups to general and mask registers only.
37297 2017-01-16 Carl Love <cel@us.ibm.com>
37299 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37300 for built-in functions
37301 vector signed char vec_nabs (vector signed char)
37302 vector signed short vec_nabs (vector signed short)
37303 vector signed int vec_nabs (vector signed int)
37304 vector signed long long vec_nabs (vector signed long long)
37305 vector float vec_nabs (vector float)
37306 vector double vec_nabs (vector double)
37307 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37309 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37310 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37311 * doc/extend.texi: Update the documentation file for the new built-in
37314 2017-01-16 Martin Sebor <msebor@redhat.com>
37316 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37319 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37321 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37322 UNSPEC_VSX__XXSPLTD to require special splat handling.
37324 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37327 * system.h: Poison strndup.
37329 2017-01-16 Alan Modra <amodra@gmail.com>
37332 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37335 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37337 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37339 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37341 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37342 call recog here. Assert that INSN_CODE (insn) is non-negative.
37344 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37347 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37349 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37350 in the currently scheduled RTL fragment.
37352 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37354 PR rtl-optimization/78751
37355 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37358 2017-01-14 Jeff Law <law@redhat.com>
37360 PR tree-optimization/79090
37361 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37362 variable length stores.
37363 (compute_trims): Delete dead assignment to *trim_tail.
37364 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37367 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37369 PR rtl-optimization/78626
37370 PR rtl-optimization/78727
37371 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37372 of a block, and split such blocks after everything else is finished.
37374 2017-01-14 Alan Modra <amodra@gmail.com>
37377 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37378 target legitimate_combined_insn.
37379 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37380 (rs6000_legitimate_combined_insn): New function.
37381 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37383 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37384 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37385 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37387 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37389 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37391 2017-01-13 Jeff Law <law@redhat.com>
37393 PR tree-optimization/33562
37394 PR tree-optimization/61912
37395 PR tree-optimization/77485
37396 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37398 (delete_dead_assignment): Likewise.
37399 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37400 statement to delete_dead_call and delete_dead_assignment.
37402 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37405 * substring-locations.c (format_warning_va): Strengthen case 1 so
37406 that both endpoints of the substring must be within the format
37407 range for just the substring to be printed.
37409 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37411 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37412 * config/i386/i386.c (ix86_target_string): Add missing options
37413 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37414 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37415 flags_other and ix86_target_other to flags2_other. Display unknown
37417 (ix86_valid_target_attribute_inner_p): Add missing options and
37418 reorder options by implied ISAs, as in ix86_target_string.
37420 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37422 * hash-table.h (hash_table::too_empty_p): New function.
37423 (hash_table::expand): Use it.
37424 (hash_table::traverse): Likewise.
37425 (hash_table::empty_slot): Use sizeof (value_type) instead of
37426 sizeof (PTR) to convert bytes to elements. Shrink the table
37427 if the current size is excessive for the current number of
37430 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37432 * ira-costs.c (record_reg_classes): Break from the inner loop
37433 early once alt_fail is known to be true. Update outer loop
37434 handling accordingly.
37436 2017-01-13 Jeff Law <law@redhat.com>
37438 * tree-ssa-dse.c (decrement_count): New function.
37439 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37440 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37441 when we know the partially dead statement is a mem* function.
37443 PR tree-optimization/61912
37444 PR tree-optimization/77485
37445 * tree-ssa-dse.c: Include expr.h.
37446 (maybe_trim_constructor_store): New function.
37447 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37449 PR tree-optimization/33562
37450 PR tree-optimization/61912
37451 PR tree-optimization/77485
37452 * doc/invoke.texi: Document new dse-max-object-size param.
37453 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37454 * tree-ssa-dse.c: Include params.h.
37455 (dse_store_status): New enum.
37456 (initialize_ao_ref_for_dse): New, partially extracted from
37458 (valid_ao_ref_for_dse, normalize_ref): New.
37459 (setup_live_bytes_from_ref, compute_trims): Likewise.
37460 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37461 (maybe_trim_partially_dead_store): Likewise.
37462 (maybe_trim_complex_store): Likewise.
37463 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37464 Track what bytes live from the original store. Return tri-state
37465 for dead, partially dead or live.
37466 (dse_dom_walker): Add constructor, destructor and new private members.
37467 (delete_dead_call, delete_dead_assignment): New extracted from
37469 (dse_optimize_stmt): Make a member of dse_dom_walker.
37470 Use initialize_ao_ref_for_dse.
37472 PR tree-optimization/33562
37473 PR tree-optimization/61912
37474 PR tree-optimization/77485
37475 * sbitmap.h (bitmap_count_bits): Prototype.
37476 (bitmap_clear_range, bitmap_set_range): Likewise.
37477 * sbitmap.c (bitmap_clear_range): New function.
37478 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37480 2017-01-13 Martin Liska <mliska@suse.cz>
37483 * function.c (set_cfun): Add new argument force.
37484 * function.h (set_cfun): Likewise.
37485 * ipa-inline-transform.c (inline_call): Use the function when
37486 strict alising from is dropped for function we inline to.
37488 2017-01-13 Richard Biener <rguenther@suse.de>
37490 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37491 for dumping GIMPLE INTEGER_CSTs.
37493 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37495 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37496 to 201112L since C++17.
37498 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37501 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37502 if -fsanitize=kernel-address is present.
37504 2017-01-13 Richard Biener <rguenther@suse.de>
37506 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37507 as _Literal ( type ) number in case usual suffixes do not
37508 preserve all information.
37510 2017-01-13 Richard Biener <rguenther@suse.de>
37512 PR tree-optimization/77283
37513 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37514 and ssa-iterators.h.
37515 (is_feasible_trace): Implement a cost model based on joiner
37518 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37521 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37522 char or short to __float128/_Float128 directly.
37524 2017-01-12 Martin Sebor <msebor@redhat.com>
37526 to -Wformat-overflow.
37527 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37528 (min_bytes_remaining): Same.
37529 (get_string_length): Same.
37530 (format_string): Same.
37531 (format_directive): Same.
37533 (pass_sprintf_length::handle_gimple_call): Same.
37535 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37537 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37538 info.nowrite calls with no lhs that can't throw. Return bool
37539 whether gsi_remove has been called or not.
37540 (pass_sprintf_length::handle_gimple_call): Return bool whether
37541 try_substitute_return_value called gsi_remove. Formatting fix.
37542 (pass_sprintf_length::execute): Don't use gsi_remove if
37543 handle_gimple_call returned true.
37546 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37547 be removed due to side-effects, don't remove following barrier nor
37548 turn the successor edge into fallthru edge.
37550 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37553 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37554 element-reversing loads and stores as not swappable.
37556 2017-01-12 Nathan Sidwell <nathan@acm.org>
37557 Nicolai Stange <nicstange@gmail.com>
37559 * combine.c (try_combine): Don't ignore result of overlap checking
37560 loop. Combine overlap & asm check into single loop.
37562 2017-01-12 Richard Biener <rguenther@suse.de>
37564 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37565 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37567 2017-01-12 Richard Biener <rguenther@suse.de>
37569 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37570 and TS_TARGET_OPTION directly derive from TS_BASE.
37571 * tree-core.h (tree_optimization_option): Derive from tree_base.
37572 (tree_target_option): Likewise.
37574 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37576 * config/i386/i386.c (memory_address_length): Increase len
37577 only when rip_relative_addr_p returns false.
37579 2017-01-11 Julia Koval <julia.koval@intel.com>
37581 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37582 (OPTION_MASK_ISA_SGX_SET): New.
37583 (ix86_handle_option): Handle OPT_msgx.
37584 * config.gcc: Added sgxintrin.h.
37585 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37586 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37587 * config/i386/i386.c (ix86_target_string): Add -msgx.
37589 (ix86_option_override_internal): Handle new options.
37590 (ix86_valid_target_attribute_inner_p): Add sgx.
37591 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37592 * config/i386/i386.opt: Add msgx.
37593 * config/i386/sgxintrin.h: New file.
37594 * config/i386/x86intrin.h: Add sgxintrin.h.
37596 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37599 * fold-const.c (maybe_nonzero_address): Return 1 for function
37601 (tree_single_nonzero_warnv_p): Don't handle function local objects
37605 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37608 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37611 * opts.c: Include "spellcheck.h"
37612 (struct string_fragment): New struct.
37613 (struct edit_distance_traits<const string_fragment &>): New
37615 (get_closest_sanitizer_option): New function.
37616 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37618 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37620 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37622 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37623 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37624 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37625 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37626 for initial die_offset if dwarf_split_debug_info.
37627 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37628 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37630 (output_skeleton_debug_sections): Formatting fix. Use
37631 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37632 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37634 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37636 * config/arm/cortex-a53.md: Add bypasses for
37637 cortex_a53_r2f_cvt.
37638 (cortex_a53_r2f): Only use for transfers.
37639 (cortex_a53_f2r): Likewise.
37640 (cortex_a53_r2f_cvt): Add reservation for conversions.
37641 (cortex_a53_f2r_cvt): Likewise.
37643 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37645 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37646 to all inlined functions, change static to extern.
37648 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37651 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37653 (arm_assemble_integer): Likewise.
37655 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37657 * config.gcc: Use new awk script to check CPU, FPU and architecture
37658 parameters for --with-... options.
37659 * config/arm/parsecpu.awk: New file
37660 * config/arm/arm-cpus.in: New file.
37661 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37663 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37665 * config/arm/t-arm: Update dependency rules.
37666 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37667 of processing .def files.
37668 * config/arm/genopt.sh: Deleted.
37669 * config/arm/gentune.sh: Deleted.
37670 * config/arm/arm-cores.def: Deleted.
37671 * config/arm/arm-arches.def: Deleted.
37672 * config/arm/arm-fpus.def: Deleted.
37673 * config/arm/arm-tune.md: Regenerated.
37674 * config/arm/arm-tables.opt: Regenerated.
37675 * config/arm/arm-cpu.h: New generated file.
37676 * config/arm/arm-cpu-data.h: New generated file.
37677 * config/arm/arm-cpu-cdata.h: New generated file.
37679 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37682 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37684 (input_varpool_node): Unpack dynamically_initialized bit.
37686 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37688 PR rtl-optimization/79032
37689 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37690 the alignment of the adjusted memory reference against that of MODE,
37691 instead of the alignment of the original memory reference.
37693 2017-01-11 Martin Jambor <mjambor@suse.cz>
37695 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37697 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37698 decorated functions.
37700 2017-01-11 Richard Biener <rguenther@suse.de>
37702 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37703 set range/nonnull info for PHI results. Do not set it on
37704 stmts marked for removal.
37706 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37708 * expr.c (store_field): In the bitfield case, fetch the return value
37709 from the registers before applying a single big-endian adjustment.
37710 Always do a final load for a BLKmode value not larger than a word.
37712 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37715 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37716 that we correctly handle column numbers greater than
37717 LINE_MAP_MAX_COLUMN_NUMBER.
37719 2017-01-10 Martin Sebor <msebor@redhat.com>
37721 PR middle-end/78245
37722 * gimple-ssa-sprintf.c (get_destination_size): Call
37723 {init,fini}object_sizes.
37724 * tree-object-size.c (addr_object_size): Adjust.
37725 (pass_through_call): Adjust.
37726 (pass_object_sizes::execute): Adjust.
37727 * tree-object-size.h (fini_object_sizes): Declare.
37729 2017-01-10 Martin Sebor <msebor@redhat.com>
37731 PR tree-optimization/78775
37732 * builtins.c (get_size_range): Move...
37733 * calls.c: ...to here.
37734 (alloc_max_size): Accept zero argument.
37735 (operand_signed_p): Remove.
37736 (maybe_warn_alloc_args_overflow): Call get_size_range.
37737 * calls.h (get_size_range): Declare.
37739 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37741 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37742 from TI's devices.csv file as of September 2016.
37743 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37745 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37747 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37748 * doc/invoke.texi: Likewise.
37749 * doc/md.texi: Likewise.
37750 * doc/objc.texi: Likewise.
37752 2017-01-10 Joshua Conner <joshconner@google.com>
37754 * config/arm/fuchsia-elf.h: New file.
37755 * config/fuchsia.h: New file.
37756 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37757 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37759 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37761 2016-01-10 Richard Biener <rguenther@suse.de>
37763 PR tree-optimization/79034
37764 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37765 Propagate out degenerate PHIs in the joiner.
37767 2017-01-10 Martin Liska <mliska@suse.cz>
37769 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37770 (sort_congruence_classes_by_decl_uid): Likewise.
37771 (sort_congruence_class_groups_by_decl_uid): Likewise.
37772 (sem_item_optimizer::merge_classes): Sort class, groups in these
37773 classes and members in the groups by DECL_UID of declarations.
37774 This would make merge operations stable.
37776 2017-01-10 Martin Liska <mliska@suse.cz>
37778 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37779 usage of m_classes_vec.
37780 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37781 (sem_item_optimizer::get_group_by_hash): Likewise.
37782 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37783 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37784 (sem_item_optimizer::verify_classes): Likewise.
37785 (sem_item_optimizer::process_cong_reduction): Likewise.
37786 (sem_item_optimizer::dump_cong_classes): Likewise.
37787 (sem_item_optimizer::merge_classes): Likewise.
37788 * ipa-icf.h (congruence_class_hash): Rename from
37789 congruence_class_group_hash. Remove declaration of m_classes_vec.
37791 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37793 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37794 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37795 * config.gcc: Add avx512vpopcntdqintrin.h.
37796 * config/i386/avx512vpopcntdqintrin.h: New.
37797 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37798 * config/i386/i386-builtin-types.def: Add new types.
37799 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37800 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37801 __builtin_ia32_vpopcountq_v8di_mask): New.
37802 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37803 __AVX512VPOPCNTDQ__.
37804 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37805 (PTA_AVX512VPOPCNTDQ): Define.
37806 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37807 TARGET_AVX512VPOPCNTDQ_P): Define.
37808 * config/i386/i386.opt: Add mavx512vpopcntdq.
37809 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37810 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37812 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37814 PR middle-end/77484
37815 * predict.def (PRED_CALL): Set to 67.
37817 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37819 * expr.c (store_field): In the bitfield case, if the value comes from
37820 a function call and is of an aggregate type returned in registers, do
37821 not modify the field mode; extract the value in all cases if the mode
37822 is BLKmode and the size is not larger than a word.
37824 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37827 * config/i386/cpuid.h: Fix undefined behavior.
37829 2017-01-04 Jeff Law <law@redhat.com>
37831 PR tree-optimization/79007
37832 PR tree-optimization/67955
37833 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37834 conservative for pt.null when flag_non_call_exceptions is on.
37836 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37838 PR translation/79019
37839 PR translation/79020
37840 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37841 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37842 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37844 * config/avr/avr.opt (maccumulate-args): Likewise.
37845 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37846 * common.opt (freport-bug): Likewise.
37847 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37848 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37849 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37850 translatable string.
37851 * config/i386/i386.c (function_value_32): Likewise.
37852 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37853 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37855 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37856 * common/config/msp430/msp430-common.c (msp430_handle_option):
37858 * symtab.c (symtab_node::verify_base): Likewise.
37859 * opts.c (set_debug_level): Likewise.
37860 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37861 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37862 missing whitespace to translatable strings.
37863 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37864 * config/sh/superh.h: Likewise.
37865 * config/i386/xopintrin.h: Likewise.
37866 * config/i386/znver1.md: Likewise.
37867 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37868 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37869 * double-int.h (struct double_int): Likewise.
37870 * double-int.c (div_and_round_double): Likewise.
37871 * wide-int.cc: Likewise.
37872 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37873 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37874 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37875 (crossjumps_occurred): ... this.
37876 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37879 PR tree-optimization/78899
37880 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37881 returning bool return struct loop *, NULL for failure and the new
37883 (versionable_outer_loop_p): Don't version outer loop if it has
37884 dont_vectorized bit set.
37885 (tree_if_conversion): When versioning outer loop, ensure
37886 tree_if_conversion is performed also on the inner loop of the
37887 non-vectorizable outer loop copy.
37888 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37889 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37890 prevent vectorization of it.
37891 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37892 the outer loop vectorization of the non-scalar version is attempted
37893 before vectorization of the inner loop in scalar version. If
37894 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37895 vectorization of its inner loop.
37896 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37897 has 2 inner loops, rename also on edges from bb whose single pred
37898 is outer_loop->header. Fix typo in function comment.
37900 2017-01-09 Martin Sebor <msebor@redhat.com>
37903 * asan.c (asan_emit_stack_protection): Increase local buffer size
37904 to avoid snprintf truncation warning.
37906 2017-01-09 Andrew Pinski <apinski@cavium.com>
37908 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
37909 to reference thunderx2t99 for the tuning structure
37910 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
37912 (thunderx2t99_extra_costs): This.
37913 * config/aarch64/aarch64-tune.md: Regenerate.
37914 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
37915 (vulcan_addrcost_table): This.
37916 (vulcan_regmove_cost): Rename to ...
37917 (thunderx2t99_regmove_cost): This.
37918 (vulcan_vector_cost): Rename to ...
37919 (thunderx2t99_vector_cost): this.
37920 (vulcan_branch_cost): Rename to ...
37921 (thunderx2t99_branch_cost): This.
37922 (vulcan_tunings): Rename to ...
37923 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
37924 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
37926 2017-01-09 Martin Jambor <mjambor@suse.cz>
37930 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
37931 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
37932 (propagate_vr_accross_jump_function): Use the above function for all
37933 value range computations for pass-through jump functions and type
37934 converasion from explicit value range values.
37935 (ipcp_propagate_stage): Do not attempt to deduce types of formal
37936 parameters from TYPE_ARG_TYPES.
37937 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
37938 (ipa_write_node_info): Stream type of the actual argument.
37939 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
37941 2017-01-09 Martin Liska <mliska@suse.cz>
37944 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
37945 (lookup_compiler): Do not show error message with have_E.
37947 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37949 PR tree-optimization/78938
37950 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
37951 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
37952 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
37953 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
37956 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37958 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
37961 2017-01-09 Richard Biener <rguenther@suse.de>
37963 PR tree-optimization/78997
37964 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
37965 name condition properly.
37967 2017-01-09 Richard Biener <rguenther@suse.de>
37970 * dwarf2out.c (is_cxx): New overload with context.
37971 (is_naming_typedef_decl): Use it.
37973 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37975 * invoke.texi (Option Summary): Correct spacing in option lists
37976 and add line breaks to fix over-long lines.
37978 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37980 PR middle-end/17660
37982 * extend.texi (Common Variable Attributes): Add xref to GCC
37983 Internals manual to explain mode attribute keywords.
37985 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37988 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
37989 and Preprocessor Options.
37990 (Options for Linking): Document -pthread here....
37991 (RS/6000 and PowerPC Options): ...not here.
37992 (Solaris 2 Options): ...or here.
37993 * doc/cppopts.texi: Document -pthread.
37995 2017-01-08 Martin Sebor <msebor@redhat.com>
37997 PR middle-end/77708
37998 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
37999 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38000 New member functions.
38001 (format_directive): Used them.
38003 (pass_sprintf_length::handle_gimple_call): Same.
38004 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38005 to avoid truncation for any argument.
38006 (extract_affine_mul): Same.
38007 * tree.c (get_file_function_name): Same.
38009 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38011 PR middle-end/77484
38012 * predict.def (PRED_INDIR_CALL): Set to 86.
38014 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38016 PR preprocessor/54124
38017 * doc/cppopts.texi: Reformat -d subtable to list the full name
38018 of the options. Add cross-reference to the docs for the general
38019 compiler -d options.
38020 * doc/invoke.texi (Developer Options): Add cross-reference to the
38021 preprocessor-specific -d option documentation.
38023 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38025 PR preprocessor/13498
38026 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38027 redudant material, and reflect new command-line options.
38028 (System Headers): Likewise.
38030 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38032 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38033 -isystem, and -idirafter. Copy-edit.
38034 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38035 default for -ftrack-macro-expansion. Delete obsolete and
38036 badly-formatted implementation details about -fdebug-cpp output.
38037 * doc/cppwarnopts.texi: Copy-edit.
38039 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38042 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38043 that the transition from a max line width >= 1<<10 to narrower
38044 lines works correctly.
38046 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38048 * doc/options.texi (PerFunction): New.
38049 * opt-functions.awk (switch_flags): Map both Optimization and
38050 PerFunction to CL_OPTIMIZATION.
38051 * opth-gen.awk: Test for PerFunction flag along with
38053 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38054 it only when the latter is present. Skip those that don't in
38055 the hash function generator.
38056 * common.opt (fvar-tracking): Mark as PerFunction instead of
38058 (fvar-tracking-assignments): Likewise.
38059 (fvar-tracking-assignments-toggle): Likewise.
38060 (fvar-tracking-uninit): Likewise.
38062 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38064 PR translation/79018
38065 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38068 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38071 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38072 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38073 TARGET_LONG_JUMP_TABLE_OFFSETS.
38074 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38075 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38076 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38077 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38078 * config/m68k/m68k.md (tablejump expander): Likewise.
38079 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38080 TARGET_LONG_JUMP_TABLE_OFFSETS.
38081 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38082 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38084 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38085 David Holsgrove <david.holsgrove@xilinx.com>
38087 * common/config/microblaze/microblaze-common.c
38088 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38089 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38091 * config/microblaze/microblaze.c (microblaze_must_save_register)
38092 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38094 (microblaze_eh_return): New function.
38095 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38096 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38097 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38098 * config/microblaze/microblaze.md (eh_return): New pattern.
38100 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38102 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38103 GCC_DIAGNOSTIC_STRINGIFY): Define.
38105 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38107 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38109 * config/arm/arm.md (<mcrr>): New.
38111 * config/arm/arm.c (arm_arch5te): New.
38112 (arm_option_override): Set arm_arch5te.
38113 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38115 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38116 (arm_mcrr_qualifiers): ... this. New.
38117 (MRRC_QUALIFIERS): Define to...
38118 (arm_mrrc_qualifiers): ... this. New.
38119 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38121 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38122 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38123 (MRRCI, mrrc, MRRC): New.
38124 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38125 VUNSPEC_MRRC2): New.
38127 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38129 * config/arm/arm.md (<mcr>): New.
38131 * config/arm/arm.c (arm_coproc_builtin_available): Add
38132 support for mcr, mrc, mcr2 and mrc2.
38133 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38134 (arm_mcr_qualifiers): ... this. New.
38135 (MRC_QUALIFIERS): Define to ...
38136 (arm_mrc_qualifiers): ... this. New.
38137 (MCR_QUALIFIERS): Define to ...
38138 (arm_mcr_qualifiers): ... this. New.
38139 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38141 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38142 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38143 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38144 VUNSPEC_MRC2): New.
38146 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38148 * config/arm/arm.md (*ldc): New.
38152 * config/arm/arm.c (arm_coproc_builtin_available): Add
38153 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38154 (arm_coproc_ldc_stc_legitimate_address): New.
38155 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38156 'qualifier_const_pointer'.
38157 (LDC_QUALIFIERS): Define to...
38158 (arm_ldc_qualifiers): ... this. New.
38159 (STC_QUALIFIERS): Define to...
38160 (arm_stc_qualifiers): ... this. New.
38161 * config/arm/arm-protos.h
38162 (arm_coproc_ldc_stc_legitimate_address): New.
38163 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38164 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38165 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38166 stc2, stcl, stc2l): New.
38167 * config/arm/constraints.md (Uz): New.
38168 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38169 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38170 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38171 VUNSPEC_STC2L): New.
38173 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38175 * config/arm/arm.md (<cdp>): New.
38176 * config/arm/arm.c (neon_const_bounds): Rename this ...
38177 (arm_const_bounds): ... this.
38178 (arm_coproc_builtin_available): New.
38179 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38180 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38181 (CDP_QUALIFIERS): Define to...
38182 (arm_cdp_qualifiers): ... this. New.
38184 (arm_expand_builtin_args): Add case for 6 arguments.
38185 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38186 (arm_const_bounds): ... this.
38187 (arm_coproc_builtin_available): New.
38188 * config/arm/arm_acle.h (__arm_cdp): New.
38190 * config/arm/arm_acle_builtins.def (cdp): New.
38192 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38193 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38194 'arm_const_bounds'.
38195 * config/arm/types.md (coproc): New.
38196 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38197 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38198 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38199 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38201 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38203 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38204 (UBINOP_QUALIFIERS): New.
38206 (acle_builtin_data): New. Change comment.
38207 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38208 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38209 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38210 arm_acle_builtins.def.
38211 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38212 (arm_init_acle_builtins): New.
38213 (CRC32_BUILTIN): Remove.
38214 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38215 crc32cb, crc32ch and crc32cw.
38216 (arm_init_crc32_builtins): Remove.
38217 (arm_init_builtins): Use arm_init_acle_builtins rather
38218 than arm_init_crc32_builtins.
38219 (arm_expand_acle_builtin): New.
38220 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38221 * config/arm/arm_acle_builtins.def: New.
38223 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38225 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38226 (arm_builtin_datum): ... this.
38227 (arm_init_neon_builtin): Rename to ...
38228 (arm_init_builtin): ... this. Add a new parameters PREFIX
38229 and USE_SIG_IN_NAME.
38230 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38231 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38232 'arm_builtin_datum'.
38233 (arm_init_vfp_builtins): Likewise.
38234 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38235 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38236 (arm_expand_neon_args): Rename to ...
38237 (arm_expand_builtin_args): ... this. Rename builtin_arg
38238 enum values and differentiate between ARG_BUILTIN_MEMORY
38239 and ARG_BUILTIN_NEON_MEMORY.
38240 (arm_expand_neon_builtin_1): Rename to ...
38241 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38242 values, arm_expand_builtin_args and add bool parameter NEON.
38243 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38244 (arm_expand_vfp_builtin): Likewise.
38245 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38247 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38249 PR middle-end/77484
38250 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38251 * predict.c (tree_estimate_probability_bb): Reverse direction of
38252 polymorphic call predictor.
38254 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38256 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38257 (determine_pass_name_match): ...this new function and...
38258 (should_skip_pass_p): ...this new function.
38260 2017-01-06 Nathan Sidwell <nathan@acm.org>
38262 * ipa-visibility.c (function_and_variable_visibility): Reformat
38263 comments and long lines. Remove extrneous if.
38264 * symtab.c (symtab_node::make_decl_local): Fix code format.
38265 (symtab_node::set_section_for_node): Fix comment typo.
38267 2017-01-06 Martin Liska <mliska@suse.cz>
38270 * lra-constraints.c: Rename invariant to lra_invariant.
38271 * predict.c (set_even_probabilities): Initialize e to NULL.
38273 2017-01-05 Martin Sebor <msebor@redhat.com>
38275 PR tree-optimization/78910
38276 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38277 (format_integer): Correct off-by-one error in the handling
38278 of precision with negative numbers in signed conversions..
38280 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38282 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38284 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38286 PR tree-optimization/71016
38287 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38288 factor_out_conditional_conversion. Formatting fix.
38289 (factor_out_conditional_conversion): Add cond_stmt argument.
38290 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38291 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38294 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38296 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38297 read-rtl-function.o, and selftest-rtl.o.
38298 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38299 (selftest::aarch64_test_loading_full_dump): New function.
38300 (selftest::aarch64_run_selftests): New function.
38301 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38302 selftest::aarch64_run_selftests.
38303 * config/i386/i386.c
38304 (selftest::ix86_test_loading_dump_fragment_1): New function.
38305 (selftest::ix86_test_loading_call_insn): New function.
38306 (selftest::ix86_test_loading_full_dump): New function.
38307 (selftest::ix86_test_loading_unspec): New function.
38308 (selftest::ix86_run_selftests): Call the new functions.
38309 * emit-rtl.c (maybe_set_max_label_num): New function.
38310 * emit-rtl.h (maybe_set_max_label_num): New decl.
38311 * function.c (instantiate_decls): Guard call to
38312 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38313 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38315 * gensupport.c (gen_reader::gen_reader): Pass "false"
38316 for new "compact" param of rtx_reader.
38317 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38318 rather than an empty string for NULL strings.
38319 * read-md.c: Potentially include config.h rather than bconfig.h.
38320 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38321 (have_error): New global, copied from errors.c.
38322 (md_reader::read_name): Rename to...
38323 (md_reader::read_name_1): ...this, adding "out_loc" param,
38324 and converting "missing name or number" to returning false, rather
38326 (md_reader::read_name): Reimplement in terms of read_name_1.
38327 (md_reader::read_name_or_nil): New function.
38328 (md_reader::read_string): Handle "(nil)" by returning NULL.
38329 (md_reader::md_reader): Add new param "compact".
38330 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38331 (md_reader::read_file): New method.
38332 * read-md.h (md_reader::md_reader): Add new param "compact".
38333 (md_reader::read_file): New method.
38334 (md_reader::is_compact): New accessor.
38335 (md_reader::read_name): Convert return type from void to file_location.
38336 (md_reader::read_name_or_nil): New decl.
38337 (md_reader::read_name_1): New decl.
38338 (md_reader::m_compact): New field.
38339 (noop_reader::noop_reader): Pass "false" for new "compact" param
38341 (rtx_reader::rtx_reader): Add new "compact" param.
38342 (rtx_reader::read_rtx_operand): Make virtual and convert return
38343 type from void to rtx.
38344 (rtx_reader::read_until): New decl.
38345 (rtx_reader::handle_any_trailing_information): New virtual function.
38346 (rtx_reader::postprocess): New virtual function.
38347 (rtx_reader::finalize_string): New virtual function.
38348 (rtx_reader::m_in_call_function_usage): New field.
38349 (rtx_reader::m_reuse_rtx_by_id): New field.
38350 * read-rtl-function.c: New file.
38351 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38352 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38353 (selftest::verify_three_block_rtl_cfg): New decl.
38354 * read-rtl-function.h: New file.
38355 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38356 For host, include function.h, memmodel.h, and emit-rtl.h.
38357 (one_time_initialization): New function.
38358 (struct compact_insn_name): New struct.
38359 (compact_insn_names): New array.
38360 (find_code): Handle insn codes in compact dumps.
38361 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38362 (bind_subst_iter_and_attr): Likewise.
38363 (add_condition_to_string): Likewise.
38364 (add_condition_to_rtx): Likewise.
38365 (apply_attribute_uses): Likewise.
38366 (add_current_iterators): Likewise.
38367 (apply_iterators): Likewise.
38368 (initialize_iterators): Guard usage of apply_subst_iterator with
38369 #ifdef GENERATOR_FILE.
38370 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38371 (md_reader::read_mapping): Likewise.
38372 (add_define_attr_for_define_subst): Likewise.
38373 (add_define_subst_attr): Likewise.
38374 (read_subst_mapping): Likewise.
38375 (check_code_iterator): Likewise.
38376 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38378 (one_time_initialization): New function.
38379 (rtx_reader::read_until): New method.
38380 (read_flags): New function.
38381 (parse_reg_note_name): New function.
38382 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38383 Handle reuse_rtx ids.
38384 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38385 Add parsing support for RTL dumps, mirroring the special-cases in
38386 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38387 values, and calling handle_any_trailing_information.
38388 (rtx_reader::read_rtx_operand): Convert return type from void
38389 to rtx, returning return_rtx. Handle case 'e'. Call
38390 finalize_string on XSTR and XTMPL fields.
38391 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38392 "(nil)" values were omitted. Call the postprocess vfunc on the
38394 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38395 class ctor. Initialize m_in_call_function_usage. Call
38396 one_time_initialization.
38397 * rtl-tests.c (selftest::test_uncond_jump): Call
38398 set_new_first_and_last_insn.
38399 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38400 * selftest-rtl.c: New file.
38401 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38402 (selftest::get_insn_by_uid): New decl.
38403 * selftest-run-tests.c (selftest::run_tests): Call
38404 read_rtl_function_c_tests.
38405 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38406 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38409 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38411 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38412 operands in a special way. Assert that pos+len <= mode precision.
38414 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38416 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38417 3 argument Alias with unlimited for the negative form.
38418 (fno-vect-cost-model): Removed.
38420 2017-01-05 Martin Liska <mliska@suse.cz>
38422 * hsa-gen.c (gen_hsa_divmod): New function.
38423 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38425 2017-01-05 Martin Liska <mliska@suse.cz>
38428 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38431 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38433 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38434 small constant length operands.
38436 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38438 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38439 between loop iterations.
38441 2017-01-05 Martin Liska <mliska@suse.cz>
38444 * gimplify.c (gimplify_decl_expr): Compare to
38445 asan_poisoned_variables instread of checking flags.
38446 (gimplify_target_expr): Likewise.
38447 (gimplify_expr): Likewise.
38448 (gimplify_function_tree): Conditionally initialize
38449 asan_poisoned_variables.
38451 2017-01-04 Jeff Law <law@redhat.com>
38453 PR tree-optimizatin/78812
38454 * rtl.h (contains_mem_rtx_p): Prototype.
38455 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38456 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38457 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38458 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38459 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38461 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38463 * input.c (assert_char_at_range): Default-initialize actual_range.
38465 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38467 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38468 to match the caller.
38470 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38472 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38473 insns after final jump in test to emit dummy move.
38475 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38477 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38478 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38480 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38482 * multiple_target.c (create_dispatcher_calls): Init e_next.
38483 * tree-ssa-loop-split.c (split_loop): Init border.
38484 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38487 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38492 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38493 (altivec_register_operand): Do not return true if the operand
38494 contains a SUBREG mixing SImode and SFmode.
38495 (vsx_register_operand): Likewise.
38496 (vsx_reg_sfsubreg_ok): New predicate.
38497 (vfloat_operand): Do not return true if the operand contains a
38498 SUBREG mixing SImode and SFmode.
38499 (vint_operand): Likewise.
38500 (vlogical_operand): Likewise.
38501 (gpc_reg_operand): Likewise.
38502 (int_reg_operand): Likewise.
38503 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38504 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38505 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38507 (rs6000_emit_move_si_sf_subreg): New helper function.
38508 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38509 fixup SUBREGs involving SImode and SFmode.
38510 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38511 numbers for the new peephole2 optimization.
38512 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38513 the GLIBC math library that do AND/IOR/XOR operations on single
38514 precision floating point.
38515 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38516 target macros to say whether we need to avoid SUBREGs mixing
38518 (TARGET_ALLOW_SF_SUBREG): Likewise.
38519 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38520 (UNSPEC_SI_FROM_SF): Likewise.
38521 (iorxor): Change spacing.
38522 (and_ior_xor): New iterator for AND, IOR, and XOR.
38523 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38524 (movdi_from_sf_zero_ext): Likewise.
38525 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38526 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38527 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38528 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38529 (fms<mode>4): Likewise.
38530 (fnma<mode>4): Likewise.
38531 (fnms<mode>4): Likewise.
38532 (nfma<mode>4): Likewise.
38533 (nfms<mode>4): Likewise.
38535 2017-01-04 Marek Polacek <polacek@redhat.com>
38538 * doc/invoke.texi: Document -Wpointer-compare.
38540 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38542 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38545 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38546 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38549 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38552 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38553 documentation of the powerpc_popcntb_ok attribute.
38554 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38555 code to issue warning messages if a requested CPU configuration is
38556 not supported by the binary (assembler and loader) toolchain.
38557 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38558 made to define a built-in function that has been disabled.
38559 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38560 made to define a built-in function that has been disabled.
38561 (altivec_init_builtins): Add comment explaining why definition
38562 of the DST built-in functions is not preceded by an assertion
38563 check. Add assertions to prevent ICE if attempts are made to
38564 define an altivec predicate or an abs* built-in function that has
38566 (htm_init_builtins): Add comment explaining why definition of the
38567 htm built-in functions is not preceded by an assertion check.
38569 2017-01-04 Jeff Law <law@redhat.com>
38571 PR tree-optimizatin/67955
38572 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38573 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38574 the points-to solution does not include pt_null. Use DECL_PT_UID
38577 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38579 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38580 Use gen_int_mode instead of gen_lopwart for const_int operands.
38582 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38584 PR tree-optimization/71563
38585 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38586 out of range value - has low bits known to be zero.
38588 2017-01-04 Alan Modra <amodra@gmail.com>
38590 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38591 * configure: Regenerate.
38592 * config.in: Regenerate.
38594 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38597 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38598 a substring of the message, but strcmp with the whole message. Ifdef
38599 ENABLE_NLS, translate the message first using dgettext.
38601 2017-01-03 Jeff Law <law@redhat.com>
38603 PR tree-optimizatin/78856
38604 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38605 (mark_threaded_blocks): Remove code to truncate thread paths that
38606 cross multiple loop headers. Instead invalidate the cached loop
38607 iteration information and handle case of a thread path walking
38608 into an irreducible region.
38610 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38613 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38614 assertions. Add support for doing the signbit if the IEEE 128-bit
38615 floating point value is in a GPR.
38616 * config/rs6000/rs6000.md (Fsignbit): Delete.
38617 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38618 Update the length attribute if the value is in a GPR.
38619 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38620 the sign or zero extension instruction, since the value is always 0/1.
38621 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38624 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38625 extracting SImode to a GPR register so that we can generate a
38626 store, limit the vector to be in a traditional Altivec register
38627 for the vextuwrx instruction.
38629 2017-01-03 Ian Lance Taylor <iant@google.com>
38631 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38633 2017-01-03 Martin Sebor <msebor@redhat.com>
38635 PR tree-optimization/78696
38636 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38637 precision. Use MPFR for %f for greater fidelity. Correct handling
38639 (pass_sprintf_length::compute_format_length): Set width and precision
38640 specified by asrerisk to void_node for vararg functions.
38641 (try_substitute_return_value): Adjust dump output.
38643 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38645 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38647 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38649 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38650 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38651 -mlra/-mno-lra was passed to the compiler.
38653 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38655 PR rtl-optimization/65618
38656 * emit-rtl.c (try_split): Move initialization of "before" and
38657 "after" to just before the call to emit_insn_after_setloc.
38659 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38661 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38663 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38665 * dwarf2out.c (gen_enumeration_type_die): When
38666 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38668 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38670 PR tree-optimization/78965
38671 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38672 Change first argument from const call_info & to call_info &. For %n
38673 set info.nowrite to false.
38675 PR middle-end/78901
38676 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38677 possibly throwing calls.
38679 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38680 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38681 and fns handling, rather than in a separate case SSA_NAME.
38683 2017-01-02 Jeff Law <law@redhat.com>
38685 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38686 fixes for first_period and second_period variables.
38688 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38691 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38692 (*insvqi_1): New insn pattern.
38693 (*insvqi_1_mem_rex64): Ditto.
38694 (*insvqi_2): Ditto.
38695 (*insvqi_3): Rename from *insvqi.
38697 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38699 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38701 * doc/cfg.texi (Edges): Remove reference to Java.
38702 (Maintaining the CFG): Ditto.
38704 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38706 PR middle-end/77674
38707 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38708 transparent aliases.
38710 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38712 PR middle-end/77484
38713 * predict.def (PRED_CALL): Update hitrate.
38714 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38715 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38716 into direct/indirect/polymorphic variants.
38718 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38720 Update copyright years.
38722 * gcc.c (process_command): Update copyright notice dates.
38723 * gcov-dump.c (print_version): Ditto.
38724 * gcov.c (print_version): Ditto.
38725 * gcov-tool.c (print_version): Ditto.
38726 * gengtype.c (create_file): Ditto.
38727 * doc/cpp.texi: Bump @copying's copyright year.
38728 * doc/cppinternals.texi: Ditto.
38729 * doc/gcc.texi: Ditto.
38730 * doc/gccint.texi: Ditto.
38731 * doc/gcov.texi: Ditto.
38732 * doc/install.texi: Ditto.
38733 * doc/invoke.texi: Ditto.
38735 Copyright (C) 2017 Free Software Foundation, Inc.
38737 Copying and distribution of this file, with or without modification,
38738 are permitted in any medium without royalty provided the copyright
38739 notice and this notice are preserved.