1 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
4 markers. Integrate source bind into debug stmt expand loop.
5 (pass_expand::execute): Check debug marker limit. Avoid deep
6 TER and expand debug locations for debug bind insns only.
7 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
9 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
10 * final.c (reemit_insn_block_notes): Take current block from
11 nonbind markers. Declare note where it's first set.
12 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
13 begin stmt markers if enabled.
14 (notice_source_line): Handle nonbind markers. Fail if their
15 location is unknown or that of builtins.
16 (rest_of_handle_final): Convert begin stmt markers to notes if
17 var-tracking didn't run.
18 (rest_of_clean_state): Skip begin stmt markers.
19 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
21 * function.c (allocate_struct_function): Set begin_stmt_markers.
22 * function.h (struct function): Add debug_marker_count counter
23 and debug_nonbind_markers flag.
24 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
25 * gimple-low.c (lower_function_body): Adjust
26 debug_nonbind_markers.
27 (lower_stmt): Drop or skip gimple debug stmts.
28 (lower_try_catch): Skip debug stmts.
29 * gimple.c (gimple_build_debug_begin_stmt): New.
30 (gimple_copy): Increment debug_marker_count if copying one.
31 * gimple.h (gimple_build_debug_begin_stmt): Declare.
32 * gimplify.c (rexpr_location): New.
33 (rexpr_has_location): New.
34 (warn_switch_unreachable_r): Handle gimple debug stmts.
35 (shortcut_cond_r): Call expr_location.
37 (find_goto_label): New.
38 (shortcut_cond_expr): Call expr_has_location, expr_location, and
40 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
42 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
43 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
44 (LANG_HOOKS_INITIALIZER): ... this.
45 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
46 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
48 (update_ebb_live_info): Skip debug insn markers.
49 * lra.c (debug_insn_static_data): Rename to...
50 (debug_bind_static_data): ... this.
51 (debug_marker_static_data): New.
52 (lra_set_insn_recog_data): Select one of the above depending
54 (lra_update_isn_regno_info): Don't assume debug insns have
56 (push_insns): Skip debug insns.
57 * lto-streamer-in.c (input_function): Drop debug stmts
58 depending on active options. Adjust debug_nonbind_markers.
59 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
60 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
61 begin stmt marker notes.
62 (print_insn): Likewise.
63 * recog.c (extract_insn): Recognize rtl for debug markers.
64 * rtl.def (DEBUG_MARKER): New.
65 * tree-inline.c: Include params.h.
66 (remap_gimple_stmt): Handle nonbind markers.
67 (maybe_move_debug_stmts_to_successors): Likewise.
68 (copy_debug_stmt): Likewise.
69 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
70 markers regardless of no side effects.
71 (tsi_link_before): Don't update container's side effects when adding
73 (tsi_link_after): Likewise.
74 (expr_first): Skip begin stmt markers.
75 (expr_last): Likewise.
76 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
77 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
78 Disregard nonbind markers.
79 * tree.c (make_node_stat): Don't set side effects for begin stmt
81 (build1_stat): Likewise.
82 * tree.def (DEBUG_BEGIN_STMT): New.
83 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
84 * var-tracking.c (delete_debug_insns): Renamed to...
85 (delete_vta_debug_insns): ... this.
86 (reemit_marker_as_note): New.
87 (vt_initialize): Reemit markers.
88 (delete_vta_debug_insns): Likewise.
89 (vt_debug_insns_local): Reemit or delete markers.
90 (variable_tracking_main_1): Likewise.
91 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
92 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
93 (gimple_debug_nonbind_marker_p): New.
94 (gimple_build_debug_bind): Adjust.
95 (gimple_build_debug_begin_stmt): New.
96 * doc/invoke.texi (max-debug-marker-count): New param.
97 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
98 (debug_parameter_ref, debug_marker): New.
99 (NOTE_INSN_BEGIN_STMT): New.
100 (DEBUG_INSN): Describe begin stmt markers.
102 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
103 without debug insns, we wouldn't, but clean up debug insns
104 after a control flow insn nevertheless.
106 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
107 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
108 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
109 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
110 (rtl_tidy_fallthru_edge): Likewise.
111 (rtl_verify_fallthru): Likewise.
112 (rtl_verify_bb_layout): Likewise.
113 (skip_insns_after_block): Likewise.
114 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
115 * dwarf2out.c: Include print-rtl.h.
116 (dwarf2out_next_real_insn): New.
117 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
118 Dump debug binds in asm comments.
119 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
120 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
121 callers to use gsi_start_nondebug_bb instead.
122 (gsi_after_labels): Skip gimple debug stmts.
123 (gsi_start_nondebug): New.
124 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
125 (proper_loop_form_for_interchange): Adjust.
126 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
127 * gimple.h (gimple_seq_last_nondebug_stmt): New.
128 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
129 (collect_fallthrough_labels): Likewise.
130 (should_warn_for_implicit_fallthrough): Likewise.
131 (warn_implicit_fallthrough_r): Likewise.
132 (expand_FALLTHROUGH_r): Likewise.
133 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
134 (graphite_copy_stmts_from_block): Skip nonbind markers.
135 * haifa-sched.c (sched_extend_bb): Skip debug insns.
136 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
137 * jump.c (clean_barriers): Skip debug insns.
138 * omp-expand.c (expand_parallel_call): Skip debug insns.
139 (expand_task_call): Likewise.
140 (remove_exit_barrier): Likewise.
141 (expand_omp_taskreg): Likewise.
142 (expand_omp_for_init_counts): Likewise.
143 (expand_omp_for_generic): Likewise.
144 (expand_omp_for_static_nochunk): Likewise.
145 (expand_omp_for_static_chunk): Likewise.
146 (expand_omp_simd): Likewise.
147 (expand_omp_taskloop_for_outer): Likewise.
148 (expand_omp_taskloop_for_inner): Likewise.
149 (expand_oacc_for): Likewise.
150 (expand_omp_sections): Likewise.
151 (expand_omp_single): Likewise.
152 (expand_omp_synch): Likewise.
153 (expand_omp_atomic_load): Likewise.
154 (expand_omp_atomic_store): Likewise.
155 (expand_omp_atomic_fetch_op): Likewise.
156 (expand_omp_atomic_pipeline): Likewise.
157 (expand_omp_atomic_mutex): Likewise.
158 (expand_omp_target): Likewise.
159 (grid_expand_omp_for_loop): Likewise.
160 (grid_expand_target_grid_body): Likewise.
161 (build_omp_regions_1): Likewise.
162 * omp-low.c (check_combined_parallel): Skip debug stmts.
163 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
164 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
165 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
167 * tree-cfg.c (make_blobs_1): Skip debug stmts.
168 (make_edges): Likewise.
169 (cleanup_dead_labels): Likewise.
170 (gimple_can_merge_blocks_p): Likewise.
171 (stmt_starts_bb_p): Likewise.
172 (gimple_block_label): Likewise.
173 (gimple_redirect_edge_and_branch): Likewise.
174 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
176 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
178 * tree-pretty-print (print_declaration): Omit initializer in slim
180 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
182 (eliminate_unnecessary_stmts): Stabilize block removal order.
183 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
184 * var-tracking.c (get_first_insn): New.
185 (vt_emit_notes): Call it.
186 (vt_initialize): Walk any insns before the first BB.
187 (delete_debug_insns): Likewise.
189 * gimple.h (enum gimple_debug_subcode): Add
190 GIMPLE_DEBUG_BEGIN_STMT.
191 (gimple_debug_begin_stmt_p): New.
192 (gimple_debug_nonbind_marker_p): New.
193 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
194 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
195 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
196 * insn-notes.def (BEGIN_STMT): New.
197 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
198 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
199 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
200 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
201 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
202 (INSN_DEBUG_MARKER_KIND): New.
203 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
204 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
205 (INSN_VAR_LOCATION_PTR): New.
206 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
208 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
209 deps only if debug bind insns are enabled.
210 (pass_expand::execute): Avoid deep TER and expand
211 debug locations for debug bind insns only.
212 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
213 debug stmts special handling down to debug bind stmts.
214 * combine.c (try_combine): Narrow debug insns special handling
215 down to debug bind insns.
216 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
217 Narrow debug insns preexisting special handling down to debug
219 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
220 handling down to debug bind insns.
221 * function.c (instantiate_virtual_regs): Skip debug markers,
222 adjust handling of debug binds.
223 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
224 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
225 * haifa-sched.c (schedule_insn): Narrow special handling of debug
226 insns to debug bind insns.
227 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
228 special handling of debug stmts to debug bind stmts.
229 * ipa-split.c (split_function): Likewise.
230 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
231 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
233 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
234 * regrename.c (build_def_use): Likewise.
235 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
236 (pass_cprop_hardreg): Narrow special casing of debug insns to
238 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
239 * reload1.c (reload): Likewise.
240 * sese.c (sese_insert_phis_for_liveouts): Narrow special
241 casing of debug stmts to debug bind stmts.
242 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
243 * ssa-iterators.h (num_imm_uses): Likewise.
244 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
245 debug stmts to debug bind stmts.
246 * tree-inline.c (tree_function_versioning): Narrow special casing
247 of debug stmts to debug bind stmts.
248 * tree-loop-distribution.c (generate_loops_for_partition):
249 Narrow special casing of debug stmts to debug bind stmts.
250 * tree-sra.c (analyze_access_subtree): Narrow special casing
251 of debug stmts to debug bind stmts.
252 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
253 stmts to debug bind stmts.
254 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
255 casing of debug stmts to debug bind stmts.
256 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
257 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
258 casing of debug stmts to debug bind stmts.
259 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
261 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
262 debug stmts to debug bind stmts.
263 (gimple_replace_ssa_lhs): Likewise.
264 (insert_debug_temp_for_var_def): Likewise.
265 (insert_debug_temps_for_defs): Likewise.
266 (reset_debug_uses): Likewise.
267 * tree-ssanames.c (release_ssa_name_fn): Likewise.
268 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
269 (adjust_debug_stmts): Likewise.
270 (adjust_phi_and_debug_stmts): Likewise.
271 (vect_do_peeling): Likewise.
272 * tree-vect-loop.c (vect_transform_loop): Likewise.
273 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
274 * var-tracking.c (adjust_mems): Narrow special casing of debug
275 insns to debug bind insns.
276 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
277 (compute_bb_dataflow, vt_find_locations): Likewise.
278 (vt_expand_loc, emit_notes_for_changes): Likewise.
279 (vt_init_cfa_base): Likewise.
280 (vt_emit_notes): Likewise.
281 (vt_initialize): Likewise.
282 (vt_finalize): Likewise.
284 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
285 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
286 (next_nonnote_nondebug_insn_bb): New.
287 (prev_nonnote_nondebug_insn_bb): New.
288 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
289 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
290 (prev_nonnote_nondebug_insn_bb): Declare.
291 (next_nonnote_nondebug_insn_bb): Declare.
292 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
293 * cfgrtl.c (get_last_bb_insn): Likewise.
294 * lra.c (push_insns): Likewise.
296 2017-12-11 David Malcolm <dmalcolm@redhat.com>
299 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
301 * selftest.c (selftest::test_runner::test_runner): New ctor.
302 (selftest::test_runner::~test_runner): New dtor.
303 * selftest.h (class selftest::test_runner): New class.
305 2017-12-11 Carl Love <cel@us.ibm.com>
307 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
308 vec_extract_fp32_from_shortl]): Add #defines.
309 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
310 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
311 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
312 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
313 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
314 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
315 * doc/extend.texi: Add documentation for the added builtins.
317 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
319 PR rtl-optimization/80693
320 PR rtl-optimization/81019
321 PR rtl-optimization/81020
322 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
323 are not mentioned in i3. Place the REG_UNUSED note on i2,
324 possibly modified to REG_DEAD, if it did not originate in i3.
326 2017-12-11 Jakub Jelinek <jakub@redhat.com>
328 * recog.c (store_data_bypass_p_1): New function.
329 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
330 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
332 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
334 PR rtl-optimization/83361
335 * ifcvt.c (if_convert): Call fixup_partitions.
337 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
339 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
340 early folding of splat_u{8,16,32}.
342 2017-12-11 Jakub Jelinek <jakub@redhat.com>
344 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
345 output_operand_lossage first argument with capital letter.
346 (aarch64_override_options): Don't start error and sorry first argument
349 2017-12-11 Andi Kleen <ak@linux.intel.com>
351 PR gcov-profile/83355
352 * auto-profile.c (string_table::get_index_by_decl): Don't
353 recurse when abstract origin points to itself.
355 2017-12-11 Bin Cheng <bin.cheng@arm.com>
357 PR tree-optimization/83320
358 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
359 (prune_datarefs_not_in_loop): Ditto.
361 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
363 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
365 2017-12-10 Jakub Jelinek <jakub@redhat.com>
367 PR tree-optimization/83337
368 * gimple-loop-interchange.cc (compute_access_stride): Handle
369 bitfield DRs properly.
371 2017-12-09 Jakub Jelinek <jakub@redhat.com>
373 PR tree-optimization/83338
374 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
375 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
376 vector of unsigned integers to vector of signed integers.
378 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
380 PR rtl-optimization/83317
381 * lra-constraints.c (process_address_1): Add insn code check.
383 2017-12-08 Michael Matz <matz@suse.de>
385 Fix PR tree-optimization/83323
386 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
387 head-controlled loops and loop BBs.
388 * common.opt (funroll-and-jam): Remove, instead ...
389 (floop-unroll-and-jam): ... reuse this option.
390 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
391 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
392 (-floop-unroll-and-jam): ... this option.
394 2017-12-08 Jakub Jelinek <jakub@redhat.com>
396 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
397 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
398 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
400 2017-12-08 Richard Biener <rguenther@suse.de>
403 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
404 handle accesses outside of zero-sized vars.
406 2017-12-08 Martin Jambor <mjambor@suse.cz>
408 PR tree-optimization/83141
409 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
410 test for MEM_REFs implicitely changing types with padding. Remove
412 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
414 2017-12-08 Jakub Jelinek <jakub@redhat.com>
416 * config/arc/arc.c (arc_attribute_table): Add exclusions to
418 * config/avr/avr.c (avr_attribute_table): Likewise.
419 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
420 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
421 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
422 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
423 of struct attribute_spec.
424 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
426 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
429 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
431 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
434 * profile-count.c (profile_count::from_gcov_type): Move from
435 profile-count.h; handle overflow.
436 * profile-count.h (profile_count::from_gcov_type): Move offline.
438 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
440 PR rtl-optimization/83304
441 * combine.c (move_deaths): If we do not know where a register died,
444 2017-12-08 Richard Biener <rguenther@suse.de>
446 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
447 Provide -fopt-info-loop feedback when we interchange in a nest.
449 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
451 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
452 for armv6 ARM CPU IDs.
454 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
456 * common/config/arm/arm-common.c: Include <algorithm>.
457 (INCLUDE_VECTOR): Define.
458 (compare_opt_names): New function.
459 (arm_rewrite_selected_arch): Only strip out extensions that can be
460 expressed through -mfpu. Sort the remaining extensions
463 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
465 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
466 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
467 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
468 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
470 2017-06-08 Tristan Gingold <gindold@adacore.com>
473 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
474 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
476 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
478 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
479 of struct attribute_spec.
481 2017-12-08 Julia Koval <julia.koval@intel.com>
483 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
484 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
485 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
486 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
487 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
488 _mm_maskz_dpwssds_epi32): Ditto.
490 2017-12-08 Richard Biener <rguenther@suse.de>
492 PR tree-optimization/81303
493 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
494 conditions try to create a comparison vector type matching
495 the data vector type.
496 (vectorizable_condition): Adjust.
497 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
498 Leave invariant conditions alone in case we can vectorize those.
500 2017-12-08 Julia Koval <julia.koval@intel.com>
502 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
503 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
504 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
505 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
506 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
508 2017-12-08 Julia Koval <julia.koval@intel.com>
510 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
511 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
512 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
513 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
514 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
515 _mm_maskz_dpbusds_epi32): New intrinsics.
517 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
519 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
521 (ld<bh>io_signed): Likewise.
522 (st<bhw>io): Likewise.
523 * config/nios2/predicates.md (ldstio_memory_operand): Allow
524 SMALL_INT12 constant integer operand.
526 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
528 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
530 (nios2_large_constant_allowed): ...this. Adjust uses.
531 (nios2_plus_symbolic_constant_p): Rename to...
532 (nios2_plus_large_constant_p): ...this. Adjust uses.
533 (nios2_legitimate_address_p): Correct CONST_INT handling.
534 (nios2_symbolic_memory_operand_p): Rename to...
535 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
536 (nios2_large_constant_p): Check for large constant integers too.
537 (nios2_split_large_constant): Handle constant integers.
538 (nios2_split_symbolic_memory_operand): Rename to...
539 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
540 (nios2_legitimize_constant_address): Handle constant integers.
541 (r0rel_constant_p): Handle small constant integers.
542 (nios2_print_operand_address): Handle r0-relative integer addresses.
543 * config/nios2/nios2-protos.h: Adjust for renamed functions.
544 * config/nios2/nios2.md: Adjust for renamed functions.
546 2017-12-07 Andrew Waterman <andrew@sifive.com>
548 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
549 (TARGET_HAVE_SRODATA_SECTION): New define.
550 (riscv_select_section): New function.
552 2017-12-08 Joseph Myers <joseph@codesourcery.com>
553 Alexander Monakov <amonakov@ispras.ru>
554 Jakub Jelinek <jakub@redhat.com>
557 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
559 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
561 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
562 a bool success value. Don't call output_operand_lossage here.
563 (aarch64_print_ldpstp_address): Return a bool success value.
564 (aarch64_print_operand_address): Call output_addr_const if
565 aarch64_print_address_internal fails.
566 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
567 'y'; call output_operand_lossage instead. Call output_operand_lossage
568 if aarch64_print_ldpstp_address fails.
570 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
572 * tree-vector-builder.h
573 (tree_vector_builder::binary_encoded_nelts): Declare.
574 * tree-vector-builder.c
575 (tree_vector_builder::binary_encoded_nelts): New function.
576 * fold-const.c (negate_expr_p): Likewise.
577 (operand_equal_p, fold_checksum_tree): Likewise.
578 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
579 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
580 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
581 (uniform_vector_p): Likewise.
582 * varasm.c (const_hash_1, compare_constant): Likewise.
583 * tree-ssa-ccp.c: Include tree-vector-builder.h.
584 (valid_lattice_transition): Operate directly on the VECTOR_CST
586 * ipa-icf.c: Include tree-vector-builder.h.
587 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
588 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
590 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
592 * tree.c (build_vector): Delete.
593 * tree.h (build_vector): Make static and move into the self-testing
596 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
598 * vector-builder.h (vector_builder::derived): New const overload.
599 (vector_builder::elt): New function.
600 * tree-vector-builder.h (tree_vector_builder::type): New function.
601 (tree_vector_builder::apply_step): Declare.
602 * tree-vector-builder.c (tree_vector_builder::apply_step): New
604 * gimple-fold.h (tree_vector_builder): Declare.
605 (gimple_build_vector): Take a tree_vector_builder instead of a
606 type and vector of elements.
607 * gimple-fold.c (gimple_build_vector): Likewise.
608 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
610 (get_initial_defs_for_reduction): Likewise.
611 (vectorizable_induction): Likewise.
613 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
615 * tree-vector-builder.h
616 (tree_vector_builder::new_binary_operation): Declare.
617 * tree-vector-builder.c
618 (tree_vector_builder::new_binary_operation): New function.
619 * fold-const.c (fold_relational_const): Use it.
620 (const_binop): Likewise. Check that both input vectors have
621 the same number of elements, thus excluding things like WIDEN_SUM.
622 Check whether it is possible to operate directly on the encodings
625 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
627 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
628 new_unary_operation, operating only on the encoded elements.
629 (const_unop): Likewise.
630 (exact_inverse): Likewise.
631 (distributes_over_addition_p): New function.
632 (const_binop): Use tree_vector_builder and new_unary_operation
633 for combinations of VECTOR_CST and INTEGER_CST. Operate only
634 on the encoded elements unless the encoding is strided and the
635 operation does not distribute over addition.
636 (fold_convert_const): Use tree_vector_builder and
637 new_unary_operation. Operate only on the encoded elements
638 for truncating integer conversions, or for non-stepped encodings.
640 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
642 * config/sparc/sparc.c: Include tree-vector-builder.h.
643 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
644 * expmed.c: Include tree-vector-builder.h.
645 (make_tree): Use tree_vector_builder instead of build_vector.
646 * fold-const.c: Include tree-vector-builder.h.
647 (const_binop): Use tree_vector_builder instead of build_vector.
648 (const_unop): Likewise.
649 (native_interpret_vector): Likewise.
650 (fold_vec_perm): Likewise.
651 (fold_ternary_loc): Likewise.
652 * gimple-fold.c: Include tree-vector-builder.h.
653 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
655 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
656 (simplify_vector_constructor): Use tree_vector_builder instead
658 * tree-vect-generic.c: Include tree-vector-builder.h.
659 (add_rshift): Use tree_vector_builder instead of build_vector.
660 (expand_vector_divmod): Likewise.
661 (optimize_vector_constructor): Likewise.
662 * tree-vect-loop.c: Include tree-vector-builder.h.
663 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
664 of build_vector. Explicitly use a stepped encoding for
666 * tree-vect-slp.c: Include tree-vector-builder.h.
667 (vect_get_constant_vectors): Use tree_vector_builder instead
669 (vect_transform_slp_perm_load): Likewise.
670 (vect_schedule_slp_instance): Likewise.
671 * tree-vect-stmts.c: Include tree-vector-builder.h.
672 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
673 (vect_gen_perm_mask_any): Likewise.
674 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
675 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
677 (build_vector_from_val): Likewise. Explicitly use a duplicate
680 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
682 * doc/generic.texi (VECTOR_CST): Describe new representation of
684 * vector-builder.h: New file.
685 * tree-vector-builder.h: Likewise.
686 * tree-vector-builder.c: Likewise.
687 * Makefile.in (OBJS): Add tree-vector-builder.o.
688 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
689 * tree-core.h (tree_base): Add a vector_cst field to the u union.
690 (tree_vector): Change the number of elements to
691 vector_cst_encoded_nelts.
692 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
693 (VECTOR_CST_ELTS): Delete.
694 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
695 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
696 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
697 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
698 (VECTOR_CST_ENCODED_ELT): Likewise.
699 (vector_cst_encoded_nelts): New function.
700 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
701 VECTOR_CST_NELTS_PER_PATTERN as arguments.
702 (vector_cst_int_elt, vector_cst_elt): Declare.
703 * tree.c: Include tree-vector-builder.h.
704 (tree_code_size): Abort if passed VECTOR_CST.
705 (tree_size): Update for new VECTOR_CST layout.
706 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
707 VECTOR_CST_NELTS_PER_PATTERN as arguments.
708 (build_vector): Use tree_vector_builder.
709 (vector_cst_int_elt, vector_cst_elt): New functions.
710 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
711 encoded elements and then create the vector in the canonical form.
712 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
713 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
714 (tree_c_tests): Call test_vector_cst_patterns.
715 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
717 (hash_tree): Likewise.
718 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
719 (streamer_write_tree_header): Likewise.
720 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
721 (streamer_alloc_tree): Likewise. Update call to make_vector.
722 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
724 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
726 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
727 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
728 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
730 2017-12-07 Bin Cheng <bin.cheng@arm.com>
731 Richard Biener <rguenther@suse.de>
733 PR tree-optimization/81303
734 * Makefile.in (gimple-loop-interchange.o): New object file.
735 * common.opt (floop-interchange): Reuse the option from graphite.
736 * doc/invoke.texi (-floop-interchange): Ditto. New document for
737 -floop-interchange and mention it for -O3.
738 * opts.c (default_options_table): Enable -floop-interchange at -O3.
739 * gimple-loop-interchange.cc: New file.
740 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
741 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
742 * passes.def (pass_linterchange): New pass.
743 * timevar.def (TV_LINTERCHANGE): New time var.
744 * tree-pass.h (make_pass_linterchange): New declaration.
745 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
746 interchange. Record IV before/after increment in new parameters.
747 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
748 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
750 (check_reduction_path): ...New function here.
751 * tree-vectorizer.h (check_reduction_path): New declaration.
753 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
756 PR rtl-optimization/80818
757 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
758 always early clobbered.
759 * lra-lives.c (process_bb_lives): Check input hard regs for early
760 clobbered non-operand hard reg.
762 2017-12-07 Jakub Jelinek <jakub@redhat.com>
765 * tree-cfg.c (verify_gimple_assign_binary): Don't require
766 types_compatible_p, just that TYPE_MODE is the same.
768 2017-12-07 Martin Sebor <msebor@redhat.com>
771 * attribs.c (empty_attribute_table): Initialize new member of
772 struct attribute_spec.
773 (decl_attributes): Add argument. Handle mutually exclusive
774 combinations of attributes.
775 (selftests::test_attribute_exclusions): New function.
776 (selftests::attribute_c_tests): Ditto.
777 * attribs.h (decl_attributes): Add default argument.
778 * selftest.h (attribute_c_tests): Declare.
779 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
780 * tree-core.h (attribute_spec::exclusions, exclude): New type and
782 * doc/extend.texi (Common Function Attributes): Update const and pure.
783 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
784 of struct attribute_spec.
785 * config/arc/arc.c (arc_attribute_table): Same.
786 * config/arm/arm.c (arm_attribute_table): Same.
787 * config/avr/avr.c ( avr_attribute_table): Same.
788 * config/bfin/bfin.c (bfin_attribute_table): Same.
789 * config/cr16/cr16.c (cr16_attribute_table): Same.
790 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
791 * config/h8300/h8300.c (h8300_attribute_table): Same.
792 * config/i386/i386.c (ix86_attribute_table): Same.
793 * config/ia64/ia64.c (ia64_attribute_table): Same.
794 * config/m32c/m32c.c (m32c_attribute_table): Same.
795 * config/m32r/m32r.c (m32r_attribute_table): Same.
796 * config/m68k/m68k.c (m68k_attribute_table): Same.
797 * config/mcore/mcore.c (mcore_attribute_table): Same.
798 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
799 * config/mips/mips.c (mips_attribute_table): Same.
800 * config/msp430/msp430.c (msp430_attribute_table): Same.
801 * config/nds32/nds32.c (nds32_attribute_table): Same.
802 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
803 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
804 * config/rl78/rl78.c (rl78__attribute_table): Same.
805 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
806 * onfig/rx/rx.c (rx_attribute_table): Same.
807 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
808 * config/sh/sh.c (sh_attribute_table): Same.
809 * config/sparc/sparc.c (sparc_attribute_table): Same.
810 * config/spu/spu.c (spu_attribute_table): Same.
811 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
812 * config/v850/v850.c (v850_attribute_table): Same.
813 * config/visium/visium.c (visium_attribute_table): Same.
815 2017-12-07 Tamar Christina <tamar.christina@arm.com>
818 * config/arm/arm.c (INCLUDE_STRING): Define.
819 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
820 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
823 2017-12-07 Michael Matz <matz@suse.de>
825 Add unroll and jam pass
827 * gimple-loop-jam.c: New file.
828 * Makefile.in (OBJS): Add gimple-loop-jam.o.
829 * common.opt (funroll-and-jam): New option.
830 * opts.c (default_options_table): Add unroll-and-jam at -O3.
831 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
832 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
833 * passes.def: Add pass_loop_jam.
834 * timevar.def (TV_LOOP_JAM): Add.
835 * tree-pass.h (make_pass_loop_jam): Declare.
836 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
837 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
838 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
839 to flow_loop_tree_node_add.
840 (duplicate_subloops, copy_loops_to): Append to sibling list.
841 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
842 * doc/invoke.texi (-funroll-and-jam): Document new option.
843 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
845 2017-12-07 Richard Biener <rguenther@suse.de>
847 PR tree-optimization/83296
848 PR tree-optimization/67769
849 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
850 flow sensitive info in an unrelated BB.
851 (value_replacement): Use reset_flow_sensitive_info.
852 (minmax_replacement): Reset flow sensitive info on the def
853 we move. Do not reset flow sensitive info in the whole BB
855 (abs_replacement): Likewise.
857 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
860 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
861 rs6000_cpu to the given -mcpu=, or to the default processor.
863 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
865 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
866 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
867 instead of rs6000_cpu_attr.
868 (rs6000_adjust_cost): Ditto.
869 (is_microcoded_insn): Ditto.
870 (rs6000_adjust_priority): Ditto.
871 (rs6000_issue_rate): Ditto.
872 (rs6000_use_sched_lookahead): Ditto.
873 (rs6000_use_sched_lookahead_guard): Ditto.
874 (rs6000_sched_reorder): Ditto.
875 (force_new_group): Ditto.
876 * config/rs6000/rs6000.md (cpu attribute): Ditto.
877 (group_ending_nop): Ditto.
879 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
881 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
882 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
883 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
884 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
886 (rs6000_reassociation_width): Ditto.
887 (rs6000_emit_epilogue): Ditto.
888 (rs6000_adjust_cost): Ditto.
889 (is_microcoded_insn): Ditto.
890 (is_cracked_insn): Ditto.
891 (rs6000_adjust_priority): Ditto.
892 (rs6000_sched_reorder): Ditto.
893 (rs6000_sched_reorder2): Ditto.
894 (insn_must_be_first_in_group): Ditto.
895 (insn_must_be_last_in_group): Ditto.
896 (rs6000_register_move_cost): Ditto.
897 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
900 2017-12-07 Julia Koval <julia.koval@intel.com>
902 * config.gcc: Add vaesintrin.h.
903 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
905 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
906 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
908 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
909 * config/i386/immintrin.h: Include vaesintrin.h.
910 * config/i386/sse.md (vaesdec_<mode>): New pattern.
911 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
912 _mm_aesdec_epi128): New intrinsics.
914 2017-12-06 David Malcolm <dmalcolm@redhat.com>
916 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
917 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
918 c-family/c-spellcheck.cc.
919 (best_macro_match::best_macro_match): Likewise.
921 (struct edit_distance_traits<cpp_hashnode *>): Move to
922 c-family/c-spellcheck.h.
923 (class best_macro_match): Likewise.
925 2017-12-06 Jakub Jelinek <jakub@redhat.com>
927 PR tree-optimization/83293
928 * gimple-ssa-strength-reduction.c (insert_initializers): Use
929 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
930 might insert into empty bb.
933 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
934 simplify for plus with :c added, and pointer_plus without that.
935 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
936 with undefined overflow and the conversion is not widening,
937 perform negation in utype and only convert to type afterwards.
938 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
939 simplify for plus with :c added, and pointer_plus without that.
940 If type is integral with undefined overflow and the conversion is
941 not widening, perform minus in utype and only convert to type
942 afterwards. Move the last pointer_diff_expr simplify into the
945 2017-12-06 Martin Sebor <msebor@redhat.com>
947 PR tree-optimization/82646
948 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
951 2017-12-06 Martin Sebor <msebor@redhat.com>
953 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
956 PR tree-optimization/83075
957 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
958 strncat/strncpy don't change length of source string.
960 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
963 2017-11-29 Martin Aberg <maberg@gaisler.com>
965 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
966 to prevent b2bst errata sequence.
967 (sqrtdf2_fix): Likewise.
969 2017-12-06 Jakub Jelinek <jakub@redhat.com>
971 PR tree-optimization/81945
972 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
973 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
974 to dest_cfun has orig_loop_num set, either remap it to the new
975 loop number if the loop got moved too, or clear it.
977 2017-12-05 Steve Ellcey <sellcey@cavium.com>
979 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
981 (thunderx2t99_nothing): New insn reservation.
982 (thunderx2t99_mrs): New insn reservation.
983 (thunderx2t99_multiple): New insn reservation.
984 (thunderx2t99_alu_basi): Add bfx to reservation.
985 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
987 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
990 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
992 2017-12-05 Bin Cheng <bin.cheng@arm.com>
994 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
995 tree-ssa-pre.c::remove_dead_inserted_code.
996 * tree-ssa-dce.h: New file.
997 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
998 (remove_dead_inserted_code): Move and rename to function
999 tree-ssa-dce.c::simple_dce_from_worklist.
1000 (pass_pre::execute): Update use.
1002 2017-12-05 Richard Biener <rguenther@suse.de>
1004 PR tree-optimization/83277
1005 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1006 to code-gen liveout vars.
1008 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1010 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1011 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1012 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1015 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1017 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1019 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1020 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1023 2017-12-05 Richard Biener <rguenther@suse.de>
1025 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1026 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1027 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1028 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1030 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1032 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1034 2017-12-05 Martin Liska <mliska@suse.cz>
1035 Jakub Jelinek <jakub@redhat.com>
1037 * doc/invoke.texi: Document the options.
1038 * flag-types.h (enum sanitize_code): Add
1039 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1040 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1041 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1042 * opts.c: Define new sanitizer options.
1043 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1044 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1046 2017-12-05 Julia Koval <julia.koval@intel.com>
1048 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1049 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1050 (ix86_handle_option): Handle -mavx512vnni.
1051 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1052 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1053 * config/i386/i386-c (__AVX512VNNI__): New.
1054 * config/i386/i386.c (ix86_target_string): Handle new option.
1055 (ix86_valid_target_attribute_inner_p): Handle new option.
1056 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1057 * config/i386/i386.opt (mavx512vnni): New option.
1059 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1062 * athlon.md: Disable for generic.
1063 * haswell.md: Enable for generic.
1064 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1065 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1067 (ix86_adjust_cost): Move generic to haswell path.
1069 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1071 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1072 instead of MEM_P in a couple more places. Fix formatting issues.
1074 2017-12-04 Jim Wilson <jimw@sifive.com>
1076 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1077 instead of GP_REG_LAST-1.
1078 (riscv_adjust_libcall_cfi_prologue): Likewise.
1079 (riscv_adjust_libcall_cri_epilogue): Likewise.
1080 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1083 2017-12-04 Luis Machado <luis.machado@linaro.org>
1085 * ipa-pure-const.c (check_decl): Add missing newline.
1086 (state_from_flags): Likewise.
1088 2017-12-04 Jeff Law <law@redhat.com>
1090 PR tree-optimizatin/78496
1091 * gimple-ssa-evrp-analyze.h
1092 (evrp_range_analyzer::get_vr_values): Simplify.
1093 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1094 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1095 and gimple-ssa-evrp-analyze.h.
1096 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1097 (simplify_stmt_for_jump_threading): Copy a blob of code from
1098 tree-vrp.c to use ranges to simplify statements.
1099 (dom_opt_dom_walker::before_dom_children): Call
1100 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1101 (dom_opt_dom_walker::after_dom_children): Similarly for
1102 evrp_range_analyzer::leave.
1103 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1106 * gimple-ssa-evrp-analyze.c
1107 (evrp_range_analyzer::extract_range_from_stmt): Always use
1108 vr_values::update_value_range so preexisting range info is
1109 medged with new range info, even if the new range is VR_VARYING.
1111 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1113 * combine.c: Adjust comment.
1114 (use_crosses_set_p): Delete.
1115 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1116 (try_combine): Ditto.
1118 2017-12-04 Richard Biener <rguenther@suse.de>
1120 PR tree-optimization/83255
1121 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1122 Re-add zero-iteration check.
1124 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1126 PR rtl-optimization/83245
1127 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1128 hard registers as earlyclobber, also if not in an asm.
1130 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1134 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1137 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1138 rs6000_cpu based on cpu_index, not tune_index.
1140 2017-12-04 Richard Biener <rguenther@suse.de>
1142 PR tree-optimization/83238
1143 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1144 code match comment, rejecting invalid SESE regions.
1146 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1148 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1149 require base operand is a REG_POINTER prior to reload on targets
1150 with non-equivalent space registers.
1152 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1154 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1155 (update_specialized_profile): Fix updating of counts.
1156 (perhaps_add_new_callers): Likewise.
1158 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1161 * x86-tune.def: Remove obsolette FIXMEs.
1162 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1163 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1164 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1166 (X86_TUNE_PAD_RETURNS): Disable for generic.
1168 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1170 PR tree-optimization/83170
1171 PR tree-optimization/83241
1172 * gimple-ssa-store-merging.c
1173 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1174 gimple_vuse (ins_stmt) in case it has changed.
1175 (imm_store_chain_info::output_merged_store): Likewise.
1177 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1181 * tree-cfg.c (pass_warn_function_return::execute): Handle
1182 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1183 with BUILTINS_LOCATION.
1187 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1188 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1189 instead of DECL_MODE.
1191 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1192 last argument to rtx pointer.
1193 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1194 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1195 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1196 depending on the chosen ISAs.
1197 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1198 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1199 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1201 * config/i386/sse.md (mov<mode>_internal): Likewise.
1202 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1204 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1206 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1208 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1210 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1211 parameters from prototype.
1212 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1213 parameters. Don't print a comment.
1214 (emit_fusion_gpr_load): Adjust.
1215 (emit_fusion_load_store): Adjust.
1216 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1217 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1218 comment on the second line.
1220 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1223 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1224 rs6000_cpu based on cpu_index, not tune_index.
1226 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1228 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1229 which_alternative instead of which_alternative + 1.
1230 (output_asm_insn): Print an extra tab if the template is short.
1232 2017-12-01 Jim Wilson <jimw@sifive.com>
1234 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1236 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1237 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1239 * doc/tm.texi: Regenerate.
1241 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1244 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1245 whether we can allocate pseudos before trying to fix an address.
1246 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1247 memory address is indexed or indirect.
1248 (floatuns_<mode>si2_hw2): Likewise.
1250 2017-12-01 Jason Merrill <jason@redhat.com>
1252 * Makefile.in (TAGS): Add c-family/*.cc.
1254 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1256 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1257 (call_value_insn): Likewise.
1258 (sibcall_insn): Likewise.
1259 (sibcall_value_insn): Likewise.
1260 (movsi_aarch64): Likewise.
1261 (movdi_aarch64): Likewise.
1262 (add_losym_): Likewise.
1263 (ldr_got_small_): Likewise.
1264 (ldr_got_small_sidi): Likewise.
1265 (ldr_got_small_28k_): Likewise.
1266 (ldr_got_small_28k_sidi): Likewise.
1267 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1268 Move output_addr_const to symbolic case. Add error check.
1270 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1272 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1273 * config/s390/s390.md: Disable MVC merging peephole if it would
1274 disable operand forwarding.
1275 (new peephole2): Split MVCs if it would turn them into up to 2
1278 2017-12-01 Richard Biener <rguenther@suse.de>
1280 PR tree-optimization/83232
1281 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1282 detection of same access. Instead of breaking the group here
1283 do not consider the duplicate. Add comment explaining real fix.
1285 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1287 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1289 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1291 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1293 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1295 * function.h (struct function): Remove cilk_frame_decl,
1296 is_cilk_function and calls_cilk_spawn fields.
1297 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1299 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1300 cilk_elemental field.
1301 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1302 * target.def: Adjust comment.
1303 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1304 Don't test cilk_elemental.
1306 PR tree-optimization/83233
1307 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1308 bswap_stat name for the struct.
1311 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1312 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1313 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1314 SWITCH_BREAK_LABEL_P set on the label.
1315 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1316 added for default case if it was missing and not all cases covered.
1317 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1318 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1320 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1321 to avoid -Wimplicit-fallthrough warning.
1322 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1323 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1325 PR tree-optimization/83221
1326 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1328 (init_reassoc): Formatting fix.
1331 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1332 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1336 * tree-cfg.c: Include asan.h.
1337 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1339 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1341 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1343 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1345 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1346 vpcmpeqd instruction.
1348 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1350 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1351 registers type for 128bit mode.
1353 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1355 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1357 * gengtype-state.c (read_state_pair): Likewise.
1358 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1359 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1360 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1361 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1362 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1363 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1364 * builtins.c (fold_builtin_3): Likewise.
1365 * graphite-scop-detection.c
1366 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1367 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1370 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1373 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1374 parameter. Return true if ignore_decl_rtl_set_p is true and other
1375 conditions are satisfied.
1376 * asan.h (asan_protect_global): Add new parameter.
1377 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1378 to asan_protect_global calls.
1380 2017-11-30 Jim Wilson <jimw@sifive.com>
1382 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1383 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1384 -mstrict-align, add info on default value. Delete redundant lines for
1385 -mabi. Add missing -mexplicit-relocs docs.
1387 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1389 * config/arc/arc.md (trap): New pattern.
1391 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1393 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1394 instruction to end into a delay slot.
1395 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1396 can be placed into a delay slot against reg_note.
1398 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1400 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1401 labels number of usages.
1403 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1405 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1407 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1409 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1412 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1413 multiplication by power of 2 constant into two shifts + comparison.
1415 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1418 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1420 2017-11-30 Richard Biener <rguenther@suse.de>
1422 PR tree-optimization/83202
1423 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1424 allow_peel argument and guard peeling.
1425 (canonicalize_loop_induction_variables): Likewise.
1426 (canonicalize_induction_variables): Pass false.
1427 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1428 peeling from cunrolli.
1430 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1432 * combine.c (try_combine): Print a message to dump file whenever
1433 I0, I1, or I2 cannot be combined into I3.
1435 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1437 PR rtl-optimization/83156
1438 PR rtl-optimization/82621
1439 * combine.c (try_combine): Don't split an I2 if one of the dests is
1440 set again before I3. Allow unused dests.
1442 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1444 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1446 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1448 PR rtl-optimization/80818
1449 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1450 recursively. Use insn code for clobber.
1451 (lra_set_insn_recog_data): Pass the new arg to
1452 collect_non_operand_hard_regs.
1453 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1455 (lra_update_insn_regno_info): Pass insn to
1456 add_regs_to_insn_regno_info.
1458 2017-11-29 Jim Wilson <jimw@sifive.com>
1459 Andrew Waterman <andrew@sifive.com>
1461 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1462 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1463 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1464 use SINGLE_SHIFT_COST.
1465 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1466 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1468 2017-11-29 Julia Koval <julia.koval@intel.com>
1470 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1471 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1472 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1473 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1474 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1475 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1476 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1477 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1478 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1479 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1480 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1481 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1482 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1483 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1484 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1485 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1486 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1487 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1488 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1489 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1490 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1491 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1492 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1493 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1494 __builtin_ia32_vpshldv_v2di_mask,
1495 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1496 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1497 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1499 2017-11-29 Julia Koval <julia.koval@intel.com>
1501 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1502 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1503 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1504 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1505 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1506 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1507 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1508 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1509 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1510 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1511 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1512 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1513 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1514 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1515 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1516 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1517 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1518 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1519 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1520 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1521 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1523 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1525 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1526 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1529 2017-11-29 Martin Aberg <maberg@gaisler.com>
1531 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1532 to prevent b2bst errata sequence.
1533 (sqrtdf2_fix): Likewise.
1535 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1537 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1538 (div_sqrt_insn_p): New function.
1539 (sparc_do_work_around_errata): Insert NOP instructions to
1540 prevent sequences that could trigger the TN-0013 errata for
1541 certain LEON3 processors.
1542 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1543 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1544 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1545 (in_branch_delay): Prevent div and sqrt in delay slot if
1547 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1549 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1551 * config/sparc/sparc.c (atomic_insn_p): New function.
1552 (sparc_do_work_around_errata): Insert NOP instructions to
1553 prevent sequences that could trigger the TN-0010 errata for
1555 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1556 instruction referable in atomic_insns_p.
1558 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1560 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1561 (atomic_compare_and_swap_leon3_1): Likewise.
1564 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1566 * config/sparc/sparc.c (fpop_insn_p): New function.
1567 (sparc_do_work_around_errata): Insert NOP instructions to
1568 prevent sequences that could trigger the TN-0012 errata for
1570 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1571 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1572 (in_branch_annul_delay): Prevent floating-point instructions
1573 in delay slot of annulled integer branch.
1575 2017-11-29 Richard Biener <rguenther@suse.de>
1577 PR tree-optimization/83202
1578 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1580 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1581 nodes vectorized to the same stmts multiple times.
1582 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1583 (vect_analyze_slp_instance): Adjust.
1584 (scalar_stmts_to_slp_tree_map_t): New typedef.
1585 (vect_schedule_slp_instance): Add a map recording the SLP node
1586 representing the vectorized stmts for a set of scalar stmts.
1587 Avoid code-generating redundancies.
1588 (vect_schedule_slp): Allocate map and pass it down.
1590 2017-11-29 Nathan Sidwell <nathan@acm.org>
1593 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1596 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1598 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1599 cases for printing LDP/STP memory addresses.
1600 (aarch64_print_address_internal): Renamed from
1601 aarch64_print_operand_address, added parameter, add Pmode check.
1602 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1603 (aarch64_print_operand_address): Indirect to
1604 aarch64_print_address_internal.
1605 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1606 'y' operand output specifier.
1608 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1611 * tree.c (build_simple_mem_ref_loc): Handle
1612 get_addr_base_and_unit_offset returning a MEM_REF.
1615 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1618 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1621 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1623 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1625 (layout::move_to_column): Likewise.
1627 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1630 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1631 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1633 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1634 Martin Jambor <mjambor@suse.cz>
1637 * tree.h (expr_type_first_operand_type_p): Declare
1638 * tree.c (expr_type_first_operand_type_p): New function.
1639 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1640 (ipa_value_from_jfunc): Adjust declaration.
1641 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1642 Use it as result type for arithmetics, unless it is NULL in which case
1643 be more conservative.
1644 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1645 ipa_get_jf_pass_through_result.
1646 (propagate_vals_across_pass_through): Likewise.
1647 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1648 is to propagate_vals_across_pass_through.
1649 (propagate_constants_across_call): Pass PARM_TYPE to
1650 propagate_scalar_across_jump_function.
1651 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1652 ipa_value_from_jfunc.
1653 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1654 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1655 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1656 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1657 target_type, pass it to ipa_value_from_jfunc.
1658 (update_indirect_edges_after_inlining): Pass parameter type to
1659 try_make_edge_direct_simple_call.
1661 2017-11-28 Jeff Law <law@redhat.com>
1663 * gimple-ssa-evrp-analyze.c
1664 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1665 refine ranges if scev_initialized_p returns true.
1666 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1668 2017-11-28 Julia Koval <julia.koval@intel.com>
1670 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1671 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1672 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1673 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1674 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1675 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1676 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1677 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1678 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1679 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1680 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1681 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1682 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1683 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1684 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1685 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1686 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1687 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1688 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1689 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1690 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1691 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1693 2017-11-28 Julia Koval <julia.koval@intel.com>
1695 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1696 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1697 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1698 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1699 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1700 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1701 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1702 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1703 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1704 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1705 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1706 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1707 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1708 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1709 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1710 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1711 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1712 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1713 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1714 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1715 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1716 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1717 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1718 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1719 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1720 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1721 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1722 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1723 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1724 __builtin_ia32_vpshld_v2di_mask): New builtins.
1725 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1726 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1728 2017-11-28 Richard Biener <rguenther@suse.de>
1730 PR tree-optimization/80776
1731 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1733 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1735 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1736 If the incoming edge is an effective fallthru because the other
1737 edge only reaches a __builtin_unreachable () then record ranges
1738 derived from the controlling condition in SSA info.
1739 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1740 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1742 2017-11-28 Olivier Hainque <hainque@adacore.com>
1744 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1745 on mingw build hosts.
1747 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1750 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1752 2017-11-28 Richard Biener <rguenther@suse.de>
1755 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1756 copies generated from memcpy use a character array as reference
1759 2017-11-28 Julia Koval <julia.koval@intel.com>
1760 Sebastian Peryt <sebastian.peryt@intel.com>
1762 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1763 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1764 cilk-common.o, cilk.h, cilk-common.c): Remove.
1766 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1767 * builtins.c (is_builtin_name): Remove cilkplus condition.
1768 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1769 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1770 cilk-builtins.def, cilkplus.def): Remove.
1771 * cif-code.def (CILK_SPAWN): Remove.
1772 * cilk-builtins.def: Delete.
1773 * cilk-common.c: Ditto.
1775 * cilkplus.def: Ditto.
1776 * config/darwin.h (fcilkplus): Delete.
1777 * cppbuiltin.c: Ditto.
1778 * doc/extend.texi: Remove cilkplus doc.
1779 * doc/generic.texi: Ditto.
1780 * doc/invoke.texi: Ditto.
1781 * doc/passes.texi: Ditto.
1782 * gcc.c (fcilkplus): Remove.
1783 * gengtype.c (cilk.h): Remove.
1784 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1786 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1788 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1790 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1791 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1792 cilkplus conditions.
1793 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1794 inline_read_section): Ditto.
1795 * ipa-inline-analysis.c (cilk.h): Remove.
1796 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1797 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1798 append_linker_options): Remove condition for fcilkplus.
1799 * lto/lto-lang.c (cilk.h): Remove.
1800 (lto_init): Remove condition for fcilkplus.
1801 * omp-expand.c (expand_cilk_for_call): Delete.
1802 (expand_omp_taskreg, expand_omp_for_static_chunk,
1803 expand_omp_for): Remove cilkplus
1805 (expand_cilk_for): Delete.
1806 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1807 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1808 execute_lower_omp, diagnose_sb_0): Ditto.
1809 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1810 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1811 * tree-nested.c: Ditto.
1812 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1813 (dump_generic_node): Ditto.
1814 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1815 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
1817 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
1819 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1821 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
1822 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
1823 for vector mode and !TARGET_SIMD.
1825 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1827 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
1829 * tree.h (SWITCH_LABELS): Remove.
1830 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
1831 assert SWITCH_BODY is non-NULL.
1832 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
1834 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
1836 PR tree-optimization/80788
1837 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
1838 has TREE_OVERFLOW set, call drop_tree_overflow.
1840 2017-11-28 Richard Biener <rguenther@suse.de>
1842 PR tree-optimization/83158
1843 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
1845 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
1848 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
1849 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
1851 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
1853 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
1854 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
1856 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1859 * varasm.c (compare_constant): Take the mode of the constants into
1860 account when comparing floating point constants.
1862 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
1864 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
1865 from explicit instantiation of debug_helper.
1866 * vec.h (DEFINE_DEBUG_VEC): Ditto.
1868 2017-11-27 Richard Biener <rguenther@suse.de>
1870 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
1873 2017-11-27 Richard Biener <rguenther@suse.de>
1875 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
1876 (type_hash_canon): Also clear int_cst_hash_table entry for
1878 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
1880 2017-11-27 Tamar Christina <tamar.christina@arm.com>
1882 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1883 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1884 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1886 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1888 * hash-map.h (gt_cleare_cache): Avoid UB.
1890 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
1892 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
1893 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
1894 (decide_unroll_runtime_iterations): Likewise.
1895 (decide_unroll_stupid): Likewise.
1897 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1900 * config/i386/i386.md: Add a loop with incssp.
1902 2017-11-27 Martin Jambor <mjambor@suse.cz>
1904 PR tree-optimization/81248
1905 * tree-sra.c (splice_param_accesses): Remove size check.
1906 (decide_one_param_reduction): Fix size check.
1907 * gimple-pretty-print.c (dump_profile): Silence warning.
1908 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
1910 2017-11-27 Jakub Jelinek <jakub@redhat.com>
1913 * dbxout.c (lastlineno): New variable.
1914 (dbx_debug_hooks): Use dbxout_switch_text_section as
1915 switch_text_section debug hook.
1916 (dbxout_function_end): Switch to current_function_section
1917 rather than function_section. If crtl->has_bb_partition,
1918 output just one N_FUN, depending on in_cold_section_p.
1919 (dbxout_source_line): Remember last lineno in lastlineno.
1920 (dbxout_switch_text_section): New function.
1921 (dbxout_function_decl): Adjust dbxout_block caller.
1922 (dbx_block_with_cold_children): New function.
1923 (dbxout_block): Return true if any LBRAC/RBRAC have been
1924 emitted. Use dbx_block_with_cold_children at depth == 0
1925 in second partition. Add PARENT_BLOCKNUM argument, pass
1926 it optionally adjusted to children. Output LBRAC/RBRAC
1927 around recursive call only if the block is in the current
1928 partition, if not and anything was output, emit empty
1930 * final.c (final_scan_insn): Compute cold_function_name
1931 before calling switch_text_section debug hook. Call
1932 that hook even if dwarf2out_do_frame if not emitting
1936 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
1937 TREE_READONLY decls.
1939 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1941 PR rtl-optimization/82488
1942 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
1944 2017-11-26 Julia Koval <julia.koval@intel.com>
1946 * config/i386/i386.c (processor_target_table): Add skylake_cost for
1948 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
1951 2017-11-26 Julia Koval <julia.koval@intel.com>
1953 * config/i386/driver-i386.c (host_detect_local_cpu):
1954 Detect skylake-avx512.
1956 2017-11-26 Julia Koval <julia.koval@intel.com>
1958 * config.gcc: Add -march=cannonlake.
1959 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
1960 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
1961 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
1962 (PTA_CANNONLAKE): New.
1963 (processor_target_table): Add cannonlake.
1964 (ix86_option_override_internal): Ditto.
1965 (fold_builtin_cpu): Ditto.
1966 (get_builtin_code_for_version): Handle cannonlake.
1967 (M_INTEL_COREI7_CANNONLAKE): New.
1968 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
1969 * doc/invoke.texi: Add -march=cannonlake.
1971 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
1973 * plugin.c (add_new_plugin): Use platform-specific library extensions.
1974 (try_init_one_plugin): Alternative implementation for MinGW.
1975 * Makefile.in (plugin_implib): New.
1976 (gengtype-lex.c): Fix broken AIX workaround.
1977 * configure: Regenerate.
1978 * doc/plugins.texi: Document support for MinGW.
1980 2017-11-25 Jakub Jelinek <jakub@redhat.com>
1982 PR rtl-optimization/81553
1983 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
1984 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
1985 is a shift where C1 has different mode than the whole shift, use C1's
1986 mode for MULT rather than the shift's mode.
1989 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
1990 builtins not enabled in the currently selected ISA.
1992 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
1994 PR tree-optimization/71026
1995 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
1996 (insert_reciprocals): Change to insert reciprocals before a division
1997 by a square and to insert the square of a reciprocal.
1998 (execute_cse_reciprocals_1): Change to consider division by a square.
1999 (register_division_in): Add importance parameter.
2001 2017-11-24 Richard Biener <rguenther@suse.de>
2003 PR tree-optimization/82402
2004 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2005 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2007 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2009 * match.pd (0-ptr): New transformation.
2011 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2014 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2015 when skipping checking.
2017 2017-11-24 Richard Biener <rguenther@suse.de>
2019 PR tree-optimization/83128
2020 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2021 (vn_reference_lookup_3): Likewise.
2023 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2026 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2027 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2030 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2031 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2032 check for the latter with gimple_call_builtin_p. Do not handle
2033 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2035 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2037 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2040 2017-11-23 Julia Koval <julia.koval@intel.com>
2042 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2043 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2044 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2045 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2046 _mm512_maskz_expandloadu_epi16): New intrinsics.
2047 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2048 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2049 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2050 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2051 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2052 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2053 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2054 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2055 _mm256_maskz_expandloadu_epi8): New intrinsics.
2056 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2057 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2058 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2059 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2060 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2061 * config/i386/sse.md (VI248_VLBW): New iterator.
2062 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2064 2017-11-23 Julia Koval <julia.koval@intel.com>
2066 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2067 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2068 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2069 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2070 _mm512_mask_compressstoreu_epi16): New.
2071 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2072 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2073 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2074 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2075 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2076 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2077 _mm256_mask_compressstoreu_epi8): New.
2078 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2079 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2080 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2081 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2082 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2083 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2084 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2085 __builtin_ia32_compresshi128_mask,
2086 __builtin_ia32_compressstoreuqi512_mask,
2087 __builtin_ia32_compressstoreuhi512_mask,
2088 __builtin_ia32_compressstoreuqi256_mask,
2089 __builtin_ia32_compressstoreuqi128_mask,
2090 __builtin_ia32_compressstoreuhi256_mask,
2091 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2092 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2094 (ix86_expand_special_args_builtin): Handle new types.
2095 (s4fma_expand): Handle new builtin array.
2096 * config/i386/immintrin.h: Include new headers.
2097 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2098 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2100 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2102 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2103 fixes. Declare temp and g variables at the top in order to avoid
2104 {} in most of the cases.
2106 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2108 * match.pd (ptr-0): New transformation.
2110 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2112 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2113 (struct addr_mode_cost_table): New.
2114 (struct tune_params): Add field addr_mode_costs.
2115 * config/arm/arm.c (generic_addr_mode_costs): New.
2116 (arm_slowmul_tune): Initialise addr_mode_costs field.
2117 (arm_fastmul_tune): Likewise.
2118 (arm_strongarm_tune): Likewise.
2119 (arm_xscale_tune): Likewise.
2120 (arm_9e_tune): Likewise.
2121 (arm_marvell_pj4_tune): Likewise.
2122 (arm_v6t2_tune): Likewise.
2123 (arm_cortex_tune): Likewise.
2124 (arm_cortex_a8_tune): Likewise.
2125 (arm_cortex_a7_tune): Likewise.
2126 (arm_cortex_a15_tune): Likewise.
2127 (arm_cortex_a35_tune): Likewise.
2128 (arm_cortex_a53_tune): Likewise.
2129 (arm_cortex_a57_tune): Likewise.
2130 (arm_exynosm1_tune): Likewise.
2131 (arm_xgene1_tune): Likewise.
2132 (arm_cortex_a5_tune): Likewise.
2133 (arm_cortex_a9_tune): Likewise.
2134 (arm_cortex_a12_tune): Likewise.
2135 (arm_cortex_a73_tune): Likewise.
2136 (arm_v7m_tune): Likewise.
2137 (arm_cortex_m7_tune): Likewise.
2138 (arm_v6m_tune): Likewise.
2139 (arm_fa726te_tune): Likewise.
2140 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2142 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2144 * config/arm/arm.c (arm_mem_costs): New function.
2145 (arm_rtx_costs_internal): Use arm_mem_costs.
2147 2017-11-23 Mark Wielaard <mark@klomp.org>
2149 * dwarf2out.c (init_sections_and_labels): Use generation to create
2150 unique ranges_section_label and ranges_base_label. Return generation.
2151 (output_rnglists): Add generation argument. Use generation to create
2152 unique ranges labels.
2153 (dwarf2out_finish): Get generation from init_sections_and_labels
2154 and pass generation to output_rnglists.
2156 2017-11-23 Mike Stump <mikestump@comcast.net>
2157 Eric Botcazou <ebotcazou@adacore.com>
2159 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2160 * cfgloop.h (struct loop): Add unroll field.
2161 * function.h (struct function): Add has_unroll bitfield.
2162 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2163 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2164 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2165 (pass_rtl_unroll_loops::gate): Likewise.
2166 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2167 for which loop->unroll==1.
2168 (decide_unroll_constant_iterations): Use note for consistency and
2169 take loop->unroll into account. Return early if loop->unroll is set.
2170 Fix thinko in existing test.
2171 (decide_unroll_runtime_iterations): Use note for consistency and
2172 take loop->unroll into account.
2173 (decide_unroll_stupid): Likewise.
2174 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2175 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2176 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2178 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2179 (print_loop): Print loop->unroll if set.
2180 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2181 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2182 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2184 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2185 loop->unroll is set and smaller than the trip count. Otherwise bypass
2186 entirely the heuristics if loop->unroll is set. Remove dead note.
2187 Fix off-by-one bug in other note.
2188 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2189 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2191 (tree_unroll_loops_completely): Make static.
2192 (pass_complete_unroll::execute): Use correct type for variable.
2193 (pass_complete_unrolli::execute): Fix formatting.
2194 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2196 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2198 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2199 enable when TARGET_PREFER_AVX128 is set.
2201 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2203 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2205 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2206 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2208 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2210 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2211 frequencies. Fix estimation of aggregate parameters.
2213 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2215 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2218 2017-11-23 Tom de Vries <tom@codesourcery.com>
2220 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2222 2017-11-23 Tom de Vries <tom@codesourcery.com>
2224 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2227 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2230 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2231 sibcall_value_pcrel_fdpic): Use local variable instead of
2233 (calli_tbr_rel): Add missing operand 2.
2234 (call_valuei_tbr_rel): Add missing operand 3.
2236 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2239 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2240 bitpos/bitsize covering the whole destination, use store_expr only if
2241 the complex mode is the same. Otherwise, use expand_normal and if
2242 it returns CONCAT, subreg each part separately instead of trying to
2243 subreg the whole result.
2245 2017-11-23 Richard Biener <rguenther@suse.de>
2247 PR tree-optimization/23094
2248 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2249 come here from walking over backedges in the first iteration.
2250 (vn_reference_lookup_3): Skip clobbers that store the same value.
2252 2017-11-23 Richard Biener <rguenther@suse.de>
2254 PR tree-optimization/81403
2255 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2256 a block we need a leader relative to.
2257 (phi_translate_1): For nary processing require a leader from
2258 get_representative_for given we run expression simplification
2259 using match-and-simplify. Remove previous fix.
2261 2017-11-22 Jeff Law <law@redhat.com>
2263 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2264 Use new method allocate_value_range rather than accessing the
2265 vrp_value_range_pool data member directly.
2266 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2267 to use extract_range_from_stmt method to avoid need for
2268 extract_range_from_assignment method.
2269 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2270 method rather than setting values_propgated data member directly.
2271 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2272 and values propagated data members and extract_range_from_assignment
2273 method. Reorder private data members to conform to standards.
2274 Add new methods set_lattice_propagation_complete and
2275 allocate_value_range.
2277 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2279 PR rtl-optimization/83030
2280 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2281 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2282 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2283 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2284 CROSSING_JUMP_P flag is already set before setting it.
2285 * cfgrtl.c (fixup_partition_crossing): Likewise.
2286 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2289 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2291 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2292 Handle the case where both arguments are using gen_const_vec_series.
2294 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2297 * pretty-print.c (pp_format): Move quoting implementation to
2298 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2299 to pass address of "quote" local.
2300 (pp_begin_quote): New function.
2301 (pp_end_quote): New function.
2302 * pretty-print.h (printer_fn): Convert penultimate param from bool
2304 (pp_begin_quote): New decl.
2305 (pp_end_quote): New decl.
2306 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2307 param from bool to bool *.
2308 * tree-diagnostic.h (default_tree_printer): Likewise.
2310 2017-11-22 Jeff Law <law@redhat.com>
2312 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2313 Initialize vr_values.
2314 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2315 vr_values via vr_values class instance rather than delegators.
2316 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2317 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2318 (evrp_range_analyzer::push_value_range): Likewise.
2319 (evrp_range_analyzer::pop_value_range): Likewise.
2320 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2321 most delegators. Those remaining are exposed as public interfaces.
2322 Make vr_values a pointer and private.
2323 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2325 (evrp_range_analyzer::get_vr_value): New method.
2326 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2327 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2328 (class evrp_dom_walker): Attach evrp_folder class, initialize
2329 it in the ctor. Remove temporary delegators.
2330 (evrp_dom_walker::before_dom_children): Call methods in attached
2331 evrp_range_analyzer class via class instance pointer. Use
2332 free value_range_constant_singleton to remove need for
2333 op_with_constant_singleton_value delegator method. Do not
2334 create a vrp_prop class instance for every call! Narrow
2335 scope of a couple variables.
2336 (evrp_dom_walker::cleanup): Call methods in attached
2337 evrp_range_analyzer class via class instance pointer.
2338 * vr-values.h (class vr_values): Privatize many methods and
2341 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2343 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2344 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2345 * expr.c (expand_expr_real_2): Likewise.
2346 * fold-const.c (const_unop): Likewise.
2347 * optabs-tree.c (optab_for_tree_code): Likewise.
2348 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2349 * tree-inline.c (estimate_operator_cost): Likewise.
2350 * tree-pretty-print.c (dump_generic_node): Likewise.
2351 (op_code_prio): Likewise.
2352 (op_symbol_code): Likewise.
2353 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2354 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2355 * internal-fn.c (direct_internal_fn_optab): New function.
2356 (direct_internal_fn_array, direct_internal_fn_supported_p
2357 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2358 * fold-const-call.c (fold_const_reduction): New function.
2359 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2361 * tree-vect-loop.c: Include internal-fn.h.
2362 (reduction_code_for_scalar_code): Rename to...
2363 (reduction_fn_for_scalar_code): ...this and return an internal
2365 (vect_model_reduction_cost): Take an internal_fn rather than
2367 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2369 (vectorizable_reduction): Use internal functions rather than tree
2370 codes for the reduction operation. Update calls to the functions
2372 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2373 Use calls to internal functions rather than REDUC tree codes.
2374 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2376 2017-11-22 Olivier Hainque <hainque@adacore.com>
2378 * config/vxworks.c (vxworks_override_options): Pick default
2379 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2380 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2381 DWARF_GNAT_ENCODINGS_DEFAULT.
2382 * config/vxworksae.h: Likewise.
2384 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2386 PR tree-optimization/83104
2387 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2390 2017-11-22 Marek Polacek <polacek@redhat.com>
2391 H.J. Lu <hongjiu.lu@intel.com>
2392 Jason Merrill <jason@redhat.com>
2397 * calls.c (initialize_argument_information): Call
2398 warn_parameter_passing_abi target hook.
2399 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2400 argument onto stack.
2401 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2402 * common.opt: Update -fabi-version description.
2403 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2404 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2406 (ix86_is_empty_record): New function.
2407 (ix86_warn_parameter_passing_abi): New function.
2408 (TARGET_EMPTY_RECORD_P): Redefine.
2409 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2410 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2411 * doc/tm.texi: Regenerated.
2412 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2413 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2414 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2415 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2416 instead of int_size_in_bytes.
2417 * expr.c (copy_blkmode_to_reg): Likewise.
2418 * function.c (aggregate_value_p): Return 0 for empty types.
2419 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2421 (locate_and_pad_parm): Call arg size_in_bytes instead
2423 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2424 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2425 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2427 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2428 (std_gimplify_va_arg_expr): Skip empty records. Call
2429 arg_size_in_bytes instead size_in_bytes.
2430 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2431 * tree-core.h (tree_type_common): Add empty_flag.
2432 (tree_decl_common): Update comments.
2433 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2435 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2436 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2438 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2439 * tree.c (default_is_empty_type): New function.
2440 (default_is_empty_record): New function.
2441 (arg_int_size_in_bytes): New function.
2442 (arg_size_in_bytes): New function.
2443 (get_ultimate_context): New function.
2444 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2445 TRANSLATION_UNIT_WARN_EMPTY_P.
2446 (default_is_empty_record, arg_int_size_in_bytes,
2447 arg_size_in_bytes, get_ultimate_context): Declare.
2449 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2451 * config/arm/arm.c (cmse_clear_registers): New function.
2452 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2453 code by call to cmse_clear_registers.
2454 (cmse_nonsecure_entry_clear_before_return): Likewise.
2456 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2458 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2459 (vdot_s32, vdotq_s32): New.
2460 (vdot_lane_u32, vdotq_lane_u32): New.
2461 (vdot_lane_s32, vdotq_lane_s32): New.
2464 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2467 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2468 for unsigned values with fewer HWIs than the precision.
2469 (test_overflow): New function.
2470 (wide_int_cc_tests): Call it.
2472 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2474 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2475 has the same mode class as Pmode.
2477 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2479 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2480 padding_bits_to_clear_ptr.
2481 (cmse_nonsecure_entry_clear_before_return): Likewise.
2483 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2485 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2486 auto_sbitap instead of integer bitfield to control register needing
2489 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2491 PR tree-optimization/83044
2492 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2493 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2494 that need the upper bound. Subtract offset from
2495 get_addr_base_and_unit_offset only if positive and subtract it
2496 before division by eltsize rather than after it.
2499 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2500 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2501 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2505 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2506 expand_binop with *mul_widen_optab, make sure at least one of the
2507 operands doesn't have VOIDmode.
2510 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2512 PR rtl-optimization/82044
2513 PR tree-optimization/82042
2514 * dse.c (record_store): Check for overflow.
2515 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2516 add_wild_read instead of clear_rhs_from_active_local_stores on
2517 overflow and log it into dump_file.
2519 2017-11-22 Richard Biener <rguenther@suse.de>
2521 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2522 fake edges to exit when looking for a place to insert.
2523 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2525 (insert_into_preds_of_block): Commit edge insertion immediately,
2526 assert that doesn't require new BBs.
2527 (fini_pre): Release expressions.
2528 (pass_pre::execute): Shuffle things around a bit, if the fn
2529 is too large do not compute AVAIL either as this is really the
2532 2017-11-22 Richard Biener <rguenther@suse.de>
2534 PR tree-optimization/83089
2535 * tree-if-conv.c (pass_if_conversion::execute): If anything
2536 changed reset SCEV and free the number of iteration estimates.
2538 2017-11-21 Martin Sebor <msebor@redhat.com>
2540 PR tree-optimization/82945
2541 * calls.h (warn_nonstring_bound): Remove unused function.
2543 2017-11-21 Martin Sebor <msebor@redhat.com>
2545 PR tree-optimization/82945
2546 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2547 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2548 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2550 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2551 * calls.h (get_attr_nonstring_decl): Declare new function.
2552 * doc/extend.texi (attribute nonstring): Update.
2553 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2554 get_attr_nonstring_decl and handle it.
2555 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2556 detection of nul-termination.
2557 (strlen_to_stridx): Change to a pointer.
2558 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2559 (pass_strlen::execute): Same.
2561 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2563 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2564 for the new option -mprefer-vector-width=[none|128|256|512].
2565 * config/i386/i386.c (ix86_target_string): remove old style options
2566 -mprefer-avx256 and make -mprefer-avx128 as alias.
2567 (ix86_option_override_internal): Apply defaults for the
2568 -mprefer-vector-width=[128|256] option.
2569 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2570 Implement macros to work with -mprefer-vector-width=.
2571 * config/i386/i386.opt: Implemented option
2572 -mprefer-vector-width=[none|128|256|512].
2573 * doc/invoke.texi: Documentation for
2574 -mprefer-vector-width=[none|128|256|512].
2576 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2578 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2580 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2582 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2584 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2586 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2588 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2589 * expr.c (expand_expr_real_2): Likewise.
2590 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2591 fold_binary_loc): Likewise.
2592 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2593 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2594 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2595 MINUS_EXPR transformations.
2596 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2597 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2598 * tree-inline.c (estimate_operator_cost): Likewise.
2599 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2600 op_symbol_code): Likewise.
2601 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2602 * vr-values.c (extract_range_from_binary_expr): Likewise.
2603 * varasm.c (initializer_constant_valid_p_1): Likewise.
2604 * tree.def: New tree code POINTER_DIFF_EXPR.
2606 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2608 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2609 integer suffix to movbe mnemonic.
2610 (*bswaphi2_movbe): Ditto.
2611 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2613 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2616 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2617 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2618 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2620 2017-11-21 Martin Liska <mliska@suse.cz>
2622 * tree-inline.c (expand_call_inline): Remove not needed
2625 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2626 Jakub Jelinek <jakub@redhat.com>
2629 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2630 Remove static keyword from f variable.
2632 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2634 PR tree-optimization/83086
2635 * gimple-ssa-store-merging.c
2636 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2637 rather than n.base_addr.
2639 2017-11-21 Martin Liska <mliska@suse.cz>
2641 PR rtl-optimization/82044
2642 PR tree-optimization/82042
2643 * dse.c (check_mem_read_rtx): Check for overflow.
2645 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2647 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2650 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2652 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2653 and its corresponding call arg location note.
2655 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2656 Andrew Burgess <andrew.burgess@embecosm.com>
2658 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2660 (arc_return_slot_offset): Likewise.
2661 (arc_eh_return_address_location): New declaration.
2662 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2663 (MUST_SAVE_REGISTER): Add exception handler case.
2664 (MUST_SAVE_RETURN_ADDR): Likewise.
2665 (arc_frame_pointer_required): Likewise.
2666 (arc_frame_pointer_needed): New function.
2667 (arc_compute_frame_size): Changed.
2668 (arc_expand_prologue): Likewise.
2669 (arc_expand_epilogue): Likewise.
2670 (arc_initial_elimination_offset): Likewise.
2671 (arc_return_slot_offset): Delete.
2672 (arc_eh_return_address_location): New function.
2673 (arc_builtin_setjmp_frame_value): Likewise.
2674 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2675 (EH_RETURN_STACKADJ_RTX): Define.
2676 (EH_RETURN_HANDLER_RTX): Likewise.
2677 * config/arc/arc.md (eh_return): Delete.
2679 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2681 * print-tree.h (debug_vec_tree): Remove prototype.
2682 * gdbinit.in (pvt): Remove macro.
2684 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2686 PR tree-optimization/83047
2687 * gimple-ssa-store-merging.c
2688 (imm_store_chain_info::output_merged_store): If the loads with the
2689 same vuse are in different basic blocks, for load_gsi pick a load
2690 location that is dominated by the other loads.
2693 * config/i386/i386.c (ix86_memmodel_check): Start
2694 -Winvalid-memory-model diagnostics with lowercase letter.
2697 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2698 set in_cold_section_p to the partition containing loc_list->first.
2699 When seeing loc_list->last_before_switch node, update secname and
2700 perform range_across_switch second partition handling only after that.
2703 * run-rtl-passes.c: Include debug.h.
2704 (run_rtl_passes): Call debug_hooks->assembly_start.
2705 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2709 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2710 OPTAB_DIRECT in calls to expand_simple_binop.
2712 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2715 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2717 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2720 * spellcheck.h (best_match::blithely_get_best_candidate): New
2723 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2725 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2726 with lower case letter.
2728 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2730 * config/i386/i386.md (bswaphi2): New expander.
2731 (*bswaphi2_movbe): New insn pattern.
2732 (bswaphi -> rorhi pepehole2): New peephole pattern.
2734 2017-11-20 Jeff Law <law@redhat.com>
2736 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2737 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2738 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2739 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2740 gimple-ssa-evrp-analyze.h.
2742 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2743 set BB_VISITED here.
2744 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2746 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2748 * tree-predcom.c: Add general comment on Store-Store chains.
2749 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2751 (get_chain_last_ref_at): Rename into...
2752 (get_chain_last_write_at): ...this.
2753 (get_chain_last_write_before_load): New function.
2754 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2755 CT_STORE_STORE when write reference is added.
2756 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2757 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2758 (initialize_root_vars_store_elim_1): Ditto.
2759 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2760 definition is created.
2761 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2762 chain by replacing it with dominant stored value.
2764 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2766 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2768 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2770 * vr-values.c (extract_range_from_binary_expr): Use a full range
2773 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2775 * config/arm/arm.md (R4_REGNUM): Define constant.
2776 (nonsecure_call_internal): Remove r4 clobber.
2777 (nonsecure_call_value_internal): Likewise.
2778 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2779 clobber and resequence match_operands.
2780 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2781 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2782 (nonsecure_call_value_reg_thumb2): Likewise.
2784 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2786 PR tree-optimization/78821
2787 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2788 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2789 to the address of the base rather than the base itself.
2790 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2791 (find_bswap_or_nop_finalize): New function.
2792 (find_bswap_or_nop): Use it.
2793 (bswap_replace): Return a tree rather than bool, change first
2794 argument from gimple * to gimple_stmt_iterator, allow inserting
2795 into an empty sequence, allow ins_stmt to be NULL - then emit
2796 all stmts into gsi. Fix up MEM_REF address gimplification.
2797 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2798 (struct store_immediate_info): Add N and INS_STMT non-static
2800 (store_immediate_info::store_immediate_info): Initialize them
2801 from newly added ctor args.
2802 (merged_store_group::apply_stores): Formatting fixes. Sort by
2804 (stmts_may_clobber_ref_p): For stores call also
2805 refs_anti_dependent_p.
2806 (gather_bswap_load_refs): New function.
2807 (imm_store_chain_info::try_coalesce_bswap): New method.
2808 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2809 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2810 (imm_store_chain_info::output_merged_store): Fail if number of
2811 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2812 and NOP_EXPR rhs_code.
2813 (pass_store_merging::process_store): Compute n and ins_stmt, if
2814 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
2815 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
2817 (pass_store_merging::execute): Calculate dominators.
2819 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
2820 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
2821 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
2822 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2823 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
2824 class pass_optimize_bswap, bswap_replace,
2825 pass_optimize_bswap::execute): Moved to ...
2826 * gimple-ssa-store-merging.c: ... this file.
2827 Include optabs-tree.h.
2828 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
2829 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2830 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
2831 anonymous namespace, remove static keywords.
2832 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
2833 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
2835 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
2838 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
2840 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
2842 * vec.h (debug_helper): New function.
2843 (DEFINE_DEBUG_VEC): New macro.
2844 * hash-set.h (debug_helper): New function.
2845 (DEFINE_DEBUG_HASH_SET): New macro.
2846 * cfg.c (debug_slim (edge)): New function.
2847 Call DEFINE_DEBUG_VEC for edges.
2848 Call DEFINE_DEBUG_HASH_SET for edges.
2849 * cfghooks.c (debug_slim (basic_block)): New function.
2850 Call DEFINE_DEBUG_VEC for basic blocks.
2851 Call DEFINE_DEBUG_HASH_SET for basic blocks.
2852 * print-tree.c (debug_slim): New function to handle trees.
2853 Call DEFINE_DEBUG_VEC for trees.
2854 Call DEFINE_DEBUG_HASH_SET for trees.
2855 (debug (vec<tree, va_gc>) &): Remove.
2856 (debug (<vec<tree, va_gc>) *): Remove.
2857 * print-rtl.c (debug_slim): New function to handle const_rtx.
2858 Call DEFINE_DEBUG_VEC for rtx_def.
2859 Call DEFINE_DEBUG_VEC for rtx_insn.
2860 Call DEFINE_DEBUG_HASH_SET for rtx_def.
2861 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
2862 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
2863 (debug (vec<rtx_insn *> *ptr): Remove.
2864 (debug_insn_vector): Remove.
2865 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
2867 2017-11-20 Tom de Vries <tom@codesourcery.com>
2869 PR rtl-optimization/82020
2870 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
2871 IF_THEN_ELSE condition.
2873 2017-11-19 Jeff Law <law@redhat.com>
2875 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2876 of degenerates resulting from ignoring an edge.
2878 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2881 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
2883 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2886 * profile-count.c (profile_count::to_sreal_scale): Fix return value
2887 for uninitialied counts.
2889 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2892 * tree-inline.c (estimate_num_insns): Set to 1 at least.
2894 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2897 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
2899 2017-11-19 Tom de Vries <tom@codesourcery.com>
2901 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
2904 2017-11-19 Tom de Vries <tom@codesourcery.com>
2906 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
2909 2017-11-19 Tom de Vries <tom@codesourcery.com>
2911 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2914 2017-11-19 Tom de Vries <tom@codesourcery.com>
2916 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
2918 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
2920 2017-11-19 Tom de Vries <tom@codesourcery.com>
2922 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
2924 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
2925 semicolon after MCORE_EXPORT_NAME call.
2927 2017-11-19 Tom de Vries <tom@codesourcery.com>
2930 * vmsdbgout.c (vmsdbgout_early_finish): New function.
2931 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
2933 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
2935 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
2936 (cgraph_edge::clone): Cleanup updating of profile.
2937 * ipa-cp.c (update_profiling_info): Likewise.
2938 * ipa-inline-transform.c (inline_transform): Likewise.
2939 * ipa-inline.c (inline_small_functions): Add missing space to dump.
2940 * ipa-split.c (execute_split_functions): Do not split when function
2942 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
2943 * profile-count.c (profile_count::dump): Add global0.
2944 (profile_count::to_cgraph_frequency): Do not ICE when entry is
2946 (profile_count::to_sreal_scale): Likewise.
2947 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
2948 (profile_count::combine_with_ipa_count): New function.
2949 * profile-count.h (profile_guessed_global0adjusted): New.
2950 (profile_count::adjusted_zero): New.
2951 (profile_count::global0adjusted): New.
2952 (profile_count::combine_with_ipa_count): New.
2953 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
2954 correct profile of return block of split functions.
2955 (copy_cfg_body): Remove unused profile_count.
2956 (copy_body): Likewise.
2957 (expand_call_inline): Update.
2958 (tree_function_versioning): Update.
2960 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
2962 * hash-set.h (hash_set::empty): New.
2963 * tree-ssa-threadbackward.h: Delete.
2964 * tree-ssa-threadbackward.c (class thread_jumps): New.
2965 Move max_threaded_paths into class.
2966 (fsm_find_thread_path): Remove arguments that are now in class.
2967 (profitable_jump_thread_path): Rename to...
2968 (thread_jumps::profitable_jump_thread_path): ...this.
2969 (convert_and_register_jump_thread_path): Rename to...
2970 (thread_jumps::convert_and_register_current_path): ...this.
2971 (check_subpath_and_update_thread_path): Rename to...
2972 (thread_jumps::check_subpath_and_update_thread_path): ...this.
2973 (register_jump_thread_path_if_profitable): Rename to...
2974 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
2975 (handle_phi): Rename to...
2976 (thread_jumps::handle_phi): ...this.
2977 (handle_assignment): Rename to...
2978 (thread_jumps::handle_assignment): ...this.
2979 (fsm_find_control_statement_thread_paths): Rename to...
2980 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
2981 (find_jump_threads_backwards): Rename to...
2982 (thread_jumps::find_jump_threads_backwards): ...this.
2983 Initialize path local data.
2984 (pass_thread_jumps::execute): Call find_jump_threads_backwards
2985 from within thread_jumps class.
2986 (pass_early_thread_jumps::execute): Same.
2988 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2990 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
2992 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2994 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
2995 to canonical location.
2997 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3000 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3002 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3004 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3006 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3007 rcrt1.o%s/grcrt1.o%s for -static-pie.
3009 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3011 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3012 ix86_shift_rotate_cost): Break out from ...
3013 (ix86_rtx_costs): ... here.
3014 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3017 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3019 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3020 when entry block was promoted unlikely.
3021 (estimate_bb_frequencies): Increase frequency scale.
3022 * profile-count.h (profile_count): Export precision info.
3024 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3026 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3027 disturbin profile of entry block.
3029 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3031 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3034 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3036 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3037 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3039 2017-11-17 Jeff Law <law@redhat.com>
3041 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3042 from evrp_dom_walker class. Various methods moved into new class.
3043 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3044 (evrp_range_analyzer::enter): New method.
3045 (evrp_range_analyzer::leave): New method.
3046 (evrp_dom_walker): Remove delegators no longer needed by this class.
3047 Replace vr_values data member with evrp_range_analyzer
3049 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3050 method extracted from evrp_dom_walker::before_dom_children.
3051 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3052 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3054 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3055 Add private copy constructor and move assignment operators.
3056 Privatize methods and class data where trivially possible.
3057 (evrp_dom_walker::cleanup): New function, extracted from
3058 execute_early_vrp. Simplify access to class data.
3060 * vr-values.h (get_output_for_vrp): Prototype.
3061 * vr-values.c (get_output_for_vrp): New function extracted from
3062 vrp_visit_assignment_or_call and extract_range_from_stmt.
3063 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3065 2017-11-17 Luis Machado <luis.machado@linaro.org>
3067 * config/aarch64/aarch64.c
3068 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3069 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3070 tune_params::AUTOPREFETCHER_WEAK.
3072 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3075 * config/arm/arm.c (arm_valid_target_attribute_rec):
3076 Parse "arch=" and "+<ext>".
3077 (arm_valid_target_attribute_tree): Re-init global options.
3078 (arm_option_override): Make non-static.
3079 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3080 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3081 (__ARM_FEATURE_CRC32): Support undef.
3082 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3083 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3085 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3087 * gdbinit.in (break-on-diagnostic): New command.
3089 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3091 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3092 return to indirect jump for EH return if control-flow
3093 protection is enabled. Change explicit 'false' argument in
3094 pro_epilogue_adjust_stack with a value of
3096 * config/i386/i386.md (simple_return_indirect_internal):
3097 Remove SImode restriction to support 64-bit.
3099 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3101 * combine.c (added_notes_insn): New.
3102 (try_combine): Handle added_notes_insn like added_links_insn.
3103 Rewrite return value code.
3104 (distribute_notes): Set added_notes_insn to the earliest insn we added
3107 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3109 PR rtl-optimization/82621
3110 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3111 dest of one of those SETs is unused.
3113 2017-11-17 Richard Biener <rguenther@suse.de>
3116 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3117 * tree-pretty-print.c (dump_generic_node): Handle
3118 annot_expr_parallel_kind.
3119 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3120 * gimplify.c (gimple_boolify): Likewise.
3122 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3124 * config.gcc (extra_headers): Add cet.h for x86 targets.
3125 * config/i386/cet.h: New file.
3126 * doc/install.texi: Add --enable-cet/--disable-cet.
3128 2017-11-17 Richard Biener <rguenther@suse.de>
3130 PR tree-optimization/83017
3131 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3132 (gen_parallel_loop): Properly count iterations.
3133 (parallelize_loops): Handle loop->can_be_parallel independent
3134 of flag_loop_parallelize_all. Make static profitability test match
3136 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3137 * invoke.texi (parloops-min-per-thread): Document.
3139 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3141 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3142 upstreaming review comments.
3144 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3146 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3147 with fast unaligned access.
3148 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3150 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3152 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3153 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3155 2017-11-17 Richard Biener <rguenther@suse.de>
3157 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3158 folding of references.
3160 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3163 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3164 of replacing call to strncmp with corresponding call to strcmp when
3167 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3169 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3170 option prefer-avx256 for skylake-avx512 configuration.
3171 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3172 (get_builtin_code_for_version): Ditto.
3174 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3175 Monk Chiang <sh.chiang04@gmail.com>
3177 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3178 (FIXED_REGISTERS): Reserve more register numbers.
3179 (CALL_USED_REGISTERS): Likewise.
3180 (REG_ALLOC_ORDER): Likewise.
3181 (REG_CLASS_CONTENTS): Likewise.
3182 (REGISTER_NAMES): Likewise.
3184 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3185 Kito Cheng <kito.cheng@gmail.com>
3187 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3189 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3191 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3193 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3194 parameter list for vec_splats.
3196 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3198 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3200 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3202 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3205 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3206 with the required version of automake 1.11.6.
3208 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3210 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3212 (DU_C2_3_power9): Correct reservation combinations.
3213 (FP_DIV_power9, VEC_DIV_power9): New.
3214 (power9-alu): Split out rotate/shift...
3215 (power9-rot): ...to here, correct dispatch resource.
3216 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3218 (power9-fp): Correct latency.
3219 (power9-sdiv): Add div/sqrt resource.
3220 (power9-ddiv): Correct latency, add div/sqrt resource.
3221 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3222 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3224 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3226 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3228 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3229 switch statement mapping KF built-ins to TF built-ins if we don't
3230 have the proper ISA 3.0 assembler support.
3232 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3234 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3235 (lower_emutls_function_body): Do not compute it.
3237 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3239 * ipa-split.c (split_bb_info): Turn time to sreal.
3240 (split_point): Likewise.
3241 (dump_split_point): Likewise.
3242 (fine_split_points): Likewise.
3243 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3245 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3247 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3250 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3252 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3254 (expensive_function_p): Remove useless assert.
3255 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3257 2017-11-16 Martin Sebor <msebor@redhat.com>
3259 PR tree-optimization/82588
3260 PR tree-optimization/82583
3261 * tree-vrp.c (check_array_ref): Handle flexible array members,
3262 string literals, and inner indices.
3263 (search_for_addr_array): Add detail to diagnostics.
3265 2017-11-16 Nathan Sidwell <nathan@acm.org>
3269 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3270 (SET_DECL_ASSEMBLER_NAME): Forward to
3271 overwrite_decl_assembler_name.
3272 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3273 (overwrite_decl_assembler_name): Declare.
3274 * tree.c (overwrite_decl_assembler_name): New.
3275 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3276 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3277 (LANG_HOOKS_INITIALIZER): Add it.
3278 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3279 * langhooks.c (lhd_set_decl_assembler_name): Use
3280 SET_DECL_ASSEMBLER_NAME.
3281 (lhd_overwrite_decl_assembler_name): Default implementation.
3283 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3284 Jackson Woodruff <jackson.woodruff@arm.com>
3286 PR tree-optimization/71026
3287 * match.pd: Canonicalize constant multiplies in division.
3289 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3291 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3292 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3293 Remove OPT_fomit_frame_pointer entry.
3294 * common/config/arc/arc-common.c: Likewise.
3295 * common/config/arm/arm-common.c: Likewise.
3296 * common/config/avr/avr-common.c: Likewise.
3297 * common/config/c6x/c6x-common.c: Likewise.
3298 * common/config/cr16/cr16-common.c: Likewise.
3299 * common/config/cris/cris-common.c: Likewise.
3300 * common/config/epiphany/epiphany-common.c: Likewise.
3301 * common/config/fr30/fr30-common.c: Likewise.
3302 * common/config/frv/frv-common.c: Likewise.
3303 * common/config/ia64/ia64-common.c: Likewise.
3304 * common/config/iq2000/iq2000-common.c: Likewise.
3305 * common/config/lm32/lm32-common.c: Likewise.
3306 * common/config/m32r/m32r-common.c: Likewise.
3307 * common/config/mcore/mcore-common.c: Likewise.
3308 * common/config/microblaze/microblaze-common.c: Likewise.
3309 * common/config/mips/mips-common.c: Likewise.
3310 * common/config/mmix/mmix-common.c: Likewise.
3311 * common/config/mn10300/mn10300-common.c: Likewise.
3312 * common/config/nios2/nios2-common.c: Likewise.
3313 * common/config/pa/pa-common.c: Likewise.
3314 * common/config/pdp11/pdp11-common.c: Likewise.
3315 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3316 * common/config/riscv/riscv-common.c: Likewise.
3317 * common/config/rs6000/rs6000-common.c: Likewise.
3318 * common/config/rx/rx-common.c: Likewise.
3319 * common/config/s390/s390-common.c: Likewise.
3320 * common/config/sh/sh-common.c: Likewise.
3321 * common/config/sparc/sparc-common.c: Likewise.
3322 * common/config/tilegx/tilegx-common.c: Likewise.
3323 * common/config/tilepro/tilepro-common.c: Likewise.
3324 * common/config/v850/v850-common.c: Likewise.
3325 * common/config/visium/visium-common.c: Likewise.
3326 * common/config/xstormy16/xstormy16-common.c: Likewise.
3327 * common/config/xtensa/xtensa-common.c: Likewise.
3328 * invoke.texi (-fomit-frame-pointer): Update documentation.
3330 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3332 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3334 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3336 * cfg.c (scale_bbs_frequencies_int,
3337 cale_bbs_frequencies_gcov_type): Remove.
3338 * cfg.h (scale_bbs_frequencies_int,
3339 cale_bbs_frequencies_gcov_type): Remove.
3341 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3343 * tree-ssa-loop-manip.c
3344 (scale_dominated_blocks_in_loop): Update to profile counts.
3345 (tree_transform_and_unroll_loop): Likewise.
3347 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3349 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3350 scale_bbs_frequencies_int.
3352 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3354 * final.c (compute_alignments): Use counts rather than frequencies.
3356 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3358 * cfgloopanal.c: Include sreal.h
3359 (average_num_loop_insns): Use counts and sreal for accounting.
3361 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3363 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3366 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3367 Kito Cheng <kito.cheng@gmail.com>
3369 * config/nds32/constraints.md: Provide more constraints.
3370 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3371 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3372 support constraints usage.
3374 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3376 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3378 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3379 Kito Cheng <kito.cheng@gmail.com>
3381 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3382 * config/nds32/nds32.opt: Refine the layout.
3383 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3384 TARGET_EXT_STRING): Support new options.
3385 * config/nds32/nds32.h: Likewise.
3386 * config/nds32/nds32.md: Likewise.
3387 * config/nds32/nds32-predicates.c: Likewise.
3388 * config/nds32/constraints.md: Likewise.
3389 * common/config/nds32/nds32-common.c: Likewise.
3391 2017-11-16 Julia Koval <julia.koval@intel.com>
3394 * config/i386/gfniintrin.h: Add sse check.
3395 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3397 2017-11-16 Julia Koval <julia.koval@intel.com>
3399 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3400 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3401 (ix86_handle_option): Handle -mavx512vbmi2.
3402 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3403 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3404 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3405 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3406 (ix86_valid_target_attribute_inner_p): Ditto.
3407 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3408 * config/i386/i386.opt (mavx512vbmi2): New option.
3409 * doc/invoke.texi: Add new option.
3411 2017-11-16 Julia Koval <julia.koval@intel.com>
3413 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3414 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3415 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3416 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3417 _mm512_gf2p8mul_epi8): New intrinsics.
3418 * config/i386/i386-builtin-types.def
3419 (V64QI_FTYPE_V64QI_V64QI): New type.
3420 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3421 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3422 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3423 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3424 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3425 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3427 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3429 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3430 explicitly as a stream of bytes.
3432 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3434 * config/rs6000/altivec.h (vec_xst_be): New #define.
3435 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3436 and externalize from *altivec_vperm_<mode>_internal.
3437 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3439 (XL_BE_V8HI): Likewise.
3440 (XL_BE_V4SI): Likewise.
3441 (XL_BE_V4SI): Likewise.
3442 (XL_BE_V2DI): Likewise.
3443 (XL_BE_V4SF): Likewise.
3444 (XL_BE_V2DF): Likewise.
3445 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3446 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3447 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3448 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3449 VSX_BUILTIN_VEC_XST_BE.
3450 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3451 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3453 (altivec_init_builtins): Replace conditional calls to def_builtin
3454 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3455 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3456 with unconditional calls. Remove calls to def_builtin for
3457 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3458 __builtin_vec_xst_be.
3459 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3460 to define_expand, and add alternate RTL generation for P8.
3461 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3462 vsx_ld_elemrev_v8hi.
3463 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3464 add alternate RTL generation for P8.
3465 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3466 vsx_ld_elemrev_v16qi.
3467 (vsx_st_elemrev_v8hi): Convert define_insn
3468 to define_expand, and add alternate RTL generation for P8.
3469 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3470 vsx_st_elemrev_v8hi.
3471 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3472 add alternate RTL generation for P8.
3473 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3474 vsx_st_elemrev_v16qi.
3476 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3479 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3480 TARGET_AVX512ER check.
3481 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3482 neither -mzeroupper nor -mno-zeroupper is used and
3483 TARGET_EMIT_VZEROUPPER is set.
3484 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3485 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3487 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3489 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3490 folding of vector compares.
3491 (fold_build_vec_cmp): New helper function.
3492 (fold_compare_helper): New helper function.
3493 (builtin_function_type): Add compare builtins to the list of functions
3494 having unsigned arguments. Cosmetic updates to comment indentation.
3495 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3496 the not+eq combination.
3498 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3500 PR tree-optimization/82726
3501 PR tree-optimization/70754
3502 * tree-predcom.c (order_drefs_by_pos): New function.
3503 (combine_chains): Move code setting has_max_use_after to...
3504 (try_combine_chains): ...here. New parameter. Sort combined chains
3505 according to position information.
3506 (tree_predictive_commoning_loop): Update call to above function.
3507 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3509 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3511 PR tree-optimization/82726
3513 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3515 PR tree-optimization/70754
3516 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3517 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3518 combined stmt before it if not NULL.
3519 (combine_chains): Process refs reversely and compute dominance point
3523 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3525 PR tree-optimization/79663
3526 * tree-predcom.c (combine_chains): Process refs in reverse order
3527 only for ZERO length chains, and add explaining comment.
3529 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3531 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3532 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3533 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3535 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3537 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3539 2017-11-15 Martin Liska <mliska@suse.cz>
3541 * tree-cfg.c (pass_warn_function_return::execute):
3542 Compare warn_return_type for greater than zero.
3544 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3548 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3549 to return true on Xeon and not on Xeon Phi.
3550 (ix86_check_avx256_register): Changed to ...
3551 (ix86_check_avx_upper_register): ... this. Add extra check for
3552 VALID_AVX512F_REG_OR_XI_MODE.
3553 (ix86_avx_u128_mode_needed): Changed
3554 ix86_check_avx256_register to ix86_check_avx_upper_register.
3555 (ix86_check_avx256_stores): Changed to ...
3556 (ix86_check_avx_upper_stores): ... this. Changed
3557 ix86_check_avx256_register to ix86_check_avx_upper_register.
3558 (ix86_avx_u128_mode_after): Changed
3559 avx_reg256_found to avx_upper_reg_found. Changed
3560 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3561 (ix86_avx_u128_mode_entry): Changed
3562 ix86_check_avx256_register to ix86_check_avx_upper_register.
3563 (ix86_avx_u128_mode_exit): Ditto.
3564 * config/i386/i386.h: (host_detect_local_cpu): New define.
3566 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3568 * config/arm/xgene1.md (xgene1): Split into automatons
3569 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3570 (xgene1_f_load): Adjust reservations and/or types.
3571 (xgene1_f_store): Likewise.
3572 (xgene1_load_pair): Likewise.
3573 (xgene1_store_pair): Likewise.
3574 (xgene1_fp_load1): Likewise.
3575 (xgene1_load1): Likewise.
3576 (xgene1_store1): Likewise.
3577 (xgene1_move): Likewise.
3578 (xgene1_alu): Likewise.
3579 (xgene1_simd): Likewise.
3580 (xgene1_bfm): Likewise.
3581 (xgene1_neon_load1): Likewise.
3582 (xgene1_neon_store1): Likewise.
3583 (xgene1_neon_logic): Likewise.
3584 (xgene1_neon_st1): Likewise.
3585 (xgene1_neon_ld1r): Likewise.
3586 (xgene1_alu_cond): Added.
3587 (xgene1_shift_reg): Likwise.
3588 (xgene1_bfx): Likewise.
3589 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3591 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3594 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3596 (can_widen_mult_without_libcall): New function.
3597 (expand_mul_overflow): If only checking unsigned mul overflow,
3598 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3599 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3600 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3601 (expand_DIVMOD): Formatting fix.
3602 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3603 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3604 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3606 PR tree-optimization/82977
3607 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3608 constructed temporary to strlen_to_stridx.put.
3610 2017-11-15 Martin Liska <mliska@suse.cz>
3612 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3613 * configure: Regenerate.
3615 2017-11-15 Martin Liska <mliska@suse.cz>
3618 * config/sh/sh-mem.cc: Use proper probability for
3621 2017-11-14 Jeff Law <law@redhat.com>
3623 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3624 the red zone for stack_clash_protection_final_dynamic_probe targets
3625 when the total dynamic stack size is zero bytes.
3627 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3628 blocks is post order.
3630 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3632 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3633 * final.c (rest_of_clean_state): Set it for the
3634 -fcompare-debug dump.
3635 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3636 class when TDF_COMPARE_DEBUG is set.
3638 * dwarf2out.c (gen_producer_string): Discard
3641 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3644 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3645 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3646 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3647 (__TGMATH_REAL_2_3): Remove macros.
3648 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3649 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3651 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3653 (remquo): Define using __TGMATH_REAL_3.
3655 2017-11-14 Jeff Law <law@redhat.com>
3657 * vr-values.c: New file with contents extracted from tree-vrp.c.
3658 * Makefile.in (OBJS): Add vr-values.o
3659 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3660 (set_value_range, set_and_canonicalize_value_range): Likewise.
3661 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3662 (value_range_constant_singleton, symbolic_range_p): Likewise.
3663 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3664 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3665 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3666 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3667 (find_case_label_range, find_case_label_index): Likewise.
3668 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3669 (range_int_cst_singleton_p, value_inside_range): Likewise.
3670 (get_single_symbol): Likewise.
3671 (switch_update): Move structure definition here.
3672 (to_remove_edges, to_update_switch_stmts): Provide externs.
3673 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3674 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3675 (vrp_val_is_min, set_value_range): Likewise.
3676 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3677 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3678 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3679 (range_is_nonnull, range_int_cst_p): Likewwise.
3680 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3681 (get_single_symbol, operand_less_p): Likewise
3682 (compare_values_warnv, compare_values): Likewise.
3683 (value_inside_range, value_range_constant_singleton): Likewise.
3684 (zero_nonzero_bitgs_from_vr): Likewise.
3685 (extract_range_from_binary_expr_1): Likewise.
3686 (overflow_comparison_p): Likewise.
3687 (to_remove_edges, to_update_switch_stmts): Likewise.
3688 (find_case_label-index, find_case_label_range): Likewise.
3689 (switch_update, set_value_range_to_nonnegative): Remove.
3690 (set_value_range_to_truthvalue): Likewise.
3691 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3692 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3693 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3694 (find_case_label_ranges, test_for_singularity): Likewise.
3695 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3696 (x_vr_values): Move to its remaining use site.
3698 2017-11-10 Jeff Law <law@redhat.com>
3700 * vr-values.h (VR_INITIALIZER): Move #define here.
3701 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3702 * Makefile.in (OBJS): Add tree-evrp.o
3703 * tree-vrp.h (assert_info): Move structure definition here.
3704 (set_value_range_to_varying): Prototype.
3705 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3706 (infer_value_range, register_edge_assert_for): Likewise.
3707 (stmt_interesting_for_vrp): Likewise.
3708 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3709 (set_value_range_to_varying): No longer static.
3710 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3711 (infer_value_range, register_edge_assert_for): Likewise.
3713 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3715 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3716 generate the XXBRD instruction.
3718 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3719 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3720 as being compatible if -mabi=ieeelongdouble.
3721 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3722 to setup float128 built-ins with hardware support.
3723 (BU_FLOAT128_HW_2): Likewise.
3724 (BU_FLOAT128_HW_3): Likewise.
3725 (BU_FLOAT128_HW_VSX_1): Likewise.
3726 (BU_FLOAT128_HW_VSX_2): Likewise.
3727 (scalar_extract_expq): Change float128 built-in functions to
3728 accommodate having both KFmode and TFmode functions. Use the
3729 KFmode variant as the default.
3730 (scalar_extract_sigq): Likewise.
3731 (scalar_test_neg_qp): Likewise.
3732 (scalar_insert_exp_q): Likewise.
3733 (scalar_insert_exp_qp): Likewise.
3734 (scalar_test_data_class_qp): Likewise.
3735 (sqrtf128_round_to_odd): Delete processing the round to odd
3736 built-in functions as special built-in functions, and define them
3737 as float128 built-ins. Use the KFmode variant as the default.
3738 (truncf128_round_to_odd): Likewise.
3739 (addf128_round_to_odd): Likewise.
3740 (subf128_round_to_odd): Likewise.
3741 (mulf128_round_to_odd): Likewise.
3742 (divf128_round_to_odd): Likewise.
3743 (fmaf128_round_to_odd): Likewise.
3744 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3745 support for KFmode and TFmode xststdcqp calls.
3746 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3747 point, switch the built-in handlers for the get/set float128
3748 exponent, get float128 mantissa, float128 test built-ins, and the
3749 float128 round to odd built-in functions. Eliminate creating the
3750 float128 round to odd built-in functions as special built-ins.
3751 (rs6000_init_builtins): Eliminate special creation of the float128
3752 round to odd built-in functions.
3753 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3754 function insns to support both TFmode and KFmode variants.
3755 (xsxsigqp_<mode>): Likewise.
3756 (xsiexpqpf_<mode>): Likewise.
3757 (xsiexpqp_<mode>): Likewise.
3758 (xststdcqp_<mode>): Likewise.
3759 (xststdcnegqp_<mode>): Likewise.
3760 (xststdcqp_<mode>): Likewise.
3762 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3764 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3765 unused path_in_freq_ptr parameter.
3766 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3768 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3770 * ipa-inline.c (edge_badness): Dump sreal frequency.
3771 (compute_inlined_call_time): Match natural implementaiton ...
3772 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3773 forgotten division by CGRAPH_FREQ_BASE.
3775 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3777 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3778 Solaris 11. Update comment.
3779 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3781 * config/sol2.h (STARTFILE_SPEC): Likewise.
3782 * configure: Regenerate.
3784 2017-11-14 Carl Love <cel@us.ibm.com>
3786 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3787 le_ and be_ prefixes to swap* variables. Remove
3788 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3791 2017-11-14 Jason Merrill <jason@redhat.com>
3793 Support GTY((cache)) on hash_map.
3794 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3795 (ggc_cache_remove): Override it instead of ggc_mx.
3796 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3797 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3798 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3799 (simple_cache_map_traits): Override maybe_mx.
3800 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3801 (hash_map): Friend gt_cleare_cache.
3802 (gt_cleare_cache): New.
3803 * tree.h (tree_cache_traits): New hash_map traits class.
3804 (tree_cache_map): New typedef.
3806 2017-11-14 Richard Biener <rguenther@suse.de>
3808 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3809 paramter and handling.
3810 (cleanup_control_flow_bb): Likewise.
3811 (cleanup_control_flow_pre): New helper performing a DFS walk
3812 to call cleanup_control_flow_bb in PRE order.
3813 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3814 via cleanup_control_flow_pre.
3816 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
3818 * config/aarch64/aarch64-simd.md
3819 (aarch64_simd_bsl<mode>_internal): Remove DImode.
3820 (*aarch64_simd_bsl<mode>_alt): Likewise.
3821 (aarch64_simd_bsldi_internal): New.
3822 (aarch64_simd_bsldi_alt): Likewise.
3824 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3826 * tracer.c (better_p): Do not compare frequencies.
3827 * reg-stack.c (better_edge): Likewise.
3828 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
3831 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3833 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
3834 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
3835 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
3836 * lto-streamer-in.c (input_function): Use update_max_bb_count.
3837 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
3838 * predict.c (maybe_hot_frequency_p): Inline to ...
3839 (maybe_hot_count_p): ... here; rewrite to counts.
3840 (counts_to_freqs): Rename to ...
3841 (update_max_bb_count): ... this one.
3842 (expensive_function_p): Use counts.
3843 (estimate_bb_frequencies): Update.
3844 (rebuild_frequencies): Update.
3845 * predict.h (counts_to_freqs): Rename to ...
3846 (update_max_bb_count): ... this one.
3847 * profile.c (compute_branch_probabilities): Add debug info
3848 * tree-inline.c (expand_call_inline): Update debug info.
3849 (optimize_inline_calls): Use update_max_bb_count..
3850 (tree_function_versioning): Use update_max_bb_count..
3851 * value-prof.c (gimple_value_profile_transformations):
3852 Do not use update_max_bb_count.
3854 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3856 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
3857 always use frequencies.
3859 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3861 * bb-reorder.c: Remove frequencies from comments.
3862 (better_edge_p): Use profile counts.
3863 (find_traces): Dump profile counts.
3864 (rotate_loop): Use profile counts.
3865 (find_traces_1_round): Likewise.
3866 (connect_better_edge_p): Use counts instead of probabilities for
3868 (copy_bb_p): Drop early check for non-0 frequency.
3869 (sanitize_hot_paths): Update comments.
3871 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3873 * ipa-split.c (struct split_point): Add count.
3874 (consider_split): Do not compute incoming frequency; compute incoming
3875 count and store it to split_point.
3876 (split_function): Set count of the call to split part correctly.
3878 2017-11-13 Carl Love <cel@us.ibm.com>
3880 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
3882 2017-11-13 Tom Tromey <tom@tromey.com>
3884 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
3886 2017-11-13 Carl Love <cel@us.ibm.com>
3888 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3889 Add support for builtins:
3890 unsigned int vec_first_{,miss}_match_{,or_eos}index,
3891 vector {un,}signed {char,int,short},
3892 vector {un,}signed {char,int,short}) arguments.
3893 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
3894 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
3895 Add BU_P9V_AV_2 expansions for the builtins.
3896 * config/rs6000/altivec.h (vec_first_match_index,
3897 vec_first_mismatch_index, vec_first_match_or_eos_index,
3898 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
3899 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
3900 new extern declaration.
3901 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
3902 * config/rs6000/vsx.md (first_match_index_<mode>,
3903 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
3904 first_mismatch_or_eos_index_<mode>): Add define expand.
3905 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
3906 * doc/extend.texi: Update the built-in documenation file for the new
3909 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
3911 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
3912 into the min/max operations for _Float<N> and _Float<N>X types.
3914 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3917 * dwarf2out.c (do_eh_frame): New static variable.
3918 (dwarf2out_begin_prologue): Set it.
3919 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
3921 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3923 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
3925 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
3927 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3930 * debug.h (dwarf2out_do_eh_frame): Declare.
3931 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
3932 (dwarf2out_do_frame): Use it.
3933 (dwarf2out_do_cfi_asm): Likewise.
3934 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
3935 (dwarf2out_assembly_start): Likewise.
3936 (dwarf2out_begin_prologue): Fix comment.
3937 * toplev.c (compile_file): Always call dwarf2out_frame_finish
3938 if the target needs either debug or unwind DWARF2 info.
3939 * lto-opts.c (lto_write_options): Do not save -fexceptions,
3940 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
3941 -ftrapping-math, -ftrapv and -fwrapv.
3943 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3945 * cgraph.c (cgraph_edge::sreal_frequency): New function.
3946 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
3947 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
3948 (estimate_edge_size_and_time): Likewise.
3949 (ipa_merge_fn_summary_after_inlining): Likewise.
3950 * ipa-inline.c (cgraph_freq_base_rec): Remove.
3951 (compute_uninlined_call_time): Use sreal_frequency.
3952 (compute_inlined_call_time): Likewise.
3953 (ipa_inline): Do not initialize cgraph_freq_base_rec.
3954 * profile-count.c: Include sreal.h.
3955 (profile_count::to_sreal_scale): New.
3956 * profile-count.h: Forward declare sreal.
3957 (profile_count::to_sreal_scale): Declare.
3959 2017-11-13 Nathan Sidwell <nathan@acm.org>
3961 * diagnostic.c (maybe_line_and_column): New.
3962 (diagnostic_get_location_text): Use it.
3963 (diagnostic_report_current_module): Likewise.
3964 (test_diagnostic_get_location_text): Add tests.
3966 2017-11-13 Luis Machado <luis.machado@linaro.org>
3968 * doc/md.texi (Specifying processor pipeline description): Fix
3969 incorrect latency for the div instruction example.
3971 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3973 PR tree-optimization/78821
3974 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
3975 that bit_not_p is the same.
3976 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
3977 (split_group): Count precisely bit_not_p bits in each statement.
3978 (invert_op): New function.
3979 (imm_store_chain_info::output_merged_store): Use invert_op to
3980 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
3981 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
3983 2017-11-13 Martin Liska <mliska@suse.cz>
3985 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
3986 (struct source_info): Likewise.
3987 (add_branch_counts): Likewise.
3988 (add_line_counts): Likewise.
3989 (function_summary): Likewise.
3990 (output_intermediate_line): Likewise.
3991 (generate_results): Likewise.
3993 2017-11-13 Martin Liska <mliska@suse.cz>
3995 * gcov.c (struct block_info): Remove typedef for block_t.
3996 (struct line_info): Likewise.
3997 (line_info::has_block): Likewise.
3998 (EXIT_BLOCK): Likewise.
3999 (unblock): Likewise.
4000 (circuit): Likewise.
4001 (get_cycles_count): Likewise.
4002 (process_file): Likewise.
4003 (read_graph_file): Likewise.
4004 (solve_flow_graph): Likewise.
4005 (find_exception_blocks): Likewise.
4006 (add_line_counts): Likewise.
4007 (accumulate_line_info): Likewise.
4008 (output_line_details): Likewise.
4010 2017-11-13 Martin Liska <mliska@suse.cz>
4012 * gcov.c (struct arc_info): Remove typedef for arc_t.
4013 (struct line_info): Likewise.
4014 (add_branch_counts): Likewise.
4015 (output_branch_count): Likewise.
4016 (function_info::~function_info): Likewise.
4017 (circuit): Likewise.
4018 (output_intermediate_line): Likewise.
4019 (read_graph_file): Likewise.
4020 (solve_flow_graph): Likewise.
4021 (find_exception_blocks): Likewise.
4022 (add_line_counts): Likewise.
4023 (accumulate_line_info): Likewise.
4024 (output_line_details): Likewise.
4025 (output_function_details): Likewise.
4027 2017-11-13 Martin Liska <mliska@suse.cz>
4029 * gcov.c (struct function_info): Remove typedef for function_t.
4030 (struct source_info): Likewise.
4031 (source_info::get_functions_at_location): Likewise.
4032 (solve_flow_graph): Likewise.
4033 (find_exception_blocks): Likewise.
4034 (add_line_counts): Likewise.
4035 (output_intermediate_file): Likewise.
4036 (process_file): Likewise.
4037 (generate_results): Likewise.
4038 (release_structures): Likewise.
4039 (read_graph_file): Likewise.
4040 (read_count_file): Likewise.
4041 (accumulate_line_counts): Likewise.
4042 (output_lines): Likewise.
4044 2017-11-13 Martin Liska <mliska@suse.cz>
4046 * gcov.c (function_info::function_info): Remove num_counts
4047 and add vector<gcov_type>.
4048 (function_info::~function_info): Use the vector.
4049 (process_file): Likewise.
4050 (read_graph_file): Likewise.
4051 (read_count_file): Likewise.
4052 (solve_flow_graph): Likewise.
4054 2017-11-13 Martin Liska <mliska@suse.cz>
4056 * gcov.c (function_info::is_artificial): New function.
4057 (process_file): Erase all artificial early.
4058 (generate_results): Skip as all artificial are already
4061 2017-11-13 Martin Liska <mliska@suse.cz>
4063 * gcov.c (read_graph_file): Store to global vector of functions.
4064 (read_count_file): Iterate the vector.
4065 (process_file): Likewise.
4066 (generate_results): Likewise.
4067 (release_structures): Likewise.
4069 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4071 PR tree-optimization/82954
4072 * gimple-ssa-store-merging.c
4073 (imm_store_chain_info::coalesce_immediate_stores): If
4074 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4076 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4078 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4079 Upddate call to ENDIAN_LANE_N.
4080 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4081 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4082 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4083 and use aarch64_endian_lane_rtx.
4084 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4086 2017-11-12 Tom de Vries <tom@codesourcery.com>
4088 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4090 2017-11-12 Tom de Vries <tom@codesourcery.com>
4092 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4094 2017-11-12 Tom de Vries <tom@codesourcery.com>
4096 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4098 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4099 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4100 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4102 2017-11-11 Martin Sebor <msebor@redhat.com>
4105 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4108 * prefic.c (translate_name): Replace strncpy with memcpy to
4109 avoid -Wstringop-truncation.
4111 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4113 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4115 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4117 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4119 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4121 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4124 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4126 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4127 enable generating XXBRH if the value is in a vector register.
4128 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4129 value is in a vector register.
4130 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4131 register to register bswap64's instead of doing the GPR sequence
4132 used on previous machines.
4133 (bswapdi2_xxbrd): New insn.
4134 (bswapdi2_reg): Disallow on ISA 3.0.
4135 (register to register bswap64 splitter): Do not split the insn on
4136 ISA 3.0 systems that use XXBRD.
4138 2017-11-10 Martin Sebor <msebor@redhat.com>
4141 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4143 (find_subframework_file): Same.
4145 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4147 * auto-profile.c (afdo_indirect_call): Drop frequency.
4148 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4149 (cgraph_node::create_edge): Drop frequency argument.
4150 (cgraph_node::create_indirect_edge): Drop frequency argument.
4151 (cgraph_edge::make_speculative): Drop frequency arguments.
4152 (cgraph_edge::resolve_speculation): Do not update frequencies
4153 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4154 (cgraph_node::dump): Check consistency in IPA mode.
4155 (cgraph_edge::maybe_hot_p): Use IPA counter.
4156 (cgraph_edge::verify_count_and_frequency): Rename to ...
4157 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4158 (cgraph_node::verify_node): Update.
4159 * cgraph.h (struct cgraph_edge): Drop frequency.
4160 (cgraph_edge::frequency): New function.
4161 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4163 (cgraph_edge::rebuild_edges): Likewise.
4164 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4165 (duplicate_thunk_for_node): Do not pass frequency.
4166 (cgraph_node::create_clone): Scale only counts.
4167 (cgraph_node::create_virtual_clone): Do not pass frequency.
4168 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4169 (cgraph_node::create_version_clone): Do not pass frequency.
4170 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4171 (cgraph_node::expand_thunk): Do not pass frequency.
4172 (cgraph_node::create_wrapper): Do not pass frequency.
4173 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4175 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4176 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4177 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4178 (gather_caller_stats): Use frequency function.
4179 (ipcp_cloning_candidate_p): Use frequency function.
4180 (ipcp_propagate_stage): Use frequency function.
4181 (get_info_about_necessary_edges): Use frequency function.
4182 (update_profiling_info): Update only IPA profile.
4183 (update_specialized_profile): Use frequency functoin.
4184 (perhaps_add_new_callers): Update only IPA profile.
4185 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4186 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4187 (dump_ipa_call_summary): Use frequency function.
4188 (estimate_edge_size_and_time): Use frequency function.
4189 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4190 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4191 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4192 (update_noncloned_counts): ... ths one; scale counts only.
4193 (clone_inlined_nodes): Do not scale frequency.
4194 (inline_call): Do not pass frequency.
4195 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4196 (compute_inlined_call_time): Use IPA profile.
4197 (want_inline_small_function_p): Use IPA profile.
4198 (want_inline_self_recursive_call_p): Use IPA profile.
4199 (edge_badness): Use IPA profile.
4200 (lookup_recursive_calls): Use IPA profile.
4201 (recursive_inlining): Do not pass frequency.
4202 (resolve_noninline_speculation): Do not update frequency.
4203 (inline_small_functions): Collect max of IPA profile.
4204 (dump_overall_stats): Dump IPA porfile.
4205 (dump_inline_stats): Dump IPA porfile.
4206 (ipa_inline): Collect IPA stats.
4207 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4208 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4209 (ipa_propagate_frequency): Use frequency function.
4210 (ipa_profile): Cleanup.
4211 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4212 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4213 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4214 (input_node): Do not stream frequency.
4215 (input_edge): Do not stream frequency.
4216 (merge_profile_summaries): Scale only IPA profiles.
4217 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4218 * predict.c (drop_profile): Do not recompute frequency.
4219 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4220 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4221 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4222 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4223 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4224 * tree-inline.c (copy_bb): Do not scale frequency.
4225 (expand_call_inline): Do not scale frequency.
4226 (tree_function_versioning): Do not scale frequency.
4227 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4229 2017-11-10 Julia Koval <julia.koval@intel.com>
4231 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4232 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4233 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4234 (_mm256_maskz_gf2p8affine_epi64_epi8)
4235 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4236 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4237 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4238 (__builtin_ia32_vgf2p8affineqb_v32qi)
4239 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4240 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4242 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4246 (arm_option_override): Refactor.
4247 (arm_option_reconfigure_globals): New.
4248 (arm_options_perform_arch_sanity_checks): New.
4249 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4251 (arm_options_perform_arch_sanity_checks): Likewise
4253 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4255 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4256 (power9-qpmul): New.
4257 * rs6000/rs6000.md ("type" attr): Add qmul.
4258 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4259 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4260 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4262 2017-11-10 Martin Sebor <msebor@redhat.com>
4265 * builtins.c (compute_objsize): Handle arrays that
4266 compute_builtin_object_size likes to fail for. Make extern.
4267 * builtins.h (compute_objsize): Declare.
4268 (check_strncpy_sizes): New function.
4269 (expand_builtin_strncpy): Call check_strncpy_sizes.
4270 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4271 -Wstringop-truncation.
4272 (gimple_fold_builtin_strncat): Same.
4273 * gimple.c (gimple_build_call_from_tree): Set call location.
4274 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4275 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4276 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4277 (handle_builtin_strlen): Use strlen_to_stridx.
4278 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4280 Use strlen_to_stridx.
4281 (pass_strlen::execute): Release strlen_to_stridx.
4282 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4283 (-Wstringop-truncation): Document new option.
4285 2017-11-10 Martin Liska <mliska@suse.cz>
4287 PR gcov-profile/82702
4288 * gcov.c (main): Handle intermediate files in a different
4290 (get_gcov_intermediate_filename): New function.
4291 (output_gcov_file): Remove support of intermediate files.
4292 (generate_results): Allocate intermediate file.
4293 (release_structures): Clean-up properly fn_end.
4294 (output_intermediate_file): Start iterating with line 1.
4296 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4298 PR tree-optimization/82929
4299 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4300 ops_swapped_p non-static data member.
4301 (store_immediate_info::store_immediate_info): Clear it.
4302 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4303 ops set ops_swapped_p.
4304 (count_multiple_uses): Handle ops_swapped_p.
4306 2017-11-10 Martin Liska <mliska@suse.cz>
4308 * coverage.c (coverage_init): Stream information about
4309 support of has_unexecuted_blocks.
4310 * doc/gcov.texi: Document that.
4311 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4312 * gcov.c (read_graph_file): Likewise.
4313 (output_line_beginning): Fix a small issue with
4316 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4318 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4319 reference of trivial component.
4321 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4324 * gimple-ssa-store-merging.c
4325 (pass_store_merging::terminate_all_aliasing_chains): For
4326 gimple_store_p stmts also call refs_output_dependent_p.
4328 PR rtl-optimization/82913
4329 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4332 2017-11-09 Jeff Law <law@redhat.com>
4334 * vr-values.h: New file with vr_values class.
4335 * tree-vrp.c: Include vr-values.h
4336 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4337 data objects into the vr_values class.
4338 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4339 (get_value_range): Make it a member function within vr_values class.
4340 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4341 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4342 (op_with_constant_singleton_value_range): Likewise.
4343 (extract_range_for_var_from_comparison_expr): Likewise.
4344 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4345 (extract_range_from_binary_expr): Likewise.
4346 (extract_range_from_unary_expr): Likewise.
4347 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4348 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4349 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4350 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4351 (compare_name_with_value, compare_names): Likewise.
4352 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4353 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4354 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4355 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4356 (extract_range_from_phi_node): Likewise.
4357 (simplify_truth_ops_using_ranges): Likewise.
4358 (simplify_div_or_mod_using_ranges): Likewise.
4359 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4360 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4361 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4362 (simplify_float_conversion_using_ranges): Likewise.
4363 (simplify_internal_call_using_ranges): Likewise.
4364 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4365 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4366 poitner via x_vr_values for calls into gimple folder.
4367 (vrp_initialize_lattice): Make this the vr_values ctor.
4368 (vrp_free_lattice): Make this the vr_values dtor.
4369 (set_vr_value): New function.
4370 (class vrp_prop): Add vr_values data member. Add various member
4371 functions as well as member functions that delegate to vr_values.
4372 (check_array_ref): Make a member function within vrp_prop class.
4373 (search_for_addr_array, vrp_initialize): Likewise.
4374 (vrp_finalize): Likewise. Revamp to avoid direct access to
4375 vr_value, values_propagated, etc.
4376 (check_array_bounds): Extract vrp_prop class instance pointer from
4377 walk info structure. Use it to call member functions.
4378 (check_all_array_refs): Make a member function within vrp_prop class.
4379 Smuggle class instance pointer via walk info structure.
4380 (x_vr_values): New local static.
4381 (vrp_valueize): Use x_vr_values to get class instance.
4382 (vr_valueize_1): Likewise.
4383 (class vrp_folder): Add vr_values data member. Add various member
4384 functions as well as member functions that delegate to vr_values.
4385 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4386 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4387 class instance from vr_values. Use it to call member functions.
4388 (vrp_dom_walker): Add vr_values data member.
4389 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4390 instance via x_vr_values.
4391 (identify_jump_threads): Accept vr_values as argument. Store
4392 it into the walker structure.
4393 (evrp_dom_walker): Add vr_values class data member. Add various
4395 (evrp_dom_walker::try_find_new_range): Use vr_values data
4396 member to access the memory allocator.
4397 (evrp_dom_walker::before_dom_children): Store vr_values class
4398 instance into the vrp_folder class.
4399 (evrp_dom_walker::push_value_range): Rework to avoid direct
4400 access to num_vr_values and vr_value.
4401 (evrp_dom_walker::pop_value_range): Likewise.
4402 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4403 Use vr_values to get to dump_all_value_ranges member function.
4404 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4405 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4406 Pass vr_values class instance down to identify_jump_threads.
4407 Remove call to vrp_free_lattice.
4408 (debug_all_value_ranges): Remove.
4410 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4411 (vrp_folder): Likewise.
4413 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4414 Get it from the existing bitmap instead.
4415 (vrp_intersect_ranges_1): Likewise.
4417 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4419 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4421 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4422 set bit_not_p to it.
4423 (imm_store_chain_info::coalesce_immediate_stores): Break group
4424 if bit_not_p is different.
4425 (count_multiple_uses, split_group,
4426 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4427 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4428 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4429 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4430 to store_immediate_info ctor.
4432 2017-11-09 Jim Wilson <jimw@sifive.com>
4434 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4435 (scan_prog_file): Likewise.
4437 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4439 * bb-reorder.c (max_entry_frequency): Remove.
4440 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4441 connect_traces, push_to_next_round_p): Remove prototypes.
4442 (find_traces_1_round): Use counts only.
4443 (push_to_next_round_p): Likewise.
4444 (find_traces): Likewise.
4445 (rotate_loop): Likewise.
4446 (find_traces_1_round): Likewise.
4447 (connect_traces): Likewise.
4448 (edge_order): Likewise.
4450 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4452 * config/arm/arm.c (output_return_instruction): Add comments to
4453 indicate requirement for cmse_nonsecure_entry return to account
4454 for the size of clearing instruction output here.
4455 (thumb_exit): Likewise.
4456 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4457 return in hardfloat mode.
4459 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4461 * config/rs6000/rs6000.c (machine_function): Add a bool,
4462 "toc_is_wrapped_separately".
4463 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4464 if it wasn't explicitly set or unset, we are optimizing for speed, and
4465 doing separate shrink-wrapping.
4466 (rs6000_get_separate_components): Enable the TOC component if
4467 saving the TOC register in the prologue.
4468 (rs6000_components_for_bb): Handle the TOC component.
4469 (rs6000_emit_prologue_components): Store the TOC register where needed.
4470 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4471 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4473 2017-11-09 Martin Jambor <mjambor@suse.cz>
4475 * ipa-param-manipulation.c: New file.
4476 * ipa-param-manipulation.h: Likewise.
4477 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4478 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4479 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4480 (ipa_parm_adjustment): Likewise.
4481 (ipa_parm_adjustment_vec): Likewise.
4482 (ipa_get_vector_of_formal_parms): Moved declaration to
4483 ipa-param-manipulation.h.
4484 (ipa_get_vector_of_formal_parm_types): Likewise.
4485 (ipa_modify_formal_parameters): Likewise.
4486 (ipa_modify_call_arguments): Likewise.
4487 (ipa_combine_adjustments): Likewise.
4488 (ipa_dump_param_adjustments): Likewise.
4489 (ipa_modify_expr): Likewise.
4490 (ipa_get_adjustment_candidate): Likewise.
4491 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4492 ipa-param-manipulation.c.
4493 (ipa_get_vector_of_formal_parm_types): Likewise.
4494 (ipa_modify_formal_parameters): Likewise.
4495 (ipa_modify_call_arguments): Likewise.
4496 (ipa_modify_expr): Likewise.
4497 (get_ssa_base_param): Likewise.
4498 (ipa_get_adjustment_candidate): Likewise.
4499 (index_in_adjustments_multiple_times_p): Likewise.
4500 (ipa_combine_adjustments): Likewise.
4501 (ipa_dump_param_adjustments): Likewise.
4502 * tree-sra.c: Also include ipa-param-manipulation.h
4503 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4506 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4507 Alan Hayward <alan.hayward@arm.com>
4508 David Sherwood <david.sherwood@arm.com>
4510 * doc/sourcebuild.texi (vect_masked_store): Document.
4512 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4513 Alan Hayward <alan.hayward@arm.com>
4514 David Sherwood <david.sherwood@arm.com>
4516 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4518 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4519 Alan Hayward <alan.hayward@arm.com>
4520 David Sherwood <david.sherwood@arm.com>
4522 * doc/sourcebuild.texi (vect_variable_length): Document.
4524 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4525 Alan Hayward <alan.hayward@arm.com>
4526 David Sherwood <david.sherwood@arm.com>
4528 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4530 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4531 Alan Hayward <alan.hayward@arm.com>
4532 David Sherwood <david.sherwood@arm.com>
4534 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4536 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4537 Alan Hayward <alan.hayward@arm.com>
4538 David Sherwood <david.sherwood@arm.com>
4540 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4541 previously undocumented selectors.
4542 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4544 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4546 * doc/rtl.texi (const_vector): Say that elements can be
4547 const_wide_ints too.
4548 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4549 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4550 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4551 * optabs.c (expand_vector_broadcast): Likewise.
4553 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4554 Alan Hayward <alan.hayward@arm.com>
4555 David Sherwood <david.sherwood@arm.com>
4557 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4558 scaled index even if the unscaled address was invalid.
4559 Don't increase the complexity of using a scale in that case.
4561 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4562 Alan Hayward <alan.hayward@arm.com>
4563 David Sherwood <david.sherwood@arm.com>
4565 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4566 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4567 UNITS_PER_WORD bytes.
4568 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4569 based on REGMODE_NATURAL_SIZE of the inner mode.
4570 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4571 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4572 * expmed.c (lowpart_bit_field_p): Divide the value up into
4573 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4574 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4575 whether something is likely to occupy more than one register.
4577 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4580 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4581 profile_count::adjust_for_ipa_scaling.
4582 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4583 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4585 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4587 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4589 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4590 (split_group): Add total_orig and total_new arguments, estimate the
4591 number of statements related to the store group without store merging
4592 and with store merging.
4593 (imm_store_chain_info::output_merged_store): Adjust split_group
4594 callers, punt if estimated number of statements with store merging
4595 is not smaller than estimated number of statements without it.
4597 (handled_load): Remove has_single_use checks.
4598 (pass_store_merging::process_store): Likewise.
4600 2017-11-09 Richard Biener <rguenther@suse.de>
4602 PR tree-optimization/82902
4603 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4605 2017-11-09 Martin Liska <mliska@suse.cz>
4608 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4611 2017-11-09 Martin Liska <mliska@suse.cz>
4613 PR tree-optimization/82669
4614 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4616 2017-11-09 Martin Liska <mliska@suse.cz>
4618 PR gcov-profile/48463
4619 * coverage.c (coverage_begin_function): Output also end locus
4620 of a function and information whether the function is
4622 * gcov-dump.c (tag_function): Parse and print the information.
4623 * gcov.c (INCLUDE_MAP): Add include.
4624 (INCLUDE_SET): Likewise.
4625 (struct line_info): Move earlier in the source file because
4626 of vector<line_info> in function_info structure.
4627 (line_info::line_info): Likewise.
4628 (line_info::has_block): Likewise.
4629 (struct source_info): Add new member index.
4630 (source_info::get_functions_at_location): New function.
4631 (function_info::group_line_p): New function.
4632 (output_intermediate_line): New function.
4633 (output_intermediate_file): Use the mentioned function.
4634 (struct function_start): New.
4635 (struct function_start_pair_hash): Likewise.
4636 (process_file): Add code that identifies group functions.
4637 Assign lines either to global or function scope.
4638 (generate_results): Skip artificial functions.
4639 (find_source): Assign index for each source file.
4640 (read_graph_file): Read new flag artificial and end_line.
4641 (add_line_counts): Assign it either to global of function scope.
4642 (accumulate_line_counts): Isolate core of the function to
4643 accumulate_line_info and call it for both function and global
4645 (accumulate_line_info): New function.
4646 (output_line_beginning): Fix GNU coding style.
4647 (print_source_line): New function.
4648 (output_line_details): Likewise.
4649 (output_function_details): Likewise.
4650 (output_lines): Iterate both source (global) scope and function
4652 (struct function_line_start_cmp): New class.
4653 * doc/gcov.texi: Reflect changes in documentation.
4655 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4658 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4659 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4660 and similarly for not instead of neg.
4662 2017-11-08 Andi Kleen <ak@linux.intel.com>
4664 * config/i386/i386.opt: Add -mforce-indirect-call.
4665 * config/i386/predicates.md: Check for flag_force_indirect_call.
4666 * doc/invoke.texi: Document -mforce-indirect-call
4668 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4670 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4672 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4673 config/riscv/riscv.c (predict.h): New include.
4674 (riscv_slow_unaligned_access_p): No longer static.
4675 (riscv_block_move_straight): Add require.
4676 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4678 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4681 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4682 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4683 nonimmediate_operand predicate for operand 1 instead of
4686 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4688 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4690 * config/aarch64/constraints.md (Uml): New constraint.
4691 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4694 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4696 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4697 of two vec_duplicates into a vec_concat.
4699 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4701 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4702 Simplify vec_merge of vec_duplicate and vec_concat.
4703 * config/aarch64/constraints.md (Utq): New constraint.
4704 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4707 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4709 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4710 Simplify vec_merge of vec_duplicate and const_vector.
4711 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4713 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4714 mode iterator. Update predicate on operand 1 to
4715 handle non-const_vec constants. Delete constraints.
4716 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4718 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4720 PR tree-optimization/78821
4721 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4723 (store_operand_info::store_operand_info): Initialize it to false.
4724 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4725 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4726 store in the group, and if chain_info is non-NULL, to ignore altogether
4728 (compatible_load_p): Fail if bit_not_p does not match.
4729 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4730 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4731 (pass_store_merging::process_store): Adjust
4732 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4733 call terminate_all_aliasing_chains newly when adding a store into
4734 a chain with non-NULL chain_info.
4736 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4738 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4740 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4742 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4744 (aarch64_layout_frame): Initialise emit_frame_chain.
4745 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4746 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4748 2017-11-08 Martin Liska <mliska@suse.cz>
4750 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4751 of gimple_call_internal_p.
4753 2017-11-07 Tom de Vries <tom@codesourcery.com>
4755 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4757 2017-11-07 Tom de Vries <tom@codesourcery.com>
4759 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4762 2017-11-08 Martin Liska <mliska@suse.cz>
4765 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4767 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4769 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4771 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4774 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4775 (*cmp<mode>_ccz_1): New insn with $k alternative.
4778 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4779 RTX_COMM_COMPARE as commutative as well.
4780 (ix86_binary_operator_ok): Formatting fix.
4781 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4782 *<code><mode>3<mask_name><round_saeonly_name>,
4783 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4784 *<s>mul<mode>3_highpart<mask_name>,
4785 *vec_widen_umult_even_v16si<mask_name>,
4786 *vec_widen_umult_even_v8si<mask_name>,
4787 *vec_widen_umult_even_v4si<mask_name>,
4788 *vec_widen_smult_even_v16si<mask_name>,
4789 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4790 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4791 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4792 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4793 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4794 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4795 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4796 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4797 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4798 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4799 ix86_binary_operator_ok. Formatting fixes.
4800 (*<plusminus_insn><mode>3<mask_name><round_name>,
4801 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4804 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4806 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4807 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4808 eq and ne if TARGET_ISEL.
4809 (cmp): New code_iterator.
4810 (UNS, UNSU_, UNSIK): New code_attrs.
4811 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4812 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4814 ("eq<mode>3"): ... this.
4815 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
4817 ("ne<mode>3"): ... this.
4819 2017-11-07 Julia Koval <julia.koval@intel.com>
4822 * common/config/i386/i386-common.c
4823 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
4824 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
4825 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
4826 * config/i386/i386.opt: Ditto.
4827 * config/i386/i386.c (ix86_target_string): Ditto.
4828 (ix86_option_override_internal): Ditto.
4829 (ix86_init_mpx_builtins): Move MPX to args2.
4830 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
4831 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4832 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4833 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4834 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4835 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4836 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
4838 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
4841 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
4842 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
4843 (zero-extendsidi peephole2): Remove peephole.
4845 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4848 * stor-layout.c: Include attribs.h.
4849 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
4850 explicit lookup of "aligned" attribute.
4852 2017-11-07 Andrew Waterman <andrew@sifive.com>
4854 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
4855 (riscv_expand_block_move): Likewise.
4856 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
4858 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
4859 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
4860 * config/riscv/riscv.c (riscv_block_move_straight): New function.
4861 (riscv_adjust_block_mem): Likewise.
4862 (riscv_block_move_loop): Likewise.
4863 (riscv_expand_block_move): Likewise.
4864 * config/riscv/riscv.md (movmemsi): New pattern.
4866 2017-11-07 Michael Clark <michaeljclark@mac.com>
4868 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
4869 (MUSL_DYNAMIC_LINKER): Likewise.
4871 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
4873 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
4874 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
4877 2017-11-07 Richard Biener <rguenther@suse.de>
4879 * match.pd: Fix build.
4881 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
4882 Jackson Woodruff <jackson.woodruff@arm.com>
4884 PR tree-optimization/71026
4885 * match.pd: Canonicalize negate in division.
4887 2017-11-07 Sudakshina Das <sudi.das@arm.com>
4890 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
4892 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4894 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
4895 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
4897 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4899 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
4900 non-scalar integral types.
4901 * match.pd (negate_expr_p): Handle MINUS_EXPR.
4902 (-(A-B), -(~A)): New transformations.
4904 2017-11-07 Tom de Vries <tom@codesourcery.com>
4906 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
4907 semicolon after "do {} while (0)".
4908 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4909 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4910 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4911 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4912 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4914 2017-11-07 Tom de Vries <tom@codesourcery.com>
4916 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
4917 after "do {} while (0)".
4918 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4919 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4920 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4921 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4922 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4924 2017-11-07 Tom de Vries <tom@codesourcery.com>
4927 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
4929 (arm_rtx_costs_internal): Add missing semicolon after
4930 HANDLE_NARROW_SHIFT_ARITH call.
4932 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4934 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4935 disable isel if it was not set explicitly.
4937 2017-11-06 James Bowman <james.bowman@ftdichip.com>
4939 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
4940 (add_type_attribute) likewise.
4942 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
4944 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
4945 of struct ix86_frame.
4946 (ix86_initial_elimination_offset): Likewise.
4947 (ix86_expand_split_stack_prologue): Likewise.
4949 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
4951 * tree-vrp.h (enum value_range_type): Update stale comment.
4953 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4954 Alan Hayward <alan.hayward@arm.com>
4955 David Sherwood <david.sherwood@arm.com>
4957 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
4958 (aarch64_expand_vec_perm_const): Take the number of units too.
4959 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
4960 (aarch64_expand_vec_perm_const): Likewise.
4961 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
4962 (vec_perm<mode>): Update accordingly.
4964 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4965 Alan Hayward <alan.hayward@arm.com>
4966 David Sherwood <david.sherwood@arm.com>
4968 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
4969 Take the number of units too.
4970 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
4971 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
4972 but check for a vector mode before rather than after the call.
4973 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4974 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
4975 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
4976 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
4977 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
4978 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
4979 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
4980 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
4981 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
4982 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
4983 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
4984 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
4985 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
4986 (aarch64_sqdmull2_n<mode>): Update accordingly.
4988 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4989 Alan Hayward <alan.hayward@arm.com>
4990 David Sherwood <david.sherwood@arm.com>
4992 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
4993 the number of units too.
4994 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
4995 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
4996 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
4997 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
4998 (vec_store_lanesxi<mode>): Update accordingly.
5000 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5001 Alan Hayward <alan.hayward@arm.com>
5002 David Sherwood <david.sherwood@arm.com>
5004 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5005 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5006 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5007 of units rather than the mode.
5008 * config/aarch64/iterators.md (nunits): New mode attribute.
5009 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5010 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5011 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5012 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5013 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5014 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5015 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5016 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5017 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5018 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5019 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5020 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5021 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5022 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5023 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5024 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5025 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5026 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5027 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5028 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5029 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5030 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5031 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5032 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5033 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5034 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5035 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5036 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5037 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5038 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5039 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5040 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5041 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5042 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5043 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5044 (aarch64_simd_vec_setv2di): Likewise.
5046 2017-11-06 Carl Love <cel@us.ibm.com>
5048 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5050 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5051 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5053 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5055 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5056 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5057 Add power 8 macro expansions.
5058 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5059 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5060 power 8 instructions. (VSX_XXBR): Add iterator.
5062 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5064 * config/arm/arm.md (predicable_short_it): Change default to "no",
5065 improve documentation, remove uses that are identical to the default.
5066 (enabled_for_depr_it): Rename to enabled_for_short_it.
5067 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5068 Remove default uses.
5069 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5070 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5071 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5072 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5074 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5077 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5078 float128 helper macros, which are no longer used after deleting
5079 the old 'q' built-in functions, and moving the round to odd
5080 built-in functions to being special built-in functions.
5081 (BU_FLOAT128_2): Likewise.
5082 (BU_FLOAT128_1_HW): Likewise.
5083 (BU_FLOAT128_2_HW): Likewise.
5084 (BU_FLOAT128_3_HW): Likewise.
5085 (FABSQ): Delete old 'q' built-in functions.
5086 (COPYSIGNQ): Likewise.
5087 (SQRTF128_ODD): Move round to odd built-in functions to be
5088 special built-in functions, so that we can handle
5089 -mabi=ieeelongdouble.
5090 (TRUNCF128_ODD): Likewise.
5091 (ADDF128_ODD): Likewise.
5092 (SUBF128_ODD): Likewise.
5093 (MULF128_ODD): Likewise.
5094 (DIVF128_ODD): Likewise.
5095 (FMAF128_ODD): Likewise.
5096 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5097 built-in names to 'f128'.
5098 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5099 old 'q' built-in functions, as the machine independent code for
5100 'f128' built-in functions handles this.
5101 (rs6000_expand_builtin): Add expansion for float128 round to odd
5102 functions, keying off on -mabi=ieeelongdouble of whether to use
5103 the KFmode or TFmode variant.
5104 (rs6000_init_builtins): Initialize the _Float128 round to odd
5106 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5107 _Float128 'q' built-in functions are now mapped into the new
5108 'f128' built-in functions.
5110 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5112 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5113 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5115 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5118 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5119 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5120 * toplev.c: Include "ipa-fnsummary.h".
5121 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5123 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5125 PR tree-optimization/82838
5126 * gimple-ssa-store-merging.c
5127 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5128 on a separate gimple_seq which is then appended to seq.
5130 2017-11-06 Jeff Law <law@redhat.com>
5133 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5134 (get_probe_interval): New functions.
5135 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5136 (ix86_adjust_stack_and_probe): Likewise.
5137 (output_adjust_stack_and_probe): Likewise.
5138 (ix86_emit_probe_stack_range): Likewise.
5139 (ix86_expand_prologue): Likewise.
5141 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5143 PR tree-optimization/82816
5144 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5145 if the modes of the two types are the same.
5146 (convert_plusminus_to_widen): Likewise.
5148 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5150 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5152 (usadv16qi): New define_expand.
5153 (usadv8hi): New define_expand.
5155 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5158 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5160 (inline_transform): Likewise.
5161 * predict.c (counts_to_freqs): Remove useless conditional.
5162 * profile-count.h (profile_count::apply_scale): Move sanity check.
5163 * tree-inline.c (copy_bb): Always scale.
5164 (copy_cfg_body): Likewise.
5166 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5169 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5171 (*cmp_ite0): Add enabled_for_depr_it attribute.
5172 (*cmp_ite1): Likewise.
5174 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5176 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5179 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5181 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5182 return the value by pointer.
5183 (extract_range_from_multiplicative_op_1): Update accordingly.
5184 Return as soon as an operation fails.
5186 2017-11-05 Tom de Vries <tom@codesourcery.com>
5189 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5190 (DEF_SANITIZER_BUILTIN): ... here.
5191 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5192 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5194 2017-11-05 Tom de Vries <tom@codesourcery.com>
5197 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5199 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5200 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5201 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5203 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5204 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5205 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5207 2017-11-05 Tom de Vries <tom@codesourcery.com>
5210 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5213 2017-11-04 Michael Clark <michaeljclark@mac.com>
5215 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5216 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5218 (*addsi3_extended): Likewise.
5219 (*addsi3_extended2): Likewise.
5220 (<optab>si3): Likewise.
5221 (<optab>di3): Likewise.
5222 (<optab><mode>3): Likewise.
5223 (<*optabe>si3_internal): Likewise.
5224 (zero_extendqi<SUPERQI:mode>2): Likewise.
5225 (*add<mode>hi3): Likewise.
5226 (*xor<mode>hi3): Likewise.
5227 (<optab>di3): Likewise.
5228 (*<optab>si3_extend): Likewise.
5229 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5230 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5231 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5233 2017-11-04 Andrew Waterman <andrew@sifive.com>
5235 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5236 TARGET_STRICT_ALIGN based upon -mtune argument.
5238 2017-11-04 Andrew Waterman <andrew@sifive.com>
5240 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5242 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5244 * config/i386/i386.c (choose_basereg): Use optional scratch
5245 register and add assertion.
5246 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5247 needed, and don't allocate stack.
5248 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5249 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5250 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5252 2017-11-03 Jeff Law <law@redhat.com>
5254 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5255 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5256 to probe at the start of a noreturn function.
5258 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5260 PR tree-optimization/78821
5261 * gimple-ssa-store-merging.c: Update the file comment.
5262 (MAX_STORE_ALIAS_CHECKS): Define.
5263 (struct store_operand_info): New type.
5264 (store_operand_info::store_operand_info): New constructor.
5265 (struct store_immediate_info): Add rhs_code and ops data members.
5266 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5267 and op1r arguments to the ctor, initialize corresponding data members.
5268 (struct merged_store_group): Add load_align_base and load_align
5270 (merged_store_group::merged_store_group): Initialize them.
5271 (merged_store_group::do_merge): Update them.
5272 (merged_store_group::apply_stores): Pick the constant for
5273 encode_tree_to_bitpos from one of the two operands, or skip
5274 encode_tree_to_bitpos if neither operand is a constant.
5275 (class pass_store_merging): Add process_store method decl. Remove
5276 bool argument from terminate_all_aliasing_chains method decl.
5277 (pass_store_merging::terminate_all_aliasing_chains): Remove
5278 var_offset_p argument and corresponding handling.
5279 (stmts_may_clobber_ref_p): New function.
5280 (compatible_load_p): New function.
5281 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5282 if there is overlap and rhs_code is not INTEGER_CST. For
5283 non-overlapping stores terminate group if rhs is not mergeable.
5284 (get_alias_type_for_stmts): Change first argument from
5285 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5286 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5287 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5289 (get_location_for_stmts): Change first argument from
5290 auto_vec<gimple *> & to vec<gimple *> &.
5291 (struct split_store): Remove orig_stmts data member, add orig_stores.
5292 (split_store::split_store): Create orig_stores rather than orig_stmts.
5293 (find_constituent_stmts): Renamed to ...
5294 (find_constituent_stores): ... this. Change second argument from
5295 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5296 to info structures rather than the statements.
5297 (split_group): Rename ALLOW_UNALIGNED argument to
5298 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5299 it. Adjust find_constituent_stores caller.
5300 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5301 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5302 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5303 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5304 (mem_valid_for_store_merging): New function.
5305 (handled_load): New function.
5306 (pass_store_merging::process_store): New method.
5307 (pass_store_merging::execute): Use process_store method. Adjust
5308 terminate_all_aliasing_chains caller.
5310 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5312 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5313 Return true for more constants, symbols and label references.
5314 (aarch64_valid_floating_const): Remove unused function.
5316 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5319 * config/aarch64/aarch64.c (aarch64_layout_frame):
5320 Undo forcing of LR at bottom of frame.
5322 2017-11-03 Jeff Law <law@redhat.com>
5325 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5326 for int_registers_saved.
5328 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5329 extracted from tree-ssa-dom.c.
5330 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5331 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5332 (record_equivalences_from_incoming_edge): Add additional argument
5333 to single_pred_edge_ignoring_loop_edges call.
5334 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5335 (uncprop_dom_walker::before_dom_children): Add additional argument
5336 to single_pred_edge_ignoring_loop_edges call.
5337 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5338 single_pred_edge_ignoring_loop_edges rather than open coding.
5339 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5341 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5343 * match.pd (-(-A)): Rewrite.
5345 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5347 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5348 (rs6000_emit_int_cmove): New declaration.
5349 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5350 (rs6000_emit_sISEL): Delete.
5351 (rs6000_emit_int_cmove): Make non-static.
5352 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5353 instead of rs6000_emit_sISEL.
5355 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5357 * asan.c (create_cond_insert_point): Maintain profile.
5358 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5360 * basic-block.h (struct basic_block_def): Remove frequency.
5361 (EDGE_FREQUENCY): Use to_frequency
5362 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5364 (find_traces): Update to use to_frequency.
5365 (find_traces_1_round): Likewise; use only IPA counts.
5366 (bb_to_key): Likewise.
5367 (connect_traces): Use IPA counts only.
5368 (copy_bb_p): Update to use to_frequency.
5369 (fix_up_crossing_landing_pad): Likewise.
5370 (sanitize_hot_paths): Likewise.
5371 * bt-load.c (basic_block_freq): Likewise.
5372 * cfg.c (init_flow): Set count_max to uninitialized.
5373 (check_bb_profile): Remove frequencies; check counts.
5374 (dump_bb_info): Do not dump frequencies.
5375 (update_bb_profile_for_threading): Update counts only.
5376 (scale_bbs_frequencies_int): Likewise.
5377 (MAX_SAFE_MULTIPLIER): Remove.
5378 (scale_bbs_frequencies_gcov_type): Update counts only.
5379 (scale_bbs_frequencies_profile_count): Update counts only.
5380 (scale_bbs_frequencies): Update counts only.
5381 * cfg.h (struct control_flow_graph): Add count-max.
5382 (update_bb_profile_for_threading): Update prototype.
5383 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5384 (find_many_sub_basic_blocks): Likewise.
5385 * cfgcleanup.c (try_forward_edges): Likewise.
5386 (try_crossjump_to_edge): Likewise.
5387 * cfgexpand.c (expand_gimple_cond): Likewise.
5388 (expand_gimple_tailcall): Likewise.
5389 (construct_init_block): Likewise.
5390 (construct_exit_block): Likewise.
5391 * cfghooks.c (verify_flow_info): Check consistency of counts.
5392 (dump_bb_for_graph): Do not dump frequencies.
5393 (split_block_1): Do not update frequencies.
5394 (split_edge): Do not update frequencies.
5395 (make_forwarder_block): Do not update frequencies.
5396 (duplicate_block): Do not update frequencies.
5397 (account_profile_record): Do not update frequencies.
5398 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5399 for global heuristics.
5400 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5401 (expected_loop_iterations_unbounded): Use counts only.
5402 * cfgloopmanip.c (scale_loop_profile): Simplify.
5403 (create_empty_loop_on_edge): Simplify
5405 (duplicate_loop_to_header_edge): Simplify
5406 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5407 (update_br_prob_note): Take care of removing note when profile
5409 (relink_block_chain): Do not dump frequency.
5410 (rtl_account_profile_record): Use to_frequency.
5411 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5412 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5413 (cgraph_update_edges_for_call_stmt_node): Likewise.
5414 (cgraph_edge::verify_count_and_frequency): Update.
5415 (cgraph_node::verify_node): Temporarily disable frequency verification.
5416 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5417 to_cgraph_frequency.
5418 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5419 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5421 (cgraph_node::expand_thunk): Update profile.
5422 * except.c (dw2_build_landing_pads): Do not update frequency.
5423 * final.c (compute_alignments): Use to_frequency.
5424 (dump_basic_block_info): Do not dump frequency.
5425 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5426 (dump_gimple_bb_header): Do not dump frequency.
5427 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5429 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5430 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5431 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5432 (init_before_recovery): Do not update frequency.
5433 (sched_create_recovery_edges): Do not update frequency.
5434 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5435 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5436 (ipa_cp_c_finalize): Set max_count to uninitialized.
5437 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5438 (param_change_prob): Use counts.
5439 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5441 * ipa-split.c (consider_split): Use to_frequency.
5442 (split_function): Use to_frequency.
5443 * ira-build.c (loop_compare_func): Likewise.
5444 (mark_loops_for_removal): Likewise.
5445 (mark_all_loops_for_removal): Likewise.
5446 * loop-doloop.c (doloop_modify): Do not update frequency.
5447 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5449 * lto-streamer-in.c (input_function): Update count_max.
5450 * omp-expand.c (expand_omp_taskreg): Update count_max.
5451 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5452 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5453 (maybe_hot_count_p): Use ipa counts only.
5454 (maybe_hot_bb_p): Simplify.
5455 (maybe_hot_edge_p): Simplify.
5456 (probably_never_executed): Do not take frequency argument.
5457 (probably_never_executed_bb_p): Do not pass frequency.
5458 (probably_never_executed_edge_p): Likewise.
5459 (combine_predictions_for_bb): Check that profile is nonzero.
5460 (propagate_freq): Do not set frequency.
5461 (drop_profile): Simplify.
5462 (counts_to_freqs): Simplify.
5463 (expensive_function_p): Use to_frequency.
5464 (propagate_unlikely_bbs_forward): Simplify.
5465 (determine_unlikely_bbs): Simplify.
5466 (estimate_bb_frequencies): Add hack to silence graphite issues.
5467 (compute_function_frequency): Use ipa counts.
5468 (pass_profile::execute): Update.
5469 (rebuild_frequencies): Use counts only.
5470 (force_edge_cold): Use counts only.
5471 * profile-count.c (profile_count::dump): Dump new count types.
5472 (profile_count::differs_from_p): Check compatiblity.
5473 (profile_count::to_frequency): New function.
5474 (profile_count::to_cgraph_frequency): New function.
5475 * profile-count.h (struct function): Declare.
5476 (enum profile_quality): Add profile_guessed_local and
5477 profile_guessed_global0.
5478 (class profile_proability): Decrease number of bits to 29;
5479 update from_reg_br_prob_note and to_reg_br_prob_note.
5480 (class profile_count: Update comment; decrease number of bits
5481 to 61. Check compatibility.
5482 (profile_count::compatible_p): New private member function.
5483 (profile_count::ipa_p): New member function.
5484 (profile_count::operator<): Handle global zero correctly.
5485 (profile_count::operator>): Handle global zero correctly.
5486 (profile_count::operator<=): Handle global zero correctly.
5487 (profile_count::operator>=): Handle global zero correctly.
5488 (profile_count::nonzero_p): New member function.
5489 (profile_count::force_nonzero): New member function.
5490 (profile_count::max): New member function.
5491 (profile_count::apply_scale): Handle IPA scalling.
5492 (profile_count::guessed_local): New member function.
5493 (profile_count::global0): New member function.
5494 (profile_count::ipa): New member function.
5495 (profile_count::to_frequency): Declare.
5496 (profile_count::to_cgraph_frequency): Declare.
5497 * profile.c (OVERLAP_BASE): Delete.
5498 (compute_frequency_overlap): Delete.
5499 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5500 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5501 * sched-ebb.c (rank): Use counts only.
5502 * shrink-wrap.c (handle_simple_exit): Use counts only.
5503 (try_shrink_wrapping): Use counts only.
5504 (place_prologue_for_one_component): Use counts only.
5505 * tracer.c (find_best_predecessor): Use to_frequency.
5506 (find_trace): Use to_frequency.
5507 (tail_duplicate): Use to_frequency.
5508 * trans-mem.c (expand_transaction): Do not update frequency.
5509 * tree-call-cdce.c: Do not update frequency.
5510 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5511 (gimple_merge_blocks): Likewise.
5512 (gimple_split_edge): Likewise.
5513 (gimple_duplicate_sese_region): Likewise.
5514 (gimple_duplicate_sese_tail): Likewise.
5515 (move_sese_region_to_fn): Likewise.
5516 (gimple_account_profile_record): Likewise.
5517 (insert_cond_bb): Likewise.
5518 * tree-complex.c (expand_complex_div_wide): Likewise.
5519 * tree-eh.c (lower_resx): Update profile.
5520 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5522 (initialize_cfun): Do not initialize frequencies
5523 (freqs_to_counts): Delete.
5524 (copy_cfg_body): Ignore count parameter.
5525 (copy_body): Update.
5526 (expand_call_inline): Update count_max.
5527 (optimize_inline_calls): Update count_max.
5528 (tree_function_versioning): Update count_max.
5529 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5530 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5532 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5533 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5534 (try_peel_loop): Likewise.
5535 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5537 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5538 (tree_transform_and_unroll_loop): Do not use frequencies
5539 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5540 Use reliable prediction only.
5541 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5542 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5543 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5544 probability scaling.
5545 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5546 not update frequency
5547 (any_remaining_duplicated_blocks): Likewise.
5548 (update_profile): Likewise.
5549 (estimated_freqs_path): Delete.
5550 (freqs_to_counts_path): Delete.
5551 (clear_counts_path): Delete.
5552 (ssa_fix_duplicate_block_edges): Likewise.
5553 (duplicate_thread_path): Likewise.
5554 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5555 * tree-tailcall.c (decrease_profile): Do not update frequency.
5556 (eliminate_tail_call): Likewise.
5557 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5558 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5559 (optimize_mask_stores): Likewise.
5560 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5561 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5562 (ubsan_expand_ptr_ifn): Update profile.
5563 * value-prof.c (gimple_ic): Simplify.
5564 * value-prof.h (gimple_ic): Update prototype.
5565 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5566 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5568 (want_inline_self_recursive_call_p): Likewise.
5569 (resolve_noninline_speculation): Only cummulate defined counts.
5570 (inline_small_functions): Use nonzero_p.
5571 (ipa_inline): Do not access freed node.
5573 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5575 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5576 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5578 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5580 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5581 non-legitimate address.
5583 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5585 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5586 (*lt0_<mode>di, *lt0_<mode>si): New.
5588 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5590 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5591 TARGET_PAIRED_FLOAT.
5593 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5594 Jim Wilson <jim.wilson@linaro.org>
5596 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5597 * config/aarch64/aarch64-tune.md: Regenerated.
5598 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5599 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5601 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5603 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5604 (arc_expand_prologue): Restore blink for millicode.
5605 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5607 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5610 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5611 gen_vec_duplicate after forcing the scalar into a register.
5613 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5615 * combine (try_combine): Print the insns input to try_combine to the
5618 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5621 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5622 Remove second argument from aarch64_process_target_attr call.
5623 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5625 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5627 (aarch64_handle_attr_arch): Remove second argument.
5628 (aarch64_handle_attr_cpu): Ditto.
5629 (aarch64_handle_attr_tune): Ditto.
5630 (aarch64_handle_attr_isa_flags): Ditto.
5631 (aarch64_process_one_target_attr): Ditto.
5632 (aarch64_process_target_attr): Ditto.
5633 (aarch64_option_valid_attribute_p): Remove second argument.
5634 on aarch64_process_target_attr call.
5636 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5638 * diagnostic.c: Include "selftest-diagnostic.h".
5639 (selftest::assert_location_text): New function.
5640 (selftest::test_diagnostic_get_location_text): New function.
5641 (selftest::diagnostic_c_tests): Call it.
5643 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5645 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5646 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5647 (class selftest::test_diagnostic_context): Move to...
5648 * selftest-diagnostic.c: New file.
5649 * selftest-diagnostic.h: New file.
5651 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5653 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5654 offset range for FT32B.
5655 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5656 * config/ft32/ft32.md: Add TARGET_NOPM.
5657 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5658 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5660 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5662 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5664 2017-11-02 Jeff Law <law@redhat.com>
5666 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5667 virtual keyword on FINAL OVERRIDE members.
5669 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5670 virtuals together. Add virtual destructor.
5671 (substitute_and_fold_engine): Similarly.
5673 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5675 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5677 2017-11-02 Richard Biener <rguenther@suse.de>
5679 PR tree-optimization/82795
5680 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5682 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5684 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5686 (gcc_GAS_CHECK_FEATURE): Remove.
5687 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5688 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5689 * configure: Regenerate.
5691 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5693 * config/arc/arc.c (hwloop_optimize): Account for empty
5696 2017-11-02 Richard Biener <rguenther@suse.de>
5699 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5700 Truncate ARRAY_REF index and element size.
5702 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5704 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5706 2017-11-01 Jeff Law <law@redhat.com>
5708 * tree-ssa-ccp.c (ccp_folder): New class derived from
5709 substitute_and_fold_engine.
5710 (ccp_folder::get_value): New member function.
5711 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5712 (ccp_fold_stmt): Remove prototype.
5713 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5714 * tree-ssa-copy.c (copy_folder): New class derived from
5715 substitute_and_fold_engine.
5716 (copy_folder::get_value): Renamed from get_value.
5717 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5718 * tree-vrp.c (vrp_folder): New class derived from
5719 substitute_and_fold_engine.
5720 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5721 (vrp_folder::get_value): New member function.
5722 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5723 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5724 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5725 provide a class interface to folder/substitute routines.
5726 (ssa_prop_fold_stmt_fn): Remove typedef.
5727 (ssa_prop_get_value_fn): Likewise.
5728 (subsitute_and_fold): Remove prototype.
5729 (replace_uses_in): Likewise.
5730 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5731 Renamed from replace_uses_in. Call the virtual member function
5732 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5733 (substitute_and_fold_dom_walker): Remove initialization of
5734 data member entries for calbacks. Add substitute_and_fold_engine
5735 member and initialize it.
5736 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5737 member functions for get_value, replace_phi_args_in c
5738 replace_uses_in, and fold_stmt calls.
5739 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5740 substitute_and_fold. Remove assert. Update ctor call.
5742 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5743 (ssa_prop_visit_phi_fn): Likewise.
5744 (class ssa_propagation_engine): New class to provide an interface
5746 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5748 (ssa_prop_visit_phi): Likewise.
5749 (ssa_propagation_engine::simulate_stmt): Moved into class.
5750 Call visit_phi/visit_stmt from the class rather than via
5751 file scoped static variables.
5752 (ssa_propagation_engine::simulate_block): Moved into class.
5753 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5754 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5755 set file scoped statics for the visit_stmt/visit_phi callbacks.
5756 * tree-complex.c (complex_propagate): New class derived from
5757 ssa_propagation_engine.
5758 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5759 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5760 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5762 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5763 ssa_propagation_engine.
5764 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5765 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5766 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5767 * tree-ssa-copy.c (copy_prop): New class derived from
5768 ssa_propagation_engine.
5769 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5770 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5771 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5772 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5773 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5774 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5775 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5777 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5779 PR rtl-optimization/82778
5780 PR rtl-optimization/82597
5781 * compare-elim.c (struct comparison): Add in_a_setter field.
5782 (find_comparison_dom_walker::before_dom_children): Remove killed
5783 bitmap and df_simulate_find_defs call, instead walk the defs.
5784 Compute last_setter and initialize in_a_setter. Merge definitions
5785 with first initialization for a few variables.
5786 (try_validate_parallel): Use insn_invalid_p instead of
5787 recog_memoized. Return insn rather than just the pattern.
5788 (try_merge_compare): Fix up comment. Don't uselessly test if
5789 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5791 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5794 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5795 Alan Hayward <alan.hayward@arm.com>
5796 David Sherwood <david.sherwood@arm.com>
5798 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5799 aarch64_hard_regno_nregs to get the number of registers
5802 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5803 Alan Hayward <alan.hayward@arm.com>
5804 David Sherwood <david.sherwood@arm.com>
5806 * config/aarch64/constraints.md (Upl): Rename to...
5808 * config/aarch64/aarch64.md
5809 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5812 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5813 Alan Hayward <alan.hayward@arm.com>
5814 David Sherwood <david.sherwood@arm.com>
5816 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
5817 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
5820 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5821 Alan Hayward <alan.hayward@arm.com>
5822 David Sherwood <david.sherwood@arm.com>
5824 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
5825 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
5826 (aarch64_evpc_dup): Generate rtl direcly, rather than using
5828 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
5829 of a single element.
5830 * config/aarch64/iterators.md: Add a comment above the permute
5831 unspecs to say that they are generated directly by
5832 aarch64_expand_vec_perm_const.
5833 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
5835 2017-11-01 Nathan Sidwell <nathan@acm.org>
5837 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
5839 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5841 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
5842 and medany code models, and describe what they do.
5844 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5846 Revert accidental duplicate:
5848 * combine.c (can_change_dest_mode): Reject changes in
5849 REGMODE_NATURAL_SIZE.
5851 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
5853 PR rtl-optimization/64682
5854 PR rtl-optimization/69567
5855 PR rtl-optimization/69737
5856 PR rtl-optimization/82683
5857 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
5858 register mentioned in the note, drop the note, unless it came from I3,
5859 in which case it should go to I3 again.
5861 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5863 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
5864 and return false if not.
5865 (clear_bytes_written_by, live_bytes_read): Update accordingly.
5867 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5869 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
5870 range is known to be empty.
5872 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5873 Alan Hayward <alan.hayward@arm.com>
5874 David Sherwood <david.sherwood@arm.com>
5876 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
5877 and CONST_VECTOR_NUNITS instead of computing the number of units from
5878 the byte sizes of the vector and element.
5879 (simplify_binary_operation_1): Likewise.
5880 (simplify_const_binary_operation): Likewise.
5881 (simplify_ternary_operation): Likewise.
5883 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5884 Alan Hayward <alan.hayward@arm.com>
5885 David Sherwood <david.sherwood@arm.com>
5887 * var-tracking.c (INT_MEM_OFFSET): Replace with...
5888 (int_mem_offset): ...this new function.
5889 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
5890 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
5891 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
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 * lower-subreg.c (interesting_mode_p): New function.
5899 (compute_costs, find_decomposable_subregs, decompose_register)
5900 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
5901 (resolve_clobber, dump_choices): Use it.
5903 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5904 Alan Hayward <alan.hayward@arm.com>
5905 David Sherwood <david.sherwood@arm.com>
5907 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
5909 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5910 Alan Hayward <alan.hayward@arm.com>
5911 David Sherwood <david.sherwood@arm.com>
5913 * alias.c (find_base_value, find_base_term): Only process integer
5914 truncations. Check the precision rather than the size.
5916 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5917 Alan Hayward <alan.hayward@arm.com>
5918 David Sherwood <david.sherwood@arm.com>
5920 * machmode.h (is_narrower_int_mode): New function
5921 * optabs.c (expand_float, expand_fix): Use it.
5922 * dwarf2out.c (rotate_loc_descriptor): Likewise.
5924 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5925 Alan Hayward <alan.hayward@arm.com>
5926 David Sherwood <david.sherwood@arm.com>
5928 * rtl.h (narrower_subreg_mode): New function.
5929 * ira-color.c (update_costs_from_allocno): Use it.
5931 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5932 Alan Hayward <alan.hayward@arm.com>
5933 David Sherwood <david.sherwood@arm.com>
5935 * optabs-query.h (convert_optab_p): New function, split out from...
5936 (convert_optab_handler): ...here.
5937 (widening_optab_handler): Delete.
5938 (find_widening_optab_handler): Remove permit_non_widening parameter.
5939 (find_widening_optab_handler_and_mode): Likewise. Provide an
5940 override that operates on mode class wrappers.
5941 * optabs-query.c (widening_optab_handler): Delete.
5942 (find_widening_optab_handler_and_mode): Remove permit_non_widening
5943 parameter. Assert that the two modes are the same class and that
5944 the "from" mode is narrower than the "to" mode. Use
5945 convert_optab_handler instead of widening_optab_handler.
5946 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
5947 instead of widening_optab_handler.
5948 * expr.c (expand_expr_real_2): Update calls to
5949 find_widening_optab_handler.
5950 * optabs.c (expand_widen_pattern_expr): Likewise.
5951 (expand_binop_directly): Take the insn_code as a parameter.
5952 (expand_binop): Only call find_widening_optab_handler for
5953 conversion optabs; use optab_handler otherwise. Update calls
5954 to find_widening_optab_handler and expand_binop_directly.
5955 Use convert_optab_handler instead of widening_optab_handler.
5956 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
5957 find_widening_optab_handler and use scalar_mode rather than
5959 (convert_plusminus_to_widen): Likewise.
5961 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5962 Alan Hayward <alan.hayward@arm.com>
5963 David Sherwood <david.sherwood@arm.com>
5965 * machmode.h (fixed_size_mode): New class.
5966 * rtl.h (get_pool_mode): Return fixed_size_mode.
5967 * gengtype.c (main): Add fixed_size_mode.
5968 * target.def (get_raw_result_mode): Return a fixed_size_mode.
5969 (get_raw_arg_mode): Likewise.
5970 * doc/tm.texi: Regenerate.
5971 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
5972 * targhooks.c (default_get_reg_raw_mode): Likewise.
5973 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
5974 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
5975 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
5976 (msp430_get_raw_result_mode): Likewise.
5977 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
5978 * dbxout.c (dbxout_parms): Require fixed-size modes.
5979 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
5980 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5981 * omp-low.c (lower_oacc_reductions): Likewise.
5982 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
5983 (simplify_subreg): Update accordingly.
5984 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
5985 (force_const_mem): Update accordingly. Return NULL_RTX for modes
5986 that aren't fixed-size.
5987 (get_pool_mode): Return a fixed_size_mode.
5988 (output_constant_pool_2): Take a fixed_size_mode.
5990 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5991 Alan Hayward <alan.hayward@arm.com>
5992 David Sherwood <david.sherwood@arm.com>
5994 * doc/rtl.texi (vec_series): Document.
5995 (const): Say that the operand can be a vec_series.
5996 * rtl.def (VEC_SERIES): New rtx code.
5997 * rtl.h (const_vec_series_p_1): Declare.
5998 (const_vec_series_p): New function.
5999 * emit-rtl.h (gen_const_vec_series): Declare.
6000 (gen_vec_series): Likewise.
6001 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6002 (gen_vec_series): Likewise.
6003 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6004 * simplify-rtx.c (simplify_unary_operation): Handle negations
6006 (simplify_binary_operation_series): New function.
6007 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6008 (test_vector_ops_series): New function.
6009 (test_vector_ops): Call it.
6010 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6011 gen_const_vec_series.
6012 (altivec_lvsr): Likewise.
6013 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6015 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6016 Alan Hayward <alan.hayward@arm.com>
6017 David Sherwood <david.sherwood@arm.com>
6019 * doc/rtl.texi (const): Update description of address constants.
6020 Say that vector constants are allowed too.
6021 * common.md (E, F): Use CONSTANT_P instead of checking for
6023 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6024 checking for CONST_VECTOR.
6025 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6027 * expr.c (expand_expr_real_2): Check for vector modes instead
6028 of checking for CONST_VECTOR.
6029 * rtl.h (const_vec_p): New function.
6030 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6031 (unwrap_const_vec_duplicate): Handle them here too.
6033 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6034 David Malcolm <dmalcolm@redhat.com>
6035 Alan Hayward <alan.hayward@arm.com>
6036 David Sherwood <david.sherwood@arm.com>
6038 * rtl.h (vec_duplicate_p): New function.
6039 * selftest-rtl.c (assert_rtx_eq_at): New function.
6040 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6041 (assert_rtx_eq_at): Declare.
6042 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6043 * selftest-run-tests.c (selftest::run_tests): Call it.
6044 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6045 (simplify_unary_operation_1): Recursively handle vector duplicates.
6046 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6048 (simplify_subreg): Handle subregs of vector duplicates.
6049 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6050 (selftest::simplify_rtx_c_tests): New functions.
6052 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6053 Alan Hayward <alan.hayward@arm.com>
6054 David Sherwood <david.sherwood@arm.com>
6056 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6057 (gen_vec_duplicate): Likewise.
6058 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6060 (gen_const_vector): ...here.
6061 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6062 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6063 whose elements are all equal.
6064 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6065 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6066 (simplify_relational_operation): Likewise.
6067 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6069 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6070 (aarch64_expand_vector_init): Likewise.
6071 * config/arm/arm.c (neon_vdup_constant): Likewise.
6072 (neon_expand_vector_init): Likewise.
6073 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6074 (arm_block_set_unaligned_vect): Likewise.
6075 (arm_block_set_aligned_vect): Likewise.
6076 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6077 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6078 (expand_vec_perm_even_odd_pack): Likewise.
6079 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6080 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6081 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6082 gen_const_vec_duplicate.
6083 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6084 * config/mips/mips.c (mips_gen_const_int_vector): Use
6085 gen_const_vec_duplicate.
6086 (mips_expand_vector_init): Use CONST0_RTX.
6087 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6088 (define_split): Use gen_const_vec_duplicate.
6089 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6090 (define_split): Use gen_const_vec_duplicate.
6091 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6092 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6093 * config/spu/spu.c (spu_const): Likewise.
6095 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6096 Alan Hayward <alan.hayward@arm.com>
6097 David Sherwood <david.sherwood@arm.com>
6099 * combine.c (can_change_dest_mode): Reject changes in
6100 REGMODE_NATURAL_SIZE.
6102 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6104 * configure.ac (--enable-libssp): New.
6105 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6106 trying to determine target-specific default. Adjust indentation.
6107 * configure: Regenerated.
6108 * doc/install.texi (Configuration): Expand --disable-libssp
6111 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6113 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6116 2017-10-31 Martin Jambor <mjambor@suse.cz>
6119 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6121 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6123 * auto-profile.c (autofdo_source_profile::read): Use
6124 UNKNOWN_LOCATION rather than 0.
6125 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6126 (warning_at): ...this overload.
6127 (warning_at_rich_loc_n): Rename to...
6128 (warning_n): ...this overload.
6129 (error_at_rich_loc): Rename to...
6130 (error_at): ...this overload.
6131 (pedwarn_at_rich_loc): Rename to...
6132 (pedwarn): ...this overload.
6133 (permerror_at_rich_loc): Rename to...
6134 (permerror): ...this overload.
6135 (inform_at_rich_loc): Rename to...
6136 (inform): ...this overload.
6137 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6138 (diagnostic_n_impl_richloc): Rename to...
6139 (diagnostic_n_impl): ...this rich_location *-based decl.
6140 (inform_at_rich_loc): Rename to...
6141 (inform): ...this, and add an assertion.
6142 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6143 (warning_at_rich_loc): Rename to...
6144 (warning_at): ...this, and add an assertion.
6145 (warning_at_rich_loc_n): Rename to...
6146 (warning_n): ...this, and add an assertion.
6147 (warning_n): Update location_t-based implementation for removal of
6148 location_t-based diagnostic_n_impl.
6149 (pedwarn_at_rich_loc): Rename to...
6150 (pedwarn): ...this, and add an assertion.
6151 (permerror_at_rich_loc): Rename to...
6152 (permerror): ...this, and add an assertion.
6153 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6154 (error_at_rich_loc): Rename to...
6155 (error_at): ...this, and add an assertion.
6156 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6157 (driver::do_spec_on_infiles): Likewise.
6158 * substring-locations.c (format_warning_va): Update for renaming
6159 of inform_at_rich_loc.
6161 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6163 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6164 _Float<N>X built-in functions so that the variant without the
6165 "__builtin_" prefix is only enabled for the GNU C and Objective C
6166 languages when they are in non-strict ANSI/ISO mode.
6167 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6168 * target.def (floatn_builtin_p): Add a target hook to control
6169 whether _Float<N> and _Float<N>X built-in functions without the
6170 "__builtin_" prefix are enabled, and return true for C and
6171 Objective C in the default hook. Include langhooks.h in
6173 * targhooks.h (default_floatn_builtin_p): Likewise.
6174 * targhooks.c (default_floatn_builtin_p): Likewise.
6175 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6176 floatn_builtin_p target hook.
6177 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6179 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6180 Eric Botcazou <ebotcazou@adacore.com>
6182 PR rtl-optimization/81803
6183 * lra-constraints.c (curr_insn_transform): Also reload the whole
6184 register for a strict subreg no wider than a word if this is for
6185 a WORD_REGISTER_OPERATIONS target.
6187 2017-10-31 Jason Merrill <jason@redhat.com>
6189 * gdbinit.in: Skip over inlines from timevar.h.
6191 2017-10-31 Martin Liska <mliska@suse.cz>
6193 * doc/gcov.texi: Document new option.
6194 * gcov.c (print_usage): Likewise print it.
6195 (process_args): Support the argument.
6196 (format_count): New function.
6197 (format_gcov): Use the function.
6199 2017-10-31 Martin Liska <mliska@suse.cz>
6201 * gcov.c (struct name_map): do not use typedef.
6202 Define operator== and operator<.
6203 (name_search): Remove.
6204 (name_sort): Remove.
6205 (main): Do not allocate names.
6206 (process_file): Add vertical space.
6207 (generate_results): Use std::find.
6208 (release_structures): Do not release memory.
6209 (find_source): Use std::find.
6211 2017-10-31 Martin Liska <mliska@suse.cz>
6213 * gcov.c (struct line_info): Remove it's typedef.
6214 (line_info::line_info): Add proper ctor.
6215 (line_info::has_block): Do not use a typedef.
6216 (struct source_info): Do not use typedef.
6217 (circuit): Likewise.
6218 (get_cycles_count): Likewise.
6219 (output_intermediate_file): Iterate via vector iterator.
6220 (add_line_counts): Use std::vector methods.
6221 (accumulate_line_counts): Likewise.
6222 (output_lines): Likewise.
6224 2017-10-31 Martin Liska <mliska@suse.cz>
6226 * gcov.c (struct source_info): Remove typedef.
6227 (source_info::source_info): Add proper ctor.
6228 (accumulate_line_counts): Use struct, not it's typedef.
6229 (output_gcov_file): Likewise.
6230 (output_lines): Likewise.
6231 (main): Do not allocate an array.
6232 (output_intermediate_file): Use size of vector container.
6233 (process_file): Resize the vector.
6234 (generate_results): Do not preallocate, use newly added vector
6236 (release_structures): Do not release sources.
6237 (find_source): Use vector methods.
6238 (add_line_counts): Do not use typedef.
6240 2017-10-31 Martin Liska <mliska@suse.cz>
6242 * doc/gcov.texi: Document that.
6243 * gcov.c (add_line_counts): Mark lines with a non-executed
6245 (output_line_beginning): Handle such lines.
6246 (output_lines): Pass new argument.
6247 (output_intermediate_file): Print it in intermediate format.
6249 2017-10-31 Martin Liska <mliska@suse.cz>
6251 * color-macros.h: New file.
6252 * diagnostic-color.c: Factor out color related to macros to
6254 * doc/gcov.texi: Document -k option.
6255 * gcov.c (INCLUDE_STRING): Include string.h.
6256 (print_usage): Add -k option.
6257 (process_args): Parse it.
6258 (pad_count_string): New function.
6259 (output_line_beginning): Likewise.
6260 (DEFAULT_LINE_START): New macro.
6261 (output_lines): Support color output.
6263 2017-10-31 Martin Liska <mliska@suse.cz>
6265 PR gcov-profile/82633
6266 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6267 their interaction with GCOV infrastructure.
6268 * configure.ac: Add -fkeep-{inline,static}-functions to
6270 * configure: Regenerate.
6272 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6275 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6277 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6280 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6281 into a register if it does not fit into an immediate offset field.
6283 2017-10-31 Olivier Hainque <hainque@adacore.com>
6285 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6287 2017-10-31 Julia Koval <julia.koval@intel.com>
6289 * config.gcc: Add gfniintrin.h.
6290 * config/i386/gfniintrin.h: New.
6291 * config/i386/i386-builtin-types.def
6292 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6293 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6294 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6295 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6296 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6297 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6298 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6299 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6300 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6301 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6302 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6303 * config/i386/immintrin.h: Include gfniintrin.h.
6304 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6306 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6308 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6310 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6312 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6313 (arm_ashldi3_1bit): Remove pattern.
6314 (ashrdi3): Remove shift by 1 expansion.
6315 (arm_ashrdi3_1bit): Remove pattern.
6316 (lshrdi3): Remove shift by 1 expansion.
6317 (arm_lshrdi3_1bit): Remove pattern.
6318 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6319 cost of ashldi3 by 1.
6320 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6321 (<shift>di3_neon): Likewise.
6323 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6325 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6326 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6327 and (*aarch64_simd_mov<VQ:mode>).
6328 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6329 pattern alternative.
6330 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6331 attributes to match pattern alternative.
6333 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6335 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6336 * config/rs6000/emmintrin.h: New file.
6337 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6339 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6341 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6342 * (movdi_vfp_cortexa8): Remove pattern.
6344 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6346 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6348 (Specific, alpha*-dec-osf5.1): Remove.
6349 (Specific, mips-sgi-irix5): Remove.
6350 (Specific, mips-sgi-irix6): Remove.
6352 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6355 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6356 arguments to clear_bit_region_be.
6358 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6360 * gimplify.c: Include memmodel.h.
6362 2017-10-30 Martin Jambor <mjambor@suse.cz>
6364 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6365 condition whether loop should be executed at all.
6367 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6369 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6370 gimple folding of vec_madd() intrinsics.
6371 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6372 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6373 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6375 2017-10-30 Richard Biener <rguenther@suse.de>
6377 PR tree-optimization/82762
6379 2017-10-23 Richard Biener <rguenther@suse.de>
6381 PR tree-optimization/82129
6383 2017-08-01 Richard Biener <rguenther@suse.de>
6385 PR tree-optimization/81181
6386 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6387 (compute_antic): ... end of iteration here.
6389 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6391 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6392 -std=iso9899:2017 and -std=gnu17.
6393 * doc/standards.texi (C Language): Document C17 support.
6394 * doc/cpp.texi (Overview): Mention -std=c17.
6395 (Standard Predefined Macros): Document C11 and C17 values of
6396 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6397 * doc/extend.texi (Inline): Do not list individual options for
6398 standards newer than C99.
6399 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6401 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6404 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6406 * asan.c (asan_finish_file): Align asan globals array by shadow
6409 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6412 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6413 (struct store_immediate_info): Add bitregion_start and bitregion_end
6415 (store_immediate_info::store_immediate_info): Add brs and bre
6416 arguments and initialize bitregion_{start,end} from those.
6417 (struct merged_store_group): Add bitregion_start, bitregion_end,
6418 align_base and mask fields. Drop unnecessary struct keyword from
6419 struct store_immediate_info. Add do_merge method.
6420 (clear_bit_region_be): Use memset instead of loop storing zeros.
6421 (merged_store_group::do_merge): New method.
6422 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6423 stores as long as the surrounding bitregions have no gaps.
6424 (merged_store_group::merge_overlapping): Use do_merge.
6425 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6426 is byte aligned, rather than requiring that start and width are
6427 byte aligned. Drop unnecessary struct keyword from
6428 struct store_immediate_info. Allocate and populate also mask array.
6429 Make start of the arrays relative to bitregion_start rather than
6430 start and size them according to bitregion_{end,start} difference.
6431 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6432 struct store_immediate_info.
6433 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6434 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6435 struct keyword from struct store_immediate_info.
6436 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6437 between stores as long as the surrounding bitregions have no gaps.
6439 (struct split_store): Add orig non-static data member.
6440 (split_store::split_store): Initialize orig to false.
6441 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6442 if there is exactly a single original stmt. Change stmts argument
6443 to pointer from reference, if NULL, don't push anything to it. Add
6444 first argument, use it to optimize skipping over orig stmts that
6445 are known to be before bitpos already. Simplify.
6446 (split_group): Return unsigned int count how many stores are or
6447 would be needed rather than a bool. Add allow_unaligned argument.
6448 Change split_stores argument from reference to pointer, if NULL,
6449 only do a dry run computing how many stores would be produced.
6450 Rewritten algorithm to use both alignment and misalign if
6451 !allow_unaligned and handle bitfield stores with gaps.
6452 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6453 from bitregion_start instead of start. Compute allow_unaligned
6454 here, if true, do 2 split_group dry runs to compute which one
6455 produces fewer stores and prefer aligned if equal. Punt if
6456 new count is bigger or equal than original before emitting any
6457 statements, rather than during that. Remove no longer needed
6458 new_ssa_names tracking. Replace num_stmts with
6459 split_stores.length (). Use 32-bit stack allocated entries
6460 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6461 if possible. Handle bitfields with gaps.
6462 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6463 Compute bitregion_{start,end} for the stores and construct
6464 store_immediate_info with that. Formatting fixes.
6466 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6469 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6470 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6472 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6474 * gimplify.c: Include tm_p.h.
6476 * common.opt (gcoff): Re-add as ignored option.
6477 (gcoff1, gcoff2, gcoff3): Likewise.
6479 * Makefile.in (OBJS): Delete sdbout.o.
6480 (GTFILES): Delete $(srcdir)/sdbout.c.
6481 * debug.h: Delete sdb_debug_hooks.
6482 * final.c: Delete sdbout.h include.
6483 (final_scan_insn): Delete SDB_DEBUG check.
6484 (rest_of_clean_state): Likewise.
6485 * output.h: Delete sdb_begin_function_line.
6488 * toplev.c: Delete sdbout.h include.
6489 (process_options): Delete SDB_DEBUG check.
6490 * tree-core.h (tree_type_common): Delete pointer field of
6492 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6493 TYPE_SYMTAB_POINTER.
6494 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6495 (TYPE_SYMTAB_IS_POINTER): Delete.
6496 (TYPE_SYMTAB_IS_DIE): Renumber.
6497 * xcoffout.c: Refer to former sdbout.c file.
6498 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6500 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6501 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6502 (Debugging Options): Delete -gcoff.
6503 (-gstabs): Delete SDB reference.
6505 (-gcoff@var{level}): Delete.
6506 * doc/passes.texi (Debugging information output): Delete SDB and
6507 sdbout.c references.
6508 * doc/tm.texi: Regenerate.
6509 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6510 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6512 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6513 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6514 (SDB_DEBUGGING_INFO): Delete.
6515 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6516 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6517 * target.def (output_source_filename): Delete COFF reference.
6519 * common.opt (gcoff): Delete.
6520 (gxcoff+): Update Negative chain.
6521 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6523 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6524 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6525 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6527 (expand_function_start): Change sdb reference to past tense.
6528 (expand_function_end): Change sdb reference to past tense.
6529 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6530 * opts.c (debug_type_names): Delete coff entry.
6531 (common_handle_option): Delete OPT_gcoff case.
6532 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6534 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6535 * config/cris/cris.h: Delete SDB reference in comment.
6536 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6537 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6538 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6539 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6541 (ix86_expand_prologue): Likewise.
6542 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6543 * config/ia64/ia64.h: Likewise.
6544 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6545 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6547 * config/mmix/mmix.h: Likewise.
6548 * config/nds32/nds32.c: Likewise.
6549 * config/stormy/storym16.h: Likewise.
6550 * config/visium/visium.h: Likewise.
6551 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6553 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6555 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6556 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6557 FRAME_POINTER_REGNUM point at high end of local var area.
6559 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6561 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6562 Move comment around. Do not reset best_edge for a copiable
6563 destination if the copy would cause a partition change.
6564 (better_edge_p): Remove redundant check.
6566 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6568 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6570 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6572 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6573 for math functions that have _Float<N> and _Float<N>X variants.
6574 (mathfn_built_in_2): Add support for math functions that have
6575 _Float<N> and _Float<N>X variants.
6576 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6577 (expand_builtin_mathfn_ternary): Add support for fma with
6578 _Float<N> and _Float<N>X variants.
6579 (expand_builtin): Likewise.
6580 (fold_builtin_3): Likewise.
6581 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6582 create math function _Float<N> and _Float<N>X variants as external
6584 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6585 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6586 the __builtin_ prefix and if not strict ansi, without the prefix.
6587 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6588 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6589 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6590 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6591 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6592 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6593 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6594 function signatures for fma _Float<N> and _Float<N>X variants.
6595 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6596 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6597 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6598 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6599 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6600 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6601 * gencfn-macros.c (print_case_cfn): Add support for math functions
6602 that have _Float<N> and _Float<N>X variants.
6603 (print_define_operator_list): Likewise.
6604 (fltfn_suffixes): Likewise.
6606 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6607 for math functions that have _Float<N> and _Float<N>X variants.
6608 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6609 and _Float<N>X variants.
6610 (COPYSIGN): Likewise.
6613 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6614 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6616 (integer_valued_read_call_p): Likewise.
6617 * fold-const-call.c (fold_const_call_ss): Likewise.
6618 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6619 _Float<N> and _Float<N>X variants.
6620 (fold_const_call_ssss): Add support for fma _Float<N> and
6621 _Float<N>X variants.
6622 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6623 support for copysign and fma _Float<N> and _Float<N>X variants.
6624 (backprop::process_builtin_call_use): Likewise.
6625 * tree-call-cdce.c (can_test_argument_range); Add support for
6626 sqrt _Float<N> and _Float<N>X variants.
6627 (edom_only_function): Likewise.
6628 (get_no_error_domain): Likewise.
6629 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6630 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6631 copysign _Float<N> and _Float<N>X variants.
6632 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6633 handled by machine independent code.
6634 (FMAF128): Likewise.
6635 * doc/cpp.texi (Common Predefined Macros): Document defining
6636 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6637 fma _Float<N> and _Float<N>X variants.
6639 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6642 * config/i386/i386-modes.def (CCFPU): Remove definition.
6643 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6644 (ix86_cc_modes_compatible): Ditto.
6645 (ix86_expand_carry_flag_compare): Ditto.
6646 (ix86_expand_int_movcc): Ditto.
6647 (ix86_expand_int_addcc): Ditto.
6648 (ix86_reverse_condition): Ditto.
6649 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6650 Return true/false for unordered/ordered fp comparisons.
6651 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6652 (ix86_prepare_fp_compare_args): Update for rename.
6653 (ix86_expand_fp_compare): Update for rename. Generate unordered
6654 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6655 (ix86_expand_sse_compare_and_jump): Ditto.
6656 * config/i386/predicates.md (fcmov_comparison_operator):
6657 Remove CCFPU mode handling.
6658 (ix86_comparison_operator): Ditto.
6659 (ix86_carry_flag_operator): Ditto.
6660 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6661 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6662 (*cmpu<mode>_cc_i387): Ditto.
6663 (FPCMP): Remove mode iterator.
6664 (unord): Remove mode attribute.
6665 (unord_subst): New define_subst transformation
6666 (unord): New define_subst attribute.
6668 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6669 (*cmpi<unord>xf_i387): Ditto.
6670 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6671 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6672 using unord_subst transformation.
6673 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6674 (round_saeonly): Also handle CCFP mode.
6675 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6676 Remove UNSPEC_SAHF unspec handling.
6678 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6680 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6682 2017-10-27 Jeff Law <law@redhat.com>
6684 * gimple-ssa-sprintf.c: Include domwalk.h.
6685 (class sprintf_dom_walker): New class, derived from dom_walker.
6686 (sprintf_dom_walker::before_dom_children): New function.
6687 (struct call_info): Moved into sprintf_dom_walker class
6688 (compute_formath_length, handle_gimple_call): Likewise.
6689 (sprintf_length::execute): Call the dominator walker rather
6690 than walking the statements.
6692 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6693 gimple statement locations.
6694 (check_array_bounds): Corresponding changes. Get the statement's
6695 location directly from wi->stmt.
6697 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6700 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6702 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6704 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6705 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6707 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6710 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6711 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6712 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6713 maybe_get_pool_constant.
6714 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6717 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6719 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6720 2.26 caveat. Update gas and gld versions.
6721 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6724 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6726 * cgraph.h (set_malloc_flag): Declare.
6727 * cgraph.c (set_malloc_flag_1): New function.
6728 (set_malloc_flag): Likewise.
6729 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6730 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6732 (read_ipa_call_summary): Add support for reading is_return_callee.
6733 (write_ipa_call_summary): Stream is_return_callee.
6734 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6735 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6736 ipa-prop.h, ipa-fnsummary.h.
6737 (pure_const_names): Change to static.
6738 (malloc_state_e): Define.
6739 (malloc_state_names): Define.
6740 (funct_state_d): Add field malloc_state.
6741 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6742 (check_retval_uses): New function.
6743 (malloc_candidate_p): Likewise.
6744 (analyze_function): Add support for malloc attribute.
6745 (pure_const_write_summary): Stream malloc_state.
6746 (pure_const_read_summary): Add support for reading malloc_state.
6747 (dump_malloc_lattice): New function.
6748 (propagate_malloc): New function.
6749 (warn_function_malloc): New function.
6750 (ipa_pure_const::execute): Call propagate_malloc and
6751 ipa_free_fn_summary.
6752 (pass_local_pure_const::execute): Add support for malloc attribute.
6753 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6754 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6756 2017-10-27 Martin Liska <mliska@suse.cz>
6758 PR gcov-profile/82457
6759 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6760 for fork-like functions to be properly instrumented.
6762 2017-10-27 Richard Biener <rguenther@suse.de>
6765 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6766 info when we redirected EH.
6768 2017-10-26 Michael Collison <michael.collison@arm.com>
6770 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6772 (<optab>_trunchf<GPI:mode>2: New pattern.
6773 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6774 * config/aarch64/iterators.md (wv): New mode attribute.
6775 (vf, VF): New mode attributes.
6776 (vgp, VGP): New mode attributes.
6777 (s): Update attribute with SImode and DImode prefixes.
6779 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6781 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6782 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6783 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6784 (nios2_option_overide): Initialize it. Don't allow R0-relative
6785 addressing with PIC.
6786 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6787 (nios2_symbolic_constant_p): Likewise.
6788 (nios2_legitimate_address_p): Likewise.
6789 (nios2_r0rel_section_name_p): New.
6790 (nios2_symbol_ref_in_r0rel_data_p): New.
6791 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6792 (r0rel_constant_p): New.
6793 (nios2_print_operand_address): Handle r0rel_constant_p.
6794 (nios2_cdx_narrow_form_p): Likewise.
6795 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6796 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6797 (Nios II Options): Document -mr0rel-sec.
6799 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6801 * config/nios2/nios2.c: Include xregex.h.
6802 (nios2_gprel_sec_regex): New.
6803 (nios2_option_overide): Initialize it. Don't allow GP-relative
6804 addressing with PIC.
6805 (nios2_small_section_name_p): Check for regex match.
6806 * config/nios2/nios2.opt (mgprel-sec=): New option.
6807 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6808 (Nios II Options): Document -mgprel-sec.
6810 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6812 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6814 2017-10-26 Tom de Vries <tom@codesourcery.com>
6816 PR tree-optimization/82707
6817 * gimple.c (gimple_copy): Fix unsharing of
6818 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
6820 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
6822 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
6823 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
6824 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
6825 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
6826 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
6827 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
6828 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
6829 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
6830 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
6831 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
6832 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
6833 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
6834 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
6835 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
6836 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
6837 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
6838 _mm512_mask_cmpunord_ps_mask): New intrinsics.
6840 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6842 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
6844 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
6845 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
6846 warning to rs6000.c. Remove the Undocumented flag, since it has
6848 (-mabi=ibmlongdouble): Likewise.
6849 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
6850 already set, set the default format for long double.
6851 (rs6000_debug_reg_global): Print whether long double is IBM or
6853 (rs6000_option_override_internal): Rework setting long double
6854 format. Only warn if the user is changing the long double default
6855 and they did not use -Wno-psabi.
6856 * doc/invoke.texi (PowerPC options): Update the documentation for
6857 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
6859 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6860 Alan Hayward <alan.hayward@arm.com>
6861 David Sherwood <david.sherwood@arm.com>
6863 * rtl.h (wider_subreg_mode): New function.
6864 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
6865 rather than an unsigned int *.
6866 * ira-color.c (regno_max_ref_width): Replace with...
6867 (regno_max_ref_mode): ...this new variable.
6868 (coalesced_pseudo_reg_slot_compare): Update accordingly.
6869 Use wider_subreg_mode.
6870 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
6871 rather than an unsigned int *.
6872 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
6873 (process_alt_operands): Likewise.
6874 (invariant_p): Likewise.
6875 * lra-spills.c (assign_mem_slot): Likewise.
6876 (add_pseudo_to_slot): Likewise.
6877 * lra.c (collect_non_operand_hard_regs): Likewise.
6878 (add_regs_to_insn_regno_info): Likewise.
6879 * reload1.c (regno_max_ref_width): Replace with...
6880 (regno_max_ref_mode): ...this new variable.
6881 (reload): Update accordingly. Update call to
6882 ira_sort_regnos_for_alter_reg.
6883 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
6884 (init_eliminable_invariants): Update to use regno_max_ref_mode.
6885 (scan_paradoxical_subregs): Likewise.
6887 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6889 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
6890 (aarch64_frame): Add emit_frame_chain boolean.
6891 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6892 Move eh_return case to aarch64_layout_frame.
6893 (aarch64_layout_frame): Initialize emit_frame_chain.
6894 (aarch64_expand_prologue): Use emit_frame_chain.
6896 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6898 * config/aarch64/aarch64.c (aarch64_layout_frame):
6899 Ensure LR is always stored at the bottom of the callee-saves.
6900 Remove rarely used frame layout which saves callee-saves at top of
6901 frame, so the store of LR can be used as a valid probe in all cases.
6903 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6905 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
6906 Improve unaligned TImode/TFmode base/offset split.
6908 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6909 Alan Hayward <alan.hayward@arm.com>
6910 David Sherwood <david.sherwood@arm.com>
6912 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
6913 * combine.c (find_single_use_1): Likewise.
6914 (expand_field_assignment): Likewise.
6915 (move_deaths): Likewise.
6916 * lra-constraints.c (simplify_operand_subreg): Likewise.
6917 (curr_insn_transform): Likewise.
6918 * lra.c (collect_non_operand_hard_regs): Likewise.
6919 (add_regs_to_insn_regno_info): Likewise.
6920 * rtlanal.c (reg_referenced_p): Likewise.
6921 (covers_regno_no_parallel_p): Likewise.
6923 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6925 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
6926 Don't print any bits outside the precision of the value.
6927 * wide-int.cc (test_printing): Add some new tests.
6929 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6931 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
6932 supports -xbrace_comment option.
6933 * configure: Regenerate.
6934 * config.in: Regenerate.
6935 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
6936 (ASM_CPU_SPEC): Use it.
6938 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6940 * target.def (static_rtx_alignment): New hook.
6941 * targhooks.h (default_static_rtx_alignment): Declare.
6942 * targhooks.c (default_static_rtx_alignment): New function.
6943 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
6944 * doc/tm.texi: Regenerate.
6945 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
6946 instead of targetm.constant_alignment. Remove call to
6948 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6949 (cris_preferred_mininum_alignment): New function, split out from...
6950 (cris_constant_alignment): ...here.
6951 (cris_static_rtx_alignment): New function.
6952 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
6954 (ix86_constant_alignment): ...here.
6955 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6956 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6957 (mmix_static_rtx_alignment): New function.
6958 * config/spu/spu.c (spu_static_rtx_alignment): New function.
6959 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6961 2017-10-26 Tamar Christina <tamar.christina@arm.com>
6964 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
6965 Add flag_trapping_math and flag_fp_int_builtin_inexact.
6967 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
6969 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
6970 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
6971 mark as a sign-extending load.
6972 (local_pic_load_u): Define.
6974 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
6977 * fold-const.c (operand_equal_for_comparison_p): Also return true
6978 if ARG0 is a simple variant of ARG1 with narrower precision.
6979 (fold_ternary_loc): Always pass unstripped operands to the predicate.
6981 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
6983 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
6985 * i386.h (processor_costs): Add gather_static, gather_per_elt,
6986 scatter_static, scatter_per_elt.
6987 * x86-tune-costs.h: Add new cost entries.
6989 2017-10-25 Richard Biener <rguenther@suse.de>
6991 * tree-ssa-sccvn.h (vn_eliminate): Declare.
6992 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
6993 class pass_fre): Move to ...
6994 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
6995 class pass_fre): ... here and adjust for statistics.
6997 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7000 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7001 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7002 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7003 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7006 2017-10-25 Richard Biener <rguenther@suse.de>
7008 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7009 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7010 eliminate_push_avail, eliminate_insert): Move inside...
7011 (class eliminate_dom_walker): ... this class in preparation
7013 (fini_eliminate): Remove by merging with ...
7014 (eliminate): ... this function. Adjust for class changes.
7015 (pass_pre::execute): Remove fini_eliminate call.
7016 (pass_fre::execute): Likewise.
7018 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7021 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7022 (VPERMI2, VPERMI2I): New mode iterators.
7023 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7024 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7026 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7027 mode iterator. Remove 3 old define_insn patterns.
7028 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7029 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7030 VPERMI2 mode iterator, remove the other two expanders.
7031 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7032 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7033 instructions, remove the other two patterns.
7034 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7035 mode iterator, remove the other two patterns.
7036 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7037 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7038 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7039 and adjust argument order accordingly.
7040 (ix86_expand_vec_perm): Adjust caller.
7041 (expand_vec_perm_1): Likewise.
7042 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7043 (expand_vec_perm_vpermt2_vpshub2): ... this.
7044 (ix86_expand_vec_perm_const_1): Adjust caller.
7045 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7048 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7049 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7050 (vec_shl_<mode>): Remove unused expander.
7051 (avx512bw_<shift_insn><mode>3): New define_insn.
7052 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7053 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7055 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7058 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7061 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7063 PR rtl-optimization/82396
7064 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7065 (autopref_multipass_init): Simplify initialization.
7066 (autopref_rank_data): Simplify sort order.
7067 * gcc/sched-int.h (autopref_multipass_data_): Remove
7068 multi_mem_insn_p, min_offset and max_offset.
7070 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7073 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7074 Check special value of flag_omit_frame_pointer.
7075 (aarch64_can_eliminate): Likewise.
7076 (aarch64_override_options_after_change_1): Simplify handling of
7077 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7079 2017-10-24 Richard Biener <rguenther@suse.de>
7081 PR tree-optimization/82697
7082 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7083 zero for conditional load and unconditional store.
7085 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7087 * doc/install.texi: Document bootstrap-cet.
7089 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7092 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7093 ENDBR instruction at function entrance if function is only
7096 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7099 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7100 patterns to better describe from which operation the CF is computed.
7101 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7102 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7103 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7104 is 0, use _0 suffixed expanders instead of emitting a comparison
7107 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7109 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7110 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7112 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7115 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7116 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7117 * loop-iv.c (iv_get_reaching_def): Likewise.
7118 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7119 variable is promoted and the partition contains undefined values.
7121 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7123 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7125 (nios2_address_cost): Define.
7126 (nios2_legitimize_address): Recognize (exp + constant) directly.
7127 (TARGET_ADDRESS_COST): Define.
7129 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7131 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7132 (nios2_symbolic_memory_operand_p): Declare.
7133 (nios2_split_large_constant): Declare.
7134 (nios2_split_symbolic_memory_operand): Declare.
7135 * config/nios2/nios2.c: Adjust includes.
7136 (nios2_symbolic_constant_allowed): New.
7137 (nios2_symbolic_constant_p): New.
7138 (nios2_plus_symbolic_constant_p): New.
7139 (nios2_valid_addr_expr_p): Recognize addresses involving
7141 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7142 (nios2_symbolic_memory_operand_p): New.
7143 (nios2_large_constant_p): New.
7144 (nios2_split_large_constant): New.
7145 (nios2_split_plus_large_constant): New.
7146 (nios2_split_symbolic_memory_operand): New.
7147 (nios2_legitimize_address): Code refactoring. Handle addresses
7148 involving symbolic constants.
7149 (nios2_emit_move_sequence): Likewise.
7150 (nios2_print_operand): Improve error output.
7151 (nios2_print_operand_address): Handle LO_SUM.
7152 (nios2_cdx_narrow_form_p): Likewise.
7153 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7154 operands involving symbolic constants.
7155 (movhi_internal, movsi_internal): Likewise.
7156 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7157 (extendhisi2, extendqi<mode>2): Likewise.
7159 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7161 * tree-pass.h (PROP_rtl_split_insns): Define.
7162 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7164 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7166 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7168 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7171 * target.def (const_not_ok_for_debug_p): Default to
7172 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7173 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7174 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7175 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7176 which targetm.const_not_ok_for_debug_p returned true.
7177 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7179 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7181 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7182 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7183 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7185 (ix86_const_not_ok_for_debug_p): New function.
7186 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7187 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7189 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7192 * system.h: Conditionally include "unique-ptr.h" if
7193 INCLUDE_UNIQUE_PTR is defined.
7194 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7195 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7197 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7199 * config/rl78/rl78.md: New define_expand "subdi3".
7201 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7204 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7205 DF_REF_INSN if DF_REF_INSN_INFO is false.
7207 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7209 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7210 xmm_move instead of sse_move.
7211 (sse_store_index): New function.
7212 (ix86_register_move_cost): Be more sensible about mismatch stall;
7213 model AVX moves correctly; make difference between sse->integer and
7215 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7216 moves; make difference between SSE and AVX.
7217 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7218 and zmm_move. Increase size of sse load and store tables;
7219 add unaligned load and store tables; add ssemmx_to_integer.
7220 * x86-tune-costs.h: Update all entries according to real
7221 move latencies from Agner Fog's manual and chip documentation.
7223 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7226 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7227 * config/i386/constraints.md (Wf): New constraint.
7228 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7229 (cmp<dwi>_doubleword): Removed.
7230 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7231 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7232 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7233 expand with cmp<dwi>_doubleword. For LTU and GEU use
7234 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7236 * common.opt (gcolumn-info): Enable by default.
7237 * doc/invoke.texi (gcolumn-info): Document new default.
7239 2017-10-23 Richard Biener <rguenther@suse.de>
7241 PR tree-optimization/82672
7242 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7243 Fold the stmt if we propagated into it.
7245 2017-10-23 Richard Biener <rguenther@suse.de>
7247 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7248 (bitmap_remove_expr_from_set): ... this. All callers call this
7249 for non-constant values.
7250 (bitmap_set_subtract): Rename to...
7251 (bitmap_set_subtract_expressions): ... this. Adjust and
7253 (bitmap_set_contains_value): Remove superfluous check.
7254 (bitmap_set_replace_value): Inline into single caller ...
7255 (bitmap_value_replace_in_set): ... here and simplify.
7256 (dependent_clean): Merge into ...
7257 (clean): ... this using an overload. Adjust.
7258 (prune_clobbered_mems): Adjust.
7259 (compute_antic_aux): Likewise.
7260 (compute_partial_antic_aux): Likewise.
7262 2017-10-23 Richard Biener <rguenther@suse.de>
7264 PR tree-optimization/82129
7266 2017-08-01 Richard Biener <rguenther@suse.de>
7268 PR tree-optimization/81181
7269 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7270 (compute_antic): ... end of iteration here.
7272 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7274 * target.def (starting_frame_offset): New hook.
7275 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7276 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7277 * doc/tm.texi.in: Regenerate.
7278 * hooks.h (hook_hwi_void_0): Declare.
7279 * hooks.c (hook_hwi_void_0): New function.
7280 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7281 STARTING_FRAME_OFFSET.
7282 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7283 * reload1.c (reload): Likewise.
7284 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7285 instead of STARTING_FRAME_OFFSET.
7286 * function.c (try_fit_stack_local): Likewise.
7287 (assign_stack_local_1): Likewise
7288 (instantiate_virtual_regs): Likewise.
7289 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7290 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7291 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7292 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7293 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7294 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7295 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7296 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7297 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7298 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7299 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7300 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7301 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7302 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7303 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7304 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7305 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7306 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7307 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7308 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7309 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7310 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7311 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7312 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7313 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7314 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7315 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7316 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7317 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7318 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7319 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7320 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7321 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7322 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7323 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7324 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7325 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7326 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7327 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7328 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7329 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7330 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7331 return a HOST_WIDE_INT.
7332 (avr_builtin_setjmp_frame_value): Use it instead of
7333 STARTING_FRAME_OFFSET.
7334 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7335 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7336 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7338 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7339 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7340 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7341 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7342 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7343 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7344 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7345 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7346 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7347 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7348 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7349 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7351 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7352 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7353 * config/mips/mips.c (mips_compute_frame_info): Refer to
7354 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7355 (mips_starting_frame_offset): New function.
7356 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7357 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7358 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7359 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7360 and return a HOST_WIDE_INT.
7361 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7362 (mmix_initial_elimination_offset): Refer to
7363 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7364 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7365 * config/pa/pa.c (pa_starting_frame_offset): New function.
7366 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7367 (pa_expand_prologue): Likewise.
7368 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7369 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7370 !FRAME_GROWS_DOWNWARD handling to...
7371 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7372 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7373 !FRAME_GROWS_DOWNWARD handling to...
7374 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7375 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7376 !FRAME_GROWS_DOWNWARD handling to...
7377 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7378 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7380 (rs6000_starting_frame_offset): New function.
7381 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7382 !FRAME_GROWS_DOWNWARD handling to...
7383 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7384 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7385 !FRAME_GROWS_DOWNWARD handling to...
7386 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7387 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7388 !FRAME_GROWS_DOWNWARD handling to...
7389 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7390 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7391 (rs6000_starting_frame_offset): New function.
7392 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7393 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7394 * config/vax/vax.c (vax_starting_frame_offset): New function.
7395 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7396 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7397 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7398 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7399 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7400 * system.h (STARTING_FRAME_OFFSET): Poison.
7402 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7404 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7405 SCALAR_TYPE_MODE instead of TYPE_MODE.
7407 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7408 Alan Hayward <alan.hayward@arm.com>
7409 David Sherwood <david.sherwood@arm.com>
7411 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7413 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7414 Alan Hayward <alan.hayward@arm.com>
7415 David Sherwood <david.sherwood@arm.com>
7417 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7419 2017-10-23 Richard Biener <rguenther@suse.de>
7421 PR tree-optimization/82129
7422 * tree-ssa-pre.c (bitmap_set_and): Remove.
7423 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7424 canonicalizing expressions in the set to those with lowest
7425 ID rather than taking that from the first edge.
7427 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7429 * combine.c (rtx_equal_for_field_assignment_p): Use
7430 byte_lowpart_offset.
7432 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7433 Alan Hayward <alan.hayward@arm.com>
7434 David Sherwood <david.sherwood@arm.com>
7436 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7437 to a SUBREG_PROMOTED_VAR.
7439 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7440 Alan Hayward <alan.hayward@arm.com>
7441 David Sherwood <david.sherwood@arm.com>
7443 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7444 (expand_debug_source_expr): Likewise.
7445 * combine.c (combine_simplify_rtx): Likewise.
7446 * cse.c (fold_rtx): Likewise.
7447 * optabs.c (expand_float): Likewise.
7448 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7449 (simplify_binary_operation_1): Likewise.
7451 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7452 Alan Hayward <alan.hayward@arm.com>
7453 David Sherwood <david.sherwood@arm.com>
7455 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7456 (record_promoted_value): Likewise.
7457 * expr.c (expand_expr_real_2): Likewise.
7458 * ree.c (update_reg_equal_equiv_notes): Likewise.
7459 (combine_set_extension): Likewise.
7460 * rtlanal.c (low_bitmask_len): Likewise.
7461 * simplify-rtx.c (neg_const_int): Likewise.
7462 (simplify_binary_operation_1): Likewise.
7464 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7465 Alan Hayward <alan.hayward@arm.com>
7466 David Sherwood <david.sherwood@arm.com>
7468 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7469 * regcprop.c (maybe_mode_change): Likewise.
7470 * reload1.c (alter_reg): Likewise.
7472 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7474 * inchash.h (inchash::hash::add_wide_int): New function.
7475 * lto-streamer-out.c (hash_tree): Use it.
7477 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7479 * inchash.h (inchash::hash::add_wide_int): Rename to...
7480 (inchash::hash::add_hwi): ...this.
7481 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7482 (polymorphic_call_target_hasher::hash): Likewise.
7483 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7484 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7485 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7486 * lto-streamer-out.c (hash_tree): Likewise.
7487 * optc-save-gen.awk: Likewise.
7488 * tree.c (add_expr): Likewise.
7490 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7493 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7494 for ordered inequality comparisons even with TARGET_IEEE_FP.
7496 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7499 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7500 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7501 Expand with cmp<dwi>_doubleword.
7503 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7505 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7506 List CET intrinsics.
7507 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7508 specific to -fcf-protection option.
7510 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7512 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7513 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7514 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7515 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7516 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7517 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7518 (extra_objs): Add cet.o for Linux/x86 targets.
7519 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7520 * config/i386/cet.c: New file.
7521 * config/i386/cetintrin.h: Likewise.
7522 * config/i386/t-cet: Likewise.
7523 * config/i386/cpuid.h (bit_SHSTK): New.
7524 (bit_IBT): Likewise.
7525 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7526 pass IBT and SHSTK bits.
7527 * config/i386/i386-builtin-types.def
7528 (VOID_FTYPE_UNSIGNED_PVOID): New.
7529 (VOID_FTYPE_UINT64_PVOID): Likewise.
7530 * config/i386/i386-builtin.def: Add CET intrinsics.
7531 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7532 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7533 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7534 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7536 * config/i386/i386.c (rest_of_insert_endbranch): New.
7537 (pass_data_insert_endbranch): Likewise.
7538 (pass_insert_endbranch): Likewise.
7539 (make_pass_insert_endbranch): Likewise.
7540 (ix86_notrack_prefixed_insn_p): Likewise.
7541 (ix86_target_string): Add -mibt, -mshstk flags.
7542 (ix86_option_override_internal): Add flag_cf_protection
7544 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7545 (ix86_print_operand): Add 'notrack' prefix output.
7546 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7547 (ix86_expand_builtin): Expand CET intrinsics.
7548 (x86_output_mi_thunk): Add 'endbranch' instruction.
7549 * config/i386/i386.h (TARGET_IBT): New.
7550 (TARGET_IBT_P): Likewise.
7551 (TARGET_SHSTK): Likewise.
7552 (TARGET_SHSTK_P): Likewise.
7553 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7554 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7555 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7556 (builtin_setjmp_setup): New pattern.
7557 (builtin_longjmp): Likewise.
7558 (rdssp<mode>): Likewise.
7559 (incssp<mode>): Likewise.
7560 (saveprevssp): Likewise.
7561 (rstorssp): Likewise.
7562 (wrss<mode>): Likewise.
7563 (wruss<mode>): Likewise.
7564 (setssbsy): Likewise.
7565 (clrssbsy): Likewise.
7566 (nop_endbr): Likewise.
7567 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7569 * config/i386/immintrin.h: Include <cetintrin.h>.
7570 * config/i386/linux-common.h
7571 (file_end_indicate_exec_stack_and_cet): New prototype.
7572 (TARGET_ASM_FILE_END): New.
7574 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7576 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7577 latencies instead of having separate table; make difference between
7578 integer and float costs.
7579 * i386.h (processor_costs): Remove scalar_stmt_cost,
7580 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7581 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7583 * x86-tune-costs.h: Remove entries which has been removed in
7584 procesor_costs from all tables; make cond_taken_branch_cost
7585 and cond_not_taken_branch_cost COST_N_INSNS based.
7587 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7589 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7591 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7593 * config/i386/i386.md (isa): Remove fma_avx512f.
7594 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7595 <avx512>_fmadd_<mode>_mask3<round_name>,
7596 <avx512>_fmsub_<mode>_mask<round_name>,
7597 <avx512>_fmsub_<mode>_mask3<round_name>,
7598 <avx512>_fnmadd_<mode>_mask<round_name>,
7599 <avx512>_fnmadd_<mode>_mask3<round_name>,
7600 <avx512>_fnmsub_<mode>_mask<round_name>,
7601 <avx512>_fnmsub_<mode>_mask3<round_name>,
7602 <avx512>_fmaddsub_<mode>_mask<round_name>,
7603 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7604 <avx512>_fmsubadd_<mode>_mask<round_name>,
7605 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7606 (*vec_widen_umult_even_v16si<mask_name>,
7607 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7608 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7610 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7612 * extend.texi: Add 'nocf_check' documentation.
7613 * gimple.texi: Add second parameter to
7614 gimple_build_call_from_tree.
7615 * invoke.texi: Add -fcf-protection documentation.
7616 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7618 2017-10-20 Richard Biener <rguenther@suse.de>
7620 PR tree-optimization/82473
7621 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7622 the largest input type.
7624 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7626 * c-attribs.c (handle_nocf_check_attribute): New function.
7627 (c_common_attribute_table): Add 'nocf_check' handling.
7628 * gimple-parser.c: Add second argument NULL to
7629 gimple_build_call_from_tree.
7630 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7631 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7633 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7634 * common.opt: Add fcf-protection flag.
7635 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7636 * flag-types.h: Add enum cf_protection_level.
7637 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7638 Add 'nocf_check' attribute propagation to gimple call.
7639 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7640 (gimple_build_call_from_tree): Update prototype.
7641 (gimple_call_nocf_check_p): New function.
7642 (gimple_call_set_nocf_check): Likewise.
7643 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7644 * ipa-icf.c: Add nocf_check attribute in statement hash.
7645 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7646 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7647 * toplev.c (process_options): Add flag_cf_protection handling.
7649 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7651 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7653 2017-10-20 Richard Biener <rguenther@suse.de>
7655 PR tree-optimization/82603
7656 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7657 remove false predicated stores.
7659 2017-10-20 Richard Biener <rguenther@suse.de>
7661 * graphite-isl-ast-to-gimple.c
7662 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7663 Remove return value and simplify, dump copied stmt after lhs
7665 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7666 Reduce dump verbosity.
7667 (gsi_insert_earliest): Likewise.
7668 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7669 * graphite.c (print_global_statistics): Adjust dumping.
7670 (print_graphite_scop_statistics): Likewise.
7671 (print_graphite_statistics): Do not dump loops here.
7672 (graphite_transform_loops): But here.
7674 2017-10-20 Nicolas Roche <roche@adacore.com>
7676 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7677 * configure: Regenerate.
7679 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7682 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7683 functions when optimizing replace GIMPLE_RETURN stmts with
7684 calls to __builtin_unreachable ().
7687 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7688 for -fsanitize=thread link of executables.
7689 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7690 link of executables.
7693 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7695 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7696 define_insns for logical vector shifts to use VI248_AVX512BW
7697 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7698 condition, useless isa and prefix attributes. Change the first
7699 2 of these define_insns to ...
7700 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7701 define_insn for avx512vl.
7702 (<shift_insn><mode>3): ... and this, new define_insn without
7703 masking for non-avx512vl.
7706 * config/i386/sse.md (*andnot<mode>3,
7707 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7708 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7709 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7710 not applied use empty suffix even for TARGET_AVX512VL.
7711 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7712 is applied, supply evex,evex or evex,evex,evex instead of just
7715 2017-10-20 Julia Koval <julia.koval@intel.com>
7717 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7718 (OPTION_MASK_ISA_GFNI_UNSET): New.
7719 (ix86_handle_option): Handle OPT_mgfni.
7720 * config/i386/cpuid.h (bit_GFNI): New.
7721 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7722 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7723 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7724 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7725 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7726 * config/i386/i386.opt: Add mgfni.
7728 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7730 * config/msp430/msp430.c (msp430_option_override): Disable
7731 -fdelete-null-pointer-checks.
7732 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7734 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7736 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7737 of x87 and SSE instructions.
7739 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7741 * asan.c (create_cond_insert_point): Do not update edge count.
7742 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7743 (afdo_propagate_circuit): Likewise.
7744 (afdo_calculate_branch_prob): Likewise.
7745 (afdo_annotate_cfg): Likewise.
7746 * basic-block.h (struct edge_def): Remove count.
7747 (edge_def::count): New accessor.
7748 * bb-reorder.c (rotate_loop): Update.
7749 (find_traces_1_round): Update.
7750 (connect_traces): Update.
7751 (sanitize_hot_paths): Update.
7752 * cfg.c (unchecked_make_edge): Update.
7753 (make_single_succ_edge): Update.
7754 (check_bb_profile): Update.
7755 (dump_edge_info): Update.
7756 (update_bb_profile_for_threading): Update.
7757 (scale_bbs_frequencies_int): Update.
7758 (scale_bbs_frequencies_gcov_type): Update.
7759 (scale_bbs_frequencies_profile_count): Update.
7760 (scale_bbs_frequencies): Update.
7761 * cfganal.c (connect_infinite_loops_to_exit): Update.
7762 * cfgbuild.c (compute_outgoing_frequencies): Update.
7763 (find_many_sub_basic_blocks): Update.
7764 * cfgcleanup.c (try_forward_edges): Update.
7765 (try_crossjump_to_edge): Update
7766 * cfgexpand.c (expand_gimple_cond): Update
7767 (expand_gimple_tailcall): Update
7768 (construct_exit_block): Update
7769 * cfghooks.c (verify_flow_info): Update
7770 (redirect_edge_succ_nodup): Update
7771 (split_edge): Update
7772 (make_forwarder_block): Update
7773 (duplicate_block): Update
7774 (account_profile_record): Update
7775 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7776 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7777 * cfgloopmanip.c (scale_loop_profile): Update.
7779 (lv_adjust_loop_entry_edge): Update.
7780 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7781 (force_nonfallthru_and_redirect): Update.
7782 (purge_dead_edges): Update.
7783 (rtl_flow_call_edges_add): Update.
7784 * cgraphunit.c (init_lowered_empty_function): Update.
7785 (cgraph_node::expand_thunk): Update.
7786 * gimple-pretty-print.c (dump_probability): Update.
7787 (dump_edge_probability): Update.
7788 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7789 * haifa-sched.c (sched_create_recovery_edges): Update.
7790 * hsa-gen.c (convert_switch_statements): Update.
7791 * ifcvt.c (dead_or_predicable): Update.
7792 * ipa-inline-transform.c (inline_transform): Update.
7793 * ipa-split.c (split_function): Update.
7794 * ipa-utils.c (ipa_merge_profiles): Update.
7795 * loop-doloop.c (add_test): Update.
7796 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7797 * lto-streamer-in.c (input_cfg): Update.
7798 (input_function): Update.
7799 * lto-streamer-out.c (output_cfg): Update.
7800 * modulo-sched.c (sms_schedule): Update.
7801 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7802 * predict.c (maybe_hot_edge_p): Update.
7803 (unlikely_executed_edge_p): Update.
7804 (probably_never_executed_edge_p): Update.
7805 (dump_prediction): Update.
7806 (drop_profile): Update.
7807 (propagate_unlikely_bbs_forward): Update.
7808 (determine_unlikely_bbs): Update.
7809 (force_edge_cold): Update.
7810 * profile.c (compute_branch_probabilities): Update.
7811 * reg-stack.c (better_edge): Update.
7812 * shrink-wrap.c (handle_simple_exit): Update.
7813 * tracer.c (better_p): Update.
7814 * trans-mem.c (expand_transaction): Update.
7815 (split_bb_make_tm_edge): Update.
7816 * tree-call-cdce.c: Update.
7817 * tree-cfg.c (gimple_find_sub_bbs): Update.
7818 (gimple_split_edge): Update.
7819 (gimple_duplicate_sese_region): Update.
7820 (gimple_duplicate_sese_tail): Update.
7821 (gimple_flow_call_edges_add): Update.
7822 (insert_cond_bb): Update.
7823 (execute_fixup_cfg): Update.
7824 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
7825 * tree-complex.c (expand_complex_div_wide): Update.
7826 * tree-eh.c (lower_resx): Update.
7827 (unsplit_eh): Update.
7828 (cleanup_empty_eh_move_lp): Update.
7829 * tree-inline.c (copy_edges_for_bb): Update.
7830 (freqs_to_counts): Update.
7831 (copy_cfg_body): Update.
7832 * tree-ssa-dce.c (remove_dead_stmt): Update.
7833 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7834 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7835 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7836 (unloop_loops): Update.
7837 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7838 * tree-ssa-loop-split.c (connect_loops): Update.
7839 (split_loop): Update.
7840 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7841 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7842 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7843 * tree-ssa-reassoc.c (branch_fixup): Update.
7844 * tree-ssa-tail-merge.c (replace_block_by): Update.
7845 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7846 (compute_path_counts): Update.
7847 (update_profile): Update.
7848 (recompute_probabilities): Update.
7849 (update_joiner_offpath_counts): Update.
7850 (estimated_freqs_path): Update.
7851 (freqs_to_counts_path): Update.
7852 (clear_counts_path): Update.
7853 (ssa_fix_duplicate_block_edges): Update.
7854 (duplicate_thread_path): Update.
7855 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7856 (case_bit_test_cmp): Update.
7857 (collect_switch_conv_info): Update.
7858 (gen_inbound_check): Update.
7859 (do_jump_if_equal): Update.
7860 (emit_cmp_and_jump_insns): Update.
7861 * tree-tailcall.c (decrease_profile): Update.
7862 (eliminate_tail_call): Update.
7863 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7864 (vect_do_peeling): Update.
7865 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7866 * ubsan.c (ubsan_expand_null_ifn): Update.
7867 (ubsan_expand_ptr_ifn): Update.
7868 * value-prof.c (gimple_divmod_fixed_value): Update.
7869 (gimple_mod_pow2): Update.
7870 (gimple_mod_subtract): Update.
7871 (gimple_ic): Update.
7872 (gimple_stringop_fixed_value): Update.
7874 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7877 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
7879 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
7881 PR rtl-optimization/82395
7882 * ira-color.c (allocno_priority_compare_func): Fix comparison step
7883 based on non_spilled_static_chain_regno_p.
7885 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7887 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
7888 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
7889 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
7891 2017-10-19 Martin Sebor <msebor@redhat.com>
7893 PR tree-optimization/82596
7894 * tree.c (array_at_struct_end_p): Handle STRING_CST.
7896 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7898 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
7899 (get_mem_refs_of_builtin_call): Likewise.
7900 * builtins.c (expand_builtin_apply): Adjust call to
7901 allocate_dynamic_stack_space.
7902 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
7903 the third argument to allocate_dynamic_stack_space, otherwise -1.
7904 (expand_builtin): Deal with all alloca variants.
7905 (is_inexpensive_builtin): Likewise.
7906 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
7907 * calls.c (special_function_p): Deal with all alloca variants.
7908 (initialize_argument_information): Adjust call to
7909 allocate_dynamic_stack_space.
7910 (expand_call): Likewise.
7911 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
7912 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
7913 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
7914 use it for the stack usage computation.
7915 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
7916 * function.c (gimplify_parameters): Call build_alloca_call_expr.
7917 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
7918 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
7919 (in_loop_p): Remove first argument and useless check.
7920 (pass_walloca::execute): Remove useless test and adjust call to above.
7921 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
7922 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
7923 (gimplify_call_expr): Deal with all alloca variants.
7924 * hsa-gen.c (gen_hsa_alloca): Likewise.
7925 (gen_hsa_insns_for_call): Likewise.
7926 * ipa-pure-const.c (special_builtin_state): Likewise.
7927 * tree-chkp.c (chkp_build_returned_bound): Likewise.
7928 * tree-object-size.c (alloc_object_size): Likewise.
7929 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7930 (call_may_clobber_ref_p_1): Likewise.
7931 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
7932 (ccp_fold_stmt): Likewise.
7933 (optimize_stack_restore): Likewise.
7934 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7935 (mark_all_reaching_defs_necessary_1): Likewise.
7936 (propagate_necessity): Likewise.
7937 (eliminate_unnecessary_stmts): Likewise.
7938 * tree.c (build_common_builtin_nodes): Build
7939 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
7940 (build_alloca_call_expr): New function.
7941 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
7942 (CASE_BUILT_IN_ALLOCA): Likewise.
7943 (build_alloca_call_expr): Declare.
7944 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
7946 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7949 * dwarf2out.c (new_die_raw): New static inline function.
7950 (new_die): Use it to create the DIE.
7951 (add_AT_external_die_ref): Likewise.
7952 (clone_die): Likewise.
7953 (clone_as_declaration): Likewise.
7954 (dwarf2out_vms_debug_main_pointer): Likewise.
7955 (base_type_die): Likewise. Remove early return for corner cases.
7956 Do not call add_pubtype on the DIE here.
7957 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
7958 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
7959 typedefs for base types with DW_AT_endianity. Make sure a DIE with
7960 native order exists for base types, attach the DIE manually and call
7961 add_pubtype on it. Do not equate a reverse order DIE to the type.
7963 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
7965 * config/arm/arm.c (align_ok_ldrd_strd): New function.
7966 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
7968 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
7970 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7972 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
7973 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
7974 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
7975 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
7976 * opts.c (sanitizer_opts): Add builtin.
7977 * ubsan.c (instrument_builtin): New function.
7978 (pass_ubsan::execute): Call it.
7979 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
7980 * doc/invoke.texi: Document -fsanitize=builtin.
7982 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
7983 builtins, store max (log2 (align), 0) into uchar field instead of
7984 align into uptr field.
7985 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
7986 store uchar 0 field instead of uptr 0 field.
7987 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
7988 instead of passing one address of struct with 2 locations pass
7989 two addresses of structs with 1 location each.
7990 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
7991 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
7992 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
7993 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
7994 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
7995 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
7996 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
7997 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
7999 2017-10-19 Martin Liska <mliska@suse.cz>
8002 * file-find.c (remove_prefix): Remove.
8003 * file-find.h (remove_prefix): Likewise.
8004 * gcc-ar.c: Remove smartness of lookup.
8006 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8008 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8009 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8010 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8012 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8015 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8016 (setcc + and to xor + setcc): New peephole2.
8018 2017-10-19 Tom de Vries <tom@codesourcery.com>
8020 * doc/sourcebuild.texi (Test Directives, Variants of
8021 dg-require-support): Add dg-require-stack-size.
8023 2017-10-19 Martin Liska <mliska@suse.cz>
8026 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8027 that have a large alignment.
8028 (gimplify_target_expr): Likewise.
8030 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8032 PR rtl-optimization/82602
8033 * ira.c (rtx_moveable_p): Return false for volatile asm.
8035 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8038 * config/i386/i386-modes.def (CCGZ): New CC mode.
8039 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8040 * config/i386/predicates.md (ix86_comparison_operator):
8042 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8043 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8044 with double-word subtraction.
8045 (put_condition_code): Handle CCGZmode.
8047 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8049 * wide-int.cc (debug (const wide_int &)): New.
8050 (debug (const wide_int *)): New.
8051 (debug (const widest_int &)): New.
8052 (debug (const widest_int *)): New.
8054 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8057 * lra-constraints.c (curr_insn_transform): Use non-input operand
8058 instead of output one for matched reload.
8060 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8062 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8063 (tree-ssa-loop-ivopts.h): New header file.
8064 (struct builtin_info): New fields.
8065 (classify_builtin_1): Compute and record base and offset parts for
8066 memset builtin partition by calling strip_offset.
8067 (offset_cmp, fuse_memset_builtins): New functions.
8068 (finalize_partitions): Fuse adjacent memset partitions by calling
8070 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8071 Expose the interface.
8072 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8074 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8076 PR tree-optimization/82574
8077 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8078 that data reference must be executed exactly once per iteration
8079 against the outermost loop in nest.
8080 (classify_partition): Update call to above function.
8082 2017-10-18 Richard Biener <rguenther@suse.de>
8084 PR tree-optimization/82591
8085 * graphite.c (graphite_transform_loops): Move code gen message
8087 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8088 Here. Handle scop_to_isl_ast failing.
8089 (scop_to_isl_ast): Limit the number of ISL operations.
8091 2017-10-18 Richard Biener <rguenther@suse.de>
8093 * graphite-isl-ast-to-gimple.c
8094 (translate_isl_ast_to_gimple::set_rename): Simplify.
8095 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8096 (graphite_copy_stmts_from_block): ... here.
8097 (copy_bb_and_scalar_dependences): Simplify.
8098 (add_parameters_to_ivs_params): Canonicalize.
8099 (generate_entry_out_of_ssa_copies): Simplify.
8100 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8101 by passing in ISL dimension.
8102 (parameter_index_in_region_1): Rename to ...
8103 (parameter_index_in_region): ... this.
8104 (extract_affine): Adjust assert, pass down parameter index.
8105 (add_param_constraints): Use range-info when available.
8106 (build_scop_context): Adjust.
8107 * sese.c (new_sese_info): Adjust.
8108 (free_sese_info): Likewise.
8109 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8110 Remove unused typedefs.
8111 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8113 2017-10-18 Martin Liska <mliska@suse.cz>
8115 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8117 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8119 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8120 (s390_sched_init): Do not reset s390_sched_state if we entered the
8121 current basic block via a fallthru edge and all others are unlikely.
8123 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8125 * config/s390/s390.c (NUM_SIDES): New variable.
8126 (LONGRUNNING_THRESHOLD): New variable.
8127 (LATENCY_FACTOR): New variable.
8128 (s390_sched_score): Decrease score for long-running instructions on
8130 (s390_sched_variable_issue): Perform bookkeeping for long-running
8133 2017-10-18 Richard Biener <rguenther@suse.de>
8135 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8136 Simplify with removal of the parameter rename map.
8137 (set_rename): Likewise.
8138 (should_copy_to_new_region): Likewise.
8139 (graphite_copy_stmts_from_block): Likewise.
8140 (copy_bb_and_scalar_dependences): Remove initialization of
8141 unused copied_bb_map.
8143 (copy_internal_parameters): Likewise.
8144 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8145 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8146 Use INTEGRAL_TYPE_P.
8147 (parameter_index_in_region_1): Rename to ...
8148 (assign_parameter_index_in_region): ... this. Assert we have
8149 a parameter we handle.
8150 (scan_tree_for_params): Adjust.
8151 * sese.h (parameter_rename_map_t): Remove.
8152 (struct sese_info_t): Remove unused parameter_rename_map and
8153 copied_bb_map members.
8154 * sese.c (new_sese_info): Adjust.
8155 (free_sese_info): Likewise.
8157 2017-10-18 Martin Liska <mliska@suse.cz>
8160 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8161 on an abnormal edge.
8163 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8165 * doc/invoke.texi (ffunction-sections and fdata-sections):
8168 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8170 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8171 the use statement can throw internally.
8173 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8175 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8176 any RTX present on the RHS of a SET.
8177 * compare-elim.c (try_eliminate_compare): Restore comment.
8179 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8181 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8182 may be also called on tcc_type nodes.
8183 * langhooks.c (lhd_tree_size): Likewise.
8185 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8187 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8188 format_warning_at_substring.
8189 (maybe_warn): Convert source_range * param to a location_t. Pass
8190 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8191 (format_directive): Remove code to extract source_ranges and
8192 source_range * in favor of just a location_t.
8193 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8195 * substring-locations.c (format_warning_va): Convert
8196 source_range * param to a location_t.
8197 (format_warning_at_substring): Likewise.
8198 * substring-locations.h (format_warning_va): Likewise.
8199 (format_warning_at_substring): Likewise.
8201 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8203 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8205 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8206 and scatter/gather ops.
8208 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8209 vec_gather_load and vec_scatter_store.
8210 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8211 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8212 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8213 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8214 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8215 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8217 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8219 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8220 Detect FTST instruction and handle its register pops. Only pop
8221 second operand if can_pop_second_op is true.
8222 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8223 set can_pop_second_op to false in the compare_for_stack_reg call.
8225 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8226 output_fp_compare for stack register operands.
8227 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8228 instructions here. Do not emit stack register pops here. Assert
8229 that FCOMPP pops next to top stack register. Rewrite function.
8231 2017-10-17 Nathan Sidwell <nathan@acm.org>
8234 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8235 use DECL_ASSEMBLER_NAME_RAW.
8238 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8241 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8242 Wilco Dijkstra <wilco.dijkstra@arm.com>
8244 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8245 converstion to Pmode from the buf_addr.
8247 2017-10-17 Richard Biener <rguenther@suse.de>
8249 * graphite-dependences.c (scop_get_reads_and_writes): Change
8250 output parameters to references.
8252 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8254 PR 71026/tree-optimization
8255 * fold-const.c (distribute_real_division): Removed.
8256 (fold_binary_loc): Remove calls to distribute_real_divison.
8258 2017-10-17 Richard Biener <rguenther@suse.de>
8260 * graphite-scop-detection.c
8261 (scop_detection::stmt_has_simple_data_refs_p): Always use
8262 the full nest as region.
8263 (try_generate_gimple_bb): Likewise.
8264 * sese.c (scalar_evolution_in_region): Simplify now that
8265 SCEV can handle instantiation in regions.
8266 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8267 in the non-loop part of a function if requested.
8269 2017-10-17 Richard Biener <rguenther@suse.de>
8271 PR tree-optimization/82563
8272 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8274 (graphite_regenerate_ast_isl): Call it.
8275 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8277 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8279 PR tree-optimization/82549
8280 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8281 Formatting fixes. Instead of calling make_bit_field_ref with negative
8284 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8286 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8287 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8288 _mm_maskz_reduce_ss): New.
8289 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8290 __builtin_ia32_reducess_mask): Ditto..
8291 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8292 * config/i386/sse.md (reduces<mode>): Renamed to ...
8293 (reduces<mode><mask_scalar_name>): ... this.
8294 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8296 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8297 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8299 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8301 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8302 * selftest-run-tests.c (selftest::run_tests): Call
8303 selftest::unique_ptr_tests_cc_tests.
8304 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8305 * unique-ptr-tests.cc: New file.
8307 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8310 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8312 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8313 (make_hard_regno_born, make_hard_regno_dead): Update
8314 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8316 2017-10-16 Jeff Law <law@redhat.com>
8318 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8320 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8322 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8324 2017-10-16 Olivier Hainque <hainque@adacore.com>
8326 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8327 with_cpu if we were configured for an e500v2 target cpu name.
8329 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8331 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8332 * doc/invoke.texi: Document +nodsp as a valid extension for
8335 2017-10-16 Martin Liska <mliska@suse.cz>
8337 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8338 (test_set_range): Likewise.
8339 (test_range_functions): Rename to ...
8340 (test_bit_in_range): ... this.
8341 (sbitmap_c_tests): Add new test.
8343 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8345 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8347 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8348 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8350 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8352 * config/aarch64/aarch64-builtins.c
8353 (aarch64_types_quadopu_lane_qualifiers): New.
8354 (TYPES_QUADOPU_LANE): New.
8355 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8356 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8357 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8358 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8359 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8360 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8361 (Vdottype, DOTPROD): New.
8362 (sur): Add SDOT and UDOT.
8364 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8366 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8367 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8368 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8370 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8371 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8372 Enable TARGET_DOTPROD.
8373 (cortex-a75.cortex-a55): Likewise.
8374 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8376 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8378 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8379 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8381 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8383 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8384 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8385 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8386 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8387 * config/arm/types.md (neon_dot, neon_dot_q): New.
8388 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8390 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8392 * config/arm/arm.h (TARGET_DOTPROD): New.
8393 * config/arm/arm.c (arm_arch_dotprod): New.
8394 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8395 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8396 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8397 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8398 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8399 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8400 * doc/invoke.texi (armv8.2-a): Document dotprod
8402 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8404 * i386.c (ix86_vec_cost): New function.
8405 (ix86_rtx_costs): Handle vector operations better.
8406 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8407 * x86-tune-costs.h: Add new costs to all tables.
8409 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8411 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8413 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8414 divsd, sqrtss and sqrtsd
8415 * x86-tune-costs.h: Add new entries to all costs.
8416 (znver1_cost): Fix to match real instruction latencies.
8418 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8419 Michael Collison <michael.collison@arm.com>
8421 * compare-elim.c: Include emit-rtl.h.
8422 (can_merge_compare_into_arith): New function.
8423 (try_validate_parallel): Likewise.
8424 (try_merge_compare): Likewise.
8425 (try_eliminate_compare): Call the above when no previous clobber
8427 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8430 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8434 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8435 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8439 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8440 to be any operand_equal_p operands. For & (B - 1) require
8441 B to be power of 2. Recognize
8442 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8444 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8447 * optabs.c (expand_memory_blockage): Fix call of
8448 targetm.have_memory_blockage.
8450 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8453 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8454 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8455 objects to extra_objs instead of overwriting it.
8457 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8459 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8460 Use any_fp_register_operand as operand[3] predicate. Simplify
8461 equality test for operands[2] and operands[4] memory location.
8462 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8463 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8464 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8465 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8466 any_fp_register_operand as operand[1] predicate. Simplify
8467 equality test for operands[0] and operands[3] memory location.
8468 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8469 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8470 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8472 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8474 * target-insns.def: Add memory_blockage.
8475 * optabs.c (expand_memory_blockage): New function.
8476 (expand_asm_memory_barrier): Rename ...
8477 (expand_asm_memory_blockage): ... to this.
8478 (expand_mem_thread_fence): Call expand_memory_blockage
8479 instead of expand_asm_memory_barrier.
8480 (expand_mem_singnal_fence): Ditto.
8481 (expand_atomic_load): Ditto.
8482 (expand_atomic_store): Ditto.
8483 * doc/md.texi (Standard Pattern Names For Generation):
8484 Document memory_blockage instruction pattern.
8486 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8488 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8489 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8490 * config/rl78/rl78.md: New define_expand "adddi3".
8492 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8494 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8497 2017-10-13 Jeff Law <law@redhat.com>
8499 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8501 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8504 * internal-fn.c (expand_mul_overflow): If both operands have
8505 the same highpart of -1 or 0 and the topmost bit of lowpart
8506 is different, overflow is if res <= 0 rather than res < 0.
8508 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8510 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8511 TARGET_P9_VECTOR code for unaligned_load case.
8513 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8515 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8517 2017-10-13 Nathan Sidwell <nathan@acm.org>
8519 * tree-core.h (tree_contains_struct): Make bool.
8520 * tree.c (tree_contains_struct): Likewise.
8521 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8522 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8523 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8524 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8525 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8527 2017-10-13 Richard Biener <rguenther@suse.de>
8529 * graphite-isl-ast-to-gimple.c
8530 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8531 parameters and dominance check.
8532 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8533 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8534 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8535 Do not update SSA form here or do intermediate IL verification.
8536 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8537 (graphite_initialize): Remove check on the number of loops in
8538 the function and inline into graphite_transform_loops.
8539 (graphite_finalize): Inline into graphite_transform_loops.
8540 (graphite_transform_loops): Perform SSA update and IL verification
8542 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8544 2017-10-13 Richard Biener <rguenther@suse.de>
8546 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8547 graphite_expression_type_precision): Avoid global constructor
8549 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8550 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8551 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8552 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8553 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8554 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8556 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8559 * config/i386/i386.h (ix86_red_zone_size): New.
8560 * config/i386/i386.md (push peephole2s): Replace
8561 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8563 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8564 Alan Hayward <alan.hayward@arm.com>
8565 David Sherwood <david.sherwood@arm.com>
8567 * combine.c (can_change_dest_mode): Reject changes in
8568 REGMODE_NATURAL_SIZE.
8570 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8571 Alan Hayward <alan.hayward@arm.com>
8572 David Sherwood <david.sherwood@arm.com>
8574 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8575 (expand_debug_source_expr): Likewise.
8576 * combine.c (combine_simplify_rtx): Likewise.
8577 * cse.c (fold_rtx): Likewise.
8578 * fwprop.c (canonicalize_address): Likewise.
8579 * targhooks.c (default_shift_truncation_mask): Likewise.
8581 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8582 Alan Hayward <alan.hayward@arm.com>
8583 David Sherwood <david.sherwood@arm.com>
8585 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8586 (widened_mode): Likewise.
8587 (expand_unop): Likewise.
8588 * ree.c (transform_ifelse): Likewise.
8589 (merge_def_and_ext): Likewise.
8590 (combine_reaching_defs): Likewise.
8591 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8593 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8594 Alan Hayward <alan.hayward@arm.com>
8595 David Sherwood <david.sherwood@arm.com>
8597 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8598 * combine.c (gen_lowpart_for_combine): Likewise.
8599 * dwarf2out.c (rtl_for_decl_location): Likewise.
8600 * final.c (alter_subreg): Likewise.
8601 * rtlhooks.c (gen_lowpart_general): Likewise.
8602 (gen_lowpart_if_possible): Likewise.
8604 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8605 Alan Hayward <alan.hayward@arm.com>
8606 David Sherwood <david.sherwood@arm.com>
8608 * calls.c (expand_call): Use subreg_lowpart_offset.
8609 * cse.c (cse_insn): Likewise.
8610 * regcprop.c (copy_value): Likewise.
8611 (copyprop_hardreg_forward_1): Likewise.
8613 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8616 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8617 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8618 =Q constraints to +Q and into insn condition add check
8619 that operands[0] and operands[1] are equal.
8620 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8621 =Q constraints to +Q and into insn condition add check
8622 that operands[0] is equal to either operands[1] or operands[2].
8625 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8626 instead of handling MINUS_EXPR twice (once for each argument),
8627 canonicalize operand order and handle just once, use rtype where
8628 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8631 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8632 any values of __C while still being pattern recognizable as a simple
8635 2017-10-13 Richard Biener <rguenther@suse.de>
8637 PR tree-optimization/82451
8639 2017-10-02 Richard Biener <rguenther@suse.de>
8641 PR tree-optimization/82355
8642 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8643 a mapping for the enclosing loop but avoid generating one for
8645 (copy_bb_and_scalar_dependences): Remove premature codegen
8646 error on PHIs in blocks duplicated into multiple places.
8647 * graphite-scop-detection.c
8648 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8649 in the region use it as loop and nest to analyze the DR in.
8650 (try_generate_gimple_bb): Likewise.
8651 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8652 (add_loop_constraints): For blocks in a loop not in the region
8653 create a dimension with a single iteration.
8654 * sese.h (gbb_loop_at_index): Remove assert.
8656 * cfgloop.c (loop_preheader_edge): For the loop tree root
8657 return the single successor of the entry block.
8658 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8659 Reset the SCEV hashtable and niters.
8660 * graphite-scop-detection.c
8661 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8662 assert that we only have POLYNOMIAL_CHREC that vary in loops
8663 contained in the region.
8664 (scop_detection::graphite_can_represent_expr): Adjust.
8665 (scop_detection::stmt_has_simple_data_refs_p): For loops
8666 not in the region set loop to NULL. The nest is now the
8667 entry edge to the region.
8668 (try_generate_gimple_bb): Likewise.
8669 * sese.c (scalar_evolution_in_region): Adjust for
8670 instantiate_scev change.
8671 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8672 Make nest parameter the edge into the region.
8673 (create_data_ref): Likewise.
8674 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8675 entry edge into a region and adjust instantiate_scev calls.
8676 (create_data_ref): Likewise.
8677 (graphite_find_data_references_in_stmt): Likewise.
8678 (find_data_references_in_stmt): Pass the loop preheader edge
8679 from the nest argument.
8680 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8681 parameter the edge into the region.
8682 (instantiate_parameters): Use the loop preheader edge as entry.
8683 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8685 (get_instantiated_value_entry): Make instantiate_below parameter
8686 the edge into the region.
8687 (instantiate_scev_name): Likewise. Adjust dominance checks,
8688 when we cannot use loop-based instantiation instantiate by
8689 walking use-def chains.
8690 (instantiate_scev_poly): Adjust.
8691 (instantiate_scev_binary): Likewise.
8692 (instantiate_scev_convert): Likewise.
8693 (instantiate_scev_not): Likewise.
8694 (instantiate_array_ref): Remove.
8695 (instantiate_scev_3): Likewise.
8696 (instantiate_scev_2): Likewise.
8697 (instantiate_scev_1): Likewise.
8698 (instantiate_scev_r): Do not blindly handle N-operand trees.
8699 Do not instantiate array-refs. Handle all constants and invariants.
8700 (instantiate_scev): Make instantiate_below parameter
8701 the edge into the region.
8702 (resolve_mixers): Use the loop preheader edge for the region
8703 parameter to instantiate_scev_r.
8704 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8706 2017-10-13 Richard Biener <rguenther@suse.de>
8708 PR tree-optimization/82525
8709 * graphite-isl-ast-to-gimple.c
8710 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8712 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8713 Fail code generation when we cannot represent the isl integer.
8714 (binary_op_to_tree): Elide modulo operations that are no-ops
8715 in the type we code generate. Remove now superfluous code
8718 2017-10-13 Richard Biener <rguenther@suse.de>
8720 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8721 (scop_detection::harmful_loop_in_region): Remove premature
8722 IV type restriction.
8723 (scop_detection::graphite_can_represent_scev): We can handle
8724 pointer IVs just fine.
8726 2017-10-13 Alan Modra <amodra@gmail.com>
8728 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8729 "Clobbers and Scratch Registers". Add paragraph on
8730 alternative to clobbers for scratch registers and OpenBLAS
8733 2017-10-13 Alan Modra <amodra@gmail.com>
8735 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8736 example of a memory input for a string of known length. Move
8737 commentary out of table. Add a number of new examples
8738 covering array memory inputs.
8740 2017-10-12 Martin Liska <mliska@suse.cz>
8742 PR tree-optimization/82493
8743 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8744 (test_range_functions): New function.
8745 (sbitmap_c_tests): Likewise.
8746 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8747 * selftest.h (sbitmap_c_tests): New function.
8749 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8751 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8753 * config/rs6000/amo.h: Fix spacing issue.
8755 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8758 * config/i386/i386.md (*ashl<mode>3_mask_1,
8759 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8760 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8763 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8765 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8766 (profile_probability): Set max_probability
8767 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8769 * profile-count.c (profile_probability::differs_from_p): Do not
8770 rely on max_probaiblity == 10000
8772 2017-10-12 Jeff Law <law@redhat.com>
8774 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8777 2017-10-12 Martin Sebor <msebor@redhat.com>
8781 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8782 (handle_alias_pairs): Call it.
8783 * common.opt (-Wattribute-alias): New option.
8784 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8785 * doc/invoke.texi (-Wattribute-alias): Document.
8787 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8790 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8792 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8794 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8795 (make_hard_regno_born, make_hard_regno_dead): Update
8796 bb_killed_pseudos and bb_gen_pseudos.
8798 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8800 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8802 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8804 * config/alpha/alpha.c (alpha_split_conditional_move):
8805 Use std::swap instead of manually swapping.
8806 (alpha_stdarg_optimize_hook): Ditto.
8807 (alpha_canonicalize_comparison): Ditto.
8809 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8811 * tree-loop-distribution.c (struct builtin_info): New struct.
8812 (struct partition): Refactor fields into struct builtin_info.
8813 (partition_free): Free struct builtin_info.
8814 (build_size_arg_loc, build_addr_arg_loc): Delete.
8815 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8816 information from struct builtin_info.
8817 (find_single_drs): New function refactored from classify_partition.
8818 Also moved builtin validity checks to this function.
8819 (compute_access_range, alloc_builtin): New functions.
8820 (classify_builtin_st, classify_builtin_ldst): New functions.
8821 (classify_partition): Refactor code into functions find_single_drs,
8822 classify_builtin_st and classify_builtin_ldst.
8823 (distribute_loop): Don't do runtime alias check when distributing
8825 (find_seed_stmts_for_distribution): New function.
8826 (pass_loop_distribution::execute): Refactor code finding seed
8827 stmts into above function. Support distribution for the innermost
8828 two-level loop nest. Adjust dump information.
8830 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8832 * tree-loop-distribution.c: Adjust the general comment.
8833 (NUM_PARTITION_THRESHOLD): New macro.
8834 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
8835 (classify_partition): Skip builtin pattern of loop nest's inner loop.
8836 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
8837 in call to build_partition_graph.
8838 (finalize_partitions): New parameter. Make loop distribution more
8839 conservative by fusing more partitions.
8840 (distribute_loop): Don't do runtime alias check in case of loop nest
8842 (find_seed_stmts_for_distribution): New function.
8843 (prepare_perfect_loop_nest): New function.
8844 (pass_loop_distribution::execute): Refactor code finding seed stmts
8845 and loop nest into above functions. Support loop nest distribution.
8846 Adjust dump information accordingly.
8848 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8850 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
8851 and set PTYPE_SEQUENTIAL for merged partition.
8853 2017-10-12 Richard Biener <rguenther@suse.de>
8855 PR tree-optimization/69728
8857 2017-09-19 Richard Biener <rguenther@suse.de>
8859 PR tree-optimization/69728
8860 * graphite-sese-to-poly.c (schedule_error): New global.
8861 (add_loop_schedule): Handle empty domain by failing the
8863 (build_original_schedule): Handle schedule_error.
8865 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
8866 domain by returning an unchanged schedule.
8868 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8870 * genrecog.c (validate_pattern): For VEC_SELECT verify that
8871 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
8873 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
8875 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
8878 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8881 * expr.c (store_field): Don't optimize away bitsize == 0 store
8882 from CALL_EXPR with addressable return type.
8884 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8886 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
8887 * config/rs6000/rs6000.md (sel): Delete mode attribute.
8888 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
8889 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
8890 TARGET_ISEL instead of TARGET_ISEL<sel>.
8892 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
8894 * config/rs6000/rs6000.c
8895 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
8897 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8899 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
8901 (reg_or_cint_operand): Fix comment.
8902 (reg_or_zero_operand): New predicate.
8903 * config/rs6000/rs6000-protos.h (output_isel): Delete.
8904 * config/rs6000/rs6000.c (output_isel): Delete.
8905 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
8906 instead of reg_or_cint_operand. Output instruction directly (not via
8908 (isel_unsigned_<mode>): Ditto.
8909 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
8910 gpc_reg_operand. Add an instruction alternative for this. Output
8911 instruction directly.
8912 (*isel_reversed_unsigned_<mode>): Ditto.
8914 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8916 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
8917 (TARGET_CANONICALIZE_COMPARISON): Define.
8919 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
8922 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8923 Check whether the dest is REG before adding REG_EQUIV note.
8925 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8928 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8930 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8931 (make_hard_regno_born, make_hard_regno_dead): Update
8932 bb_killed_pseudos and bb_gen_pseudos.
8934 2017-10-11 Nathan Sidwell <nathan@acm.org>
8936 * incpath.h (enum incpath_kind): Name enum, prefix values.
8937 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
8938 * incpath.c (heads, tails): Use INC_MAX.
8939 (add_env_var_paths, add_standard_paths): Use incpath_kind.
8940 (merge_include_chains, split_quote_chain,
8941 register_include_chains): Update incpath_kind names.
8942 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
8943 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
8945 (add_framework_path, darwin_register_objc_includes): Likewise.
8946 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
8948 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8950 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
8951 Do not use float_operator operator predicate.
8952 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
8953 * config/i386/predicates.md (float_operator): Remove predicate.
8955 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8957 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
8958 (*jccxf_i387): Ditto.
8959 (*jcc<mode>_i387): Ditto.
8960 (*jccu<mode>_i387): Ditto.
8961 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
8962 (*jcc_*_i387 splitters): Remove.
8963 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
8964 * config/i386/i386.c (ix86_split_fp_branch): Remove.
8965 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
8968 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8970 * profile-count.h (slow_safe_scale_64bit): New function.
8971 (safe_scale_64bit): New inline.
8972 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
8973 * profile-count.c: Include wide-int.h
8974 (slow_safe_scale_64bit): New.
8976 2017-10-11 Nathan Sidwell <nathan@acm.org>
8978 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
8979 HAS_DECL_ASSEMBLER_NAME_P.
8980 * gimple-expr.c (gimple_decl_printable_name: Check
8981 HAS_DECL_ASSEMBLER_NAME_P too.
8982 * ipa-utils.h (type_in_anonymous_namespace_p): Check
8983 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
8984 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
8985 * passes.c (rest_of_decl_compilation): Check
8986 HAS_DECL_ASSEMBLER_NAME_P too.
8987 * recog.c (verify_changes): Likewise.
8988 * tree-pretty-print.c (dump_decl_name): Likewise.
8989 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
8991 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
8992 (SET_DECL_ASSEMBLER_NAME): Use it.
8993 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
8994 (COPY_DECL_ASSEMBLER_NAME): Likewise.
8995 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
8997 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8999 * config.gcc (i386, x86_64): Add extra objects.
9000 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9001 (ix86_min_insn_size): Declare.
9002 (ix86_issue_rate): Declare.
9003 (ix86_adjust_cost): Declare.
9004 (ia32_multipass_dfa_lookahead): Declare.
9005 (ix86_macro_fusion_p): Declare.
9006 (ix86_macro_fusion_pair_p): Declare.
9007 (ix86_bd_has_dispatch): Declare.
9008 (ix86_bd_do_dispatch): Declare.
9009 (ix86_core2i7_init_hooks): Declare.
9010 (ix86_atom_sched_reorder): Declare.
9011 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9012 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9013 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9014 (rip_relative_addr_p): Rename to ...
9015 (ix86_rip_relative_addr_p): ... this one; export.
9016 (memory_address_length): Update.
9017 (ix86_issue_rate): Move to x86-tune-sched.c.
9018 (ix86_flags_dependent): Move to x86-tune-sched.c.
9019 (ix86_agi_dependent): Move to x86-tune-sched.c.
9020 (exact_dependency_1): Move to x86-tune-sched.c.
9021 (exact_store_load_dependency): Move to x86-tune-sched.c.
9022 (ix86_adjust_cost): Move to x86-tune-sched.c.
9023 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9024 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9025 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9026 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9027 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9028 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9029 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9030 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9031 (min_insn_size): Rename to ...
9032 (ix86_min_insn_size): ... this one; export.
9033 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9034 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9035 (core2i7_first_cycle_multipass_backtrack): Move to
9036 x86-tune-sched-core.c.
9037 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9038 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9039 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9040 (ix86_avoid_jump_mispredicts): Update.
9041 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9042 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9043 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9044 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9045 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9046 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9047 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9048 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9049 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9050 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9051 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9052 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9053 (BIG): Move to ix86-tune-sched-bd.c.
9054 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9055 (enum insn_path): Move to ix86-tune-sched-bd.c.
9056 (get_mem_group): Move to ix86-tune-sched-bd.c.
9057 (is_cmp): Move to ix86-tune-sched-bd.c.
9058 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9059 (is_branch): Move to ix86-tune-sched-bd.c.
9060 (is_prefetch): Move to ix86-tune-sched-bd.c.
9061 (init_window): Move to ix86-tune-sched-bd.c.
9062 (allocate_window): Move to ix86-tune-sched-bd.c.
9063 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9064 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9065 (process_end_window): Move to ix86-tune-sched-bd.c.
9066 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9067 (find_constant): Move to ix86-tune-sched-bd.c.
9068 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9069 (has_immediate): Move to ix86-tune-sched-bd.c.
9070 (get_insn_path): Move to ix86-tune-sched-bd.c.
9071 (get_insn_group): Move to ix86-tune-sched-bd.c.
9072 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9073 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9074 (add_insn_window): Move to ix86-tune-sched-bd.c.
9075 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9076 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9077 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9078 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9079 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9080 (do_dispatch): Move to ix86-tune-sched-bd.c.
9081 (has_dispatch): Move to ix86-tune-sched-bd.c.
9082 * config/i386/t-i386: Add new object files.
9083 * config/i386/x86-tune-costs.h: New file.
9084 * config/i386/x86-tune-sched-atom.c: New file.
9085 * config/i386/x86-tune-sched-bd.c: New file.
9086 * config/i386/x86-tune-sched-core.c: New file.
9087 * config/i386/x86-tune-sched.c: New file.
9089 2017-10-11 Liu Hao <lh_mouse@126.com>
9091 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9092 the generic version below instead.
9093 (should_colorize): Recognize Windows consoles as terminals
9095 * pretty-print.c [__MINGW32__] (write_all): New function.
9096 [__MINGW32__] (find_esc_head): Likewise.
9097 [__MINGW32__] (find_esc_terminator): Likewise.
9098 [__MINGW32__] (eat_esc_sequence): Likewise.
9099 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9101 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9104 2017-10-11 Richard Biener <rguenther@suse.de>
9106 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9107 Properly call analyze_scalar_evolution with the loop of the stmt.
9109 2017-10-11 Richard Biener <rguenther@suse.de>
9111 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9112 * tree-core.h (tree_base): Add chrec_var union member.
9113 * tree.h (CHREC_VAR): Remove.
9114 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9115 * tree-chrec.h (build_polynomial_chrec): Adjust.
9116 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9117 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9119 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9121 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9122 * match.pd: ... here.
9123 ((T) X == (T) Y): Relax condition.
9125 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9127 PR tree-optimization/82472
9128 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9130 (break_alias_scc_partitions): Update postorder number.
9132 2017-10-11 Martin Liska <mliska@suse.cz>
9135 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9137 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9139 2017-10-11 Martin Liska <mliska@suse.cz>
9141 * print-rtl.c (print_insn): Move declaration of idbuf
9142 to same scope as name.
9144 2017-10-11 Martin Liska <mliska@suse.cz>
9149 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9152 2017-10-11 Martin Liska <mliska@suse.cz>
9155 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9158 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9160 * config.gcc (default_gnu_indirect_function): Default to yes for
9161 arm*-*-linux* with glibc.
9163 2017-10-11 Richard Biener <rguenther@suse.de>
9165 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9166 default-defs and types we do not want to analyze.
9167 (interpret_loop_phi): Replace unreachable code with an assert.
9168 (compute_scalar_evolution_in_loop): Remove and inline ...
9169 (analyze_scalar_evolution_1): ... here, replacing condition with
9170 what makes the intent clearer. Remove handling of cases
9171 get_scalar_evolution now handles.
9173 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9175 PR rtl-optimization/81434
9176 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9177 comment for main loop. In sched_group_found if, also add checks for
9178 pass and min_cost_group.
9180 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9182 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9183 (rs6000_insn_cost): New function.
9184 * config/rs6000/rs6000.md (cost): New attribute.
9186 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9187 H.J. Lu <hongjiu.lu@intel.com>
9191 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9192 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9193 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9194 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9195 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9196 OPTION_MASK_ISA_MMX and some other bit set require both
9197 mmx and the ISAs without the mmx bit.
9198 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9199 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9200 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9201 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9202 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9203 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9204 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9205 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9206 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9207 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9208 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9209 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9210 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9211 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9212 Add OPTION_MASK_ISA_MMX.
9214 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9216 * config.gcc (armv7*-*-freebsd*): New target.
9217 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9219 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9221 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9222 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9225 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9227 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9228 a template parameter.
9229 (WIDE_INT_REF_FOR): Update accordingly.
9230 * tree.h (wi::int_traits <const_tree>): Delete.
9231 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9232 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9233 (wi::tree_to_wide_ref): New typedef.
9234 (wi::to_wide): New function.
9235 * calls.c (get_size_range): Use wi::to_wide when operating on
9237 * cgraph.c (cgraph_node::create_thunk): Likewise.
9238 * config/i386/i386.c (ix86_data_alignment): Likewise.
9239 (ix86_local_alignment): Likewise.
9240 * dbxout.c (stabstr_O): Likewise.
9241 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9242 * expr.c (const_vector_from_tree): Likewise.
9243 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9244 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9245 (fold_negate_expr_1, int_const_binop_1, const_binop)
9246 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9247 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9248 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9249 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9250 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9251 (fold_not_const, round_up_loc): Likewise.
9252 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9253 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9254 (alloca_call_type): Likewise.
9255 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9256 * godump.c (go_output_typedef): Likewise.
9257 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9258 * internal-fn.c (get_min_precision): Likewise.
9259 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9260 * ipa-polymorphic-call.c
9261 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9262 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9263 (ipa_modify_call_arguments): Likewise.
9264 * match.pd: Likewise.
9265 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9266 * print-tree.c (print_node_brief, print_node): Likewise.
9267 * stmt.c (expand_case): Likewise.
9268 * stor-layout.c (layout_type): Likewise.
9269 * tree-affine.c (tree_to_aff_combination): Likewise.
9270 * tree-cfg.c (group_case_labels_stmt): Likewise.
9271 * tree-data-ref.c (dr_analyze_indices): Likewise.
9272 (prune_runtime_alias_test_list): Likewise.
9273 * tree-dump.c (dequeue_and_dump): Likewise.
9274 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9275 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9276 * tree-pretty-print.c (dump_generic_node): Likewise.
9277 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9278 (simple_iv_with_niters): Likewise.
9279 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9280 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9281 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9282 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9283 (refine_value_range_using_guard, number_of_iterations_ne_max)
9284 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9285 (get_cst_init_from_scev, record_nonwrapping_iv)
9286 (scev_var_range_cant_overflow): Likewise.
9287 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9288 * tree-ssa-pre.c (compute_avail): Likewise.
9289 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9290 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9291 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9292 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9293 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9294 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9295 (dump_case_nodes, try_switch_expansion): Likewise.
9296 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9297 (vect_do_peeling): Likewise.
9298 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9299 * tree-vect-stmts.c (vectorizable_load): Likewise.
9300 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9301 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9302 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9303 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9304 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9305 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9306 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9307 (evrp_dom_walker::before_dom_children): Likewise.
9308 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9309 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9310 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9311 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9312 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9313 (get_range_pos_neg): Likewise.
9314 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9315 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9316 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9317 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9318 * config/avr/avr.c (avr_fold_builtin): Likewise.
9319 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9320 * config/msp430/msp430.c (msp430_attr): Likewise.
9321 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9322 * config/powerpcspe/powerpcspe-c.c
9323 (altivec_resolve_overloaded_builtin): Likewise.
9324 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9325 (rs6000_expand_ternop_builtin): Likewise.
9326 * config/rs6000/rs6000-c.c
9327 (altivec_resolve_overloaded_builtin): Likewise.
9328 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9329 (rs6000_expand_ternop_builtin): Likewise.
9330 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9332 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9334 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9335 when copying loop nest with only one inner loop.
9337 2017-10-10 Richard Biener <rguenther@suse.de>
9339 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9340 blocks if SCEV is active.
9341 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9343 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9344 (scev_initialize): Assert we are not yet initialized.
9346 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9348 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9349 inner loop's exit stmt by making it always exit the loop, otherwise
9350 we would generate an infinite empty loop.
9352 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9354 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9355 renaming variables in new preheader if it's deleted.
9357 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9359 * tree-loop-distribution.c (struct partition): Remove unused field
9360 loops of the structure.
9361 (partition_alloc, partition_free): Ditto.
9362 (build_rdg_partition_for_vertex): Ditto.
9364 2017-10-09 Jeff Law <law@redhat.com>
9366 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9367 return type to match prototype and documentation.
9369 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9371 * config/rs6000/rs6000.c (processor_costs): Move to ...
9372 * config/rs6000/rs6000.h: ... here.
9373 (rs6000_cost): Declare.
9375 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9377 * except.c (setjmp_fn): New global variable.
9378 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9379 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9380 if DONT_USE_BUILTIN_SETJMP is defined.
9382 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9384 * target.def (insn_cost): New hook.
9385 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9386 * doc/tm.texi: Regenerate.
9387 * rtlanal.c (insn_cost): Use the new hook.
9389 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9391 * combine.c (combine_validate_cost): Compute the new insn_cost,
9392 not just pattern_cost.
9393 (try_combine): Adjust comment.
9395 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9397 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9399 * combine.c (uid_insn_cost): Adjust comment.
9400 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9402 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9403 * dse.c (find_shift_sequence): Ditto.
9404 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9405 (bb_valid_for_noce_process_p): Use pattern_cost.
9406 * rtl.h (insn_rtx_cost): Delete.
9407 (pattern_cost): New prototype.
9408 (insn_cost): New prototype.
9409 * rtlanal.c (insn_rtx_cost): Rename to...
9410 (pattern_cost): ... this.
9413 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9415 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9416 (*jcc<mode>_0_r_i387): Ditto.
9417 (*jccxf_r_i387): Ditto.
9418 (*jcc<mode>_r_i387): Ditto.
9419 (*jccu<mode>_r_i387): Ditto.
9420 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9421 (*jcc): Rename from *jcc_1.
9423 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9425 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9426 deferred rescans after the lvx/stvx recombination pre-pass.
9428 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9430 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9431 memory operation instruction support.
9432 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9433 (rs6000-ibm-aix[789]*): Likewise.
9434 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9435 Document new functions.
9437 2017-10-09 Richard Biener <rguenther@suse.de>
9439 PR tree-optimization/82397
9440 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9441 equality only for semantically equal trees.
9443 2017-10-09 Richard Biener <rguenther@suse.de>
9445 PR tree-optimization/82449
9446 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9447 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9448 allow constant addresses.
9449 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9452 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9454 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9457 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9460 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9463 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9466 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9468 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9471 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9472 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9474 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9476 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9477 (WI_BINARY_PREDICATE_RESULT): Likewise.
9478 (wi::binary_traits::operator_result): New type.
9479 (wi::binary_traits::predicate_result): Likewise.
9480 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9481 (generic_wide_int::operator==, generic_wide_int::operator!=)
9482 (generic_wide_int::operator&, generic_wide_int::and_not)
9483 (generic_wide_int::operator|, generic_wide_int::or_not)
9484 (generic_wide_int::operator^, generic_wide_int::operator+
9485 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9487 (operator~, unary operator-, operator==, operator!=, operator&)
9488 (operator|, operator^, operator+, binary operator-, operator*): New
9490 * expr.c (get_inner_reference): Use wi::bit_and_not.
9491 * fold-const.c (fold_binary_loc): Likewise.
9492 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9493 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9494 (bit_value_binop): Likewise.
9495 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9496 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9497 (extract_range_from_binary_expr_1): Likewise.
9498 (masked_increment): Likewise.
9499 (simplify_bit_ops_using_ranges): Likewise.
9501 2017-10-09 Martin Jambor <mjambor@suse.cz>
9504 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9505 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9506 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9508 (hsa_fixup_mov_insn_type): New function.
9509 (hsa_op_with_type::get_in_type): Use it.
9510 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9511 immediates in an assert.
9512 (hsa_op_with_type::extend_int_to_32bit): New method.
9513 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9514 types. Convert to dest type if necessary.
9515 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9516 (reg_for_gimple_ssa): Pass false as min32int to
9517 hsa_type_for_scalar_tree_type.
9518 (gen_hsa_addr): Fixup type when creating addresable temporary.
9519 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9520 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9521 necessary. Call hsa_fixup_mov_insn_type.
9522 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9523 extend operands and convert to dest type if necessary.
9524 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9525 to dest type if necessary.
9526 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9527 if conversion nt necessary and size matches.
9528 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9529 to dest type if necessary.
9530 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9531 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9533 (gen_hsa_clrsb): Likewise.
9534 (gen_hsa_ffs): Likewise.
9535 (gen_hsa_divmod): Extend operands and convert to dest type if
9537 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9539 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9541 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9542 Remove empty default arguments. Use a brace block as output
9544 (conditional return): Ditto.
9546 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9547 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9549 (group_ending_nop): Ditto.
9550 (doloop_end): Ditto.
9551 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9552 (splitters for those): Ditto.
9554 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9556 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9557 a conditional jump (and the compare for it) so that pc_rtx is the
9559 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9560 for the deleted and renamed ctr<mode>_internal[234] patterns.
9561 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9562 Delete second conditional return pattern.
9563 (ctr<mode>_internal2): Delete this second bdnz pattern.
9564 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9565 (ctr<mode>_internal4): Delete this second bdz pattern.
9567 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9569 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9570 (always_initialized_rtx_for_ssa_name_p): New predicate.
9571 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9572 (finish_out_of_ssa): Free new field of SA.
9573 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9574 * tree-ssa-coalesce.c: Include tree-ssa.h.
9575 (get_parm_default_def_partitions): Remove extern keyword.
9576 (get_undefined_value_partitions): New function.
9577 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9578 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9580 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9582 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9584 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9586 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9588 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9589 for targets that preffer 128bit.
9591 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9593 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9595 2017-10-08 Olivier Hainque <hainque@adacore.com>
9597 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9598 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9600 (thumb_set_return_address): Likewise.
9602 2017-10-08 Olivier Hainque <hainque@adacore.com>
9604 * common/config/arm/arm-common.c (arm_except_unwind_info):
9605 Handle DWARF2_UNWIND_INFO.
9607 2017-10-07 Michael Collison <michael.collison@arm.com>
9609 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9612 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9614 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9615 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9616 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9617 defined, force the creation of a new block for a dispatch label.
9619 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9621 * invoke.texi (Wsuggest-attribute=cold): Document.
9622 * common.opt (Wsuggest-attribute=cold): New
9623 * ipa-pure-const.c (warn_function_cold): New function.
9624 * predict.c (compute_function_frequency): Use it.
9625 * predict.h (warn_function_cold): Declare.
9627 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9629 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9632 2017-10-06 Martin Liska <mliska@suse.cz>
9634 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9635 keyword for member functions.
9636 (struct sanopt_tree_couple): New struct.
9637 (struct sanopt_tree_couple_hash): New function.
9638 (struct sanopt_ctx): Add new hash_map.
9639 (has_dominating_ubsan_ptr_check): New function.
9640 (record_ubsan_ptr_check_stmt): Likewise.
9641 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9642 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9643 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9645 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9648 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9649 aarch64_simd_valid_immediate on CONST_VECTORs.
9650 (aarch64_reg_or_bic_imm): Likewise.
9652 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9654 PR rtl-optimization/82396
9655 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9657 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9659 * graphite-dependences.c (scop_get_reads): Move code to...
9660 (scop_get_must_writes): Move code to...
9661 (scop_get_may_writes): Move code to...
9662 (scop_get_reads_and_writes): ... here.
9663 (scop_get_dependences): Call scop_get_reads_and_writes.
9665 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9667 PR tree-optimization/82434
9668 * fold-const.h (can_native_encode_type_p,
9669 can_native_encode_string_p): Remove.
9670 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9671 don't encode anything, just return what would be otherwise returned.
9672 (native_encode_fixed, native_encode_complex, native_encode_vector):
9674 (native_encode_string): Likewise. Inline by hand
9675 can_native_encode_string_p.
9676 (can_native_encode_type_p): Remove.
9677 (can_native_encode_string_p): Remove.
9678 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9679 STRING_CSTs using can_native_encode_string_p, test all
9680 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9681 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9682 argument from native_encode_expr.
9683 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9684 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9687 2017-10-06 Richard Biener <rguenther@suse.de>
9689 PR tree-optimization/82397
9690 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9691 operand_equal_p but rely on data_ref_compare_tree for detecting
9693 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9694 to match up with dr_group_sort_cmp.
9696 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9699 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9701 * config/s390/s390-builtin-types.def: Regenerate.
9703 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9706 * config/s390/s390-builtin-types.def: Regenerate.
9707 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9708 Change flag from B_VXE to B_VX.
9709 (s390_vec_min_dbl): Remove B_VXE flag.
9711 2017-10-06 Richard Biener <rguenther@suse.de>
9713 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9714 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9715 translate_isl_ast_to_gimple::is_valid_rename,
9716 translate_isl_ast_to_gimple::get_rename,
9717 translate_isl_ast_to_gimple::get_def_bb_for_const,
9718 translate_isl_ast_to_gimple::get_new_name,
9719 translate_isl_ast_to_gimple::collect_all_ssa_names,
9720 translate_isl_ast_to_gimple::copy_loop_phi_args,
9721 translate_isl_ast_to_gimple::collect_all_ssa_names,
9722 translate_isl_ast_to_gimple::copy_loop_phi_args,
9723 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9724 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9725 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9726 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9727 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9728 translate_isl_ast_to_gimple::copy_cond_phi_args,
9729 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9730 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9731 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9732 translate_isl_ast_to_gimple::rename_uses,
9733 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9734 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9735 (set_rename_for_each_def): Likewise.
9736 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9737 here. Handle rewriting SCEV analyzable uses here.
9738 (copy_bb_and_scalar_dependences): Generate code for PHI
9740 (graphite_regenerate_ast_isl): Adjust.
9741 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9742 (add_write, add_read): New functions.
9743 (build_cross_bb_scalars_def): Use it and simplify.
9744 (build_cross_bb_scalars_use): Likewise.
9745 (graphite_find_cross_bb_scalar_vars): Inline into...
9746 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9747 simulating out-of-SSA. Compute liveout and add dependencies.
9748 (build_scops): Force an empty entry block.
9749 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9751 (sese_build_liveouts): Declare.
9752 (sese_trivially_empty_bb_p): Likewise.
9753 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9754 compute liveout and debug_liveout.
9755 (sese_bad_liveouts_use): Remove.
9756 (sese_reset_debug_liveouts_bb): Likewise.
9757 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9758 (sese_build_liveouts): Build liveout and debug_liveout and store
9760 (new_sese_info): Adjust.
9761 (free_sese_info): Likewise.
9762 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9763 do not build liveout here.
9764 (move_sese_in_condition): Adjust region entry.
9765 (scev_analyzable_p): Match up with chrec_apply requirements.
9766 (sese_trivially_empty_bb_p): New.
9767 * tree-into-ssa.c (get_reaching_def): Properly support generating
9768 default-defs for incremental rewrite of anonymous names.
9770 2017-10-06 Richard Biener <rguenther@suse.de>
9772 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9773 precision do not perform modulo reduction.
9775 2017-10-06 Richard Biener <rguenther@suse.de>
9777 PR tree-optimization/82436
9778 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9779 conservatively choose the vectorization factor when checking
9780 whether we can perform the required load permutation.
9781 (vect_transform_slp_perm_load): Assert when we may not fail.
9783 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9785 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9786 message for incompatible -msdata=* and -mcall-* options.
9788 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9790 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9791 rate for post-reload scheduling.
9793 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9795 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9797 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9799 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9800 to improve monte carlo in scimark.
9802 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9804 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9805 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9806 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9807 pentium4_cost, nocona_cost): Set reassociation width to 1.
9808 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9809 width to 2 for fp operations and 1 otherwise.
9810 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9812 (atom_cost): Set reassociation width to 2.
9813 (slm_cost, generic_cost): Set fp reassociation width
9814 to 2 and 1 otherwise.
9815 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
9816 (core_cost): Set fp reassociation width to 4 and vector to 2.
9817 (ix86_reassociation_width): Rewrite using cost table; special case
9818 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
9819 and TARGET_AVX128_OPTIMAL.
9820 * config/i386/i386.h (processor_costs): Add
9821 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
9822 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
9823 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
9824 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
9825 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
9826 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
9828 2017-10-05 Nathan Sidwell <nathan@acm.org>
9830 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
9832 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9834 * config/arm/arm.c (arm_test_fpu_data): New.
9835 (arm_run_selftests): Call arm_test_fpu_data.
9837 2017-10-04 Nathan Sidwell <nathan@acm.org>
9839 * toplev.c (toplev::main): Remove excess parens on pretty_printer
9841 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
9843 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9845 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
9846 check type for aarch64_simd_valid_immediate.
9847 (aarch64_output_simd_mov_immediate): Update prototype.
9848 (aarch64_simd_valid_immediate): Update prototype.
9849 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
9850 support for ORR-immediate.
9851 (and<mode>3): modified pattern to add support for BIC-immediate.
9852 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
9853 now checks for valid immediate for BIC and ORR based on new enum
9855 (aarch64_output_simd_mov_immediate): Function now used to output
9856 BIC/ORR imm as well based on new enum argument.
9857 * config/aarch64/constraints.md (Do): New vector immediate constraint.
9859 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
9860 (aarch64_reg_or_bic_imm): Likewise.
9862 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9864 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
9865 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
9867 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9871 PR rtl-optimization/82396
9872 * haifa-sched.c (autopref_multipass_init): Simplify
9874 (autopref_rank_data): Simplify sort order.
9875 * sched-int.h (autopref_multipass_data_): Remove
9876 multi_mem_insn_p, min_offset and max_offset.
9878 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9880 * doc/sourcebuild.texi: Document vect_peeling_profitable.
9882 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9884 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
9887 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9889 * doc/sourcebuild.texi: Document vect_long_mult.
9891 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
9893 PR tree-optimization/82413
9894 * fold-const.c (build_range_check): Use widest_int when comparing
9895 the maximum ETYPE value with HIGH.
9897 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9899 PR rtl-optimization/82396
9900 * haifa-sched.c (autopref_multipass_init): Simplify
9902 (autopref_rank_data): Simplify sort order.
9903 * sched-int.h (autopref_multipass_data_): Remove
9904 multi_mem_insn_p, min_offset and max_offset.
9906 2017-10-04 Jakub Jelinek <jakub@redhat.com>
9908 PR tree-optimization/82381
9909 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
9910 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
9913 PR tree-optimization/82374
9914 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
9915 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
9916 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
9917 current_function_decl to the new decl.
9919 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9921 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
9922 helper macro for IEEE float128 hardware built-in functions.
9923 (SQRTF128_ODD): Add built-in functions with the round-to-odd
9925 (TRUNCF128_ODD): Likewise.
9926 (ADDF128_ODD): Likewise.
9927 (SUBF128_ODD): Likewise.
9928 (MULF128_ODD): Likewise.
9929 (DIVF128_ODD): Likewise.
9930 (FMAF128_ODD): Likewise.
9931 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
9932 UNSPEC_TRUNC_ROUND_TO_ODD.
9933 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
9934 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
9935 floating point round to odd instructions.
9936 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
9937 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
9938 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
9939 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
9940 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
9941 (trunc<mode>sf2_hw): Change the truncate with round to odd
9942 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
9943 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
9944 to odd hardware instructions.
9945 (sub<mode>3_odd): Likewise.
9946 (mul<mode>3_odd): Likewise.
9947 (div<mode>3_odd): Likewise.
9948 (sqrt<mode>2_odd): Likewise.
9949 (fma<mode>4_odd): Likewise.
9950 (fms<mode>4_odd): Likewise.
9951 (nfma<mode>4_odd): Likewise.
9952 (nfms<mode>4_odd): Likewise.
9953 (trunc<mode>df2_odd): Change the truncate with round to odd
9954 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
9956 * doc/extend.texi (PowerPC built-in functions): Update documentation
9957 for existing IEEE float128-bit built-in functions. Add built-in
9958 functions that generate the IEEE 128-bit floating point round to
9961 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
9963 PR rtl-optimization/77729
9964 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
9965 to (X&(C1&~C2))|C2 transformations.
9967 2017-10-03 Martin Jambor <mjambor@suse.cz>
9969 PR tree-optimization/82363
9970 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
9971 mismatch, mark lacc written regardless of racc.
9973 2017-10-03 Jakub Jelinek <jakub@redhat.com>
9975 PR tree-optimization/82381
9976 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
9977 stmt_to_insert nor wheather SSA_NAMEs are default defs.
9978 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
9979 fallthrough into reassoc_stmt_dominates_stmt_p.
9982 * combine.c (combine_instructions): Don't combine in unreachable
9985 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9988 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
9989 function to not use the have_cpu variable. Do not set cpu_index,
9990 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
9992 (rs6000_valid_attribute_p): Remove duplicate initializations of
9993 old_optimize and func_optimize.
9994 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
9995 (rs6000_activate_target_options): Make global.
9996 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
9999 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10001 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10002 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10003 Set *poffset to 0 and *psize and *pmax_size to -1 if
10004 *poffset + *psize overflows in HOST_WIDE_INT.
10006 PR tree-optimization/82387
10007 PR tree-optimization/82388
10008 PR tree-optimization/82389
10009 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10010 instead of live_bytes non-NULL.
10012 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10015 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10018 2017-10-02 Richard Biener <rguenther@suse.de>
10020 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10021 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10022 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10024 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10026 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10027 requested precision matches the type's.
10028 * calls.c (alloc_max_size): Calculate the new candidate size as
10029 a widest_int and use wi::to_widest when comparing it with the
10030 current candidate size.
10031 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10032 zero rather than integer_zero_node.
10033 * match.pd: Check for a no-op conversion before using wi::add
10034 rather than after. Use tree_to_uhwi when summing small shift
10035 counts into an unsigned int.
10037 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10038 Alan Hayward <alan.hayward@arm.com>
10039 David Sherwood <david.sherwood@arm.com>
10042 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10043 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10044 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10045 POINTER_AND_FP_REGS.
10047 2017-10-02 Richard Biener <rguenther@suse.de>
10049 PR tree-optimization/82355
10050 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10051 a mapping for the enclosing loop but avoid generating one for
10052 the loop tree root.
10053 (copy_bb_and_scalar_dependences): Remove premature codegen
10054 error on PHIs in blocks duplicated into multiple places.
10055 * graphite-scop-detection.c
10056 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10057 in the region use it as loop and nest to analyze the DR in.
10058 (try_generate_gimple_bb): Likewise.
10059 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10060 (add_loop_constraints): For blocks in a loop not in the region
10061 create a dimension with a single iteration.
10062 * sese.h (gbb_loop_at_index): Remove assert.
10064 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10066 * omp-expand.c (adjust_context_scope): New function.
10067 (expand_parallel_call): Call adjust_context_scope.
10069 2017-10-01 Jeff Law <law@redhat.com>
10071 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10072 dom_opt_dom_walker class with direct access to private members.
10073 Add comments. Call test_for_singularity.
10074 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10075 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10076 m_dummy_cond anymore.
10077 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10079 (pass_dominator:execute): Build the dummy_cond here and pass it
10080 to the dom_opt_dom_walker ctor.
10081 (test_for_singularity): New function.
10083 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10084 Maya Rashish <coypu@sdf.org>
10086 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10087 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10088 (alpha*-*-netbsd*) Use nbsd_tm_file.
10089 (arm*-*-netbsdelf*) Likewise.
10090 (i[34567]86-*-netbsdelf*) Likewise.
10091 (x86_64-*-netbsd*) Likewise.
10092 (mips*-*-netbsd*) Likewise.
10093 (powerpc-*-netbsd*) Likewise.
10094 (sh*-*-netbsd*) Likewise.
10095 (sparc-*-netbsdelf*) Likewise.
10096 (sparc64-*-netbsd*) Likewise.
10097 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10099 (vax-*-netbsdelf*) Likewise.
10100 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10101 (UINT_FAST8_TYPE) Likewise.
10102 (INT_FAST16_TYPE) Check CHAR_FAST16.
10103 (UINT_FAST16_TYPE) Likewise.
10105 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10108 * config/i386/i386.md
10109 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10110 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10111 *divmodsi4_zext_2): New define_insn_and_split.
10112 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10113 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10114 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10115 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10116 New define_insn_and_split.
10117 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10118 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10119 operands[1] having DImode when mode is SImode.
10121 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10122 always SImode for DIV and MOD in REG_EQUAL notes.
10124 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10126 PR middle-end/82319
10127 * match.pd: Fix handling of NaNs in pattern.
10129 2017-09-29 Jeff Law <law@redhat.com>
10131 * sbitmap.c (bitmap_bit_in_range_p): New function.
10132 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10133 * tree-ssa-dse.c (live_bytes_read): New function.
10134 (dse_classify_store): Ignore reads of dead bytes.
10136 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10137 typos and whitespace errors.
10138 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10139 * config/s390/s390.c (s390_emit_prologue): Likewise.
10141 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10144 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10145 with a symbol for LRA.
10147 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10149 PR rtl-optimization/82338
10150 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10152 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10154 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10155 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10156 (qsort_chk): Declare.
10157 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10158 (qsort_chk): New function.
10160 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10162 PR tree-optimization/82337
10163 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10164 phi definition if the PHI result appears in an abnormal PHI.
10165 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10166 basis appears in an abnormal PHI.
10168 2017-09-29 Richard Biener <rguenther@suse.de>
10170 * graphite-isl-ast-to-gimple.c
10171 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10172 (binary_op_to_tree): Use it.
10173 (get_rename_from_scev): Likewise.
10174 (copy_loop_phi_nodes): Likewise.
10175 (copy_bb_and_scalar_dependences): Likewise.
10176 (translate_pending_phi_nodes): Likewise.
10178 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10181 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10182 for movabsq $(i32 << shift), r64.
10184 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10186 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10187 index when encoding %esp as %rsp to avoid 0x67 prefix.
10189 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10191 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10192 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10194 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10196 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10197 Extensions with more than 16 double VFP registers.
10198 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10199 to_clear_mask and all code related to it. Replace the remaining
10200 entry by a sbitmap and adapt code accordingly.
10202 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10204 * brig-builtins.def: Change pure attributes to const.
10206 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10208 * config.gcc (default_gnu_indirect_function): Default to yes for
10209 sparc*-*-linux* with glibc.
10211 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10213 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10214 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10215 when creating .init_array and .fini_array sections with priority
10218 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10221 * config/aarch64/aarch64.c
10222 (aarch64_builtin_support_vector_misalignment): Always return false
10223 when misalignment is unknown.
10225 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10227 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10228 this function to return false if the definition used by the swap
10229 instruction is artificial, or if the memory address from which the
10230 constant value is loaded is not represented by a base address held
10231 in a register or if the base address register is a frame or stack
10232 pointer. Additionally, return false if the base address of the
10233 loaded constant is a SYMBOL_REF but is not considered to be a
10235 (replace_swapped_load_constant): New function.
10236 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10237 loaded constant vector with a load of a swapped constant vector.
10239 2017-09-27 Carl Love <cel@us.ibm.com>
10241 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10242 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10243 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10246 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10248 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10249 first, always call autopref_rank_data otherwise.
10251 2017-09-27 Richard Biener <rguenther@suse.de>
10253 * graphite-scop-detection.c (find_scop_parameters): Move
10254 loop bound handling ...
10255 (gather_bbs::before_dom_children): ... here, avoiding the need
10256 to build scop_info->loop_nest.
10257 (record_loop_in_sese): Remove.
10258 * sese.h (sese_info_t::loop_nest): Remove.
10259 * sese.c (new_sese_info): Do not allocate loop_nest.
10260 (free_sese_info): Do not free loop_nest.
10262 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10265 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10266 lhs from calls if the lhs has addressable type.
10268 2017-09-27 Richard Biener <rguenther@suse.de>
10270 * graphite.h (scop::max_alias_set): New member.
10271 * graphite-scop-detection.c: Remove references to non-existing
10272 --param in comments.
10273 (build_alias_sets): Record the maximum alias set used for drs.
10274 (build_scops): Support zero as unlimited for
10275 --param graphite-max-arrays-per-scop.
10276 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10277 and inline into ...
10278 (build_poly_sr_1): ... here. Compute alias set based on the
10279 maximum alias set used for drs rather than
10280 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10282 2017-09-27 Richard Biener <rguenther@suse.de>
10284 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10285 --param loop-block-tile-size=0 to disable tiling.
10287 2017-09-27 Richard Biener <rguenther@suse.de>
10289 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10290 (graphite-max-nb-scop-params): Document special value zero.
10291 * domwalk.h (dom_walker::STOP): New symbolical constant.
10292 (dom_walker::dom_walker): Add optional parameter for bb to
10294 (dom_walker::~dom_walker): Declare.
10295 (dom_walker::before_dom_children): Document STOP return value.
10296 (dom_walker::m_user_bb_to_rpo): New member.
10297 (dom_walker::m_bb_to_rpo): Likewise.
10298 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10299 mapping here if not provided by the user.
10300 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10301 provided by the user.
10302 (dom_walker::STOP): Define.
10303 (dom_walker::walk): Do not compute bb to RPO mapping here.
10304 Support STOP return value from before_dom_children to stop
10306 * graphite-optimize-isl.c (optimize_isl): If the schedule
10307 is the same still generate code if -fgraphite-identity
10308 or -floop-parallelize-all are given.
10309 * graphite-scop-detection.c: Include cfganal.h.
10310 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10312 (gather_bbs::before_dom_children): Return STOP for BBs
10314 (build_scops): Compute bb to RPO mapping and pass it to
10315 the domwalk. Treat --param graphite-max-nb-scop-params=0
10316 as not limiting the number of params.
10317 * graphite.c (graphite_initialize): Remove limit on the
10318 number of basic-blocks in a function.
10319 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10320 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10321 default value of 10.
10323 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10325 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10326 Adjust code to eliminate needing to do the shift right 32-bits
10327 operation after XSCVDPSPN.
10329 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10331 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10332 ((X / Y) != 0 -> X >= Y): Likewise.
10334 2017-09-26 Carl Love <cel@us.ibm.com>
10336 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10337 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10338 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10339 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10340 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10341 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10342 definitions and overloading.
10343 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10344 statement for P9V_BUILTIN_XST_LEN_R.
10345 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10346 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10347 define_expand and define_insn for the instructions and builtins.
10348 * doc/extend.texi: Update the built-in documentation file for the new
10349 built-in functions.
10350 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10351 define_insn for the instructions
10353 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10356 * gcc/config/netbsd-protos.h: New file.
10357 * gcc/config/netbsd.c: New file.
10358 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10359 * gcc/config/t-netbsd: New file.
10360 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10361 (tmake_file) Add t-netbsd.
10362 (extra_objs) Add netbsd.o.
10364 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10368 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10370 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10372 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10373 sign extension from a vector register to a GPR by doing a 32-bit
10374 direct move and then an EXTSW.
10375 (extendsi<mode>2 splitter): Likewise.
10376 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10377 right or vector extract after doing XSCVDPSPN. Use
10378 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10380 (movdi_from_sf_zero_ext): Likewise.
10381 (reload_gpr_from_vsxsf): Likewise.
10382 (p8_mfvsrd_4_disf): Delete, no longer used.
10383 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10384 then needing to move the SFmode to a GPR to use the XSCVDPSP
10385 instruction instead of FRSP and XSCVDPSPN.
10386 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10387 it is adjacent to the other XSCVSPDP insns.
10388 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10389 SFmode to be in traditional Altivec registers.
10390 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10391 (vsx_xscvspdpn): Likewise.
10392 (vsx_xscvspdpn_scalar): Likewise.
10394 2017-09-26 Martin Jambor <mjambor@suse.cz>
10396 * tree-sra.c (compare_access_positions): Put integral types first,
10397 stabilize sorting of integral types, remove conditions putting
10398 non-full-precision integers last.
10399 (sort_and_splice_var_accesses): Disable scalarization if a
10400 non-integert would be represented by a non-full-precision integer.
10402 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10404 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10405 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10406 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10407 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10408 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10409 conditionals inside the function instead of around it. Call
10410 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10411 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10413 2017-09-26 Richard Biener <rguenther@suse.de>
10415 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10417 (scop_detection::build_scop_breadth): ... this. Removed.
10418 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10419 (scop_detection::harmful_stmt_in_bb): Likewise.
10420 (scop_detection::graphite_can_represent_stmt): Likewise.
10421 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10422 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10423 (scop_detection::can_represent_loop_1): ... this. Removed.
10424 (scop_detection::harmful_loop_in_region): Simplify after inlining
10425 the above and remove more quadraticness.
10426 (build_scops): Adjust.
10427 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10430 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10433 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10434 REGNO (base) == SP_REG if base is a REG.
10436 PR middle-end/35691
10437 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10438 if it is different SSA_NAME.
10439 (optimize_range_tests_cmp_bitwise): New function.
10440 (optimize_range_tests): Call it.
10442 2017-09-26 Richard Biener <rguenther@suse.de>
10444 PR tree-optimization/82321
10445 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10446 for the def being inside the loop.
10448 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10450 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10452 * config/s390/s390-builtins.def: Fix constraint on op4.
10454 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10456 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10457 independent expanders.
10458 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10459 ("vec_ordered", "vec_unordered"): New expanders.
10461 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10463 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10466 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10468 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10469 vec_unpacks_lo_v16qi.
10470 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10472 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10474 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10475 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10476 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10478 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10480 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10482 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10484 ("*vec_slb<mode>"): New insn pattern.
10485 ("vec_shr_<mode>"): New expander.
10486 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10487 and force the shift count operand to V16QImode.
10488 ("vec_srb<mode>"): Set shift count mode to V16QI.
10490 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10492 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10493 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10494 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10496 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10499 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10501 2017-09-26 Richard Biener <rguenther@suse.de>
10503 PR tree-optimization/82320
10504 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10507 2017-09-25 Jeff Law <law@redhat.com>
10509 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10510 prototype for new argument.
10511 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10512 mostly extracted from rs6000_emit_allocate_stack.
10513 (rs6000_emit_probe_stack_range_stack_clash): New function.
10514 (rs6000_emit_allocate_stack): Call
10515 rs6000_emit_probe_stack_range_stack_clash as needed.
10516 (rs6000_emit_probe_stack_range): Add additional argument
10517 to call to gen_probe_stack_range{si,di}.
10518 (output_probe_stack_range): New.
10519 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10520 (output_probe_stack_range_stack_clash): New.
10521 (rs6000_emit_prologue): Emit notes into dump file as requested.
10522 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10523 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10524 Add additional operand and pass it to output_probe_stack_range.
10526 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10528 PR tree-optimization/82163
10529 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10530 (checking_verify_loop_closed_ssa): New parameter.
10531 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10532 (check_loop_closed_ssa_stmt): Delete.
10533 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10534 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10535 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10538 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10540 * brig-builtins.def: Treat HSAIL barrier builtins as
10541 setjmp/longjump style functions.
10543 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10545 * target.def (constant_alignment): New hook.
10546 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10547 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10548 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10549 * doc/tm.texi: Regenerate.
10550 * targhooks.h (default_constant_alignment): Declare.
10551 (constant_alignment_word_strings): Likewise.
10552 * targhooks.c (default_constant_alignment): New function.
10553 (constant_alignment_word_strings): Likewise.
10554 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10555 instead of CONSTANT_ALIGNMENT.
10556 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10557 (force_const_mem): Likewise.
10558 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10559 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10560 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10561 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10562 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10564 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10565 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10566 constant_alignment_word_strings.
10567 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10568 (CONSTANT_ALIGNMENT): Likewise.
10569 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10570 (arm_constant_alignment): New function.
10571 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10572 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10573 constant_alignment_word_strings.
10574 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10575 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10576 constant_alignment_word_strings.
10577 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10578 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10579 (cris_constant_alignment): New function.
10580 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10581 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10582 (epiphany_constant_alignment): New function.
10583 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10584 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10585 constant_alignment_word_strings.
10586 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10587 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10588 constant_alignment_word_strings.
10589 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10590 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10591 constant_alignment_word_strings.
10592 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10593 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10594 * config/i386/i386.c (ix86_constant_alignment): Make static.
10595 Use the same interface as the target hook.
10596 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10597 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10598 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10599 constant_alignment_word_strings.
10600 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10601 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10602 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10603 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10604 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10605 constant_alignment_word_strings.
10606 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10607 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10608 constant_alignment_word_strings.
10609 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10610 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10611 constant_alignment_word_strings.
10612 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10613 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10615 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10616 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10617 * config/mips/mips.c (mips_constant_alignment): New function.
10618 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10619 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10620 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10621 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10622 (mmix_constant_alignment): Make static. Use the same interface
10623 as the target hook.
10624 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10625 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10626 constant_alignment_word_strings.
10627 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10628 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10629 constant_alignment_word_strings.
10630 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10631 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10632 constant_alignment_word_strings.
10633 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10634 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10635 (rs6000_constant_alignment): New function.
10636 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10637 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10638 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10639 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10640 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10641 (rs6000_constant_alignment): New function.
10642 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10643 * config/s390/s390.c (s390_constant_alignment): New function.
10644 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10645 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10646 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10647 constant_alignment_word_strings.
10648 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10649 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10650 (sparc_constant_alignment): New function.
10651 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10652 * config/spu/spu.c (spu_constant_alignment): New function.
10653 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10654 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10655 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10656 constant_alignment_word_strings.
10657 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10658 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10659 constant_alignment_word_strings.
10660 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10661 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10662 constant_alignment_word_strings.
10663 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10664 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10665 (visium_constant_alignment): New function.
10666 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10667 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10668 (xtensa_constant_alignment): New function.
10669 * system.h (CONSTANT_ALIGNMENT): Poison.
10671 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10673 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10674 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10675 (rs6000_builtin_valid_without_lhs): New helper function.
10676 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10677 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10679 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10681 * target.h (vec_perm_indices): Use unsigned short rather than
10683 (auto_vec_perm_indices): Likewise.
10684 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10685 Use unsigned int rather than unsigned char.
10686 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10688 2017-09-25 Richard Biener <rguenther@suse.de>
10690 * cfgloop.h (sort_sibling_loops): Declare.
10691 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10692 (sort_sibling_loops): New function sorting the sibling loop list
10694 * graphite.c (graphite_transform_loops): Sort sibling loops.
10696 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10698 * target.def (vec_perm_const_ok): Change sel parameter to
10700 * optabs-query.c (can_vec_perm_p): Update accordingly.
10701 * doc/tm.texi: Regenerate.
10702 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10703 auto_vec_perm_indices and remove separate nelt field.
10704 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10705 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10706 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10707 (aarch64_expand_vec_perm_const): Update accordingly.
10708 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10709 to vec_perm_indices.
10710 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10711 auto_vec_perm_indices and remove separate nelt field.
10712 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10713 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10714 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10716 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10717 to vec_perm_indices.
10718 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10719 sel to vec_perm_indices.
10720 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10721 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10722 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10724 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10727 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10730 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10731 on the FUNCTION_DECL function context if it has a DIE that is a
10734 2017-09-25 Richard Biener <rguenther@suse.de>
10736 PR tree-optimization/82285
10737 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10740 2017-09-25 Tom de Vries <tom@codesourcery.com>
10744 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10747 2017-09-25 Richard Biener <rguenther@suse.de>
10749 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10750 ISL errors other than isl_error_quota happen. Dump if the
10751 schedule is the same.
10752 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10753 errors instead of aborting inside ISL.
10755 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10758 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10759 of libgcc_eh for m64.
10760 * config/i386/darwin64.h: Likewise.
10762 2017-09-25 Richard Biener <rguenther@suse.de>
10764 PR middle-end/82144
10765 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10766 attribute for incomplete types nor twice for complete ones.
10768 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10771 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10772 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10775 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10778 * config/i386/i386.opt (mprefer-avx256): Use
10779 ix86_target_flags variable.
10780 * config/i386/i386.c (ix86_target_string): Move
10781 -mprefer-avx256 to flag2_opts.
10783 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10785 PR middle-end/35691
10786 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10787 and x != -1 | y != -1 into (x & y) != -1.
10789 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10791 * config.gcc: Add new case statement to set
10792 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10793 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10794 s390x-*-linux* case statements. Added aarch64 to the list of
10795 supported architectures.
10797 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10799 PR tree-optimization/82289
10800 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10801 STMT_VINFO_RELEVANT_P.
10803 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10804 Alan Hayward <alan.hayward@arm.com>
10805 David Sherwood <david.sherwood@arm.com>
10807 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10808 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10809 (extract_range_from_binary_expr_1): Don't call
10810 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10812 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10813 Alan Hayward <alan.hayward@arm.com>
10814 David Sherwood <david.sherwood@arm.com>
10816 * target.def (preferred_vector_alignment): New hook.
10817 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
10819 * doc/tm.texi: Regenerate.
10820 * targhooks.h (default_preferred_vector_alignment): Declare.
10821 * targhooks.c (default_preferred_vector_alignment): New function.
10822 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
10824 (DR_TARGET_ALIGNMENT): New macro.
10825 (aligned_access_p): Update commentary.
10826 (vect_known_alignment_in_bytes): New function.
10827 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
10829 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
10830 Calculate the misalignment based on the target alignment rather than
10832 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
10833 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
10834 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
10835 the target alignment, rather than masking the element misalignment
10836 with the number of elements in a vector. Also use the target
10837 alignment when calculating the maximum number of peels.
10838 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
10839 instead of TYPE_ALIGN_UNIT.
10840 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
10841 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
10842 (vect_create_addr_base_for_vector_ref): Update call accordingly.
10843 (vect_create_data_ref_ptr): Likewise.
10844 (vect_setup_realignment): Realign by ANDing with
10845 -DR_TARGET_MISALIGNMENT.
10846 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
10847 the number of peels based on DR_TARGET_ALIGNMENT.
10848 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
10849 with the guaranteed alignment boundary when deciding whether
10851 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
10852 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
10853 (ensure_base_align): Remove stmt_info parameter. Get the
10854 target base alignment from DR_TARGET_ALIGNMENT.
10855 (vectorizable_store): Update call accordingly. Interpret
10856 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
10858 (vectorizable_load): Likewise.
10860 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10861 Alan Hayward <alan.hayward@arm.com>
10862 David Sherwood <david.sherwood@arm.com>
10864 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
10865 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
10866 (vect_enhance_data_refs_alignment): Likewise.
10868 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10870 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10871 error. Only quit immediately if parsing is complete.
10872 (BEGIN): Initialize fatal_err and parse_done.
10873 (begin fpu, end fpu): Check number of arguments.
10874 (begin arch, end arch): Likewise.
10875 (begin cpu, end cpu): Likewise.
10876 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10877 (optalias): Likewise.
10879 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10881 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10882 * config/arm/arm-isa.h: Delete. Move definitions to ...
10883 * arm-cpus.in: ... here. Use new feature and fgroup values.
10884 * config/arm/arm.c (arm_option_override): Use lower case for feature
10886 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10887 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10888 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10890 (print_isa_bits_for): New function.
10891 (gen_isa): New function.
10892 (gen_comm_data): Use print_isa_bits_for.
10893 (define feature): New keyword.
10894 (define fgroup): New keyword.
10895 * config/arm/t-arm (TM_H): Remove.
10896 (GTM_H): Add arm-isa.h.
10897 (arm-isa.h): Add rule to generate file.
10898 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10899 case for feature bit names.
10901 2017-09-22 Richard Biener <rguenther@suse.de>
10903 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
10905 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
10906 print of no dependency loops ...
10907 * graphite.c (graphite_transform_loops): ... here.
10908 (canonicalize_loop_closed_ssa_form): Work from inner to outer
10910 (same_close_phi_node, remove_duplicate_close_phi,
10911 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
10912 (canonicalize_loop_closed_ssa): ... here and simplify.
10913 * graphite-optimize-isl.c: Include tree-vectorizer.h.
10914 (optimize_isl): Use dump_printf_loc to tell when we stopped
10915 optimizing because of an ISL timeout.
10917 2017-09-22 Richard Biener <rguenther@suse.de>
10919 PR tree-optimization/82291
10920 * tree-if-conv.c (predicate_mem_writes): Make sure to
10921 remove writes in blocks predicated with false.
10923 2017-09-22 Richard Biener <rguenther@suse.de>
10925 * sese.c: Include cfganal.h.
10926 (if_region_set_false_region): Remove.
10927 (create_if_region_on_edge): Likewise.
10928 (move_sese_in_condition): Re-implement without destroying
10931 2017-09-22 Richard Biener <rguenther@suse.de>
10933 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
10934 Verify both BBs contain loop PHI nodes before dispatching to
10935 copy_loop_phi_args.
10936 (graphite_regenerate_ast_isl): Do not recompute dominators,
10937 do not verify three times. Restructure for clarity.
10938 * graphite-scop-detection.c (same_close_phi_node,
10939 remove_duplicate_close_phi, make_close_phi_nodes_unique,
10940 defined_in_loop_p, canonicalize_loop_closed_ssa,
10941 canonicalize_loop_closed_ssa_form): Simplify, remove excess
10942 checking and SSA rewrite, move to ...
10943 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
10944 (graphite_initialize): Do not pass in ctx, do not reset the
10945 SCEV cache, compute only dominators.
10946 (graphite_transform_loops): Allocate ISL ctx after
10947 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
10948 Maintain post-dominators only around build_scops.
10949 * sese.c (if_region_set_false_region): Make static. Free
10950 and recompute dominators.
10951 (move_sese_in_condition): Assert we don't get called with
10952 post-dominators computed.
10953 * sese.h (if_region_set_false_region): Remove.
10955 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
10957 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
10958 mode attribute for TARGET_AVX512VL.
10960 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
10962 * config/i386/i386.opt (mprefer-avx256): New option.
10963 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
10965 (ix86_preferred_simd_mode): Return 256-bit AVX modes
10966 for TARGET_PREFER_AVX256.
10967 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
10969 2017-09-21 Jeff Law <law@redhat.com>
10971 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
10972 Fix dump output if the only stack space is for pushed registers.
10974 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10976 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
10979 2017-09-21 Martin Sebor <msebor@redhat.com>
10982 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
10983 code (in C++) or code that triggers warnings.
10985 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
10987 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
10989 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
10991 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
10992 * sched-rgn.c: Ditto.
10993 * sel-sched-ir.c: Ditto.
10995 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
10997 * toplev.h (set_random_seed): Adjust return type.
10998 * toplev.c (init_local_tick): Move eager initialization of random_seed
10999 to get_random_seed. Adjust comment.
11000 (init_random_seed): Inline to get_random_seed, delete.
11001 (get_random_seed): Initialize random_seed lazily.
11002 (set_random_seed): Do not return previous value.
11003 (print_switch_value): Do not call get_random_seed.
11005 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11007 * cgraph.c (delete_function_version): New, broken out from...
11008 (cgraph_node::delete_function_version): ...here. Rename to
11009 cgraph_node::delete_function_version_by_decl. Update all uses.
11010 (cgraph_node::remove): Call delete_function_version.
11012 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11015 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11016 VAR_DECLs to which addressable non-volatile parameters are mapped
11017 and for id->retvar after the return value assignment. Clear
11018 id->retval and id->retbnd after inlining.
11020 2017-09-21 Richard Biener <rguenther@suse.de>
11022 PR tree-optimization/82276
11023 PR tree-optimization/82244
11024 * tree-vrp.c (build_assert_expr_for): Set
11025 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11027 (remove_range_assertions): Revert earlier change.
11029 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11032 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11034 2017-09-21 Richard Biener <rguenther@suse.de>
11036 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11037 Restore valid IL after code generation errors.
11038 * graphite.c (graphite_transform_loops): Diagnose code
11039 generation issues as MSG_MISSED_OPTIMIZATION and continue
11040 with processing SCOPs.
11042 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11043 Alan Hayward <alan.hayward@arm.com>
11044 David Sherwood <david.sherwood@arm.com>
11046 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11047 rather than choosing between plus_constant and gen_rtx_<CODE>.
11048 * expr.c (emit_push_insn): Likewise.
11049 (expand_expr_real_2): Likewise.
11051 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11052 Alan Hayward <alan.hayward@arm.com>
11053 David Sherwood <david.sherwood@arm.com>
11055 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11057 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11058 Alan Hayward <alan.hayward@arm.com>
11059 David Sherwood <david.sherwood@arm.com>
11061 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11062 calling tree_to_uhwi.
11064 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11065 Alan Hayward <alan.hayward@arm.com>
11066 David Sherwood <david.sherwood@arm.com>
11068 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11069 INTEGER_CST rather than a negative test for ADDR_EXPR.
11071 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11072 Alan Hayward <alan.hayward@arm.com>
11073 David Sherwood <david.sherwood@arm.com>
11075 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11076 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11078 2017-09-21 Richard Biener <rguenther@suse.de>
11080 PR tree-optimization/71351
11081 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11082 graphite_create_new_loop_guard): Remove, fold remaining parts
11084 (translate_isl_ast_node_for): ... here and simplify.
11086 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11089 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11090 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11091 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11092 alternative always use QI mode, for -Os imov (=R,R) alternative
11093 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11096 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11097 Jeff Law <law@redhat.com>
11099 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11100 (allocate_stack_space): New function, partially extracted from
11101 s390_emit_prologue.
11102 (s390_emit_prologue): Track offset to most recent stack probe.
11103 Code to allocate space moved into allocate_stack_space.
11104 Dump actions when no stack is allocated.
11105 (s390_prologue_plus_offset): New function.
11106 (s390_emit_stack_probe): Likewise.
11108 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11110 * common.opt (Wa, Wl, Wp, g, gz=): Add
11112 (gno-column-info): Remove.
11113 (gcolumn-info): Drop RejectNegative.
11114 (gno-): New prefix.
11115 (gno-record-gcc-switches): Remove.
11116 (grecord-gcc-switches): Drop RejectNegative.
11117 (gno-split-dwarf): Remove.
11118 (gsplit-dwarf): Drop RejectNegative.
11119 (gno-strict-dwarf): Remove.
11120 (gstrict-dwarf): Drop RejectNegative.
11121 * config/darwin.opt (gfull, gused): Add RejectNegative.
11122 * dwarf2out.c (gen_producer_string): Drop
11123 gno-record-gcc-switches handler.
11124 * optc-gen.awk: Add g to prefixes with negative forms.
11125 * opts-common.c (remapping_prefix_p): New.
11126 (find_opt): Check it.
11127 (generate_canonical_option): Test g prefix.
11128 (option_map): Add -gno- mapping.
11129 (add_misspelling_candidates): Check remapping_prefix_p.
11131 2017-09-20 Jeff Law <law@redhat.com>
11133 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11134 thinko in stack clash protection support.
11136 * explow.c (compute_stack_clash_protection_loop_data): Use
11137 CONST_INT_P instead of explicit test. Verify object is a
11138 CONST_INT_P before looking at INTVAL.
11139 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11140 instead of explicit test.
11142 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11145 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11146 address instead of to r1 and r11.
11148 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11150 * config.gcc: Support "knm".
11151 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11152 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11154 * config/i386/i386.c (m_KNM): Define.
11155 (processor_target_table): Add "knm".
11157 (ix86_option_override_internal): Add "knm".
11158 (ix86_issue_rate): Add PROCESSOR_KNM.
11159 (ix86_adjust_cost): Ditto.
11160 (ia32_multipass_dfa_lookahead): Ditto.
11161 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11162 (fold_builtin_cpu): Add M_INTEL_KNM.
11163 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11164 (processor_type): Add PROCESSOR_KNM.
11165 * config/i386/x86-tune.def: Add m_KNM.
11166 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11168 2017-09-20 Richard Biener <rguenther@suse.de>
11170 PR tree-optimization/80213
11171 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11172 are allowed in empty BBs as well.
11173 (canonicalize_loop_closed_ssa): Also look for other complex
11175 (scop_detection::get_sese): Include the loop-closed PHI block
11177 (scop_detection::merge_sese): Remove code adding extra blocks.
11178 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11179 (build_scops): Assert the final returned scop is invalid.
11181 2017-09-20 Richard Biener <rguenther@suse.de>
11183 PR tree-optimization/82264
11184 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11186 (vn_phi_lookup): Likewise.
11187 (vn_phi_insert): Likewise.
11189 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11191 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11192 that fits into uhwi or shwi, add DW_AT_const_value regardless
11193 of early_dwarf without going through RTL, using add_AT_unsigned
11196 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11197 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11198 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11199 (DEBUG_MACRO_SECTION): Likewise.
11200 (DEBUG_LTO_MACRO_SECTION): Likewise.
11201 (DEBUG_STR_DWO_SECTION): Likewise.
11202 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11203 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11204 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11205 (DEBUG_LTO_LINE_STR_SECTION): Define.
11206 (init_sections_and_labels): Initialize debug_line_str_section
11207 variable. Initialize debug_loc_section for -gdwarf-5 to
11208 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11210 2017-09-20 Richard Biener <rguenther@suse.de>
11212 * graphite-sese-to-poly.c (extract_affine): Properly handle
11213 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11215 2017-09-20 Richard Biener <rguenther@suse.de>
11217 PR tree-optimization/81373
11218 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11219 Force SESE live-out defs to be handled even if they are
11222 2017-09-19 Jeff Law <law@redhat.com>
11224 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11225 nothing for stack adjustments with REG_STACK_CHECK.
11226 * sched-deps.c (parse_add_or_inc): Reject insns with
11227 REG_STACK_CHECK from dependency breaking.
11228 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11229 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11230 * reg-notes.def (STACK_CHECK): New note.
11232 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11233 (ix86_expand_prologue): Dump stack clash info as needed.
11234 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11236 * function.c (dump_stack_clash_frame_info): New function.
11237 * function.h (dump_stack_clash_frame_info): Prototype.
11238 (enum stack_clash_probes): New enum.
11240 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11241 flag_stack_clash_protection.
11242 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11243 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11244 (arm_frame_pointer_required): Likewise.
11245 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11246 (ia64_expand_prologue): Likewise.
11247 * config/mips/mips.c (mips_expand_prologue): Likewise.
11248 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11249 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11250 (sparc_flat_expand_prologue): Likewise.
11251 * config/spu/spu.c (spu_expand_prologue): Likewise.
11253 * explow.c: Include "params.h".
11254 (anti_adjust_stack_and_probe_stack_clash): New function.
11255 (get_stack_check_protect): Likewise.
11256 (compute_stack_clash_protection_loop_data): Likewise.
11257 (emit_stack_clash_protection_loop_start): Likewise.
11258 (emit_stack_clash_protection_loop_end): Likewise.
11259 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11260 Use anti_adjust_stack_and_probe_stack_clash.
11261 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11262 (emit_stack_clash_protection_loop_start): Likewise.
11263 (emit_stack_clash_protection_loop_end): Likewise.
11264 * rtl.h (get_stack_check_protect): Prototype.
11265 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11266 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11267 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11269 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11271 * doc/tm.texi: Rebuilt.
11272 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11273 get_stack_check_protect.
11274 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11275 * config/arm/arm.c (arm_expand_prologue): Likewise.
11276 (arm_frame_pointer_required): Likewise.
11277 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11278 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11279 * config/mips/mips.c (mips_expand_prologue): Likewise.
11280 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11281 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11282 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11283 (sparc_flat_expand_prologue): Likewise.
11285 * common.opt (-fstack-clash-protection): New option.
11286 * flag-types.h (enum stack_check_type): Note difference between
11287 -fstack-check= and -fstack-clash-protection.
11288 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11289 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11290 * toplev.c (process_options): Issue warnings/errors for cases
11291 not handled with -fstack-clash-protection.
11292 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11293 (-fstack-check): Note additional problem with -fstack-check=generic.
11294 Note that -fstack-check is primarily for Ada and refer users
11295 to -fstack-clash-protection for stack-clash-protection.
11296 Document new params for stack clash protection.
11298 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11300 * config/i386/i386.c (ix86_split_long_move): Do not handle
11301 address used for LEA in a special way.
11303 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11305 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11307 2017-09-19 Martin Sebor <msebor@redhat.com>
11310 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11311 of incompatible types.
11313 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11315 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11316 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11317 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11318 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11320 2017-09-19 Richard Biener <rguenther@suse.de>
11322 PR tree-optimization/82244
11323 * tree-vrp.c (remove_range_assertions): Do not propagate
11324 a constant to abnormals but replace the assert with a copy.
11326 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11328 PR rtl-optimization/57878
11329 PR rtl-optimization/68988
11330 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11331 avoidance test involving non_reload_pseudos. Move frequency test
11332 below the general fragmentation avoidance test.
11334 2017-09-19 Richard Biener <rguenther@suse.de>
11336 PR tree-optimization/69728
11337 * graphite-sese-to-poly.c (schedule_error): New global.
11338 (add_loop_schedule): Handle empty domain by failing the
11340 (build_original_schedule): Handle schedule_error.
11342 2017-09-19 Richard Biener <rguenther@suse.de>
11344 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11345 Do not iterate to sibling loops but only to siblings of inner
11348 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11351 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11352 that the registers are different.
11354 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11356 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11357 to processor_model and "amdfam17h" to arch_names_table.
11358 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11360 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11363 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11365 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11366 Alan Hayward <alan.hayward@arm.com>
11367 David Sherwood <david.sherwood@arm.com>
11369 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11371 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11373 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11374 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11375 vect_schedule_slp_instance.
11376 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11377 Update call to vect_slp_analyze_node_operations. Simplify return
11379 (vect_slp_analyze_bb_1): Update call accordingly.
11380 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11381 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11382 (vect_schedule_slp): Update call accordingly.
11384 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11385 Alan Hayward <alan.hayward@arm.com>
11386 David Sherwood <david.sherwood@arm.com>
11388 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11389 with types that aren't in fact scalar.
11391 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11393 * tree-vect-slp.c (vect_record_max_nunits): New function,
11395 (vect_build_slp_tree_1): ...here.
11396 (vect_build_slp_tree_2): Call it for phis too.
11398 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11400 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11401 to vect_get_vec_def_for_operand when getting the mask operand.
11403 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11404 Alan Hayward <alan.hayward@arm.com>
11405 David Sherwood <david.sherwood@arm.com>
11407 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11410 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11411 Alan Hayward <alan.hayward@arm.com>
11412 David Sherwood <david.sherwood@arm.com>
11414 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11415 calculation for vector booleans.
11417 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11418 Alan Hayward <alan.hayward@arm.com>
11419 David Sherwood <david.sherwood@arm.com>
11421 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11423 (vect_transform_stmt): ...here.
11424 (vect_analyze_stmt): Use it instead of calling
11425 vectorizable_live_operation directly.
11427 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11429 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11430 non-SIMT targets in acc vector loops.
11432 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11434 * configure.ac: Add arc and check if assembler supports gdwarf2.
11435 * configure: Regenerate.
11437 2017-09-18 Richard Biener <rguenther@suse.de>
11439 PR tree-optimization/82220
11440 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11441 epilogue niters from the min_profitable_iters compute.
11443 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11446 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11447 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11448 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11450 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11453 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11454 switching to a new text section.
11456 2017-09-18 Richard Biener <rguenther@suse.de>
11458 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11460 (build_alias_set): Reject aliases with no access function.
11462 2017-09-18 Richard Biener <rguenther@suse.de>
11464 PR tree-optimization/79622
11465 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11467 (build_cross_bb_scalars_use): Likewise.
11469 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11471 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11473 2017-09-18 Alan Modra <amodra@gmail.com>
11476 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11477 stack_pointer_rtx for count 0. Update comments. Break up
11478 large rtl expression.
11480 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11482 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11483 Increase to 20 bytes.
11484 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11485 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11486 or avx version of the stub.
11488 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11491 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11492 compute the minimum stack alignment. Also update preferred stack
11493 boundary for leaf functions.
11495 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11497 PR tree-optimization/82228
11498 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11501 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11503 * common/config/nds32/nds32-common.c
11504 (nds32_option_optimization_table): Refine formatting.
11505 (nds32_option_optimization_table): Use -fsched-pressure and
11506 -fomit-frame-pointer for specific optimization level.
11508 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11510 * config/nds32/nds32.c: Refine formatting and comments.
11511 * config/nds32/nds32.h: Likewise.
11512 * config/nds32/nds32.md: Likewise.
11513 * config/nds32/nds32-cost.c: Likewise.
11514 * config/nds32/nds32-isr.c: Likewise.
11515 * config/nds32/nds32-md-auxiliary.c: Likewise.
11516 * config/nds32/nds32-multiple.md: Likewise.
11517 * config/nds32/nds32-predicates.c: Likewise.
11519 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11520 Jakub Jelinek <jakub@redhat.com>
11522 Add support for -std=c++2a.
11523 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11525 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11527 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11530 * doc/extend.texi (Common Function Attributes): Add
11531 references to ARM, AArch64, and S/390 specific attributes.
11532 (Function Specific Option Pragmas): Add AArch64 and S/390
11533 to list of back ends that support the target pragma.
11535 2017-09-15 Nathan Sidwell <nathan@acm.org>
11537 * doc/standards.texi: Fix C++17 description. Update URLs for
11540 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11542 * common.opt (Wcast-align=strict): New warning option.
11543 * doc/invoke.texi: Document -Wcast-align=strict.
11545 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11547 * cgraph.h (cgraph_thunk_info): Add comments.
11548 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11549 assert for VIRTUAL_* arguments stricter.
11551 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11553 PR tree-optimization/71026
11554 * match.pd: Move RDIV patterns from fold-const.c
11555 * fold-const.c (distribute_real_division): Removed.
11556 (fold_binary_loc): Remove calls to distribute_real_divison.
11558 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11560 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11561 c++1z and gnu++1z as deprecated. Change other references to
11562 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11563 Change -Wc++1z-compat to -Wc++17-compat.
11564 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11565 * dwarf2out.c (highest_c_language): Handle C++17.
11566 (gen_compile_unit_die): Likewise.
11568 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11570 PR rtl-optimization/82192
11571 * combine.c (make_extraction): Don't look through non-paradoxical
11572 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11575 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11576 Alan Hayward <alan.hayward@arm.com>
11577 David Sherwood <david.sherwood@arm.com>
11579 * target.def (function_arg_offset): New hook.
11580 * targhooks.h (default_function_arg_offset): Declare.
11581 * targhooks.c (default_function_arg_offset): New function.
11582 * function.c (locate_and_pad_parm): Use
11583 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11584 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11585 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11586 * doc/tm.texi: Regenerate.
11587 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11588 * config/spu/spu.c (spu_function_arg_offset): New function.
11589 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11590 * system.h (FUNCTION_ARG_OFFSET): Poison.
11592 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11593 Alan Hayard <alan.hayward@arm.com>
11594 David Sherwood <david.sherwood@arm.com>
11596 * target.def (truly_noop_truncation): New hook.
11597 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11598 than TRULY_NOOP_TRUNCATION.
11599 * hooks.h (hook_bool_uint_uint_true): Declare.
11600 * hooks.c (hook_bool_uint_uint_true): New function.
11601 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11602 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11603 * doc/tm.texi: Regenerate.
11604 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11605 rather than TRULY_NOOP_TRUNCATION in comments.
11606 (simplify_comparison): Likewise.
11607 (record_truncated_value): Likewise.
11608 * expmed.c (extract_bit_field_1): Likewise.
11609 (extract_split_bit_field): Likewise.
11610 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11611 instead of TRULY_NOOP_TRUNCATION.
11612 * function.c (assign_parm_setup_block): Likewise.
11613 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11614 * rtlhooks.c: Include target.h.
11615 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11616 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11617 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11618 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11619 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11620 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11621 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11622 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11623 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11624 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11625 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11626 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11627 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11628 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11629 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11630 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11631 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11632 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11633 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11634 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11635 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11636 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11637 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11638 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11639 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11640 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11641 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11642 rather than TRULY_NOOP_TRUNCATION in comments.
11643 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11644 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11645 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11646 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11647 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11648 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11649 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11650 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11651 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11652 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11653 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11654 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11655 rather than TRULY_NOOP_TRUNCATION in comments.
11656 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11657 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11658 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11659 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11660 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11661 TRULY_NOOP_TRUNCATION condition.
11662 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11663 (TRULY_NOOP_TRUNCATION): Delete.
11664 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11665 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11666 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11667 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11668 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11669 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11670 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11671 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11672 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11673 rather than TRULY_NOOP_TRUNCATION in comments.
11674 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11675 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11676 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11677 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11678 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11679 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11681 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11684 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11685 (*cmp_ior): Likewise.
11686 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11687 (*ior_scc_scc_cmp): Likewise.
11688 (*and_scc_scc): Likewise.
11689 (*and_scc_scc_cmp): Likewise.
11691 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11692 Alan Hayard <alan.hayward@arm.com>
11693 David Sherwood <david.sherwood@arm.com>
11695 * target.def (can_change_mode_class): New hook.
11696 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11697 (hard_regno_nregs): Likewise.
11698 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11699 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11700 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11701 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11702 (LOAD_EXTEND_OP): Update accordingly.
11703 * doc/tm.texi: Regenerate.
11704 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11705 CANNOT_CHANGE_MODE_CLASS.
11706 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11707 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11708 * combine.c (simplify_set): Update accordingly.
11709 * emit-rtl.c (validate_subreg): Likewise.
11710 * recog.c (general_operand): Likewise.
11711 * regcprop.c (mode_change_ok): Likewise.
11712 * reload1.c (choose_reload_regs): Likewise.
11713 (inherit_piecemeal_p): Likewise.
11714 * rtlanal.c (simplify_subreg_regno): Likewise.
11715 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11716 instead of CANNOT_CHANGE_MODE_CLASS.
11717 (reload_cse_simplify_operands): Likewise.
11718 * reload.c (push_reload): Use targetm.can_change_mode_class
11719 instead of CANNOT_CHANGE_MODE_CLASS.
11720 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11721 REG_CANNOT_CHANGE_MODE_P.
11722 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11723 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11724 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11725 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11726 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11727 (arm_can_change_mode_class): New function.
11728 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11729 than CANNOT_CHANGE_MODE_CLASS in comments.
11730 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11731 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11732 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11733 (ix86_can_change_mode_class): ...this new function, inverting the
11734 sense of the return value.
11735 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11736 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11737 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11738 (ia64_can_change_mode_class): New function.
11739 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11740 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11741 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11742 (m32c_can_change_mode_class): ...this new function, inverting the
11743 sense of the return value.
11744 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11745 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11746 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11747 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11748 (mips_can_change_mode_class): ...this new function, inverting the
11749 sense of the return value.
11750 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11751 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11752 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11753 (msp430_can_change_mode_class): New function.
11754 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11755 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11756 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11757 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11758 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11759 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11760 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11761 (pa_cannot_change_mode_class): Replace with...
11762 (pa_can_change_mode_class): ...this new function, inverting the
11763 sense of the return value.
11764 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11765 than CANNOT_CHANGE_MODE_CLASS in comments.
11766 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11767 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11768 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11769 (pdp11_cannot_change_mode_class): Replace with...
11770 (pdp11_can_change_mode_class): ...this new function, inverting the
11771 sense of the return value.
11772 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11773 * config/powerpcspe/powerpcspe-protos.h
11774 (rs6000_cannot_change_mode_class_ptr): Delete.
11775 * config/powerpcspe/powerpcspe.c
11776 (rs6000_cannot_change_mode_class_ptr): Delete.
11777 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11778 (rs6000_option_override_internal): Assign to
11779 targetm.can_change_mode_class instead of
11780 rs6000_cannot_change_mode_class_ptr.
11781 (rs6000_cannot_change_mode_class): Replace with...
11782 (rs6000_can_change_mode_class): ...this new function, inverting the
11783 sense of the return value.
11784 (rs6000_debug_cannot_change_mode_class): Replace with...
11785 (rs6000_debug_can_change_mode_class): ...this new function.
11786 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11787 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11788 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11789 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11790 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11792 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11793 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11794 (rs6000_option_override_internal): Assign to
11795 targetm.can_change_mode_class instead of
11796 rs6000_cannot_change_mode_class_ptr.
11797 (rs6000_cannot_change_mode_class): Replace with...
11798 (rs6000_can_change_mode_class): ...this new function, inverting the
11799 sense of the return value.
11800 (rs6000_debug_cannot_change_mode_class): Replace with...
11801 (rs6000_debug_can_change_mode_class): ...this new function.
11802 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11803 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11804 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11805 (s390_can_change_mode_class): ...this new function, inverting the
11806 sense of the return value.
11807 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11808 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11809 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11810 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11811 (sh_cannot_change_mode_class): Replace with...
11812 (sh_can_change_mode_class): ...this new function, inverting the
11813 sense of the return value.
11814 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11815 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11816 (sparc_can_change_mode_class): New function.
11817 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11818 * config/spu/spu.c (spu_can_change_mode_class): New function.
11819 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11820 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11821 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11822 (visium_can_change_mode_class): New function.
11823 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
11825 2017-09-15 Richard Biener <rguenther@suse.de>
11827 PR tree-optimization/82217
11828 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
11829 but not undefined case.
11831 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11834 * postreload.c (reload_cse_simplify_operands): Skip
11835 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
11837 2017-09-15 Richard Biener <rguenther@suse.de>
11839 PR tree-optimization/68823
11840 * graphite-scop-detection.c (build_alias_set): If we have a
11841 possible dependence check whether we can handle them by just
11842 looking at the DRs DR_ACCESS_FNs.
11843 (build_scops): If build_alias_set fails, fail the SCOP.
11845 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11847 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
11848 to support float128 built-in functions that require the ISA 3.0
11850 (BU_FLOAT128_3_HW): Likewise.
11851 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11852 built-in functions.
11853 (FMAF128): Likewise.
11855 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
11856 support for built-in functions that need the ISA 3.0 IEEE 128-bit
11857 floating point instructions.
11858 (rs6000_invalid_builtin): Likewise.
11859 (rs6000_builtin_mask_names): Likewise.
11860 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
11861 (RS6000_BTM_FLOAT128_HW): Likewise.
11862 (RS6000_BTM_COMMON): Likewise.
11863 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
11865 * doc/extend.texi (RS/6000 built-in functions): Document the
11866 IEEE 128-bit floating point square root and fused multiply-add
11867 built-in functions.
11869 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
11871 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
11872 reg (r2) isn't in the set of registers defined in the prologue.
11874 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11875 Alan Hayward <alan.hayward@arm.com>
11876 David Sherwood <david.sherwood@arm.com>
11878 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
11879 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
11880 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
11881 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
11882 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
11884 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
11885 max_vectorization_factor.
11886 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
11888 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11889 Alan Hayward <alan.hayward@arm.com>
11890 David Sherwood <david.sherwood@arm.com>
11892 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
11893 (vect_worthwhile_without_simd_p): Declare.
11894 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
11895 (vectorizable_reduction): Use it.
11896 * tree-vect-stmts.c (vectorizable_shift): Likewise.
11897 (vectorizable_operation): Likewise.
11899 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11900 Alan Hayward <alan.hayward@arm.com>
11901 David Sherwood <david.sherwood@arm.com>
11903 * tree-vectorizer.h (vect_get_num_copies): New function.
11904 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
11905 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11906 (vectorizable_induction): Likewise.
11907 (vectorizable_live_operation): Likewise.
11908 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11909 (vectorizable_bswap): Likewise.
11910 (vectorizable_call): Likewise.
11911 (vectorizable_conversion): Likewise.
11912 (vectorizable_assignment): Likewise.
11913 (vectorizable_shift): Likewise.
11914 (vectorizable_operation): Likewise.
11915 (vectorizable_store): Likewise.
11916 (vectorizable_load): Likewise.
11917 (vectorizable_condition): Likewise.
11918 (vectorizable_comparison): Likewise.
11919 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
11921 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11922 Alan Hayward <alan.hayward@arm.com>
11923 David Sherwood <david.sherwood@arm.com>
11925 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
11926 of vect_init_vector.
11928 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11929 Alan Hayward <alan.hayward@arm.com>
11930 David Sherwood <david.sherwood@arm.com>
11932 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
11933 an inline wrapper that provides a location.
11934 (gimple_build_vector): Likewise.
11935 * gimple-fold.c (gimple_build_vector_from_val): New function.
11936 (gimple_build_vector): Likewise.
11937 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
11938 functions to build the initial value. Always return a gimple value.
11939 (get_initial_defs_for_reduction): Likewise. Only compute
11941 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
11942 vect_init_vector on the results from get_initial_def(s)_for_reduction.
11943 (vectorizable_induction): Use gimple_build_vector rather than
11946 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11947 Alan Hayward <alan.hayward@arm.com>
11948 David Sherwood <david.sherwood@arm.com>
11950 * target.h (vec_perm_indices): New typedef.
11951 (auto_vec_perm_indices): Likewise.
11952 * optabs-query.h: Include target.h
11953 (can_vec_perm_p): Take a vec_perm_indices *.
11954 * optabs-query.c (can_vec_perm_p): Likewise.
11955 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
11956 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11957 * tree-vect-generic.c (lower_vec_perm): Likewise.
11958 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
11959 (vect_grouped_load_supported): Likewise.
11960 (vect_shift_permute_load_chain): Likewise.
11961 (vect_permute_store_chain): Use auto_vec_perm_indices.
11962 (vect_permute_load_chain): Likewise.
11963 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
11964 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
11965 Update uses of can_vec_perm_p.
11966 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
11967 mode with a number of elements. Take a vec_perm_indices *.
11968 (vect_create_epilog_for_reduction): Update accordingly.
11969 Use auto_vec_perm_indices.
11970 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
11971 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
11972 (vect_transform_slp_perm_load): Likewise.
11973 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
11974 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
11975 (vect_gen_perm_mask_checked): Likewise.
11976 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
11977 (vect_gen_perm_mask_checked): Likewise.
11978 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
11979 (vectorizable_store): Likewise.
11980 (vectorizable_load): Likewise.
11981 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
11982 (vectorizable_bswap): Likewise.
11984 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11985 Alan Hayward <alan.hayward@arm.com>
11986 David Sherwood <david.sherwood@arm.com>
11988 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
11989 * tree.c (build_vector): Likewise.
11990 (build_vector_from_ctor): Update accordingly.
11991 (build_vector_from_val): Likewise.
11992 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11993 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11994 * tree-vect-generic.c (add_rshift): Likewise.
11995 (expand_vector_divmod): Likewise.
11996 (optimize_vector_constructor): Likewise.
11997 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
11998 (vect_transform_slp_perm_load): Likewise.
11999 (vect_schedule_slp_instance): Likewise.
12000 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12001 (vectorizable_call): Likewise.
12002 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12003 * expmed.c (make_tree): Likewise.
12004 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12005 a vector passed to build_vector.
12006 (fold_convert_const): Likewise.
12007 (exact_inverse): Likewise.
12008 (fold_ternary_loc): Likewise.
12009 (fold_relational_const): Likewise.
12010 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12011 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12012 (const_unop): Likewise. Store the reduction accumulator in a
12013 variable rather than an array.
12014 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12015 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12016 the new vector, rather than constructing it after the input arrays.
12017 (native_interpret_vector): Use auto_vec<tree> when building
12018 a vector passed to build_vector. Add elements in order.
12019 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12020 auto_vec<tree> when building a vector passed to build_vector.
12021 (vect_create_epilog_for_reduction): Likewise.
12022 (vectorizable_induction): Likewise.
12023 (get_initial_def_for_reduction): Likewise. Fix indentation of
12025 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12027 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12028 passed to build_vector.
12030 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12031 Alan Hayward <alan.hayward@arm.com>
12032 David Sherwood <david.sherwood@arm.com>
12034 * tree-core.h (tree_base::u): Add an "nelts" field.
12035 (tree_vector): Use VECTOR_CST_NELTS as the length.
12036 * tree.c (tree_size): Likewise.
12037 (make_vector): Initialize VECTOR_CST_NELTS.
12038 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12039 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12040 TYPE_VECTOR_SUBPARTS.
12041 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12042 as the number of units, setting it from VECTOR_CST_NELTS.
12043 (const_vector_from_tree): Likewise.
12044 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12045 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12046 (fold_negate_expr_1): Likewise.
12047 (fold_convert_const): Likewise.
12048 (const_binop): Likewise. Differentiate the number of output and
12050 (const_unop): Likewise.
12051 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12052 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12053 in cases that did the opposite.
12055 2017-09-14 Richard Biener <rguenther@suse.de>
12057 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12060 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12062 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12064 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12067 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12068 if and where to split a bb, except for splitting before debug insn
12069 sequences followed by non-label real insn. Delete debug insns
12070 in between basic blocks.
12072 * combine.c (make_compound_operation_int): Formatting fixes.
12074 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12075 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12076 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12077 * config/sol2.h (LINK_EH_SPEC): Likewise.
12078 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12079 * config/s390/linux.h (LINK_SPEC): Likewise.
12080 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12081 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12082 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12083 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12084 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12085 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12086 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12087 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12089 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12091 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12093 (ENDFILE_LINUX_SPEC): Likewise.
12094 (LINK_EH_SPEC): Likewise.
12095 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12096 (LINK_OS_LINUX_SPEC32): Likewise.
12097 (LINK_OS_LINUX_SPEC64): Likewise.
12098 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12099 (LINK_OS_LINUX_SPEC): Likewise.
12101 2017-09-13 Martin Liska <mliska@suse.cz>
12103 PR middle-end/82154
12104 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12105 CASE_HIGH is NULL_TREE.
12107 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12108 Alan Hayward <alan.hayward@arm.com>
12109 David Sherwood <david.sherwood@arm.com>
12111 * target.def (secondary_memory_needed): New hook.
12112 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12113 instead of SECONDARY_MEMORY_NEEDED.
12114 (secondary_memory_needed_mode): Likewise.
12115 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12116 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12117 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12118 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12119 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12120 * doc/tm.texi: Regenerate.
12121 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12122 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12123 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12124 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12125 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12126 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12127 mode argument first and change the reg_class arguments to reg_class_t.
12128 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12129 Make static. Update the call to inline_secondary_memory_needed.
12130 (ix86_register_move_cost): Update the call to
12131 inline_secondary_memory_needed.
12132 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12133 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12135 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12136 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12138 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12139 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12140 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12141 and match hook interface. Add comment from mips.h.
12142 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12143 * config/mmix/mmix.md (truncdfsf2): Refer to
12144 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12146 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12147 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12148 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12149 (pa_secondary_memory_needed): New function.
12150 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12151 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12152 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12153 (pdp11_secondary_memory_needed): Make static and match hook interface.
12154 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12155 * config/powerpcspe/powerpcspe-protos.h
12156 (rs6000_secondary_memory_needed_ptr): Delete.
12157 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12159 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12160 (rs6000_option_override_internal): Assign to
12161 targetm.secondary_memory_needed rather than
12162 rs6000_secondary_memory_needed_ptr.
12163 (rs6000_secondary_memory_needed): Match hook interface.
12164 (rs6000_debug_secondary_memory_needed): Likewise.
12165 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12166 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12167 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12168 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12169 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12170 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12172 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12173 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12174 (rs6000_option_override_internal): Assign to
12175 targetm.secondary_memory_needed rather than
12176 rs6000_secondary_memory_needed_ptr.
12177 (rs6000_secondary_memory_needed): Match hook interface.
12178 (rs6000_debug_secondary_memory_needed): Likewise.
12179 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12180 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12181 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12182 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12183 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12184 (sparc_secondary_memory_needed): New function.
12185 * lra-constraints.c (check_and_process_move): Refer to
12186 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12188 (curr_insn_transform): Likewise.
12189 (process_alt_operands): Use targetm.secondary_memory_needed
12190 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12191 (check_secondary_memory_needed_p): Likewise.
12192 (choose_split_class): Likewise.
12193 * reload.c: Unconditionally include code that was previously
12194 conditional on SECONDARY_MEMORY_NEEDED.
12195 (push_secondary_reload): Use targetm.secondary_memory_needed
12196 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12197 (push_reload): Likewise.
12198 * reload1.c: Unconditionally include code that was previously
12199 conditional on SECONDARY_MEMORY_NEEDED.
12200 (choose_reload_regs): Use targetm.secondary_memory_needed
12201 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12202 (gen_reload): Likewise.
12203 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12205 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12206 Alan Hayward <alan.hayward@arm.com>
12207 David Sherwood <david.sherwood@arm.com>
12209 * target.def (secondary_memory_needed_mode): New hook:
12210 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12211 * targhooks.h (default_secondary_memory_needed_mode): New function.
12212 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12213 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12214 * doc/tm.texi: Regenerate.
12215 * lra-constraints.c (check_and_process_move): Use
12216 targetm.secondary_memory_needed_mode instead of
12217 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12218 (curr_insn_transform): Likewise.
12219 * reload.c (get_secondary_mem): Likewise.
12220 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12221 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12223 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12224 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12225 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12226 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12227 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12229 * config/powerpcspe/powerpcspe-protos.h
12230 (rs6000_secondary_memory_needed_mode): Delete.
12231 * config/powerpcspe/powerpcspe.c
12232 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12233 (rs6000_secondary_memory_needed_mode): Make static.
12234 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12235 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12237 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12239 (rs6000_secondary_memory_needed_mode): Make static.
12240 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12241 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12242 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12243 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12244 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12246 (sparc_secondary_memory_needed_mode): New function.
12247 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12249 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12251 * config/aarch64/constraints.md (Umq): New constraint.
12252 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12254 (mov<mode>): Update condition.
12256 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12258 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12259 when bitposition is the same.
12261 2017-09-13 Richard Biener <rguenther@suse.de>
12263 * dwarf2out.c (output_die_symbol): Remove.
12264 (output_die): Do not output a DIEs symbol.
12266 2017-09-13 Richard Biener <rguenther@suse.de>
12268 PR middle-end/82128
12269 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12270 default-def to avoid breaking iterator update with the weird
12271 interaction with cgraph_update_edges_for_call_stmt_node.
12273 2017-09-13 Richard Biener <rguenther@suse.de>
12275 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12276 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12277 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12278 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12280 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12282 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12283 Disable pc relative literal load irrespective of
12284 TARGET_FIX_ERR_A53_84341 for default.
12286 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12288 * config/sparc/sparc.c (output_return): Output the source location of
12289 the insn in the delay slot, if any.
12290 (output_sibcall): Likewise.
12292 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12295 * common.opt (-static-pie): New alias.
12296 (shared): Negate static-pie.
12297 (-no-pie): Update help text.
12299 (static-pie): New option.
12300 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12301 -static-pie support.
12302 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12303 (LINK_EH_SPEC): Likewise.
12304 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12305 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12306 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12307 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12308 (init_gcc_specs): Likewise.
12309 (init_spec): Likewise.
12310 (display_help): Update help message for -pie.
12311 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12314 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12316 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12317 (movdi_aarch64): Likewise.
12318 (movti_aarch64): Likewise.
12320 2017-09-12 Simon Wright <simon@pushface.org>
12323 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12324 calculation of the minor version, always output as 0.
12326 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12329 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12330 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12331 on it early, rather than manual conversion late. For
12332 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12333 instead of performing manual conversion.
12335 2017-09-12 Carl Love <cel@us.ibm.com>
12337 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12338 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12340 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12341 VMULOSW): Add definitions.
12342 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12343 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12344 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12345 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12346 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12348 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12350 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12352 (movti_aarch64): Likewise.
12353 (movdf_aarch64): Likewise.
12354 (movtf_aarch64): Likewise.
12355 (load_pairdi): Likewise.
12356 (store_pairdi): Likewise.
12357 (load_pairdf): Likewise.
12358 (store_pairdf): Likewise.
12359 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12360 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12361 (ldr_got_small_<mode>): Likewise.
12362 (ldr_got_small_28k_<mode>): Likewise.
12363 (ldr_got_tiny): Likewise.
12364 * config/aarch64/iterators.md (ldst_sz): New.
12365 (ldpstp_sz): Likewise.
12366 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12368 (thunderx_load): Split load_8 to load_16.
12369 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12371 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12372 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12373 (xgene1_store_pair): Split store_8 to store_16.
12374 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12375 (falkor_st_0_st_sd): Split store_8 to store_16.
12377 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12379 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12380 and store1/2/3/4 to store_4/8/12/16.
12381 * config/aarch64/aarch64.md: Update for rename.
12382 * config/arm/arm.md: Likewise.: Likewise.
12383 * config/arm/arm.c: Likewise.
12384 * config/arm/thumb1.md: Likewise.
12385 * config/arm/thumb2.md: Likewise.
12386 * config/arm/vfp.md: Likewise.
12387 * config/arm/arm-generic.md: Likewise.
12388 * config/arm/arm1020e.md: Likewise.
12389 * config/arm/arm1026ejs.md: Likewise.
12390 * config/arm/arm1136jfs.md: Likewise.
12391 * config/arm/arm926ejs.md: Likewise.
12392 * config/arm/cortex-a15.md: Likewise.
12393 * config/arm/cortex-a17.md: Likewise.
12394 * config/arm/cortex-a5.md: Likewise.
12395 * config/arm/cortex-a53.md: Likewise.
12396 * config/arm/cortex-a57.md: Likewise.
12397 * config/arm/cortex-a7.md: Likewise.
12398 * config/arm/cortex-a8.md: Likewise.
12399 * config/arm/cortex-a9.md: Likewise.
12400 * config/arm/cortex-m4.md: Likewise.
12401 * config/arm/cortex-m7.md: Likewise.
12402 * config/arm/cortex-r4.md: Likewise.
12403 * config/arm/exynos-m1.md: Likewise.
12404 * config/arm/fa526.md: Likewise.
12405 * config/arm/fa606te.md: Likewise.
12406 * config/arm/fa626te.md: Likewise.
12407 * config/arm/fa726te.md: Likewise.
12408 * config/arm/fmp626.md: Likewise.
12409 * config/arm/iwmmxt.md: Likewise.
12410 * config/arm/ldmstm.md: Likewise.
12411 * config/arm/marvell-pj4.md: Likewise.
12412 * config/arm/xgene1.md: Likewise.
12413 * config/aarch64/thunderx.md: Likewise.
12414 * config/aarch64/thunderx2t99.md: Likewise.
12415 * config/aarch64/falkor.md: Likewise.
12417 2017-09-12 Martin Liska <mliska@suse.cz>
12419 * attribs.c (private_lookup_attribute): New function.
12420 * attribs.h (private_lookup_attribute): Declared here.
12421 (lookup_attribute): Called from this place.
12423 2017-09-12 Richard Biener <rguenther@suse.de>
12425 PR tree-optimization/82157
12426 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12427 stmts with side-effects.
12429 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12430 Alan Hayward <alan.hayward@arm.com>
12431 David Sherwood <david.sherwood@arm.com>
12433 * target.def (hard_regno_nregs): New hook.
12434 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12435 * targhooks.h (default_hard_regno_nregs): Declare.
12436 * targhooks.c (default_hard_regno_nregs): New function.
12437 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12438 (TARGET_HARD_REGNO_NREGS): ...this hook.
12439 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12440 (CLASS_MAX_NREGS): Likewise.
12441 * doc/tm.texi: Regenerate.
12442 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12443 instead of HARD_REGNO_NREGS.
12444 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12445 HARD_REGNO_NREGS in the comment.
12446 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12447 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12448 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12449 Return an unsigned int.
12450 (TARGET_HARD_REGNO_NREGS): Redefine.
12451 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12452 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12453 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12454 (arc_hard_regno_nregs): New function.
12455 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12456 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12457 (arm_hard_regno_nregs): New function.
12458 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12459 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12460 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12461 (TARGET_HARD_REGNO_NREGS): Redefine.
12462 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12463 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12464 (HARD_REGNO_NREGS): Delete.
12465 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12466 (cr16_hard_regno_nregs): New function.
12467 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12468 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12469 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12470 (cris_hard_regno_nregs): New function.
12471 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12472 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12473 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12474 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12475 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12476 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12477 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12478 (frv_hard_regno_nregs): Make static. Take and return an
12480 (frv_class_max_nregs): Remove outdated copy of documentation.
12481 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12482 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12483 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12484 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12485 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12486 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12487 (TARGET_HARD_REGNO_NREGS): Redefine.
12488 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12489 (CLASS_MAX_NREGS): Update comment.
12490 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12491 (ia64_hard_regno_nregs): New function.
12492 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12493 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12494 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12495 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12496 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12498 (m32c_hard_regno_nregs): Likewise. Make static.
12499 (TARGET_HARD_REGNO_NREGS): Redefine.
12500 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12501 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12502 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12503 (m68k_hard_regno_nregs): New function.
12504 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12505 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12506 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12507 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12508 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12509 Take and return an unsigned int.
12510 (TARGET_HARD_REGNO_NREGS): Redefine.
12511 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12512 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12513 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12514 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12515 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12516 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12517 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12518 (msp430_hard_regno_nregs): Make static. Take and return an
12520 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12521 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12522 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12523 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12524 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12525 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12526 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12527 (TARGET_HARD_REGNO_NREGS): Redefine.
12528 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12529 (PA_HARD_REGNO_NREGS): ...this.
12530 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12531 (PA_HARD_REGNO_NREGS): ...this.
12532 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12533 (pa_hard_regno_nregs): New function.
12534 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12535 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12536 (pdp11_hard_regno_nregs): New function.
12537 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12538 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12539 (rs6000_hard_regno_nregs_hook): New function.
12540 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12541 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12542 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12543 Take and return an unsigned int. Move earlier in file.
12544 (TARGET_HARD_REGNO_NREGS): Redefine.
12545 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12546 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12547 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12548 (rl78_hard_regno_nregs): Make static. Take and return an
12550 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12551 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12552 (rs6000_hard_regno_nregs_hook): New function.
12553 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12554 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12555 (TARGET_HARD_REGNO_NREGS): Redefine.
12556 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12557 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12558 instead of HARD_REGNO_NREGS.
12559 (s390_hard_regno_nregs): New function.
12560 (s390_hard_regno_mode_ok): Add comment from s390.h.
12561 (TARGET_HARD_REGNO_NREGS): Redefine.
12562 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12563 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12564 (sh_hard_regno_nregs): New function.
12565 (sh_pass_in_reg_p): Use it.
12566 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12567 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12568 (sparc_hard_regno_nregs): New function.
12569 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12570 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12571 (spu_function_arg_advance): Use it, supplying a valid register number.
12572 (TARGET_HARD_REGNO_NREGS): Redefine.
12573 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12574 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12575 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12576 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12577 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12578 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12579 (CLASS_MAX_NREGS): Remove copy of old documentation.
12580 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12581 (visium_hard_regno_nregs): New function.
12582 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12583 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12584 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12585 xtensa_hard_regno_nregs): New function.
12586 * system.h (HARD_REGNO_NREGS): Poison.
12588 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12590 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12591 hard_regno_nregs instead of HARD_REGNO_NREGS.
12592 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12593 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12594 (c6x_expand_epilogue): Likewise.
12595 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12596 (frv_read_iacc_argument): Likewise.
12597 * config/sh/sh.c: Include regs.h.
12598 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12599 (regs_used): Likewise.
12600 (output_stack_adjust): Likewise.
12601 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12602 * expmed.c: Include regs.h.
12603 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12604 * ree.c: Include regs.h.
12605 (combine_reaching_defs): Use hard_regno_nregs instead of
12607 (add_removable_extension): Likewise.
12609 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12611 * regs.h (hard_regno_nregs): Turn into a function.
12612 (end_hard_regno): Update accordingly.
12613 * caller-save.c (setup_save_areas): Likewise.
12614 (save_call_clobbered_regs): Likewise.
12615 (replace_reg_with_saved_mem): Likewise.
12616 (insert_restore): Likewise.
12617 (insert_save): Likewise.
12618 * combine.c (can_change_dest_mode): Likewise.
12619 (move_deaths): Likewise.
12620 (distribute_notes): Likewise.
12621 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12622 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12623 (rs6000_split_multireg_move): Likewise.
12624 (rs6000_register_move_cost): Likewise.
12625 (rs6000_memory_move_cost): Likewise.
12626 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12627 (rs6000_split_multireg_move): Likewise.
12628 (rs6000_register_move_cost): Likewise.
12629 (rs6000_memory_move_cost): Likewise.
12630 * cselib.c (cselib_reset_table): Likewise.
12631 (cselib_lookup_1): Likewise.
12632 * emit-rtl.c (set_mode_and_regno): Likewise.
12633 * function.c (aggregate_value_p): Likewise.
12634 * ira-color.c (setup_profitable_hard_regs): Likewise.
12635 (check_hard_reg_p): Likewise.
12636 (calculate_saved_nregs): Likewise.
12637 (assign_hard_reg): Likewise.
12638 (improve_allocation): Likewise.
12639 (calculate_spill_cost): Likewise.
12640 * ira-emit.c (modify_move_list): Likewise.
12641 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12642 (ira_hard_reg_in_set_p): Likewise.
12643 * ira.c (setup_reg_mode_hard_regset): Likewise.
12644 (clarify_prohibited_class_mode_regs): Likewise.
12645 (check_allocation): Likewise.
12646 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12647 (lra_setup_reg_renumber): Likewise.
12648 (setup_try_hard_regno_pseudos): Likewise.
12649 (spill_for): Likewise.
12650 (assign_hard_regno): Likewise.
12651 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12652 * lra-constraints.c (in_class_p): Likewise.
12653 (lra_constraint_offset): Likewise.
12654 (simplify_operand_subreg): Likewise.
12655 (lra_constraints): Likewise.
12656 (split_reg): Likewise.
12657 (split_if_necessary): Likewise.
12658 (invariant_p): Likewise.
12659 (inherit_in_ebb): Likewise.
12660 * lra-lives.c (process_bb_lives): Likewise.
12661 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12662 (get_hard_regs): Likewise.
12663 (do_remat): Likewise.
12664 * lra-spills.c (assign_spill_hard_regs): Likewise.
12665 * mode-switching.c (create_pre_exit): Likewise.
12666 * postreload.c (reload_combine_recognize_pattern): Likewise.
12667 * recog.c (peep2_find_free_register): Likewise.
12668 * regcprop.c (kill_value_regno): Likewise.
12669 (set_value_regno): Likewise.
12670 (copy_value): Likewise.
12671 (maybe_mode_change): Likewise.
12672 (find_oldest_value_reg): Likewise.
12673 (copyprop_hardreg_forward_1): Likewise.
12674 * regrename.c (check_new_reg_p): Likewise.
12675 (regrename_do_replace): Likewise.
12676 * reload.c (push_reload): Likewise.
12677 (combine_reloads): Likewise.
12678 (find_dummy_reload): Likewise.
12679 (operands_match_p): Likewise.
12680 (find_reloads): Likewise.
12681 (find_equiv_reg): Likewise.
12682 (reload_adjust_reg_for_mode): Likewise.
12683 * reload1.c (count_pseudo): Likewise.
12684 (count_spilled_pseudo): Likewise.
12685 (find_reg): Likewise.
12686 (clear_reload_reg_in_use): Likewise.
12687 (free_for_value_p): Likewise.
12688 (allocate_reload_reg): Likewise.
12689 (choose_reload_regs): Likewise.
12690 (reload_adjust_reg_for_temp): Likewise.
12691 (emit_reload_insns): Likewise.
12692 (delete_output_reload): Likewise.
12693 * rtlanal.c (subreg_get_info): Likewise.
12694 * sched-deps.c (sched_analyze_reg): Likewise.
12695 * sel-sched.c (init_regs_for_mode): Likewise.
12696 (mark_unavailable_hard_regs): Likewise.
12697 (choose_best_reg_1): Likewise.
12698 (verify_target_availability): Likewise.
12699 * valtrack.c (dead_debug_insert_temp): Likewise.
12700 * var-tracking.c (track_loc_p): Likewise.
12701 (emit_note_insn_var_location): Likewise.
12702 * varasm.c (make_decl_rtl): Likewise.
12703 * reginfo.c (choose_hard_reg_mode): Likewise.
12704 (init_reg_modes_target): Refer directly to
12705 this_target_regs->x_hard_regno_nregs.
12707 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12709 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12710 instead of hard_regno_nregs.
12712 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12714 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12715 end_hard_regno instead of hard_regno_nregs.
12716 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12717 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12718 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12719 * ira-color.c (improve_allocation): Likewise.
12720 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12721 * lra-lives.c (mark_regno_live): Likewise.
12722 (mark_regno_dead): Likewise.
12723 * lra-remat.c (operand_to_remat): Likewise.
12724 * lra.c (collect_non_operand_hard_regs): Likewise.
12725 * postreload.c (reload_combine_note_store): Likewise.
12726 (move2add_valid_value_p): Likewise.
12727 * reload.c (regno_clobbered_p): Likewise.
12729 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12731 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12733 * config/v850/v850.c (v850_reorg): Likewise.
12734 * reload.c (refers_to_regno_for_reload_p): Likewise.
12735 (find_equiv_reg): Likewise.
12736 * reload1.c (reload_reg_reaches_end_p): Likewise.
12738 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12740 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12742 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12743 * config/arm/arm.c (output_move_neon): Likewise.
12744 (arm_attr_length_move_neon): Likewise.
12745 (neon_split_vcombine): Likewise.
12746 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12747 (c6x_mark_reg_written): Likewise.
12748 (c6x_dwarf_register_span): Likewise.
12749 * config/i386/i386.c (ix86_save_reg): Likewise.
12750 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12751 (rws_access_reg): Likewise.
12752 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12753 * mode-switching.c (create_pre_exit): Likewise.
12754 * ree.c (combine_reaching_defs): Likewise.
12755 (add_removable_extension): Likewise.
12756 * regcprop.c (find_oldest_value_reg): Likewise.
12757 (copyprop_hardreg_forward_1): Likewise.
12758 * reload.c (reload_inner_reg_of_subreg): Likewise.
12759 (push_reload): Likewise.
12760 (combine_reloads): Likewise.
12761 (find_dummy_reload): Likewise.
12762 (reload_adjust_reg_for_mode): Likewise.
12763 * reload1.c (find_reload_regs): Likewise.
12764 (forget_old_reloads_1): Likewise.
12765 (reload_reg_free_for_value_p): Likewise.
12766 (reload_adjust_reg_for_temp): Likewise.
12767 (emit_reload_insns): Likewise.
12768 (delete_output_reload): Likewise.
12769 * sel-sched.c (choose_best_reg_1): Likewise.
12770 (choose_best_pseudo_reg): Likewise.
12772 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12773 Alan Hayward <alan.hayward@arm.com>
12774 David Sherwood <david.sherwood@arm.com>
12776 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12777 * target.def (slow_unaligned_access): New hook.
12778 * targhooks.h (default_slow_unaligned_access): Declare.
12779 * targhooks.c (default_slow_unaligned_access): New function.
12780 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12781 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12782 * doc/tm.texi: Regenerate.
12783 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12784 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12785 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12787 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12788 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12790 (rs6000_slow_unaligned_access): New function.
12791 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12792 (expand_block_compare): Likewise.
12793 (expand_strn_compare): Likewise.
12794 (rs6000_rtx_costs): Likewise.
12795 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12796 (riscv_slow_unaligned_access): Likewise.
12797 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12798 (riscv_slow_unaligned_access_p): ...this and make static.
12799 (riscv_option_override): Update accordingly.
12800 (riscv_slow_unaligned_access): New function.
12801 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12802 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12803 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12804 (rs6000_slow_unaligned_access): New function.
12805 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12806 (rs6000_rtx_costs): Likewise.
12807 * config/rs6000/rs6000-string.c (expand_block_compare)
12808 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12809 of SLOW_UNALIGNED_ACCESS.
12810 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12811 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12812 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12813 of SLOW_UNALIGNED_ACCESS.
12814 * expmed.c (simple_mem_bitfield_p): Likewise.
12815 * expr.c (alignment_for_piecewise_move): Likewise.
12816 (emit_group_load_1): Likewise.
12817 (emit_group_store): Likewise.
12818 (copy_blkmode_from_reg): Likewise.
12819 (emit_push_insn): Likewise.
12820 (expand_assignment): Likewise.
12821 (store_field): Likewise.
12822 (expand_expr_real_1): Likewise.
12823 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12824 * lra-constraints.c (simplify_operand_subreg): Likewise.
12825 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12826 * gimple-ssa-store-merging.c: Likewise in block comment at start
12828 * tree-ssa-strlen.c: Include target.h.
12829 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
12830 of SLOW_UNALIGNED_ACCESS.
12831 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
12833 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12835 PR rtl-optimization/82185
12836 * expmed.c (emit_store_flag_int): Only test tem if it has been
12839 2017-09-12 Richard Biener <rguenther@suse.de>
12841 PR middle-end/82149
12842 * match.pd ((FTYPE) N CMP CST): Fix typo.
12844 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
12846 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
12848 (mips_near_type_p): Add 'short_call' attribute as a synonym
12850 * doc/extend.texi (short_call): Document new function attribute.
12852 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12855 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
12856 assertion check that in the condition.
12857 (get_atomic_generic_size): Likewise. Before testing if parameter
12858 has pointer type, if it has array type, call for C++
12859 default_conversion to perform array-to-pointer conversion.
12861 2017-09-12 Richard Biener <rguenther@suse.de>
12863 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
12864 for operations we cannot scalarize.
12866 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
12868 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
12869 vectors heap vectors. Clean up comments.
12870 Make visited_bbs a reference.
12871 (profitable_jump_thread_path): Make GC
12872 vectors heap vectors. Clean up comments.
12874 (convert_and_register_jump_thread_path): Make GC vectors heap
12876 (check_subpath_and_update_thread_path): Same. Clean up comments.
12877 Make visited_bbs a reference.
12878 (handle_phi): Abstract common code to to
12879 register_jump_thread_path_if_profitable.
12880 Rename VAR_BB to DEF_BB.
12882 Make GC vectors heap vectors.
12883 Make visited_bbs a reference.
12884 (handle_assignment): Same.
12885 (register_jump_thread_path_if_profitable): New.
12886 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
12888 Make GC vectors heap vectors. Clean up comments.
12889 Make visited_bbs a reference.
12890 (find_jump_threads_backwards): Make visited_bbs live in the stack.
12891 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
12894 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
12897 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
12898 words of E_DImode object are reachable by xtensa_uimm8x4 access.
12900 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
12902 Revert r251800 and r251799.
12904 2017-09-11 Martin Jambor <mjambor@suse.cz>
12907 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
12908 arguments in advance.
12909 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
12910 use it to find predecessor edges.
12911 (naive_outof_ssa): Collect vector of predecessors.
12913 2017-09-08 Jason Merrill <jason@redhat.com>
12915 PR c++/70029 - ICE with ref-qualifier and -flto
12916 * langhooks.h (struct lang_hooks_for_types): Add
12917 copy_lang_qualifiers.
12918 * attribs.c (build_type_attribute_qual_variant): Use it.
12919 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
12921 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
12922 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
12924 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
12927 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
12928 (*mulsi3_sp64): New instruction.
12929 (mulsi3): New expander.
12931 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
12933 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
12935 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12937 * sancov.c: Include memmodel.h.
12939 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
12942 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
12945 2017-09-07 Carl Love <cel@us.ibm.com>
12947 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
12948 the sldi instruction.
12950 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
12952 * sancov.c: Include tm_p.h.
12954 2017-09-07 Jakub Jelinek <jakub@redhat.com>
12957 * output.h (switch_to_other_text_partition): New declaration.
12958 * varasm.c (switch_to_other_text_partition): New function.
12959 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
12960 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
12961 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
12962 to the other text partition before emitting LCL label and switch back
12963 after emitting the word after it.
12965 2017-09-07 Richard Biener <rguenther@suse.de>
12967 * passes.def (pass_split_crit_edges): Remove instance before PRE.
12968 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
12969 critical edges here, after loop init.
12970 (pass_data_pre): Remove PROP_no_crit_edges flags.
12971 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
12972 for valueization of call args to avoid leaking VN_TOP.
12973 (visit_use): Assert we do not visit default defs.
12974 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
12975 Use error_mark_node to more easily detect leaking VN_TOP.
12976 All default-defs are varying, not VN_TOP. Mark them visited.
12977 (run_scc_vn): Make code match comment.
12979 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12981 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
12982 OPTION_MASK_FLOAT128_KEYWORD.
12983 (POWERPC_MASKS): Likewise.
12984 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
12985 support for the -mfloat128-type option, and make -mfloat128
12986 default on PowerPC Linux systems. Define or undefine
12987 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
12988 Define __float128 to be __ieee128 if IEEE 128-bit support is
12989 enabled, or undefine it.
12990 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
12991 Delete defining __FLOAT128_TYPE__.
12992 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
12993 -mfloat128-type option and make -mfloat128 default on PowerPC
12995 (TARGET_FLOAT128_TYPE): Likewise.
12996 (-mfloat128-type): Likewise.
12997 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12998 Delete the -mfloat128-type option and make -mfloat128 default on
12999 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13000 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13001 128-bit floating point is enabled. Change tests from using
13002 -mfloat128-type to -mfloat128.
13003 (rs6000_mangle_type): Use the correct mangling for the __float128
13004 type even if normal long double is restricted to 64-bits.
13005 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13007 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13008 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13009 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13011 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13012 documentation for -mfloat128.
13014 2017-09-06 Olivier Hainque <hainque@adacore.com>
13016 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13018 2017-09-06 Wish Wu <wishwu007@gmail.com>
13019 Jakub Jelinek <jakub@redhat.com>
13021 * asan.c (initialize_sanitizer_builtins): Add
13022 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13023 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13024 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13025 BT_FN_VOID_UINT64_PTR variables.
13026 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13027 (BT_FN_VOID_UINT16_UINT16): Likewise.
13028 (BT_FN_VOID_UINT32_UINT32): Likewise.
13029 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13030 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13031 (BT_FN_VOID_UINT64_PTR): Likewise.
13032 * common.opt (flag_sanitize_coverage): New variable.
13033 (fsanitize-coverage=trace-pc): Remove.
13034 (fsanitize-coverage=): Add.
13035 * flag-types.h (enum sanitize_coverage_code): New enum.
13036 * fold-const.c (fold_range_test): Disable non-short-circuit
13037 optimization if flag_sanitize_coverage.
13038 (fold_truth_andor): Likewise.
13039 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13040 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13041 (coverage_sanitizer_opts): New array.
13042 (get_closest_sanitizer_option): Add OPTS argument, handle also
13043 OPT_fsanitize_coverage_.
13044 (parse_sanitizer_options): Adjusted to also handle
13045 OPT_fsanitize_coverage_.
13046 (common_handle_option): Add OPT_fsanitize_coverage_.
13047 * sancov.c (instrument_comparison, instrument_switch): New function.
13048 (sancov_pass): Add trace-cmp support.
13049 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13050 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13051 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13052 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13053 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13054 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13055 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13056 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13057 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13058 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13060 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13062 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13063 error. Only quit immediately if parsing is complete.
13064 (BEGIN): Initialize fatal_err and parse_done.
13065 (begin fpu, end fpu): Check number of arguments.
13066 (begin arch, end arch): Likewise.
13067 (begin cpu, end cpu): Likewise.
13068 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13069 (optalias): Likewise.
13071 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13073 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13074 * config/arm/arm-isa.h: Delete. Move definitions to ...
13075 * arm-cpus.in: ... here. Use new feature and fgroup values.
13076 * config/arm/arm.c (arm_option_override): Use lower case for feature
13078 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13079 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13080 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13082 (print_isa_bits_for): New function.
13083 (gen_isa): New function.
13084 (gen_comm_data): Use print_isa_bits_for.
13085 (define feature): New keyword.
13086 (define fgroup): New keyword.
13087 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13088 (arm-isa.h): Add rule to generate file.
13089 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13090 case for feature bit names.
13092 2017-09-06 Richard Biener <rguenther@suse.de>
13094 * tree-ssa-pre.c (NECESSARY): Remove.
13095 (create_expression_by_pieces): Do not touch pass-local flags.
13096 (insert_into_preds_of_block): Likewise.
13097 (do_pre_regular_insertion): Likewise.
13098 (eliminate_insert): Likewise.
13099 (eliminate_dom_walker::before_dom_children): Likewise.
13100 (fini_eliminate): Do not look at inserted_exprs.
13101 (mark_operand_necessary): Remove.
13102 (remove_dead_inserted_code): Replace with simple work-list
13103 algorithm based on inserted_exprs and SSA uses.
13104 (pass_pre::execute): Re-order fini_eliminate and
13105 remove_dead_inserted_code.
13107 2017-09-06 Olivier Hainque <hainque@adacore.com>
13109 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13110 for VxWorks 7. Adjust surrounding comments.
13112 2017-09-06 Richard Biener <rguenther@suse.de>
13114 * gimple-ssa-strength-reduction.c
13115 (find_candidates_dom_walker::before_dom_children): Also allow
13118 2017-09-06 Richard Biener <rguenther@suse.de>
13120 PR tree-optimization/82108
13121 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13122 for gap in the non-permutation SLP case.
13124 2017-09-06 Martin Jambor <mjambor@suse.cz>
13126 PR tree-optimization/82078
13127 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13128 add_access_to_work_queue...
13129 (build_accesses_from_assign): ...here.
13130 (propagate_all_subaccesses): Make sure racc is the group
13131 representative, if there is one.
13133 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13135 PR middle-end/82095
13136 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13139 2017-09-06 Richard Biener <rguenther@suse.de>
13141 * gimple-ssa-strength-reduction.c
13142 (find_candidates_dom_walker::before_doom_children): Use a
13143 type and not a mode check.
13145 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13148 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13150 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13152 2017-09-05 Jeff Law <law@redhat.com>
13154 PR tree-optimization/64910
13155 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13156 cases where we have 3 or more operands.
13158 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13160 PR middle-end/81768
13161 * omp-low.c (lower_omp_for): Recompute tree invariant if
13162 gimple_omp_for_initial/final is ADDR_EXPR.
13164 PR middle-end/81768
13165 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13166 into gimple val before gimplification fo the COND_EXPR.
13168 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13170 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13171 REGION_COPY argument.
13172 (thread_through_all_blocks): Remove unused argument to
13173 duplicate_thread_path.
13175 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13176 Alan Hayward <alan.hayward@arm.com>
13177 David Sherwood <david.sherwood@arm.com>
13179 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13180 Take a scalar_mode rather than a machine_mode.
13181 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13182 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13183 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13184 (aarch64_gen_adjusted_ldpstp): Likewise.
13185 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13187 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13188 Alan Hayward <alan.hayward@arm.com>
13189 David Sherwood <david.sherwood@arm.com>
13191 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13192 Take a scalar_int_mode instead of a machine_mode.
13193 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13194 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13195 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13196 (aarch64_simd_attr_length_rglist): Delete.
13197 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13198 a scalar_int_mode instead of a machine_mode.
13199 (aarch64_add_offset): Likewise.
13200 (aarch64_internal_mov_immediate): Likewise
13201 (aarch64_add_constant_internal): Likewise.
13202 (aarch64_add_constant): Likewise.
13203 (aarch64_movw_imm): Likewise.
13204 (aarch64_rtx_arith_op_extract_p): Likewise.
13205 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13206 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13207 Remove assert that the mode isn't a vector.
13208 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13209 (aarch64_expand_mov_immediate): Update calls after above changes.
13210 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13211 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13212 (aarch64_move_imm): Likewise.
13213 (aarch64_can_const_movi_rtx_p): Likewise.
13214 (aarch64_strip_extend): Likewise.
13215 (aarch64_extr_rtx_p): Likewise.
13216 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13217 a CONST_INT when the mode parameter is VOIDmode.
13218 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13220 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13222 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13223 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13224 (bitwise_type_for_mode): Update accordingly.
13226 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13228 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13229 * stor-layout.c (mode_for_size_tree): Likewise.
13230 (mode_for_array): Update accordingly.
13231 (layout_decl): Likewise.
13232 (compute_record_mode): Likewise. Only set the mode once.
13234 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13236 * target.def (get_mask_mode): Change return type to opt_mode.
13238 * doc/tm.texi: Regenerate.
13239 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13240 * targhooks.c (default_get_mask_mode): Likewise.
13241 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13242 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13243 targetm.get_mask_mode.
13244 * tree.c (build_truth_vector_type): Likewise.
13246 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13248 * machmode.h (mode_for_vector): Return an opt_mode.
13249 * stor-layout.c (mode_for_vector): Likewise.
13250 (mode_for_int_vector): Update accordingly.
13251 (layout_type): Likewise.
13252 * config/i386/i386.c (emit_memmov): Likewise.
13253 (ix86_expand_set_or_movmem): Likewise.
13254 (ix86_expand_vector_init): Likewise.
13255 (ix86_get_mask_mode): Likewise.
13256 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13258 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13259 * expmed.c (extract_bit_field_1): Likewise.
13260 * expr.c (expand_expr_real_2): Likewise.
13261 * optabs-query.c (can_vec_perm_p): Likewise.
13262 (can_vec_mask_load_store_p): Likewise.
13263 * optabs.c (expand_vec_perm): Likewise.
13264 * targhooks.c (default_get_mask_mode): Likewise.
13265 * tree-vect-stmts.c (vectorizable_store): Likewise.
13266 (vectorizable_load): Likewise.
13267 (get_vectype_for_scalar_type_and_size): Likewise.
13269 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13271 * machmode.h (mode_for_int_vector): New function.
13272 * stor-layout.c (mode_for_int_vector): Likewise.
13273 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13274 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13275 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13276 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13277 (s390_expand_vcond): Likewise.
13279 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13281 * machmode.h (opt_machine_mode): New type.
13282 (opt_mode<T>): Allow construction from anything that can be
13284 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13285 (mode_for_size): Return an opt_machine_mode.
13286 * stor-layout.c (mode_for_size): Likewise.
13287 (mode_for_size_tree): Update call accordingly.
13288 (bitwise_mode_for_mode): Likewise.
13289 (make_fract_type): Likewise.
13290 (make_accum_type): Likewise.
13291 * caller-save.c (replace_reg_with_saved_mem): Update call
13293 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13294 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13295 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13296 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13297 * expmed.c (extract_bit_field_1): Likewise.
13298 * reload.c (get_secondary_mem): Likewise.
13299 * varasm.c (assemble_integer): Likewise.
13300 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13303 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13305 * machmode.h (decimal_float_mode_for_size): New function.
13306 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13307 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13308 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13309 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13310 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13311 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13313 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13315 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13316 (get_builtin_sync_mode): Likewise.
13317 (expand_ifn_atomic_compare_exchange): Likewise.
13318 (expand_builtin_atomic_clear): Likewise.
13319 (expand_builtin_atomic_test_and_set): Likewise.
13320 (fold_builtin_atomic_always_lock_free): Likewise.
13321 * calls.c (compute_argument_addresses): Likewise.
13322 (emit_library_call_value_1): Likewise.
13323 (store_one_arg): Likewise.
13324 * combine.c (combine_instructions): Likewise.
13325 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13326 * config/arm/arm.c (arm_function_value): Likewise.
13327 (aapcs_allocate_return_reg): Likewise.
13328 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13329 * config/i386/i386.c (construct_container): Likewise.
13330 (ix86_gimplify_va_arg): Likewise.
13331 (ix86_expand_sse_cmp): Likewise.
13332 (emit_memmov): Likewise.
13333 (emit_memset): Likewise.
13334 (expand_small_movmem_or_setmem): Likewise.
13335 (ix86_expand_pextr): Likewise.
13336 (ix86_expand_pinsr): Likewise.
13337 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13338 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13340 * config/mips/mips.c (mips_function_value_1) Likewise.
13341 (mips_block_move_straight): Likewise.
13342 (mips_expand_ins_as_unaligned_store): Likewise.
13343 * config/powerpcspe/powerpcspe.c
13344 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13345 (rs6000_darwin64_record_arg_flush): Likewise.
13346 * config/rs6000/rs6000.c
13347 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13348 (rs6000_darwin64_record_arg_flush): Likewise.
13349 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13350 (sparc_function_value_1): Likewise.
13351 * config/spu/spu.c (adjust_operand): Likewise.
13352 (spu_emit_branch_or_set): Likewise.
13353 (arith_immediate_p): Likewise.
13354 * emit-rtl.c (gen_lowpart_common): Likewise.
13355 * expr.c (expand_expr_real_1): Likewise.
13356 * function.c (assign_parm_setup_block): Likewise.
13357 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13358 * reload1.c (alter_reg): Likewise.
13359 * stor-layout.c (mode_for_vector): Likewise.
13360 (layout_type): Likewise.
13362 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13364 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13365 (spu_convert_move): Likewise.
13366 * lower-subreg.c (resolve_simple_move): Likewise.
13368 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13371 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13372 define_insn to a define_expand.
13373 (altivec_vsum2sws_direct): New define_insn.
13374 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13376 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13378 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13381 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13386 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13387 realign stack if stack alignment needed is less than incoming
13390 2017-09-05 Marek Polacek <polacek@redhat.com>
13393 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13396 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13398 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13400 2017-09-05 Richard Biener <rguenther@suse.de>
13402 PR tree-optimization/82084
13403 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13405 2017-09-05 Richard Biener <rguenther@suse.de>
13407 PR tree-optimization/82102
13408 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13410 2017-09-05 Martin Liska <mliska@suse.cz>
13412 PR tree-optimization/82032
13413 * tree-cfg.c (generate_range_test): New function.
13414 * tree-cfg.h (generate_range_test): Declared here.
13415 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13416 (cleanup_control_expr_graph): Use it.
13417 * tree-switch-conversion.c (try_switch_expansion): Remove
13419 (emit_case_nodes): Use generate_range_test.
13421 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13424 * config/i386/i386.md (*<btsc><mode>_mask): Add
13425 TARGET_USE_BT to insn constraint.
13426 (*btr<mode>_mask): Ditto.
13428 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13430 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13431 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13433 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13436 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13437 TARGET_NEON and TARGET_IWMMXT.
13438 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13439 TARGET_NEON and TARGET_IWMMXT.
13440 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13442 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13444 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13445 (ix86_rewrite_tls_address): Ditto.
13446 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13447 (ix86_rewrite_tls_address_1): Ditto.
13448 (ix86_rewrite_tls_address): Ditto.
13449 * config/i386/predicates.md (tls_address_pattern): New predicate.
13450 * config/i386/i386.md (TLS address splitter): New splitter.
13452 2017-09-04 Richard Biener <rguenther@suse.de>
13454 PR tree-optimization/82084
13455 * fold-const.h (can_native_encode_string_p): Declare.
13456 * fold-const.c (can_native_encode_string_p): Factor out from ...
13457 (native_encode_string): ... here.
13458 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13459 vectorizing stores from constants we later cannot handle.
13461 2017-09-04 Marek Polacek <polacek@redhat.com>
13464 * doc/invoke.texi: Update -Wtautological-compare documentation.
13466 2017-09-04 Jeff Law <law@redhat.com>
13468 PR tree-optimization/64910
13469 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13470 swap the first and last operand if the last is a constant.
13472 2017-09-04 Marek Polacek <polacek@redhat.com>
13475 * convert.c (do_narrow): When sanitizing signed integer overflows,
13476 bail out for signed types.
13477 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13479 2017-09-04 Richard Biener <rguenther@suse.de>
13481 PR tree-optimization/82060
13482 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13483 Move devirtualization after stmt folding and before EH/AB/noreturn
13484 cleanup to get the stmt refs canonicalized. Use a bool instead
13485 of gimple_modified_p since that doesn't work for NOPs. Schedule
13486 NOPs generated by folding for removal.
13488 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13489 Alan Hayward <alan.hayward@arm.com>
13490 David Sherwood <david.sherwood@arm.com>
13492 * coretypes.h (pad_direction): New enum.
13493 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13494 (FUNCTION_ARG_PADDING): Likewise.
13495 * target.def (function_arg_padding): New hook.
13496 * targhooks.h (default_function_arg_padding): Declare.
13497 * targhooks.c (default_function_arg_padding): New function.
13498 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13499 (TARGET_FUNCTION_ARG_PADDING): ...this.
13500 * doc/tm.texi: Regenerate.
13501 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13502 instead of direction.
13503 (compute_argument_addresses): Likewise.
13504 (load_register_parameters): Likewise.
13505 (emit_library_call_value_1): Likewise.
13506 (store_one_arg): Use targetm.calls.function_arg_padding instead
13507 of FUNCTION_ARG_PADDING.
13508 (must_pass_in_stack_var_size_or_pad): Likewise.
13509 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13510 (emit_group_store): Likewise.
13511 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13512 instead of FUNCTION_ARG_PADDING.
13513 (emit_push_insn): Likewise, and propagate enum change throughout
13515 * function.h (direction): Delete.
13516 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13518 * function.c (assign_parm_find_stack_rtl): Likewise.
13519 (assign_parm_setup_block_p): Likewise.
13520 (assign_parm_setup_block): Likewise.
13521 (gimplify_parameters): Likewise.
13522 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13523 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13525 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13526 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13527 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13528 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13529 (aarch64_function_arg_padding): ...this new function.
13530 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13531 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13532 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13533 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13534 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13535 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13536 (arm_pad_arg_upward): Replace with...
13537 (arm_function_arg_padding): ...this new function.
13538 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13540 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13541 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13542 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13543 (ia64_hpux_function_arg_padding): Replace with...
13544 (ia64_function_arg_padding): ...this new function. Use pad_direction
13545 instead of direction. Check for TARGET_HPUX.
13546 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13547 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13548 (iq2000_function_arg_padding): New function.
13549 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13550 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13551 (mips_function_arg_padding): ...this new function.
13552 (mips_pad_reg_upward): Update accordingly.
13553 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13554 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13555 targetm.calls.function_arg_padding.
13556 (FUNCTION_ARG_PADDING): Delete.
13557 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13558 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13559 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13560 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13561 (nios2_block_reg_padding): Return pad_direction instead of direction.
13562 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13563 instead of direction.
13564 (nios2_function_arg_padding): Likewise. Make static.
13565 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13566 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13567 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13568 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13569 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13570 (pa_function_arg_padding): Make static. Return pad_direction instead
13572 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13573 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13574 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13575 instead of direction. Use targetm.calls.function_arg_padding.
13576 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13577 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13578 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13579 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13580 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13582 (function_arg_padding): Rename to...
13583 (rs6000_function_arg_padding): ...this. Make static. Return
13584 pad_direction instead of direction.
13585 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13586 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13587 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13588 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13589 instead of direction. Use targetm.calls.function_arg_padding.
13590 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13591 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13592 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13593 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13594 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13595 (function_arg_padding): Rename to...
13596 (rs6000_function_arg_padding): ...this. Make static. Return
13597 pad_direction instead of direction.
13598 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13599 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13600 * config/s390/s390.c (s390_function_arg_padding): New function.
13601 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13602 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13603 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13604 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13605 (function_arg_padding): Rename to...
13606 (sparc_function_arg_padding): ...this. Make static. Return
13607 pad_direction instead of direction.
13608 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13609 * config/spu/spu.c (spu_function_arg_padding): New function.
13610 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13611 * system.h (FUNCTION_ARG_PADDING): Poison.
13613 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13614 Alan Hayward <alan.hayward@arm.com>
13615 David Sherwood <david.sherwood@arm.com>
13617 * target.def (modes_tieable_p): New hook.
13618 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13619 (TARGET_MODES_TIEABLE_P): ...this.
13620 * doc/tm.texi.in: Regenerate.
13621 * hooks.h (hook_bool_mode_mode_true): Declare.
13622 * hooks.c (hook_bool_mode_mode_true): New function.
13623 * combine.c (subst): Use targetm.modes_tieable_p instead of
13625 * dse.c (find_shift_sequence): Likewise.
13626 * expmed.c (extract_low_bits): Likewise.
13627 * lower-subreg.c: Include target.h.
13628 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13630 * rtlanal.c (rtx_cost): Likewise.
13631 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13632 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13633 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13634 (TARGET_MODES_TIEABLE_P): Redefine.
13635 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13636 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13637 (TARGET_MODES_TIEABLE_P): Redefine.
13638 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13639 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13640 (arc_modes_tieable_p): New function.
13641 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13642 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13643 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13644 (arm_modes_tieable_p): Make static.
13645 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13646 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13647 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13648 (TARGET_MODES_TIEABLE_P): Redefine.
13649 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13650 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13651 (TARGET_MODES_TIEABLE_P): Redefine.
13652 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13653 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13654 (cr16_modes_tieable_p): New function.
13655 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13656 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13657 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13658 (TRULY_NOOP_TRUNCATION): Update comment.
13659 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13660 (TRULY_NOOP_TRUNCATION): Update comment.
13661 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13662 (frv_modes_tieable_p): New function.
13663 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13664 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13665 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13666 (TARGET_MODES_TIEABLE_P): Redefine.
13667 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13668 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13669 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13670 (TARGET_MODES_TIEABLE_P): Redefine.
13671 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13672 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13673 (ia64_modes_tieable_p): New function.
13674 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13675 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13676 (iq2000_modes_tieable_p): New function.
13677 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13678 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13679 (lm32_modes_tieable_p): New function.
13680 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13681 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13682 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13683 (TARGET_MODES_TIEABLE_P): Redefine.
13684 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13685 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13686 (m32r_modes_tieable_p): New function.
13687 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13688 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13689 (m68k_modes_tieable_p): New function.
13690 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13691 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13692 (mcore_modes_tieable_p): New function.
13693 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13694 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13696 (TARGET_MODES_TIEABLE_P): Redefine.
13697 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13698 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13699 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13700 (TARGET_MODES_TIEABLE_P): Redefine.
13701 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13702 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13703 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13704 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13705 (mn10300_modes_tieable_p): ...this and make static.
13706 (TARGET_MODES_TIEABLE_P): Redefine.
13707 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13708 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13709 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13710 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13711 (msp430_modes_tieable_p): Make static.
13712 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13713 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13714 (TARGET_MODES_TIEABLE_P): Redefine.
13715 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13716 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13717 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13718 (TARGET_MODES_TIEABLE_P): Redefine.
13719 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13720 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13721 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13722 (TARGET_MODES_TIEABLE_P): Redefine.
13723 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13724 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13725 (pdp11_modes_tieable_p): New function.
13726 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13727 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13728 (rs6000_modes_tieable_p): New function.
13729 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13730 * config/powerpcspe/powerpcspe.md: Update comment.
13731 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13732 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13733 (TARGET_MODES_TIEABLE_P): Redefine.
13734 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13735 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13736 (rl78_modes_tieable_p): New function.
13737 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13738 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13739 (rs6000_modes_tieable_p): New function.
13740 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13741 * config/rs6000/rs6000.md: Update comment.
13742 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13743 * config/rx/rx.c (rx_modes_tieable_p): New function.
13744 (TARGET_MODES_TIEABLE_P): Redefine.
13745 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13746 * config/s390/s390.c (s390_modes_tieable_p): New function.
13747 (TARGET_MODES_TIEABLE_P): Redefine.
13748 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13749 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13750 (sh_modes_tieable_p): New function.
13751 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13752 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13753 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13754 (sparc_modes_tieable_p): Make static.
13755 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13756 * config/spu/spu.c (spu_modes_tieable_p): New function.
13757 (TARGET_MODES_TIEABLE_P): Redefine.
13758 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13759 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13760 (TARGET_MODES_TIEABLE_P): Redefine.
13761 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13762 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13763 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13764 * config/v850/v850.c (v850_modes_tieable_p): New function.
13765 (TARGET_MODES_TIEABLE_P): Redefine.
13766 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13767 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13768 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13769 (visium_modes_tieable_p): New function.
13770 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13771 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13772 (xtensa_modes_tieable_p): New function.
13773 * system.h (MODES_TIEABLE_P): Poison.
13775 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13776 Alan Hayward <alan.hayward@arm.com>
13777 David Sherwood <david.sherwood@arm.com>
13779 * target.def (hard_regno_mode_ok): New hook.
13780 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13781 (TARGET_HARD_REGNO_MODE_OK): ...this.
13782 * doc/tm.texi.in: Regenerate.
13783 * hooks.h (hook_bool_uint_mode_true): Declare.
13784 * hooks.c (hook_bool_uint_mode_true): New function.
13785 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13786 HARD_REGNO_MODE_OK.
13787 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13788 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13789 instead of HARD_REGNO_MODE_OK.
13790 * caller-save.c: Include target.h.
13791 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13792 HARD_REGNO_MODE_OK.
13793 * combine.c (can_combine_p): Likewise.
13794 (combinable_i3pat): Likewise.
13795 (can_change_dest_mode): Likewise.
13796 * expr.c (init_expr_target): Likewise.
13797 (convert_move): Likewise.
13798 (convert_modes): Likewise.
13799 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13800 (setup_prohibited_mode_move_regs): Likewise.
13801 * ira.h (target_ira): Likewise.
13802 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13803 * lra-constraints.c (process_alt_operands): Likewise.
13804 (split_reg): Likewise.
13805 * recog.c (peep2_find_free_register): Likewise.
13806 * ree.c (combine_reaching_defs): Likewise.
13807 * regcprop.c (maybe_mode_change): Likewise.
13808 * reginfo.c (init_reg_sets_1): Likewise.
13809 (choose_hard_reg_mode): Likewise.
13810 (simplifiable_subregs): Likewise.
13811 * regrename.c (check_new_reg_p): Likewise.
13812 * reload.c (find_valid_class): Likewise.
13813 (find_valid_class_1): Likewise.
13814 (reload_inner_reg_of_subreg): Likewise.
13815 (push_reload): Likewise.
13816 (combine_reloads): Likewise.
13817 (find_dummy_reload): Likewise.
13818 (find_reloads): Likewise.
13819 * reload1.c (find_reg): Likewise.
13820 (set_reload_reg): Likewise.
13821 (allocate_reload_reg): Likewise.
13822 (choose_reload_regs): Likewise.
13823 (reload_adjust_reg_for_temp): Likewise.
13824 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
13825 (simplify_subreg_regno): Likewise.
13826 * sel-sched.c (init_regs_for_mode): Likewise.
13827 * varasm.c (make_decl_rtl): Likewise.
13828 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
13829 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
13830 HARD_REGNO_MODE_OK.
13831 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
13832 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
13833 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13834 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
13835 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
13836 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13837 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
13838 (arc_mode_class): Delete.
13839 (HARD_REGNO_MODE_OK): Delete.
13840 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13841 (arc_hard_regno_mode_ok): Rename old array to...
13842 (arc_hard_regno_mode_ok_modes): ...this.
13843 (arc_conditional_register_usage): Update accordingly.
13844 (arc_mode_class): Make static.
13845 (arc_hard_regno_mode_ok): New function.
13846 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
13847 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
13848 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13849 (arm_hard_regno_mode_ok): Make static.
13850 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
13851 HARD_REGNO_MODE_OK.
13852 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
13853 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
13854 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
13856 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13857 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
13858 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
13859 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
13860 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
13861 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13862 * config/bfin/predicates.md (valid_reg_operand): Use
13863 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
13864 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
13865 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
13866 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13867 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
13868 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
13869 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13870 (cr16_hard_regno_mode_ok): Make static and return a bool.
13871 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
13872 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13873 (cris_hard_regno_mode_ok): New function.
13874 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
13875 (epiphany_mode_class): Delete.
13876 (HARD_REGNO_MODE_OK): Delete.
13877 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
13878 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13879 (hard_regno_mode_ok): Rename to...
13880 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
13881 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
13882 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
13883 HARD_REGNO_MODE_OK.
13884 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
13885 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
13886 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13887 (frv_hard_regno_mode_ok): Make static and return a bool.
13888 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
13889 HARD_REGNO_MODE_OK.
13890 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
13891 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
13892 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
13893 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
13895 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13896 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
13897 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
13898 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
13900 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13901 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
13902 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13903 (ia64_hard_regno_mode_ok): New function.
13904 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
13905 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13906 (iq2000_hard_regno_mode_ok): New function.
13907 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
13908 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13909 (lm32_hard_regno_mode_ok): New function.
13910 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
13911 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
13912 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
13913 instead of HARD_REGNO_MODE_OK.
13914 (m32c_hard_regno_ok): Rename to...
13915 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
13916 (m32c_cannot_change_mode_class): Update accordingly.
13917 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13918 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
13919 (m32r_mode_class): Delete.
13920 (HARD_REGNO_MODE_OK): Delete.
13921 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13922 (m32r_hard_regno_mode_ok): Rename to...
13923 (m32r_hard_regno_modes): ...this.
13924 (m32r_mode_class): Make static.
13925 (m32r_hard_regno_mode_ok): New function.
13926 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
13927 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
13928 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13929 (m68k_hard_regno_mode_ok): Make static.
13930 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
13931 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13932 (mcore_hard_regno_mode_ok): New function.
13933 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
13934 (HARD_REGNO_MODE_OK): Delete.
13935 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
13937 (microblaze_hard_regno_mode_ok_p): ...this and make static.
13938 (microblaze_hard_regno_mode_ok): New function.
13939 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13940 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
13941 (mips_hard_regno_mode_ok): Delete.
13942 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
13943 (mips_hard_regno_mode_ok_p): ...this and make static.
13944 (mips_hard_regno_mode_ok_p): Rename to...
13945 (mips_hard_regno_mode_ok_uncached): ...this.
13946 (mips_hard_regno_mode_ok): New function.
13947 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
13948 of HARD_REGNO_MODE_OK.
13949 (mips_option_override): Update after above name changes.
13950 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13951 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
13952 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
13953 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
13954 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
13955 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13956 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
13957 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
13958 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
13959 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13960 (msp430_hard_regno_mode_ok): Make static and return a bool.
13961 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
13962 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
13963 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
13965 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13966 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
13967 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
13968 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
13969 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
13970 (PA_HARD_REGNO_MODE_OK): ...this
13971 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
13972 (PA_HARD_REGNO_MODE_OK): ...this.
13973 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13974 (pa_hard_regno_mode_ok): New function.
13975 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
13976 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13977 (pdp11_hard_regno_mode_ok): New function.
13978 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
13979 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
13981 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
13983 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13984 (rs6000_hard_regno_mode_ok): Rename to...
13985 (rs6000_hard_regno_mode_ok_uncached): ...this.
13986 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13987 (rs6000_hard_regno_mode_ok): New function.
13988 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
13989 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
13990 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
13991 (riscv_hard_regno_mode_ok): ...this and make static.
13992 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13993 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
13994 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
13995 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13996 (rl78_hard_regno_mode_ok): Make static and return bool.
13997 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
13998 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14000 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14002 (rs6000_hard_regno_mode_ok): Rename to...
14003 (rs6000_hard_regno_mode_ok_uncached): ...this.
14004 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14005 (rs6000_hard_regno_mode_ok): New function.
14006 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14007 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14008 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14009 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14010 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14011 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14012 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14013 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14014 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14015 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14016 (sh_hard_regno_mode_ok): Make static.
14017 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14018 instead of HARD_REGNO_MODE_OK.
14019 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14020 (sparc_mode_class): Delete.
14021 (HARD_REGNO_MODE_OK): Delete.
14022 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14023 (hard_regno_mode_classes): Make static.
14024 (sparc_mode_class): Likewise.
14025 (sparc_hard_regno_mode_ok): New function.
14026 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14027 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14028 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14030 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14031 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14032 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14033 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14034 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14035 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14036 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14037 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14038 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14039 (visium_hard_regno_mode_ok): New function.
14040 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14041 instead of HARD_REGNO_MODE_OK.
14042 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14043 (HARD_REGNO_MODE_OK): Delete.
14044 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14045 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14046 (xtensa_option_override): Update accordingly.
14047 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14048 (xtensa_hard_regno_mode_ok): New function.
14049 * system.h (HARD_REGNO_MODE_OK): Poison.
14051 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14052 Alan Hayward <alan.hayward@arm.com>
14053 David Sherwood <david.sherwood@arm.com>
14055 * target.def (hard_regno_call_part_clobbered): New hook.
14056 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14057 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14058 * doc/tm.texi: Regenerate.
14059 * hooks.h (hook_bool_uint_mode_false): Declare.
14060 * hooks.c (hook_bool_uint_mode_false): New function.
14061 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14062 * cselib.c (cselib_process_insn): Use
14063 targetm.hard_regno_call_part_clobbered instead of
14064 HARD_REGNO_CALL_PART_CLOBBERED.
14065 * ira-conflicts.c (ira_build_conflicts): Likewise.
14066 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14067 * lra-constraints.c (need_for_call_save_p): Likewise.
14068 * lra-lives.c: Include target.h.
14069 (check_pseudos_live_through_calls): Use
14070 targetm.hard_regno_call_part_clobbered instead of
14071 HARD_REGNO_CALL_PART_CLOBBERED.
14072 * regcprop.c: Include target.h.
14073 (copyprop_hardreg_forward_1): Use
14074 targetm.hard_regno_call_part_clobbered instead of
14075 HARD_REGNO_CALL_PART_CLOBBERED.
14076 * reginfo.c (choose_hard_reg_mode): Likewise.
14077 * regrename.c (check_new_reg_p): Likewise.
14078 * reload.c (find_equiv_reg): Likewise.
14079 * reload1.c (emit_reload_insns): Likewise.
14080 * sched-deps.c (deps_analyze_insn): Likewise.
14081 * sel-sched.c (init_regs_for_mode): Likewise.
14082 (mark_unavailable_hard_regs): Likewise.
14083 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14084 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14085 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14087 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14088 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14089 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14091 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14093 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14094 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14095 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14097 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14098 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14099 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14101 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14102 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14104 * config/powerpcspe/powerpcspe.c
14105 (rs6000_hard_regno_call_part_clobbered): New function.
14106 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14107 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14108 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14110 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14111 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14112 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14114 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14115 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14116 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14118 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14119 Alan Hayward <alan.hayward@arm.com>
14120 David Sherwood <david.sherwood@arm.com>
14122 * rtl.h (subreg_memory_offset): Declare.
14123 * emit-rtl.c (subreg_memory_offset): New function.
14124 * expmed.c (store_bit_field_1): Use it.
14125 * expr.c (undefined_operand_subword_p): Likewise.
14126 * simplify-rtx.c (simplify_subreg): Likewise.
14128 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14130 PR rtl-optimization/57448
14133 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14134 using atomic_load pattern.
14135 (expand_atomic_store): Likewise.
14137 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14140 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14141 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14142 handling. Use replace_call_with_value with NULL instead of
14143 gsi_replace, unlink_stmt_vdef and release_defs.
14145 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14148 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14150 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14153 * rtl.h (emit_library_call_value_1): Declare.
14154 (emit_library_call): Replace declaration with a series of overloads.
14155 Remove the parameter count argument.
14156 (emit_library_call_value): Likewise.
14157 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14158 with an "rtx_mode_t *".
14159 (emit_library_call_value): Delete.
14160 (emit_library_call): Likewise.
14161 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14162 (asan_emit_allocas_unpoison): Likewise.
14163 * builtins.c (expand_builtin_powi): Likewise.
14164 (expand_asan_emit_allocas_unpoison): Likewise.
14165 * cfgexpand.c (expand_main_function): Likewise.
14166 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14167 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14168 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14169 * config/arm/arm.c (arm_trampoline_init): Likewise.
14170 (arm_call_tls_get_addr): Likewise.
14171 (arm_expand_divmod_libfunc): Likewise.
14172 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14173 (smulsi3_highpart): Likewise.
14174 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14175 (c6x_expand_compare): Likewise.
14176 (c6x_expand_movmem): Likewise.
14177 * config/frv/frv.c (frv_trampoline_init): Likewise.
14178 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14179 (ix86_expand_divmod_libfunc): Likewise.
14180 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14181 (ia64_expand_compare): Likewise.
14182 (ia64_profile_hook): Likewise.
14183 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14184 (nonlocal_goto): Likewise.
14185 (restore_stack_nonlocal): Likewise.
14186 * config/m32r/m32r.c (block_move_call): Likewise.
14187 (m32r_trampoline_init): Likewise.
14188 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14189 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14190 (m68k_call_m68k_read_tp): Likewise.
14191 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14192 (microblaze_expand_divide): Likewise.
14193 * config/mips/mips.h (mips_args): Likewise.
14194 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14195 (MIPS_ICACHE_SYNC): Likewise.
14196 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14197 (nios2_trampoline_init): Likewise.
14198 * config/pa/pa.c (hppa_tls_call): Likewise.
14199 (pa_trampoline_init): Likewise.
14200 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14201 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14202 (expand_strn_compare): Likewise.
14203 (rs6000_generate_compare): Likewise.
14204 (rs6000_expand_float128_convert): Likewise.
14205 (output_profile_hook): Likewise.
14206 (rs6000_trampoline_init): Likewise.
14207 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14208 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14209 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14210 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14211 (rs6000_generate_compare): Likewise.
14212 (rs6000_expand_float128_convert): Likewise.
14213 (output_profile_hook): Likewise.
14214 (rs6000_trampoline_init): Likewise.
14215 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14216 * config/sh/sh.c (sh_trampoline_init): Likewise.
14217 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14218 (sparc_emit_float_lib_cmp): Likewise.
14219 (sparc32_initialize_trampoline): Likewise.
14220 (sparc64_initialize_trampoline): Likewise.
14221 (sparc_profile_hook): Likewise.
14222 * config/spu/spu.c (ea_load_store): Likewise.
14223 * config/spu/spu.md (floatunssidf2): Likewise.
14224 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14225 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14226 * config/visium/visium.c (expand_block_move_4): Likewise.
14227 (expand_block_move_2): Likewise.
14228 (expand_block_move_1): Likewise.
14229 (expand_block_set_4): Likewise.
14230 (expand_block_set_2): Likewise.
14231 (expand_block_set_1): Likewise.
14232 (visium_trampoline_init): Likewise.
14233 (visium_profile_hook): Likewise.
14234 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14235 (xtensa_setup_frame_addresses): Likewise.
14236 (xtensa_trampoline_init): Likewise.
14237 * except.c (sjlj_emit_function_enter): Likewise.
14238 (sjlj_emit_function_exit): Likewise.
14239 * explow.c (allocate_dynamic_stack_space): Likewise.
14240 (probe_stack_range): Likewise.
14241 * expr.c (convert_mode_scalar): Likewise.
14242 * optabs.c (expand_binop): Likewise.
14243 (expand_twoval_binop_libfunc): Likewise.
14244 (expand_unop): Likewise.
14245 (prepare_cmp_insn): Likewise.
14246 (prepare_float_lib_cmp): Likewise.
14247 (expand_float): Likewise.
14248 (expand_fix): Likewise.
14249 (expand_fixed_convert): Likewise.
14250 (maybe_emit_sync_lock_test_and_set): Likewise.
14251 (expand_atomic_compare_and_swap): Likewise.
14252 (expand_mem_thread_fence): Likewise.
14253 (expand_atomic_fetch_op): Likewise.
14255 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14257 * doc/generic.texi (OpenACC): Adjust URL.
14258 * doc/invoke.texi (C Dialect Options): Ditto.
14260 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14262 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14263 predicate for operand 1. Add (m,<S>) constraint.
14264 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14265 Prevent memory operand 1 with register operand 2.
14267 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14269 PR rtl-optimization/82024
14270 * combine.c (try_combine): If the combination result is a PARALLEL,
14271 and we only need to retain the SET in there that would be placed
14272 at I2, check that we can place that at I3 instead, before doing so.
14274 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14277 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14279 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14280 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14283 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14284 Jeff Law <law@redhat.com>
14286 * varasm.c (bss_initializer_p): Do not put constants into .bss
14287 (categorize_decl_for_section): Handle bss_initializer_p returning
14288 false when DECL_INITIAL is NULL.
14290 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14293 * config/s390/s390.c (s390_can_inline_p): New function.
14295 2017-09-01 Jeff Law <law@redhat.com>
14297 PR tree-optimization/82052
14298 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14299 Always initialize the returned slot after a hash table miss
14300 when INSERT is true.
14302 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14304 * config/s390/s390.md (mem_signal_fence): Remove.
14305 * doc/md.texi (mem_signal_fence): Remove.
14306 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14308 * target-insns.def (mem_signal_fence): Remove.
14310 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14313 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14316 * asan.c (create_odr_indicator): Strip name encoding from assembler
14317 name before appending it after __odr_asan_.
14319 2017-09-01 Martin Liska <mliska@suse.cz>
14321 PR tree-optimization/82059
14322 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14323 frequency only when an edge is redirected.
14325 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14327 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14328 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14329 (arc_conditional_register_usage): Remove ARC600 lp_count
14331 (arc_file_start): Emit Tag_ARC_CPU_variation.
14332 (arc_can_use_doloop_p): New conditions to use ZOLs.
14333 (hwloop_fail): New function.
14334 (hwloop_optimize): Likewise.
14335 (hwloop_pattern_reg): Likewise.
14336 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14337 (arc_reorg_loops): New function, calls reorg_loops.
14338 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14339 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14341 (arc_loop_hazard): Remove function, functionality moved into
14343 (arc_hazard): Remove arc_loop_hazard call.
14344 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14345 into hwloop_optimize.
14346 (arc_label_align): Remove ZOL handling.
14347 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14348 * config/arc/arc.md (doloop_begin): Remove pattern.
14349 (doloop_begin_i): Likewise.
14350 (doloop_end_i): Likewise.
14351 (doloop_fallback): Likewise.
14352 (doloop_fallback_m): Likewise.
14353 (doloop_end): Reimplement expand.
14354 (arc_lp): New pattern for LP instruction.
14355 (loop_end): New pattern.
14356 (loop_fail): Likewise.
14357 (decrement_and_branch_until_zero): Likewise.
14358 * config/arc/arc.opt (mlpc-width): New option.
14359 * doc/invoke.texi (mlpc-width): Document option.
14361 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14363 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14364 (arc_ccfsm_advance): Fix checking for delay slots.
14365 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14367 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14369 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14371 (movhi_insn): Update store instruction constraint which are saving
14372 6-bit short immediates.
14373 (movsi_insn): Consider also short scaled load operations.
14374 (zero_extendhisi2_i): Use Usd constraint instead of T.
14375 (extendhisi2_i): Add q constraint.
14376 (arc_clzsi2): Add type and length attributes.
14377 (arc_ctzsi2): Likewise.
14378 * config/arc/constraints.md (Usc): Update constraint, the
14379 assembler can parse two relocations for a single instruction.
14381 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14383 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14384 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14386 2017-08-31 Olivier Hainque <hainque@adacore.com>
14388 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14389 match as powerpc-wrs-vxworks*.
14391 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14393 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14394 register constraint for by-element operand.
14395 (aarch64_mls_elt_merge<mode>): Likewise.
14397 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14399 * config/arc/arc.c (arc_can_follow_jump): Check for short
14402 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14404 * config.gcc: Use g.opt for arc.
14405 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14406 functionality moved to ...
14407 (legitimate_scaled_address_p): New function, ...here.
14408 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14409 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14410 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14412 (arc_override_options): Handle G option.
14413 (arc_output_pic_addr_const): Correct function definition.
14414 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14415 (arc_decl_anon_ns_mem_p): Delete.
14416 (arc_in_small_data_p): Overhaul this function to take into
14417 consideration the value given via G option.
14418 (arc_rewrite_small_data_1): Renamed and corrected old
14419 arc_rewrite_small_data function.
14420 (arc_rewrite_small_data): New function.
14421 (small_data_pattern): Don't use pic_offset_table_rtx.
14422 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14423 * config/arc/simdext.md (movmisalignv2hi): Use
14424 prepare_move_operands function.
14426 (movmisalign*): Likewise.
14427 * doc/invoke.texi (ARC options): Document -G option.
14429 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14431 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14433 * config/arc/arc.c (arc_print_operand): Output scalled address for
14434 sdata whenever is possible.
14435 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14436 load/stores are available.
14437 (compact_sda_memory_operand): Check for the alignment required by
14438 code density instructions.
14439 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14441 * config/arc/constraints.md (Usd): Update constraint.
14442 (Us0): New constraint.
14443 (Usc): Update constraint.
14445 2017-08-31 Richard Biener <rguenther@suse.de>
14447 PR middle-end/82054
14448 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14449 function only once.
14451 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14453 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14454 Resize type_signature.
14456 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14457 Alan Hayward <alan.hayward@arm.com>
14458 David Sherwood <david.sherwood@arm.com>
14460 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14461 subregs whose inner modes can be stored in GPRs.
14462 (aarch64_classify_index): Likewise.
14464 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14465 Alan Hayward <alan.hayward@arm.com>
14466 David Sherwood <david.sherwood@arm.com>
14468 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14469 (V_INT_EQUIV): ...this.
14470 (v_cmp_result): Rename to...
14471 (v_int_equiv): ...this.
14472 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14473 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14474 (copysign<mode>3): Likewise.
14475 (aarch64_simd_bsl<mode>_internal): Likewise.
14476 (aarch64_simd_bsl<mode>): Likewise.
14477 (vec_cmp<mode><mode>): Likewise.
14478 (vcond<mode><mode>): Likewise.
14479 (vcond<v_cmp_mixed><mode>): Likewise.
14480 (vcondu<mode><v_cmp_mixed>): Likewise.
14481 (aarch64_cm<optab><mode>): Likewise.
14482 (aarch64_cmtst<mode>): Likewise.
14483 (aarch64_fac<optab><mode>): Likewise.
14484 (vec_perm_const<mode>): Likewise.
14485 (vcond_mask_<mode><v_cmp_result>): Rename to...
14486 (vcond_mask_<mode><v_int_equiv>): ...this.
14487 (vec_cmp<mode><v_cmp_result>): Rename to...
14488 (vec_cmp<mode><v_int_equiv>): ...this.
14490 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14491 Alan Hayward <alan.hayward@arm.com>
14492 David Sherwood <david.sherwood@arm.com>
14494 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14496 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14497 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14498 (UNSPEC_LD4_DREG): New unspecs.
14499 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14500 (aarch64_ld2<mode>_dreg_be): Replace with...
14501 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14503 (aarch64_ld3<mode>_dreg_le)
14504 (aarch64_ld3<mode>_dreg_be): Replace with...
14505 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14507 (aarch64_ld4<mode>_dreg_le)
14508 (aarch64_ld4<mode>_dreg_be): Replace with...
14509 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14512 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14514 PR tree-optimization/81987
14515 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14516 insert an initializer in a location not dominated by the stride
14519 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14521 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14522 on the entire header of the finally block in the fallthru case.
14524 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14526 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14528 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14530 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14531 rs6000_emit_move_from_cr and call renamed function.
14532 (rs6000_emit_prologue): Call renamed functions.
14533 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14534 movesi_from_cr, remove volatile CRs.
14536 2017-08-30 Jon Beniston <jon@beniston.com>
14537 Richard Biener <rguenther@suse.de>
14539 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14540 of VECTOR_MODE_P check.
14541 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14542 element vector types.
14544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14546 * df.h (df_read_modify_subreg_p): Remove in favor of...
14547 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14548 const_rtx instead of an rtx.
14549 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14550 * df-problems.c (df_word_lr_mark_ref): Likewise.
14551 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14552 (mark_pseudo_reg_dead): Likewise.
14553 (mark_ref_dead): Likewise.
14554 * reginfo.c (init_subregs_of_mode): Likewise.
14555 * sched-deps.c (sched_analyze_1): Likewise.
14556 * df-scan.c (df_def_record_1): Likewise.
14557 (df_uses_record): Likewise.
14558 (df_read_modify_subreg_p): Remove in favor of...
14559 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14560 const_rtx instead of an rtx.
14562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14563 Alan Hayward <alan.hayward@arm.com>
14564 David Sherwood <david.sherwood@arm.com>
14566 * rtl.h (partial_subreg_p): New function.
14567 * caller-save.c (save_call_clobbered_regs): Use it.
14568 * calls.c (expand_call): Likewise.
14569 * combine.c (combinable_i3pat): Likewise.
14570 (simplify_set): Likewise.
14571 (make_extraction): Likewise.
14572 (make_compound_operation_int): Likewise.
14573 (gen_lowpart_or_truncate): Likewise.
14574 (force_to_mode): Likewise.
14575 (make_field_assignment): Likewise.
14576 (reg_truncated_to_mode): Likewise.
14577 (record_truncated_value): Likewise.
14578 (move_deaths): Likewise.
14579 * cse.c (record_jump_cond): Likewise.
14580 (cse_insn): Likewise.
14581 * cselib.c (cselib_lookup_1): Likewise.
14582 * expmed.c (extract_bit_field_using_extv): Likewise.
14583 * function.c (assign_parm_setup_reg): Likewise.
14584 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14585 * ira-build.c (create_insn_allocnos): Likewise.
14586 * lra-coalesce.c (merge_pseudos): Likewise.
14587 * lra-constraints.c (match_reload): Likewise.
14588 (simplify_operand_subreg): Likewise.
14589 (curr_insn_transform): Likewise.
14590 * lra-lives.c (process_bb_lives): Likewise.
14591 * lra.c (new_insn_reg): Likewise.
14592 (lra_substitute_pseudo): Likewise.
14593 * regcprop.c (mode_change_ok): Likewise.
14594 (maybe_mode_change): Likewise.
14595 (copyprop_hardreg_forward_1): Likewise.
14596 * reload.c (push_reload): Likewise.
14597 (find_reloads): Likewise.
14598 (find_reloads_subreg_address): Likewise.
14599 * reload1.c (alter_reg): Likewise.
14600 (eliminate_regs_1): Likewise.
14601 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14603 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14605 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14606 back to if statements, including unpack.
14608 2017-08-30 Martin Liska <mliska@suse.cz>
14610 PR inline-asm/82001
14611 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14613 (func_checker::compare_asm_inputs_outputs): ... this function.
14614 (func_checker::compare_gimple_asm): Use the function to compare
14615 also ASM constrains.
14616 * ipa-icf-gimple.h: Rename the function.
14618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14619 Alan Hayward <alan.hayward@arm.com>
14620 David Sherwood <david.sherwood@arm.com>
14622 * coretypes.h (complex_mode): New type.
14623 * gdbhooks.py (build_pretty_printer): Handle it.
14624 * machmode.h (complex_mode): New class.
14625 (complex_mode::includes_p): New function.
14626 (is_complex_int_mode): Likewise.
14627 (is_complex_float_mode): Likewise.
14628 * genmodes.c (get_mode_class): Handle complex mode classes.
14629 * function.c (expand_function_end): Use is_complex_int_mode.
14631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14632 Alan Hayward <alan.hayward@arm.com>
14633 David Sherwood <david.sherwood@arm.com>
14635 * coretypes.h (scalar_mode_pod): New typedef.
14636 * gdbhooks.py (build_pretty_printer): Handle it.
14637 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14638 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14639 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14640 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14641 as_a <scalar_mode>.
14643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14644 Alan Hayward <alan.hayward@arm.com>
14645 David Sherwood <david.sherwood@arm.com>
14647 * machmode.h (mode_for_vector): Take a scalar_mode instead
14649 * stor-layout.c (mode_for_vector): Likewise.
14650 * explow.c (promote_mode): Use as_a <scalar_mode>.
14651 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14654 Alan Hayward <alan.hayward@arm.com>
14655 David Sherwood <david.sherwood@arm.com>
14657 * target.def (preferred_simd_mode): Take a scalar_mode
14658 instead of a machine_mode.
14659 * targhooks.h (default_preferred_simd_mode): Likewise.
14660 * targhooks.c (default_preferred_simd_mode): Likewise.
14661 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14662 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14663 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14664 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14665 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14666 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14667 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14668 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14670 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14671 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14672 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14673 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14674 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14675 * doc/tm.texi: Regenerate.
14676 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14680 Alan Hayward <alan.hayward@arm.com>
14681 David Sherwood <david.sherwood@arm.com>
14683 * target.def (scalar_mode_supported_p): Take a scalar_mode
14684 instead of a machine_mode.
14685 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14686 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14687 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14688 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14689 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14690 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14691 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14692 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14693 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14694 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14695 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14696 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14697 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14698 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14700 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14701 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14702 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14703 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14704 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14706 * doc/tm.texi: Regenerate.
14708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14709 Alan Hayward <alan.hayward@arm.com>
14710 David Sherwood <david.sherwood@arm.com>
14712 * coretypes.h (opt_scalar_mode): New typedef.
14713 * gdbhooks.py (build_pretty_printers): Handle it.
14714 * machmode.h (mode_iterator::get_2xwider): Add overload for
14716 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14718 * expr.c (convert_mode_scalar): Likewise.
14719 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14720 * optabs.c (expand_float): Likewise.
14721 (expand_fix): Likewise.
14722 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14725 Alan Hayward <alan.hayward@arm.com>
14726 David Sherwood <david.sherwood@arm.com>
14728 * optabs.c (expand_float): Explicitly check for scalars before
14729 using a branching expansion.
14730 (expand_fix): Likewise.
14732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14733 Alan Hayward <alan.hayward@arm.com>
14734 David Sherwood <david.sherwood@arm.com>
14736 * expr.c (convert_mode): Split scalar handling out into...
14737 (convert_mode_scalar): ...this new function. Treat the modes
14740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14741 Alan Hayward <alan.hayward@arm.com>
14742 David Sherwood <david.sherwood@arm.com>
14744 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14746 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14749 Alan Hayward <alan.hayward@arm.com>
14750 David Sherwood <david.sherwood@arm.com>
14752 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14753 rather than a machine_mode.
14754 (fixed_from_string): Likewise.
14755 (fixed_convert): Likewise.
14756 (fixed_convert_from_int): Likewise.
14757 (fixed_convert_from_real): Likewise.
14758 (real_convert_from_fixed): Likewise.
14759 * fixed-value.c (fixed_from_double_int): Likewise.
14760 (fixed_from_string): Likewise.
14761 (fixed_convert): Likewise.
14762 (fixed_convert_from_int): Likewise.
14763 (fixed_convert_from_real): Likewise.
14764 (real_convert_from_fixed): Likewise.
14765 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14768 Alan Hayward <alan.hayward@arm.com>
14769 David Sherwood <david.sherwood@arm.com>
14771 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14772 of separate mode class checks. Do not allow vector modes here.
14773 (immed_wide_int_const): Use as_a <scalar_mode>.
14774 * explow.c (trunc_int_for_mode): Likewise.
14775 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14776 (wi::shwi): Likewise.
14777 (wi::min_value): Likewise.
14778 (wi::max_value): Likewise.
14779 * dwarf2out.c (loc_descriptor): Likewise.
14780 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14781 for CONST_WIDE_INT.
14783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14784 Alan Hayward <alan.hayward@arm.com>
14785 David Sherwood <david.sherwood@arm.com>
14787 * tree.h (SCALAR_TYPE_MODE): New macro.
14788 * expr.c (expand_expr_addr_expr_1): Use it.
14789 (expand_expr_real_2): Likewise.
14790 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14791 (fold_convert_const_fixed_from_int): Likewise.
14792 (fold_convert_const_fixed_from_real): Likewise.
14793 (native_encode_fixed): Likewise
14794 (native_encode_complex): Likewise
14795 (native_encode_vector): Likewise.
14796 (native_interpret_fixed): Likewise.
14797 (native_interpret_real): Likewise.
14798 (native_interpret_complex): Likewise.
14799 (native_interpret_vector): Likewise.
14800 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14801 (simd_clone_adjust_argument_types): Likewise.
14802 (simd_clone_init_simd_arrays): Likewise.
14803 (simd_clone_adjust): Likewise.
14804 * stor-layout.c (layout_type): Likewise.
14805 * tree.c (build_minus_one_cst): Likewise.
14806 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14807 * tree-inline.c (estimate_move_cost): Likewise.
14808 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14809 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14810 (vectorizable_reduction): Likewise.
14811 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14812 (vect_recog_mixed_size_cond_pattern): Likewise.
14813 (check_bool_pattern): Likewise.
14814 (adjust_bool_pattern): Likewise.
14815 (search_type_for_mask_1): Likewise.
14816 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
14817 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14818 (vectorizable_load): Likewise.
14819 (vectorizable_store): Likewise.
14820 * ubsan.c (ubsan_encode_value): Likewise.
14821 * varasm.c (output_constant): Likewise.
14823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14824 Alan Hayward <alan.hayward@arm.com>
14825 David Sherwood <david.sherwood@arm.com>
14827 * coretypes.h (scalar_mode): New class.
14828 * machmode.h (scalar_mode): Likewise.
14829 (scalar_mode::includes_p): New function.
14830 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
14831 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
14832 * genmodes.c (get_mode_class): Handle remaining scalar modes.
14833 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
14834 * expmed.c (store_bit_field_1): Likewise.
14835 (extract_bit_field_1): Likewise.
14836 * expr.c (write_complex_part): Likewise.
14837 (read_complex_part): Likewise.
14838 (emit_move_complex_push): Likewise.
14839 (expand_expr_real_2): Likewise.
14840 * function.c (assign_parm_setup_reg): Likewise.
14841 (assign_parms_unsplit_complex): Likewise.
14842 * optabs.c (expand_binop): Likewise.
14843 * rtlanal.c (subreg_get_info): Likewise.
14844 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14845 * varasm.c (output_constant_pool_2): Likewise.
14847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14848 Alan Hayward <alan.hayward@arm.com>
14849 David Sherwood <david.sherwood@arm.com>
14851 * expmed.c (extract_high_half): Use scalar_int_mode and remove
14853 (expmed_mult_highpart_optab): Likewise.
14854 (expmed_mult_highpart): Likewise.
14856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14857 Alan Hayward <alan.hayward@arm.com>
14858 David Sherwood <david.sherwood@arm.com>
14860 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
14861 instead of a machine_mode.
14862 (builtin_memset_read_str): Likewise.
14863 * builtins.c (c_readstr): Likewise.
14864 (builtin_memcpy_read_str): Likewise.
14865 (builtin_strncpy_read_str): Likewise.
14866 (builtin_memset_read_str): Likewise.
14867 (builtin_memset_gen_str): Likewise.
14868 (expand_builtin_signbit): Use scalar_int_mode for local variables.
14869 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
14870 instead of a machine_mode.
14871 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
14873 (make_extraction): Likewise.
14874 (try_widen_shift_mode): Take and return scalar_int_modes instead
14876 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
14877 a scalar_int_mode instead of a machine_mode.
14878 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
14879 (avr_addr_space_pointer_mode): Likewise.
14880 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
14881 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
14882 (msp430_unwind_word_mode): Likewise.
14883 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
14884 (spu_addr_space_pointer_mode): Likewise.
14885 (spu_addr_space_address_mode): Likewise.
14886 (spu_libgcc_cmp_return_mode): Likewise.
14887 (spu_libgcc_shift_count_mode): Likewise.
14888 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
14889 (rl78_addr_space_pointer_mode): Likewise.
14890 (fl78_unwind_word_mode): Likewise.
14891 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
14893 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
14894 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
14895 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
14896 (mips_valid_pointer_mode): Likewise.
14897 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
14898 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
14899 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
14901 (ft32_addr_space_address_mode): Likewise.
14902 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
14903 scalar_int_mode instead of a machine_mode.
14904 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
14906 (m32c_addr_space_address_mode): Likewise.
14907 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
14908 (rs6000_eh_return_filter_mode): Likewise.
14909 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
14910 (rs6000_eh_return_filter_mode): Likewise.
14911 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
14912 (s390_libgcc_shift_count_mode): Likewise.
14913 (s390_unwind_word_mode): Likewise.
14914 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
14916 * target.def (mode_rep_extended): Likewise.
14917 (valid_pointer_mode): Likewise.
14918 (addr_space.valid_pointer_mode): Likewise.
14919 (eh_return_filter_mode): Return a scalar_int_mode rather than
14921 (libgcc_cmp_return_mode): Likewise.
14922 (libgcc_shift_count_mode): Likewise.
14923 (unwind_word_mode): Likewise.
14924 (addr_space.pointer_mode): Likewise.
14925 (addr_space.address_mode): Likewise.
14926 * doc/tm.texi: Regenerate.
14927 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
14929 (do_jump): Use scalar_int_mode for local variables.
14930 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
14931 rather than a machine_mode.
14932 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
14933 (scompare_loc_descriptor_wide): Likewise.
14934 (scompare_loc_descriptor_narrow): Likewise.
14935 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
14937 * except.c (sjlj_emit_dispatch_table): Likewise.
14938 (expand_builtin_eh_copy_values): Likewise.
14939 * explow.c (convert_memory_address_addr_space_1): Likewise.
14940 Take a scalar_int_mode rather than a machine_mode.
14941 (convert_memory_address_addr_space): Take a scalar_int_mode rather
14942 than a machine_mode.
14943 (memory_address_addr_space): Use scalar_int_mode for local variables.
14944 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
14945 rather than a machine_mode.
14946 * expmed.c (mask_rtx): Likewise.
14947 (init_expmed_one_conv): Likewise.
14948 (expand_mult_highpart_adjust): Likewise.
14949 (extract_high_half): Likewise.
14950 (expmed_mult_highpart_optab): Likewise.
14951 (expmed_mult_highpart): Likewise.
14952 (expand_smod_pow2): Likewise.
14953 (expand_sdiv_pow2): Likewise.
14954 (emit_store_flag_int): Likewise.
14955 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
14957 (extract_low_bits): Likewise.
14958 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
14960 * expr.c (pieces_addr::adjust): Likewise.
14961 (can_store_by_pieces): Likewise.
14962 (store_by_pieces): Likewise.
14963 (clear_by_pieces_1): Likewise.
14964 (expand_expr_addr_expr_1): Likewise.
14965 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
14966 (expand_expr_real_1): Likewise.
14967 (try_casesi): Likewise.
14968 * final.c (shorten_branches): Likewise.
14969 * fold-const.c (fold_convert_const_int_from_fixed): Change the
14970 type of "mode" to machine_mode.
14971 * internal-fn.c (expand_arith_overflow_result_store): Take a
14972 scalar_int_mode rather than a machine_mode.
14973 (expand_mul_overflow): Use scalar_int_mode for local variables.
14974 * loop-doloop.c (doloop_modify): Likewise.
14975 (doloop_optimize): Likewise.
14976 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
14977 than a machine_mode.
14978 (expand_doubleword_shift_condmove): Likewise.
14979 (expand_doubleword_shift): Likewise.
14980 (expand_doubleword_clz): Likewise.
14981 (expand_doubleword_popcount): Likewise.
14982 (expand_doubleword_parity): Likewise.
14983 (expand_absneg_bit): Use scalar_int_mode for local variables.
14984 (prepare_float_lib_cmp): Likewise.
14985 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
14986 rather than a machine_mode.
14987 (convert_memory_address_addr_space): Likewise.
14988 (get_mode_bounds): Likewise.
14989 (get_address_mode): Return a scalar_int_mode rather than a
14991 * rtlanal.c (get_address_mode): Likewise.
14992 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
14993 than a machine_mode.
14994 * targhooks.c (default_mode_rep_extended): Likewise.
14995 (default_valid_pointer_mode): Likewise.
14996 (default_addr_space_valid_pointer_mode): Likewise.
14997 (default_eh_return_filter_mode): Return a scalar_int_mode rather
14998 than a machine_mode.
14999 (default_libgcc_cmp_return_mode): Likewise.
15000 (default_libgcc_shift_count_mode): Likewise.
15001 (default_unwind_word_mode): Likewise.
15002 (default_addr_space_pointer_mode): Likewise.
15003 (default_addr_space_address_mode): Likewise.
15004 * targhooks.h (default_eh_return_filter_mode): Likewise.
15005 (default_libgcc_cmp_return_mode): Likewise.
15006 (default_libgcc_shift_count_mode): Likewise.
15007 (default_unwind_word_mode): Likewise.
15008 (default_addr_space_pointer_mode): Likewise.
15009 (default_addr_space_address_mode): Likewise.
15010 (default_mode_rep_extended): Take a scalar_int_mode rather than
15012 (default_valid_pointer_mode): Likewise.
15013 (default_addr_space_valid_pointer_mode): Likewise.
15014 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15016 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15017 rather than a machine_mode.
15018 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15019 for local variables.
15020 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15021 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15022 than a machine_mode.
15024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15025 Alan Hayward <alan.hayward@arm.com>
15026 David Sherwood <david.sherwood@arm.com>
15028 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15029 the mode argument to scalar_int_mode.
15030 (do_jump_by_parts_zero_rtx): Likewise.
15031 (do_jump_by_parts_equality_rtx): Likewise.
15032 (do_jump_by_parts_greater): Take a mode argument.
15033 (do_jump_by_parts_equality): Likewise.
15034 (do_jump_1): Update calls accordingly.
15036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15037 Alan Hayward <alan.hayward@arm.com>
15038 David Sherwood <david.sherwood@arm.com>
15040 * is-a.h (safe_dyn_cast): New function.
15041 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15042 (jump_table_for_label): Likewise.
15043 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15044 instead of an rtx_insn *.
15045 (shorten_branches): Use dyn_cast instead of LABEL_P and
15046 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15047 rtx_jump_table_data::get_data_mode.
15048 (final_scan_insn): Likewise.
15050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15051 Alan Hayward <alan.hayward@arm.com>
15052 David Sherwood <david.sherwood@arm.com>
15054 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15055 trying to combine a full-register integer set with a subreg
15058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15059 Alan Hayward <alan.hayward@arm.com>
15060 David Sherwood <david.sherwood@arm.com>
15062 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15063 that is always either address_mode or pointer_mode.
15065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15066 Alan Hayward <alan.hayward@arm.com>
15067 David Sherwood <david.sherwood@arm.com>
15069 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15070 when the two are known to be equal.
15072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15073 Alan Hayward <alan.hayward@arm.com>
15074 David Sherwood <david.sherwood@arm.com>
15076 * simplify-rtx.c (simplify_const_unary_operation): Use
15077 is_a <scalar_int_mode> instead of checking for a nonzero
15078 precision. Forcibly convert op_mode to a scalar_int_mode
15079 in that case. More clearly differentiate the operand and
15080 result modes and use the former when deciding what the value
15081 of a count-bits operation should be. Use is_int_mode instead
15082 of checking for a MODE_INT. Remove redundant check for whether
15083 this mode has a zero precision.
15085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15086 Alan Hayward <alan.hayward@arm.com>
15087 David Sherwood <david.sherwood@arm.com>
15089 * optabs.c (widen_leading): Change the type of the mode argument
15090 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15091 (widen_bswap): Likewise.
15092 (expand_parity): Likewise.
15093 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15094 (expand_ffs): Likewise.
15095 (epand_unop): Check for scalar integer modes before calling the
15098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15099 Alan Hayward <alan.hayward@arm.com>
15100 David Sherwood <david.sherwood@arm.com>
15102 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15104 (expand_expr_real_1): Update call accordingly.
15106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15107 Alan Hayward <alan.hayward@arm.com>
15108 David Sherwood <david.sherwood@arm.com>
15110 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15111 value_mode arguments. Use scalar_int_mode internally.
15112 (store_bit_field_1): Rename the new integer mode from imode
15113 to op0_mode and use it instead of GET_MODE (op0). Update calls
15114 to store_split_bit_field, store_bit_field_using_insv and
15115 store_fixed_bit_field.
15116 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15117 Use scalar_int_mode internally. Use a bit count rather than a mode
15118 when calculating the largest bit size for get_best_mode.
15119 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15120 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15121 Remove assertion that OP0 has a scalar integer mode.
15122 (store_split_bit_field): Add op0_mode and value_mode arguments.
15123 Update calls to extract_fixed_bit_field.
15124 (extract_bit_field_using_extv): Add an op0_mode argument.
15125 Use scalar_int_mode internally.
15126 (extract_bit_field_1): Rename the new integer mode from imode to
15127 op0_mode and use it instead of GET_MODE (op0). Update calls to
15128 extract_split_bit_field, extract_bit_field_using_extv and
15129 extract_fixed_bit_field.
15130 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15131 to extract_split_bit_field and extract_fixed_bit_field_1.
15132 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15133 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15134 on the target mode.
15135 (extract_split_bit_field): Add an op0_mode argument. Update call
15136 to extract_fixed_bit_field.
15138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15139 Alan Hayward <alan.hayward@arm.com>
15140 David Sherwood <david.sherwood@arm.com>
15142 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15143 * explow.c (hard_function_value): Likewise.
15144 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15145 convert_to_mode call outside the loop.
15146 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15147 for the mode iterator. Require the mode specified by max_pieces
15149 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15151 (copy_blkmode_to_reg): Likewise.
15152 (set_storage_via_setmem): Likewise.
15153 * optabs.c (prepare_cmp_insn): Likewise.
15154 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15155 * stor-layout.c (finish_bitfield_representative): Likewise.
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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15162 * expr.c (convert_move): Use them.
15163 (convert_modes): Likewise.
15164 (store_expr_with_bounds): Likewise.
15166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15167 Alan Hayward <alan.hayward@arm.com>
15168 David Sherwood <david.sherwood@arm.com>
15170 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15171 parameter for the mode of "x". Remove the "known_x", "known_mode"
15172 and "known_ret" arguments. Change the type of the mode argument
15173 to scalar_int_mode.
15174 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15175 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15176 (reg_num_sign_bit_copies_for_combine): Likewise.
15177 * rtlanal.c (nonzero_bits1): Likewise.
15178 (num_sign_bit_copies1): Likewise.
15179 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15180 (reg_num_sign_bit_copies_general): Likewise.
15181 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15182 (reg_nonzero_bits_general): Likewise.
15184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15185 Alan Hayward <alan.hayward@arm.com>
15186 David Sherwood <david.sherwood@arm.com>
15188 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15189 than in subroutines. Return 1 for non-integer modes.
15190 (cached_num_sign_bit_copies): Change the type of the mode parameter
15191 to scalar_int_mode.
15192 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15193 classes. Handle CONST_INT_P first and then check whether X also
15194 has a scalar integer mode. Check the same thing for inner registers
15195 of a SUBREG and for values that are being extended or truncated.
15197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15198 Alan Hayward <alan.hayward@arm.com>
15199 David Sherwood <david.sherwood@arm.com>
15201 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15202 in subroutines. Return the mode mask for non-integer modes.
15203 (cached_nonzero_bits): Change the type of the mode parameter
15204 to scalar_int_mode.
15205 (nonzero_bits1): Likewise. Remove early exit for other mode
15206 classes. Handle CONST_INT_P first and then check whether X
15207 also has a scalar integer mode.
15209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15210 Alan Hayward <alan.hayward@arm.com>
15211 David Sherwood <david.sherwood@arm.com>
15213 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15214 Return a scalar_int_mode and assert that the size is greater than
15216 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15217 (op_by_pieces_d::op_by_pieces_d): Likewise.
15218 (op_by_pieces_d::run): Likewise.
15219 (can_store_by_pieces): Likewise.
15221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15222 Alan Hayward <alan.hayward@arm.com>
15223 David Sherwood <david.sherwood@arm.com>
15225 * combine.c (extract_left_shift): Add a mode argument and update
15227 (make_compound_operation_int): Change the type of the mode parameter
15228 to scalar_int_mode and update the call to extract_left_shift.
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 * combine.c (simplify_and_const_int): Change the type of the mode
15235 parameter to scalar_int_mode.
15236 (simplify_and_const_int_1): Likewise. Update recursive call.
15238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15239 Alan Hayward <alan.hayward@arm.com>
15240 David Sherwood <david.sherwood@arm.com>
15242 * combine.c (simplify_compare_const): Check that the mode is a
15243 scalar_int_mode (rather than VOIDmode) before testing its
15245 (simplify_comparison): Move COMPARISON_P handling out of the
15246 loop and restrict the latter part of the loop to scalar_int_modes.
15247 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15248 and when considering SUBREG_REGs. Use is_int_mode instead of
15249 checking GET_MODE_CLASS against MODE_INT.
15251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15252 Alan Hayward <alan.hayward@arm.com>
15253 David Sherwood <david.sherwood@arm.com>
15255 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15256 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15257 shift_unit_mode and for modes involved in scalar shifts.
15259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15260 Alan Hayward <alan.hayward@arm.com>
15261 David Sherwood <david.sherwood@arm.com>
15263 * combine.c (force_int_to_mode): New function, split out from...
15264 (force_to_mode): ...here. Keep xmode up-to-date and use it
15265 instead of GET_MODE (x).
15267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15268 Alan Hayward <alan.hayward@arm.com>
15269 David Sherwood <david.sherwood@arm.com>
15271 * optabs-query.h (extraction_insn::struct_mode): Change type to
15272 opt_scalar_int_mode and update comment.
15273 (extraction_insn::field_mode): Change type to scalar_int_mode.
15274 (extraction_insn::pos_mode): Likewise.
15275 * combine.c (make_extraction): Update accordingly.
15276 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15277 (get_optab_extraction_insn): Likewise.
15278 * recog.c (simplify_while_replacing): Likewise.
15279 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15280 parameter to opt_scalar_int_mode.
15282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15283 Alan Hayward <alan.hayward@arm.com>
15284 David Sherwood <david.sherwood@arm.com>
15286 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15287 to a scalar_int_mode instead of a machine_mode.
15288 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15289 (get_best_mode): Return a boolean and use a pointer argument to store
15290 the selected mode. Replace the limit mode parameter with a bit limit.
15291 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15292 for the values returned by bit_field_mode_iterator::next_mode.
15293 (store_bit_field): Update call to get_best_mode.
15294 (store_fixed_bit_field): Likewise.
15295 (extract_fixed_bit_field): Likewise.
15296 * expr.c (optimize_bitfield_assignment_op): Likewise.
15297 * fold-const.c (optimize_bit_field_compare): Likewise.
15298 (fold_truth_andor_1): Likewise.
15299 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15300 Update for new type of m_mode.
15301 (get_best_mode): As above.
15303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15304 Alan Hayward <alan.hayward@arm.com>
15305 David Sherwood <david.sherwood@arm.com>
15307 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15308 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15309 (store_bit_field): Check is_a <scalar_int_mode> before calling
15310 strict_volatile_bitfield_p.
15311 (extract_bit_field): 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 * target.def (cstore_mode): Return a scalar_int_mode.
15318 * doc/tm.texi: Regenerate.
15319 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15320 * targhooks.h (default_cstore_mode): Likewise.
15321 * targhooks.c (default_cstore_mode): Likewise, using a forced
15323 * expmed.c (emit_cstore): Expect the target of the cstore to be
15326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15327 Alan Hayward <alan.hayward@arm.com>
15328 David Sherwood <david.sherwood@arm.com>
15330 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15332 (niter_desc): Likewise mode.
15333 (iv_analyze): Add a mode parameter.
15335 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15336 and change its type to scalar_int_mode.
15337 * loop-iv.c: Update commentary at head of file.
15338 (iv_constant): Pass the mode paraeter before the rtx it describes
15339 and change its type to scalar_int_mode. Remove VOIDmode handling.
15340 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15341 (iv_extend): Likewise.
15342 (shorten_into_mode): Likewise.
15343 (iv_add): Use scalar_int_mode.
15344 (iv_mult): Likewise.
15345 (iv_shift): Likewise.
15346 (canonicalize_iv_subregs): Likewise.
15347 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15348 it describes and change its mode to scalar_int_mode. Also change
15349 the type of the returned inner_mode to scalar_int_mode.
15350 (get_biv_step): Likewise, turning outer_mode from a pointer
15351 into a direct parameter. Update call to get_biv_step_1.
15352 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15353 iv_constant and get_biv_step.
15354 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15355 and change its type to scalar_int_mode. Don't initialise iv->mode
15356 to VOIDmode and remove later checks for its still being VOIDmode.
15357 Update calls to iv_analyze_op and iv_analyze_expr. Check
15358 is_a <scalar_int_mode> when changing the mode under consideration.
15359 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15360 Update call to iv_analyze_expr.
15361 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15362 inner register is not also a scalar_int_mode. Update call to
15364 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15365 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15366 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15367 separate mode class checks. Update calls to iv_analyze. Remove
15368 fix-up of VOIDmodes after iv_analyze_biv.
15369 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15370 don't have a scalar_int_mode. Update call to biv_p.
15372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15373 Alan Hayward <alan.hayward@arm.com>
15374 David Sherwood <david.sherwood@arm.com>
15376 * cfgexpand.c (convert_debug_memory_address): Use
15377 as_a <scalar_int_mode>.
15378 * combine.c (expand_compound_operation): Likewise.
15379 (make_extraction): Likewise.
15380 (change_zero_ext): Likewise.
15381 (simplify_comparison): Likewise.
15382 * cse.c (cse_insn): Likewise.
15383 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15384 (mem_loc_descriptor): Likewise.
15385 (loc_descriptor): Likewise.
15386 * expmed.c (init_expmed_one_mode): Likewise.
15387 (synth_mult): Likewise.
15388 (emit_store_flag_1): Likewise.
15389 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15390 of a comparison with size.
15391 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15392 (reduce_to_bit_field_precision): Likewise.
15393 * function.c (expand_function_end): Likewise.
15394 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15395 * loop-doloop.c (doloop_modify): Likewise.
15396 * optabs.c (expand_binop): Likewise.
15397 (expand_unop): Likewise.
15398 (expand_copysign_absneg): Likewise.
15399 (prepare_cmp_insn): Likewise.
15400 (maybe_legitimize_operand): Likewise.
15401 * recog.c (const_scalar_int_operand): Likewise.
15402 * rtlanal.c (get_address_mode): Likewise.
15403 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15404 (simplify_cond_clz_ctz): Likewise.
15405 * tree-nested.c (get_nl_goto_field): Likewise.
15406 * tree.c (build_vector_type_for_mode): Likewise.
15407 * var-tracking.c (use_narrower_mode): Likewise.
15409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15410 Alan Hayward <alan.hayward@arm.com>
15411 David Sherwood <david.sherwood@arm.com>
15413 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15414 * builtins.c (expand_builtin_signbit): Use it.
15415 * cfgexpand.c (expand_debug_expr): Likewise.
15416 * dojump.c (do_jump): Likewise.
15417 (do_compare_and_jump): Likewise.
15418 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15419 * expmed.c (make_tree): Likewise.
15420 * expr.c (expand_expr_real_2): Likewise.
15421 (expand_expr_real_1): Likewise.
15422 (try_casesi): Likewise.
15423 * fold-const-call.c (fold_const_call_ss): Likewise.
15424 * fold-const.c (unextend): Likewise.
15425 (extract_muldiv_1): Likewise.
15426 (fold_single_bit_test): Likewise.
15427 (native_encode_int): Likewise.
15428 (native_encode_string): Likewise.
15429 (native_interpret_int): Likewise.
15430 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15431 * internal-fn.c (expand_addsub_overflow): Likewise.
15432 (expand_neg_overflow): Likewise.
15433 (expand_mul_overflow): Likewise.
15434 (expand_arith_overflow): Likewise.
15435 * match.pd: Likewise.
15436 * stor-layout.c (layout_type): Likewise.
15437 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15438 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15439 * tree-ssanames.c (get_range_info): Likewise.
15440 * tree-switch-conversion.c (array_value_type) Likewise.
15441 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15442 (vect_recog_divmod_pattern): Likewise.
15443 (vect_recog_mixed_size_cond_pattern): Likewise.
15444 * tree-vrp.c (extract_range_basic): Likewise.
15445 (simplify_float_conversion_using_ranges): Likewise.
15446 * tree.c (int_fits_type_p): Likewise.
15447 * ubsan.c (instrument_bool_enum_load): Likewise.
15448 * varasm.c (mergeable_string_section): Likewise.
15449 (narrowing_initializer_constant_valid_p): Likewise.
15450 (output_constant): Likewise.
15452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15453 Alan Hayward <alan.hayward@arm.com>
15454 David Sherwood <david.sherwood@arm.com>
15456 * machmode.h (NARROWEST_INT_MODE): New macro.
15457 * expr.c (alignment_for_piecewise_move): Use it instead of
15458 GET_CLASS_NARROWEST_MODE (MODE_INT).
15459 (push_block): Likewise.
15460 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15462 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * postreload.c (move2add_valid_value_p): Change the type of the
15469 mode parameter to scalar_int_mode.
15470 (move2add_use_add2_insn): Add a mode parameter and use it instead
15472 (move2add_use_add3_insn): Likewise.
15473 (reload_cse_move2add): Update accordingly.
15475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15476 Alan Hayward <alan.hayward@arm.com>
15477 David Sherwood <david.sherwood@arm.com>
15479 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15481 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15482 * optabs.c (expand_unop): Likewise.
15484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15485 Alan Hayward <alan.hayward@arm.com>
15486 David Sherwood <david.sherwood@arm.com>
15488 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15489 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15490 (popcount_loc_descriptor): Likewise.
15491 (bswap_loc_descriptor): Likewise.
15492 (rotate_loc_descriptor): Likewise.
15493 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15494 calling the functions above.
15496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15497 Alan Hayward <alan.hayward@arm.com>
15498 David Sherwood <david.sherwood@arm.com>
15500 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15502 (try_combine): Likewise.
15503 (simplify_if_then_else): Likewise.
15504 * cse.c (cse_insn): Likewise.
15505 * dwarf2out.c (mem_loc_descriptor): Likewise.
15506 * emit-rtl.c (gen_lowpart_common): Likewise.
15507 * simplify-rtx.c (simplify_truncation): Likewise.
15508 (simplify_binary_operation_1): Likewise.
15509 (simplify_const_relational_operation): Likewise.
15510 (simplify_ternary_operation): Likewise.
15511 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15513 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15514 Alan Hayward <alan.hayward@arm.com>
15515 David Sherwood <david.sherwood@arm.com>
15517 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15518 * reload.c (push_reload): Likewise.
15519 (find_reloads): Likewise.
15521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15522 Alan Hayward <alan.hayward@arm.com>
15523 David Sherwood <david.sherwood@arm.com>
15525 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15526 (make_compound_operation_int): Likewise.
15527 (change_zero_ext): Likewise.
15528 * expr.c (convert_move): Likewise.
15529 (convert_modes): Likewise.
15530 * fwprop.c (forward_propagate_subreg): Likewise.
15531 * loop-iv.c (get_biv_step_1): Likewise.
15532 * optabs.c (widen_operand): Likewise.
15533 * postreload.c (move2add_valid_value_p): Likewise.
15534 * recog.c (simplify_while_replacing): Likewise.
15535 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15536 (simplify_binary_operation_1): Likewise. Remove redundant
15537 mode equality check.
15539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15540 Alan Hayward <alan.hayward@arm.com>
15541 David Sherwood <david.sherwood@arm.com>
15543 * combine.c (combine_simplify_rtx): Add checks for
15544 is_a <scalar_int_mode>.
15545 (simplify_if_then_else): Likewise.
15546 (make_field_assignment): Likewise.
15547 (simplify_comparison): Likewise.
15548 * ifcvt.c (noce_try_bitop): Likewise.
15549 * loop-invariant.c (canonicalize_address_mult): Likewise.
15550 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15553 Alan Hayward <alan.hayward@arm.com>
15554 David Sherwood <david.sherwood@arm.com>
15556 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15557 is_a <scalar_int_mode> instead of != BLKmode.
15559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15560 Alan Hayward <alan.hayward@arm.com>
15561 David Sherwood <david.sherwood@arm.com>
15563 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15564 instead of != VOIDmode.
15565 * combine.c (if_then_else_cond): Likewise.
15566 (change_zero_ext): Likewise.
15567 * dwarf2out.c (mem_loc_descriptor): Likewise.
15568 (loc_descriptor): Likewise.
15569 * rtlanal.c (canonicalize_condition): Likewise.
15570 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15573 Alan Hayward <alan.hayward@arm.com>
15574 David Sherwood <david.sherwood@arm.com>
15576 * simplify-rtx.c (simplify_binary_operation_1): Use
15577 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15580 Alan Hayward <alan.hayward@arm.com>
15581 David Sherwood <david.sherwood@arm.com>
15583 * wide-int.h (int_traits<unsigned char>) New class.
15584 (int_traits<unsigned short>) Likewise.
15585 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15586 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15588 * combine.c (set_nonzero_bits_and_sign_copies): Use
15589 is_a <scalar_int_mode>.
15590 (find_split_point): Likewise.
15591 (combine_simplify_rtx): Likewise.
15592 (simplify_logical): Likewise.
15593 (expand_compound_operation): Likewise.
15594 (expand_field_assignment): Likewise.
15595 (make_compound_operation): Likewise.
15596 (extended_count): Likewise.
15597 (change_zero_ext): Likewise.
15598 (simplify_comparison): Likewise.
15599 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15600 (ucompare_loc_descriptor): Likewise.
15601 (minmax_loc_descriptor): Likewise.
15602 (mem_loc_descriptor): Likewise.
15603 (loc_descriptor): Likewise.
15604 * expmed.c (init_expmed_one_mode): Likewise.
15605 * lra-constraints.c (lra_constraint_offset): Likewise.
15606 * optabs.c (prepare_libcall_arg): Likewise.
15607 * postreload.c (move2add_note_store): Likewise.
15608 * reload.c (operands_match_p): Likewise.
15609 * rtl.h (load_extend_op): Likewise.
15610 * rtlhooks.c (gen_lowpart_general): Likewise.
15611 * simplify-rtx.c (simplify_truncation): Likewise.
15612 (simplify_unary_operation_1): Likewise.
15613 (simplify_binary_operation_1): Likewise.
15614 (simplify_const_binary_operation): Likewise.
15615 (simplify_const_relational_operation): Likewise.
15616 (simplify_subreg): Likewise.
15617 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15618 * var-tracking.c (adjust_mems): Likewise.
15619 (prepare_call_arguments): Likewise.
15621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15622 Alan Hayward <alan.hayward@arm.com>
15623 David Sherwood <david.sherwood@arm.com>
15625 * machmode.h (is_int_mode): New fuction.
15626 * combine.c (find_split_point): Use it.
15627 (combine_simplify_rtx): Likewise.
15628 (simplify_if_then_else): Likewise.
15629 (simplify_set): Likewise.
15630 (simplify_shift_const_1): Likewise.
15631 (simplify_comparison): Likewise.
15632 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15633 * cse.c (notreg_cost): Likewise.
15634 (cse_insn): Likewise.
15635 * cselib.c (cselib_lookup_1): Likewise.
15636 * dojump.c (do_jump_1): Likewise.
15637 (do_compare_rtx_and_jump): Likewise.
15638 * dse.c (get_call_args): Likewise.
15639 * dwarf2out.c (rtl_for_decl_init): Likewise.
15640 (native_encode_initializer): Likewise.
15641 * expmed.c (emit_store_flag_1): Likewise.
15642 (emit_store_flag): Likewise.
15643 * expr.c (convert_modes): Likewise.
15644 (store_field): Likewise.
15645 (expand_expr_real_1): Likewise.
15646 * fold-const.c (fold_read_from_constant_string): Likewise.
15647 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15648 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15649 * optabs.c (expand_binop): Likewise.
15650 (expand_unop): Likewise.
15651 (expand_abs_nojump): Likewise.
15652 (expand_one_cmpl_abs_nojump): Likewise.
15653 * simplify-rtx.c (mode_signbit_p): Likewise.
15654 (val_signbit_p): Likewise.
15655 (val_signbit_known_set_p): Likewise.
15656 (val_signbit_known_clear_p): Likewise.
15657 (simplify_relational_operation_1): Likewise.
15658 * tree.c (vector_type_mode): Likewise.
15660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15661 Alan Hayward <alan.hayward@arm.com>
15662 David Sherwood <david.sherwood@arm.com>
15664 * machmode.h (smallest_mode_for_size): Fix formatting.
15665 (smallest_int_mode_for_size): New function.
15666 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15667 instead of smallest_mode_for_size.
15668 * combine.c (make_extraction): Likewise.
15669 * config/arc/arc.c (arc_expand_movmem): Likewise.
15670 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15671 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15672 * config/s390/s390.c (s390_expand_insv): Likewise.
15673 * config/sparc/sparc.c (assign_int_registers): Likewise.
15674 * config/spu/spu.c (spu_function_value): Likewise.
15675 (spu_function_arg): Likewise.
15676 * coverage.c (get_gcov_type): Likewise.
15677 (get_gcov_unsigned_t): Likewise.
15678 * dse.c (find_shift_sequence): Likewise.
15679 * expmed.c (store_bit_field_1): Likewise.
15680 * expr.c (convert_move): Likewise.
15681 (store_field): Likewise.
15682 * internal-fn.c (expand_arith_overflow): Likewise.
15683 * optabs-query.c (get_best_extraction_insn): Likewise.
15684 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15685 * stor-layout.c (layout_type): Likewise.
15686 (initialize_sizetypes): Likewise.
15687 * targhooks.c (default_get_mask_mode): Likewise.
15688 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15691 Alan Hayward <alan.hayward@arm.com>
15692 David Sherwood <david.sherwood@arm.com>
15694 * machmode.h (opt_mode::else_blk): New function.
15695 (int_mode_for_mode): Declare.
15696 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15697 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15699 * cfgexpand.c (expand_debug_expr): Likewise.
15700 * combine.c (gen_lowpart_or_truncate): Likewise.
15701 (gen_lowpart_for_combine): Likewise.
15702 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15703 * config/avr/avr.c (avr_to_int_mode): Likewise.
15704 (avr_out_plus_1): Likewise.
15705 (avr_out_plus): Likewise.
15706 (avr_out_round): Likewise.
15707 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15708 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15709 (s390_expand_vcond): Likewise.
15710 * config/spu/spu.c (spu_split_immediate): Likewise.
15711 (spu_expand_mov): Likewise.
15712 * dse.c (get_stored_val): Likewise.
15713 * expmed.c (store_bit_field_1): Likewise.
15714 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15716 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15717 (extract_low_bits): Likewise.
15718 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15719 handling rather than repeating the check.
15720 (emit_group_store): Likewise.
15721 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15722 * optabs.c (expand_absneg_bit): Likewise.
15723 (expand_copysign_absneg): Likewise.
15724 (expand_copysign_bit): Likewise.
15725 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15726 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15727 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15728 * var-tracking.c (prepare_call_arguments): Likewise.
15729 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15730 int_mode_for_mode instead of mode_for_size.
15731 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15734 Alan Hayward <alan.hayward@arm.com>
15735 David Sherwood <david.sherwood@arm.com>
15737 * machmode.h (int_mode_for_size): New function.
15738 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15739 instead of mode_for_size.
15740 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15742 * combine.c (expand_field_assignment): Use int_mode_for_size
15743 instead of mode_for_size.
15744 (make_extraction): Likewise.
15745 (simplify_shift_const_1): Likewise.
15746 (simplify_comparison): Likewise.
15747 * dojump.c (do_jump): Likewise.
15748 * dwarf2out.c (mem_loc_descriptor): Likewise.
15749 * emit-rtl.c (init_derived_machine_modes): Likewise.
15750 * expmed.c (flip_storage_order): Likewise.
15751 (convert_extracted_bit_field): Likewise.
15752 * expr.c (copy_blkmode_from_reg): Likewise.
15753 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15754 * internal-fn.c (expand_mul_overflow): Likewise.
15755 * lower-subreg.c (simple_move): Likewise.
15756 * optabs-libfuncs.c (init_optabs): Likewise.
15757 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15758 * tree.c (vector_type_mode): Likewise.
15759 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15760 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15761 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15762 * tree-vect-stmts.c (vectorizable_load): Likewise.
15763 (vectorizable_store): Likewise.
15765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15766 Alan Hayward <alan.hayward@arm.com>
15767 David Sherwood <david.sherwood@arm.com>
15769 * coretypes.h (pod_mode): New type.
15770 (scalar_int_mode_pod): New typedef.
15771 * machmode.h (pod_mode): New class.
15772 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15773 * genmodes.c (emit_mode_int_n): Update accordingly.
15774 * lower-subreg.h (target_lower_subreg): Change type to
15775 scalar_int_mode_pod.
15776 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15777 scalar_int_mode_pod.
15779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15780 Alan Hayward <alan.hayward@arm.com>
15781 David Sherwood <david.sherwood@arm.com>
15783 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15784 machine_mode to scalar_int_mode.
15785 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15786 (rs6000_option_override_internal): Remove cast to int.
15787 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15788 machine_mode to scalar_int_mode.
15789 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15790 (rs6000_option_override_internal): Remove cast to int.
15791 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15792 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15794 * config/s390/s390.c (s390_expand_builtin): Likewise.
15795 * coretypes.h (scalar_int_mode): New type.
15796 (opt_scalar_int_mode): New typedef.
15797 * machmode.h (scalar_int_mode): New class.
15798 (scalar_int_mode::includes_p): New function.
15799 (byte_mode): Change type to scalar_int_mode.
15800 (word_mode): Likewise.
15801 (ptr_mode): Likewise.
15802 * emit-rtl.c (byte_mode): Likewise.
15803 (word_mode): Likewise.
15804 (ptr_mode): Likewise.
15805 (init_derived_machine_modes): Update accordingly.
15806 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15807 and MODE_PARTIAL_INT.
15808 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15809 opt_scalar_int_mode.
15811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15812 Alan Hayward <alan.hayward@arm.com>
15813 David Sherwood <david.sherwood@arm.com>
15815 * target.def (libgcc_floating_mode_supported_p): Take a
15817 * doc/tm.texi: Regenerate.
15818 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
15820 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15821 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
15824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15825 Alan Hayward <alan.hayward@arm.com>
15826 David Sherwood <david.sherwood@arm.com>
15828 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
15829 * doc/tm.texi: Regenerate.
15830 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
15831 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
15832 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
15833 * targhooks.h (default_floatn_mode): Likewise.
15834 * targhooks.c (default_floatn_mode): Likewise.
15835 * tree.c (build_common_tree_nodes): Update accordingly.
15837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15838 Alan Hayward <alan.hayward@arm.com>
15839 David Sherwood <david.sherwood@arm.com>
15841 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
15842 (mode_iterator::iterate_p): Likewise.
15843 (mode_iterator::get_wider): Likewise.
15844 * expr.c (init_expr_target): Use opt_scalar_float_mode.
15846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15847 Alan Hayward <alan.hayward@arm.com>
15848 David Sherwood <david.sherwood@arm.com>
15850 * coretypes.h (opt_scalar_float_mode): New typedef.
15851 * machmode.h (float_mode_for_size): New function.
15852 * emit-rtl.c (double_mode): Delete.
15853 (init_emit_once): Use float_mode_for_size.
15854 * stor-layout.c (layout_type): Likewise.
15855 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
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 * output.h (assemble_real): Take a scalar_float_mode.
15862 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
15863 * config/arm/arm.md (consttable_4): Likewise.
15864 (consttable_8): Likewise.
15865 (consttable_16): Likewise.
15866 * config/mips/mips.md (consttable_float): Likewise.
15867 * config/s390/s390.c (s390_output_pool_entry): Likewise.
15868 * varasm.c (assemble_real): Take a scalar_float_mode.
15869 (output_constant_pool_2): Update accordingly.
15870 (output_constant): Likewise.
15872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15873 Alan Hayward <alan.hayward@arm.com>
15874 David Sherwood <david.sherwood@arm.com>
15876 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
15877 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
15878 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
15879 (native_encode_real): Likewise.
15880 (native_interpret_real): Likewise.
15881 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
15882 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15885 Alan Hayward <alan.hayward@arm.com>
15886 David Sherwood <david.sherwood@arm.com>
15888 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
15889 <scalar_float_mode>. Simplify.
15890 (gen_extend_conv_libfunc): Likewise.
15892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15893 Alan Hayward <alan.hayward@arm.com>
15894 David Sherwood <david.sherwood@arm.com>
15896 * coretypes.h (scalar_float_mode): New type.
15897 * machmode.h (mode_traits::from_int): Use machine_mode if
15898 USE_ENUM_MODES is defined.
15899 (is_a): New function.
15901 (dyn_cast): Likewise.
15902 (scalar_float_mode): New class.
15903 (scalar_float_mode::includes_p): New function.
15904 (is_float_mode): Likewise.
15905 * gdbhooks.py (MachineModePrinter): New class.
15906 (build_pretty_printer): Use it for scalar_float_mode.
15907 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
15908 (format_helper::format_helper): Turn into a template.
15909 * genmodes.c (get_mode_class): New function.
15910 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
15911 or machine_mode if none.
15912 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
15913 as_a <scalar_float_mode>.
15914 * dwarf2out.c (mem_loc_descriptor): Likewise.
15915 (insert_float): Likewise.
15916 (add_const_value_attribute): Likewise.
15917 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15918 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
15919 (expand_unop): Update accordingly.
15920 (expand_abs_nojump): Likewise.
15921 (expand_copysign_absneg): Take a scalar_float_mode.
15922 (expand_copysign_bit): Likewise.
15923 (expand_copysign): Update accordingly.
15925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15926 Alan Hayward <alan.hayward@arm.com>
15927 David Sherwood <david.sherwood@arm.com>
15929 * coretypes.h (opt_mode): New class.
15930 * machmode.h (opt_mode): Likewise.
15931 (opt_mode::else_void): New function.
15932 (opt_mode::require): Likewise.
15933 (opt_mode::exists): Likewise.
15934 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
15935 (GET_MODE_2XWIDER_MODE): Likewise.
15936 (mode_iterator::get_wider): Update accordingly.
15937 (mode_iterator::get_2xwider): Likewise.
15938 (mode_iterator::get_known_wider): Likewise, turning into a template.
15939 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
15940 forcing a wider mode to exist.
15941 * config/cr16/cr16.h (LONG_REG_P): Likewise.
15942 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15943 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
15944 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
15945 * lower-subreg.c (init_lower_subreg): Likewise.
15946 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
15947 on the final iteration.
15948 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
15949 a wider mode exists before asking for a move pattern.
15950 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
15951 forcing a wider mode to exist.
15952 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
15953 returning false if no such mode exists.
15954 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
15955 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
15956 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
15957 Avoid checking for a MODE_INT if we already know the mode is not a
15959 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
15960 forcing a wider mode to exist.
15961 (expmed_mult_highpart_optab): Likewise.
15962 (expmed_mult_highpart): Likewise.
15963 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
15965 * lto-streamer-in.c (lto_input_mode_table): Likewise.
15966 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
15967 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15968 * internal-fn.c (expand_mul_overflow): Update use of
15969 GET_MODE_2XWIDER_MODE.
15970 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15971 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
15972 GET_MODE_WIDER_MODE.
15973 (convert_plusminus_to_widen): Likewise.
15974 * tree-switch-conversion.c (array_value_type): Likewise.
15975 * var-tracking.c (emit_note_insn_var_location): Likewise.
15976 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15977 Return false inside rather than outside the loop if no wider mode
15979 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
15980 and GET_MODE_2XWIDER_MODE
15981 (can_compare_p): Use else_void.
15982 * gdbhooks.py (OptMachineModePrinter): New class.
15983 (build_pretty_printer): Use it for opt_mode.
15985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15986 Alan Hayward <alan.hayward@arm.com>
15987 David Sherwood <david.sherwood@arm.com>
15989 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
15990 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
15992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15993 Alan Hayward <alan.hayward@arm.com>
15994 David Sherwood <david.sherwood@arm.com>
15996 * machmode.h (mode_traits): New structure.
15997 (get_narrowest_mode): New function.
15998 (mode_iterator::start): Likewise.
15999 (mode_iterator::iterate_p): Likewise.
16000 (mode_iterator::get_wider): Likewise.
16001 (mode_iterator::get_known_wider): Likewise.
16002 (mode_iterator::get_2xwider): Likewise.
16003 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16004 (FOR_EACH_MODE): Likewise.
16005 (FOR_EACH_MODE_FROM): Likewise.
16006 (FOR_EACH_MODE_UNTIL): Likewise.
16007 (FOR_EACH_WIDER_MODE): Likewise.
16008 (FOR_EACH_2XWIDER_MODE): Likewise.
16009 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16010 * combine.c (simplify_comparison): Likewise
16011 * config/i386/i386.c (type_natural_mode): Likewise.
16012 * cse.c (cse_insn): Likewise.
16013 * dse.c (find_shift_sequence): Likewise.
16014 * emit-rtl.c (init_derived_machine_modes): Likewise.
16015 (init_emit_once): Likewise.
16016 * explow.c (hard_function_value): Likewise.
16017 * expmed.c (extract_fixed_bit_field_1): Likewise.
16018 (extract_bit_field_1): Likewise.
16019 (expand_divmod): Likewise.
16020 (emit_store_flag_1): Likewise.
16021 * expr.c (init_expr_target): Likewise.
16022 (convert_move): Likewise.
16023 (alignment_for_piecewise_move): Likewise.
16024 (widest_int_mode_for_size): Likewise.
16025 (emit_block_move_via_movmem): Likewise.
16026 (copy_blkmode_to_reg): Likewise.
16027 (set_storage_via_setmem): Likewise.
16028 (compress_float_constant): Likewise.
16029 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16030 * optabs-query.c (get_best_extraction_insn): Likewise.
16031 * optabs.c (expand_binop): Likewise.
16032 (expand_twoval_unop): Likewise.
16033 (expand_twoval_binop): Likewise.
16034 (widen_leading): Likewise.
16035 (widen_bswap): Likewise.
16036 (expand_parity): Likewise.
16037 (expand_unop): Likewise.
16038 (prepare_cmp_insn): Likewise.
16039 (prepare_float_lib_cmp): Likewise.
16040 (expand_float): Likewise.
16041 (expand_fix): Likewise.
16042 (expand_sfix_optab): Likewise.
16043 * postreload.c (move2add_use_add2_insn): Likewise.
16044 * reg-stack.c (reg_to_stack): Likewise.
16045 * reginfo.c (choose_hard_reg_mode): Likewise.
16046 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16047 * stor-layout.c (mode_for_size): Likewise.
16048 (smallest_mode_for_size): Likewise.
16049 (mode_for_vector): Likewise.
16050 (finish_bitfield_representative): Likewise.
16051 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16052 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16053 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16054 * var-tracking.c (prepare_call_arguments): Likewise.
16056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16057 Alan Hayward <alan.hayward@arm.com>
16058 David Sherwood <david.sherwood@arm.com>
16060 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16061 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16062 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16063 * machmode.h (mode_size): Move earlier in file.
16064 (mode_precision): Likewise.
16065 (mode_inner): Likewise.
16066 (mode_nunits): Likewise.
16067 (mode_unit_size): Likewise.
16068 (unit_unit_precision): Likewise.
16069 (mode_wider): Likewise.
16070 (mode_2xwider): Likewise.
16071 (machine_mode): New class.
16072 (mode_to_bytes): New function.
16073 (mode_to_bits): Likewise.
16074 (mode_to_precision): Likewise.
16075 (mode_to_inner): Likewise.
16076 (mode_to_unit_size): Likewise.
16077 (mode_to_unit_precision): Likewise.
16078 (mode_to_nunits): Likewise.
16079 (GET_MODE_SIZE): Use mode_to_bytes.
16080 (GET_MODE_BITSIZE): Use mode_to_bits.
16081 (GET_MODE_PRECISION): Use mode_to_precision.
16082 (GET_MODE_INNER): Use mode_to_inner.
16083 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16084 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16085 (GET_MODE_NUNITS): Use mode_to_nunits.
16086 * system.h (ALWAYS_INLINE): New macro.
16087 * config/powerpcspe/powerpcspe-c.c
16088 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16089 int for arg1_mode and arg2_mode.
16091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16092 Alan Hayward <alan.hayward@arm.com>
16093 David Sherwood <david.sherwood@arm.com>
16095 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16096 Prefix mode names with E_ in case statements.
16097 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16098 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16099 (aarch64_split_simd_move): Likewise.
16100 (aarch64_gen_storewb_pair): Likewise.
16101 (aarch64_gen_loadwb_pair): Likewise.
16102 (aarch64_gen_store_pair): Likewise.
16103 (aarch64_gen_load_pair): Likewise.
16104 (aarch64_get_condition_code_1): Likewise.
16105 (aarch64_constant_pool_reload_icode): Likewise.
16106 (get_rsqrte_type): Likewise.
16107 (get_rsqrts_type): Likewise.
16108 (get_recpe_type): Likewise.
16109 (get_recps_type): Likewise.
16110 (aarch64_gimplify_va_arg_expr): Likewise.
16111 (aarch64_simd_container_mode): Likewise.
16112 (aarch64_emit_load_exclusive): Likewise.
16113 (aarch64_emit_store_exclusive): Likewise.
16114 (aarch64_expand_compare_and_swap): Likewise.
16115 (aarch64_gen_atomic_cas): Likewise.
16116 (aarch64_emit_bic): Likewise.
16117 (aarch64_emit_atomic_swap): Likewise.
16118 (aarch64_emit_atomic_load_op): Likewise.
16119 (aarch64_evpc_trn): Likewise.
16120 (aarch64_evpc_uzp): Likewise.
16121 (aarch64_evpc_zip): Likewise.
16122 (aarch64_evpc_ext): Likewise.
16123 (aarch64_evpc_rev): Likewise.
16124 (aarch64_evpc_dup): Likewise.
16125 (aarch64_gen_ccmp_first): Likewise.
16126 (aarch64_gen_ccmp_next): Likewise.
16127 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16128 (alpha_emit_xfloating_libcall): Likewise.
16129 (emit_insxl): Likewise.
16130 (alpha_arg_type): Likewise.
16131 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16132 (arc_preferred_simd_mode): Likewise.
16133 (arc_secondary_reload): Likewise.
16134 (get_arc_condition_code): Likewise.
16135 (arc_print_operand): Likewise.
16136 (arc_legitimate_constant_p): Likewise.
16137 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16138 * config/arc/arc.md (casesi_load): Likewise.
16139 (casesi_compact_jump): Likewise.
16140 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16141 (cc_use_register): Likewise.
16142 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16143 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16144 (arm_init_iwmmxt_builtins): Likewise.
16145 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16146 (neon_expand_vector_init): Likewise.
16147 (arm_attr_length_move_neon): Likewise.
16148 (maybe_get_arm_condition_code): Likewise.
16149 (arm_emit_vector_const): Likewise.
16150 (arm_preferred_simd_mode): Likewise.
16151 (arm_output_iwmmxt_tinsr): Likewise.
16152 (thumb1_output_casesi): Likewise.
16153 (thumb2_output_casesi): Likewise.
16154 (arm_emit_load_exclusive): Likewise.
16155 (arm_emit_store_exclusive): Likewise.
16156 (arm_expand_compare_and_swap): Likewise.
16157 (arm_evpc_neon_vuzp): Likewise.
16158 (arm_evpc_neon_vzip): Likewise.
16159 (arm_evpc_neon_vrev): Likewise.
16160 (arm_evpc_neon_vtrn): Likewise.
16161 (arm_evpc_neon_vext): Likewise.
16162 (arm_validize_comparison): Likewise.
16163 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16164 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16165 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16166 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16167 (c6x_preferred_simd_mode): Likewise.
16168 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16169 (epiphany_rtx_costs): Likewise.
16170 * config/epiphany/predicates.md (proper_comparison_operator):
16172 * config/frv/frv.c (condexec_memory_operand): Likewise.
16173 (frv_emit_move): Likewise.
16174 (output_move_single): Likewise.
16175 (output_condmove_single): Likewise.
16176 (frv_hard_regno_mode_ok): Likewise.
16177 (frv_matching_accg_mode): Likewise.
16178 * config/h8300/h8300.c (split_adds_subs): Likewise.
16179 (h8300_rtx_costs): Likewise.
16180 (h8300_print_operand): Likewise.
16181 (compute_mov_length): Likewise.
16182 (output_logical_op): Likewise.
16183 (compute_logical_op_length): Likewise.
16184 (compute_logical_op_cc): Likewise.
16185 (h8300_shift_needs_scratch_p): Likewise.
16186 (output_a_shift): Likewise.
16187 (compute_a_shift_length): Likewise.
16188 (compute_a_shift_cc): Likewise.
16189 (expand_a_rotate): Likewise.
16190 (output_a_rotate): Likewise.
16191 * config/i386/i386.c (classify_argument): Likewise.
16192 (function_arg_advance_32): Likewise.
16193 (function_arg_32): Likewise.
16194 (function_arg_64): Likewise.
16195 (function_value_64): Likewise.
16196 (ix86_gimplify_va_arg): Likewise.
16197 (ix86_legitimate_constant_p): Likewise.
16198 (put_condition_code): Likewise.
16199 (split_double_mode): Likewise.
16200 (ix86_avx256_split_vector_move_misalign): Likewise.
16201 (ix86_expand_vector_logical_operator): Likewise.
16202 (ix86_split_idivmod): Likewise.
16203 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16204 (ix86_build_const_vector): Likewise.
16205 (ix86_build_signbit_mask): Likewise.
16206 (ix86_match_ccmode): Likewise.
16207 (ix86_cc_modes_compatible): Likewise.
16208 (ix86_expand_branch): Likewise.
16209 (ix86_expand_sse_cmp): Likewise.
16210 (ix86_expand_sse_movcc): Likewise.
16211 (ix86_expand_int_sse_cmp): Likewise.
16212 (ix86_expand_vec_perm_vpermi2): Likewise.
16213 (ix86_expand_vec_perm): Likewise.
16214 (ix86_expand_sse_unpack): Likewise.
16215 (ix86_expand_int_addcc): Likewise.
16216 (ix86_split_to_parts): Likewise.
16217 (ix86_vectorize_builtin_gather): Likewise.
16218 (ix86_vectorize_builtin_scatter): Likewise.
16219 (avx_vpermilp_parallel): Likewise.
16220 (inline_memory_move_cost): Likewise.
16221 (ix86_tieable_integer_mode_p): Likewise.
16222 (x86_maybe_negate_const_int): Likewise.
16223 (ix86_expand_vector_init_duplicate): Likewise.
16224 (ix86_expand_vector_init_one_nonzero): Likewise.
16225 (ix86_expand_vector_init_one_var): Likewise.
16226 (ix86_expand_vector_init_concat): Likewise.
16227 (ix86_expand_vector_init_interleave): Likewise.
16228 (ix86_expand_vector_init_general): Likewise.
16229 (ix86_expand_vector_set): Likewise.
16230 (ix86_expand_vector_extract): Likewise.
16231 (emit_reduc_half): Likewise.
16232 (ix86_emit_i387_round): Likewise.
16233 (ix86_mangle_type): Likewise.
16234 (ix86_expand_round_sse4): Likewise.
16235 (expand_vec_perm_blend): Likewise.
16236 (canonicalize_vector_int_perm): Likewise.
16237 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16238 (expand_vec_perm_1): Likewise.
16239 (expand_vec_perm_interleave3): Likewise.
16240 (expand_vec_perm_even_odd_pack): Likewise.
16241 (expand_vec_perm_even_odd_1): Likewise.
16242 (expand_vec_perm_broadcast_1): Likewise.
16243 (ix86_vectorize_vec_perm_const_ok): Likewise.
16244 (ix86_expand_vecop_qihi): Likewise.
16245 (ix86_expand_mul_widen_hilo): Likewise.
16246 (ix86_expand_sse2_abs): Likewise.
16247 (ix86_expand_pextr): Likewise.
16248 (ix86_expand_pinsr): Likewise.
16249 (ix86_preferred_simd_mode): Likewise.
16250 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16251 * config/i386/sse.md (*andnot<mode>3): Likewise.
16252 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16253 (*<code><mode>3): Likewise.
16254 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16255 (ia64_expand_atomic_op): Likewise.
16256 (ia64_arg_type): Likewise.
16257 (ia64_mode_to_int): Likewise.
16258 (ia64_scalar_mode_supported_p): Likewise.
16259 (ia64_vector_mode_supported_p): Likewise.
16260 (expand_vec_perm_broadcast): Likewise.
16261 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16262 (iq2000_function_arg_advance): Likewise.
16263 (iq2000_function_arg): Likewise.
16264 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16265 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16266 (m68k_libcall_value): Likewise.
16267 (m68k_function_value): Likewise.
16268 (sched_attr_op_type): Likewise.
16269 * config/mcore/mcore.c (mcore_output_move): Likewise.
16270 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16272 (microblaze_function_arg): Likewise.
16273 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16274 (mips_print_operand): Likewise.
16275 (mips_mode_ok_for_mov_fmt_p): Likewise.
16276 (mips_vector_mode_supported_p): Likewise.
16277 (mips_preferred_simd_mode): Likewise.
16278 (mips_expand_vpc_loongson_even_odd): Likewise.
16279 (mips_expand_vec_unpack): Likewise.
16280 (mips_expand_vi_broadcast): Likewise.
16281 (mips_expand_vector_init): Likewise.
16282 (mips_expand_vec_reduc): Likewise.
16283 (mips_expand_msa_cmp): Likewise.
16284 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16285 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16286 (cc_flags_for_mode): Likewise.
16287 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16288 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16289 (nds32_output_casesi_pc_relative): Likewise.
16290 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16291 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16292 (nvptx_gen_unpack): Likewise.
16293 (nvptx_gen_pack): Likewise.
16294 (nvptx_gen_shuffle): Likewise.
16295 (nvptx_gen_wcast): Likewise.
16296 (nvptx_preferred_simd_mode): Likewise.
16297 * config/pa/pa.c (pa_secondary_reload): Likewise.
16298 * config/pa/predicates.md (base14_operand): Likewise.
16299 * config/powerpcspe/powerpcspe-c.c
16300 (altivec_resolve_overloaded_builtin): Likewise.
16301 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16303 (rs6000_preferred_simd_mode): Likewise.
16304 (output_vec_const_move): Likewise.
16305 (rs6000_expand_vector_extract): Likewise.
16306 (rs6000_split_vec_extract_var): Likewise.
16307 (reg_offset_addressing_ok_p): Likewise.
16308 (rs6000_legitimate_offset_address_p): Likewise.
16309 (rs6000_legitimize_address): Likewise.
16310 (rs6000_emit_set_const): Likewise.
16311 (rs6000_const_vec): Likewise.
16312 (rs6000_emit_move): Likewise.
16313 (spe_build_register_parallel): Likewise.
16314 (rs6000_darwin64_record_arg_recurse): Likewise.
16315 (swap_selector_for_mode): Likewise.
16316 (spe_init_builtins): Likewise.
16317 (paired_init_builtins): Likewise.
16318 (altivec_init_builtins): Likewise.
16319 (do_load_for_compare): Likewise.
16320 (rs6000_generate_compare): Likewise.
16321 (rs6000_expand_float128_convert): Likewise.
16322 (emit_load_locked): Likewise.
16323 (emit_store_conditional): Likewise.
16324 (rs6000_output_function_epilogue): Likewise.
16325 (rs6000_handle_altivec_attribute): Likewise.
16326 (rs6000_function_value): Likewise.
16327 (emit_fusion_gpr_load): Likewise.
16328 (emit_fusion_p9_load): Likewise.
16329 (emit_fusion_p9_store): Likewise.
16330 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16331 (fusion_gpr_mem_load): Likewise.
16332 (fusion_addis_mem_combo_load): Likewise.
16333 (fusion_addis_mem_combo_store): Likewise.
16334 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16335 (fusion_gpr_mem_load): Likewise.
16336 (fusion_addis_mem_combo_load): Likewise.
16337 (fusion_addis_mem_combo_store): Likewise.
16338 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16340 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16341 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16342 (rs6000_preferred_simd_mode): Likewise.
16343 (output_vec_const_move): Likewise.
16344 (rs6000_expand_vector_extract): Likewise.
16345 (rs6000_split_vec_extract_var): Likewise.
16346 (reg_offset_addressing_ok_p): Likewise.
16347 (rs6000_legitimate_offset_address_p): Likewise.
16348 (rs6000_legitimize_address): Likewise.
16349 (rs6000_emit_set_const): Likewise.
16350 (rs6000_const_vec): Likewise.
16351 (rs6000_emit_move): Likewise.
16352 (rs6000_darwin64_record_arg_recurse): Likewise.
16353 (swap_selector_for_mode): Likewise.
16354 (paired_init_builtins): Likewise.
16355 (altivec_init_builtins): Likewise.
16356 (rs6000_expand_float128_convert): Likewise.
16357 (emit_load_locked): Likewise.
16358 (emit_store_conditional): Likewise.
16359 (rs6000_output_function_epilogue): Likewise.
16360 (rs6000_handle_altivec_attribute): Likewise.
16361 (rs6000_function_value): Likewise.
16362 (emit_fusion_gpr_load): Likewise.
16363 (emit_fusion_p9_load): Likewise.
16364 (emit_fusion_p9_store): Likewise.
16365 * config/rx/rx.c (rx_gen_move_template): Likewise.
16366 (flags_from_mode): Likewise.
16367 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16368 (s390_slb_comparison): Likewise.
16369 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16370 (s390_vector_mode_supported_p): Likewise.
16371 (s390_cc_modes_compatible): Likewise.
16372 (s390_match_ccmode_set): Likewise.
16373 (s390_canonicalize_comparison): Likewise.
16374 (s390_emit_compare_and_swap): Likewise.
16375 (s390_branch_condition_mask): Likewise.
16376 (s390_rtx_costs): Likewise.
16377 (s390_secondary_reload): Likewise.
16378 (__SECONDARY_RELOAD_CASE): Likewise.
16379 (s390_expand_cs): Likewise.
16380 (s390_preferred_simd_mode): Likewise.
16381 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16382 * config/sh/sh.c (sh_print_operand): Likewise.
16383 (dump_table): Likewise.
16384 (sh_secondary_reload): Likewise.
16385 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16386 * config/sh/sh.md (casesi_worker_1): Likewise.
16387 (casesi_worker_2): Likewise.
16388 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16389 (fcc_comparison_operator): Likewise.
16390 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16391 (emit_soft_tfmode_cvt): Likewise.
16392 (sparc_preferred_simd_mode): Likewise.
16393 (output_cbranch): Likewise.
16394 (sparc_print_operand): Likewise.
16395 (sparc_expand_vec_perm_bmask): Likewise.
16396 (vector_init_bshuffle): Likewise.
16397 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16398 (spu_vector_mode_supported_p): Likewise.
16399 (spu_expand_insv): Likewise.
16400 (spu_emit_branch_or_set): Likewise.
16401 (spu_handle_vector_attribute): Likewise.
16402 (spu_builtin_splats): Likewise.
16403 (spu_builtin_extract): Likewise.
16404 (spu_builtin_promote): Likewise.
16405 (spu_expand_sign_extend): Likewise.
16406 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16407 (tilegx_simd_int): Likewise.
16408 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16409 (tilepro_simd_int): Likewise.
16410 * config/v850/v850.c (const_double_split): Likewise.
16411 (v850_print_operand): Likewise.
16412 (ep_memory_offset): Likewise.
16413 * config/vax/vax.c (vax_rtx_costs): Likewise.
16414 (vax_output_int_move): Likewise.
16415 (vax_output_int_add): Likewise.
16416 (vax_output_int_subtract): Likewise.
16417 * config/visium/predicates.md (visium_branch_operator): Likewise.
16418 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16419 (visium_print_operand_address): Likewise.
16420 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16421 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16422 (xtensa_expand_conditional_branch): Likewise.
16423 (xtensa_copy_incoming_a7): Likewise.
16424 (xtensa_output_literal): Likewise.
16425 * dfp.c (decimal_real_maxval): Likewise.
16426 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16429 Alan Hayward <alan.hayward@arm.com>
16430 David Sherwood <david.sherwood@arm.com>
16432 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16433 (mode_nunits_inline): Likewise.
16434 (mode_inner_inline): Likewise.
16435 (mode_unit_size_inline): Likewise.
16436 (mode_unit_precision_inline): Likewise.
16437 (emit_insn_modes_h): Likewise. Also emit a #define of the
16439 (emit_mode_wider): Add an E_ prefix to mode names.
16440 (emit_mode_complex): Likewise.
16441 (emit_mode_inner): Likewise.
16442 (emit_mode_adjustments): Likewise.
16443 (emit_mode_int_n): Likewise.
16444 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16445 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16446 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16447 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16448 (CRC32_BUILTIN, ENTRY): Likewise.
16449 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16450 (aarch64_pop_regs): Likewise.
16451 (aarch64_process_components): Likewise.
16452 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16453 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16454 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16455 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16456 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16457 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16459 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16460 (pcmpistr): Likewise.
16461 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16462 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16463 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16465 * config/rl78/rl78.c (mduc_regs): Likewise.
16466 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16467 (htm_expand_builtin): Likewise.
16468 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16469 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16470 * config/xtensa/xtensa.c (print_operand): Likewise.
16471 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16472 (NUM_MODE_VECTOR_INT): Likewise.
16473 * genoutput.c (null_operand): Likewise.
16474 (output_operand_data): Likewise.
16475 * genrecog.c (print_parameter_value): Likewise.
16476 * lra.c (debug_operand_data): Likewise.
16478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16480 * dwarf2out.c (scompare_loc_descriptor_wide)
16481 (scompare_loc_descriptor_narrow): New functions, split out from...
16482 (scompare_loc_descriptor): ...here.
16483 * expmed.c (emit_store_flag_int): New function, split out from...
16484 (emit_store_flag): ...here.
16486 2017-08-30 Richard Biener <rguenther@suse.de>
16488 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16489 (dwarf2out_early_finish): Move setting of AT_pubnames from
16490 early debug output to early finish.
16492 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16494 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16495 and -mdata-region to the assembler.
16497 2017-08-30 Richard Biener <rguenther@suse.de>
16499 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16501 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16502 (dwarf2out_early_global_decl): Only generate a DIE for the
16503 abstract origin if it doesn't already exist or is a declaration DIE.
16504 (resolve_addr): Do not add the linkage name twice when
16505 generating a stub DIE for the DW_TAG_GNU_call_site target.
16507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16509 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16510 Use machine_mode rather than int for arg1_mode.
16512 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16515 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16516 that the second argument of the built-in functions to unpack
16517 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16518 switch statement instead a lot of if statements.
16519 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16520 Allow 64-bit values to be in Altivec registers as well as
16521 traditional floating point registers.
16522 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16524 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16526 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16527 MAX_REGS_PER_ADDRESS == 1.
16529 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16531 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16532 * config/i386/i386.c (ix86_option_override_internal): Simplify
16533 setting of opts->x_flag_entry.
16535 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16536 Jakub Jelinek <jakub@redhat.com>
16537 Richard Biener <rguenther@suse.de>
16539 PR tree-optimization/81503
16540 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16541 folded constant fits in the target type; reorder tests for clarity.
16543 2017-08-29 Martin Liska <mliska@suse.cz>
16545 * passes.def: Include pass_lower_switch.
16546 * stmt.c (dump_case_nodes): Remove and move to
16547 tree-switch-conversion.
16548 (case_values_threshold): Likewise.
16549 (expand_switch_as_decision_tree_p): Likewise.
16550 (emit_case_decision_tree): Likewise.
16551 (expand_case): Likewise.
16552 (balance_case_nodes): Likewise.
16553 (node_has_low_bound): Likewise.
16554 (node_has_high_bound): Likewise.
16555 (node_is_bounded): Likewise.
16556 (emit_case_nodes): Likewise.
16557 (struct simple_case_node): New struct.
16558 (add_case_node): Remove.
16559 (emit_case_dispatch_table): Use vector instead of case_list.
16560 (reset_out_edges_aux): Remove.
16561 (compute_cases_per_edge): Likewise.
16562 (expand_case): Build list of simple_case_node.
16563 (expand_sjlj_dispatch_table): Use it.
16564 * tree-switch-conversion.c (struct case_node): Moved from
16565 stmt.c and adjusted.
16566 (emit_case_nodes): Likewise.
16567 (node_has_low_bound): Likewise.
16568 (node_has_high_bound): Likewise.
16569 (node_is_bounded): Likewise.
16570 (case_values_threshold): Likewise.
16571 (reset_out_edges_aux): Likewise.
16572 (compute_cases_per_edge): Likewise.
16573 (add_case_node): Likewise.
16574 (dump_case_nodes): Likewise.
16575 (balance_case_nodes): Likewise.
16576 (expand_switch_as_decision_tree_p): Likewise.
16577 (emit_jump): Likewise.
16578 (emit_case_decision_tree): Likewise.
16579 (try_switch_expansion): Likewise.
16580 (do_jump_if_equal): Likewise.
16581 (emit_cmp_and_jump_insns): Likewise.
16582 (fix_phi_operands_for_edge): New function.
16583 (record_phi_operand_mapping): Likewise.
16584 (class pass_lower_switch): New pass.
16585 (pass_lower_switch::execute): New function.
16586 (make_pass_lower_switch): Likewise.
16587 (conditional_probability):
16588 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16589 * tree-pass.h: Add make_pass_lower_switch.
16591 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16594 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16597 2017-08-29 Richard Biener <rguenther@suse.de>
16599 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16600 we do not add a DW_AT_inline attribute twice.
16601 (gen_subprogram_die): Remove code setting DW_AT_inline on
16602 DECL_ABSTRACT_P nodes.
16604 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16606 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16607 calls to internal functions.
16608 (gimplify_modify_expr): Likewise.
16609 * tree-call-cdce.c (use_internal_fn): Likewise.
16610 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16611 (convert_to_divmod): Set the nothrow flag.
16612 * tree-if-conv.c (predicate_mem_writes): Likewise.
16613 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16614 (vectorizable_call): Likewise.
16615 (vectorizable_store): Likewise.
16616 (vectorizable_load): Likewise.
16617 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16618 (vect_recog_mask_conversion_pattern): Likewise.
16620 2017-08-29 Martin Liska <mliska@suse.cz>
16623 * gcc.c (driver_handle_option): Add new argument.
16624 * opts-common.c (handle_option): Pass
16625 target_option_override_hook.
16626 * opts-global.c (lang_handle_option): Add new option.
16627 (set_default_handlers): Add new argument.
16628 (decode_options): Likewise.
16629 * opts.c (target_handle_option): Likewise.
16630 (common_handle_option): Call target_option_override_hook.
16631 * opts.h (struct cl_option_handler_func): Add hook for
16632 target option override.
16633 (struct cl_option_handlers): Likewise.
16634 (set_default_handlers): Add new argument.
16635 (decode_options): Likewise.
16636 (common_handle_option): Likewise.
16637 (target_handle_option): Likewise.
16638 * toplev.c (toplev::main): Pass targetm.target_option.override
16641 2017-08-29 Richard Biener <rguenther@suse.de>
16642 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16644 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16645 life to the active subtree.
16647 2017-08-28 Jeff Law <law@redhat.com>
16649 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16650 derive_equivalences.
16651 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16652 Code moved into....
16653 (edge_info::derive_equivalences): New private member function
16655 * tree-ssa-dom.c (class edge_info): Changed from a struct
16656 to a class. Add ctor/dtor, methods and data members.
16657 (edge_info::edge_info): Renamed from allocate_edge_info.
16658 Initialize additional members.
16659 (edge_info::~edge_info): New.
16660 (free_dom_edge_info): Delete the edge info.
16661 (record_edge_info): Use new class & associated member functions.
16662 Tighten forms for testing for edge equivalences.
16663 (record_temporary_equivalences): Iterate over the simple
16664 equivalences rather than assuming there's only one per edge.
16665 (cprop_into_successor_phis): Iterate over the simple
16666 equivalences rather than assuming there's only one per edge.
16667 (optimize_stmt): Use operand_equal_p rather than pointer
16668 equality for mini-DSE code.
16670 2017-08-28 Nathan Sidwell <nathan@acm.org>
16672 * gcc.c (execute): Fold SIGPIPE handling into switch
16673 statement. Adjust internal error message.
16675 2017-08-28 Richard Biener <rguenther@suse.de>
16678 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16679 Do nothing for removed DIEs.
16681 2017-08-28 Richard Biener <rguenther@suse.de>
16683 PR tree-optimization/81977
16684 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16687 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16690 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16691 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16692 using targetm.gen_mem_thread_fence.
16694 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16697 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16698 predicate to register_operand. Reorder operands.
16699 (*btr<mode>): Ditto.
16700 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16701 (*btr<mode>_mask): Ditto.
16703 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16705 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16706 * config/rs6000/xmmintrin.h: New file.
16707 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16709 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16712 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16713 parameter and_insn and return it.
16714 (recombine_lvx_pattern): Insert a copy to ensure availability of
16715 the base register of the copied masking operation at the point of
16716 the instruction replacement.
16717 (recombine_stvx_pattern): Likewise.
16719 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16721 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16722 undocumented switches.
16723 (-mpower9-dform-vector): Likewise.
16724 (-mpower9-dform): Likewise.
16725 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16726 comments to delete references to -mpower9-dform* switches.
16727 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16728 Delete reference to -mpower9-dform* switches, test for
16729 -mpower9-vector instead.
16730 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16731 (OTHER_P9_VECTOR_MASKS): Likewise.
16732 (POWERPC_MASKS): Likewise.
16733 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16734 tests against -mpower9-dform* to -mpower9-vector. Delete code
16735 that checked for -mpower9-dform* consistancy with other options.
16736 Add test for -mpower9-misc to enable other power9 switches.
16737 (rs6000_init_hard_regno_mode_ok): Likewise.
16738 (rs6000_option_override_internal): Likewise.
16739 (rs6000_emit_prologue): Likewise.
16740 (rs6000_emit_epilogue): Likewise.
16741 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16742 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16743 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16745 (emit_fusion_p9_store): Likewise.
16746 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16747 resetting these macros if the assembler does not support ISA 3.0
16749 (TARGET_P9_DFORM_VECTOR): Likewise.
16750 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16751 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16753 2017-08-25 Alan Modra <amodra@gmail.com>
16755 PR rtl-optimization/81747
16756 * cse.c (cse_extended_basic_block): Don't attempt to record
16757 equivalences for degenerate conditional branches that branch
16758 to their fall-through.
16760 2017-08-24 Martin Sebor <msebor@redhat.com>
16762 PR middle-end/81908
16763 * gimple-fold.c (size_must_be_zero_p): New function.
16764 (gimple_fold_builtin_memory_op): Call it.
16766 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16768 * config/rs6000/mm_malloc.h: New file.
16770 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16772 PR tree-optimization/81913
16773 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16774 analysis when either IVs in condition can wrap.
16776 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16778 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16779 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16781 2017-08-24 Richard Biener <rguenther@suse.de>
16784 * targhooks.c (default_target_can_inline_p): Properly
16785 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16786 is present and always compare.
16787 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16788 infer -mfpmath=sse from TARGET_SSE_P.
16789 (ix86_can_inline_p): Properly use target_option_default_node when
16790 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16792 2017-08-24 Richard Biener <rguenther@suse.de>
16795 * dwarf2out.c (output_die): Handle flag_generate_offload like
16797 (output_comp_unit): Likewise.
16798 (gen_array_type_die): Likewise.
16799 (dwarf2out_early_finish): Likewise.
16800 (note_variable_value_in_expr): Likewise.
16801 (dwarf2out_finish): Likewise. Adjust assert.
16802 * cgraphunit.c (symbol_table::compile): Move setting of
16803 flag_generate_offload earlier ...
16804 (symbol_table::finalize_compilation_unit): ... here, before
16805 early debug finalization.
16807 2017-08-24 Richard Biener <rguenther@suse.de>
16809 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16810 and ipa-fnsummary.h.
16811 (ix86_can_inline_p): When ix86_fpmath flags do not match
16812 check whether the callee uses FP math at all.
16814 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
16816 PR middle-end/81931
16817 * tree-ssanames.c (get_nonzero_bits): Use element_precision
16818 instead of TYPE_PRECISION.
16820 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
16821 Alan Hayward <alan.hayward@arm.com>
16822 David Sherwood <david.sherwood@arm.com>
16824 * combine.c (make_extraction): Use subreg_offset_from_lsb.
16826 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
16828 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
16830 (ix86_frame::stack_realign_allocate): New field.
16831 (struct machine_frame_state): Modify comments.
16832 (machine_frame_state::sp_realigned_fp_end): New field.
16833 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
16834 layout calculation.
16835 (sp_valid_at): Add assertion to assure no attempt to access invalid
16836 offset of a realigned stack.
16837 (fp_valid_at): Likewise.
16838 (choose_baseaddr): Modify comments.
16839 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
16840 ix86_expand_prologue.
16841 (ix86_expand_prologue): Modify stack realignment and allocation.
16842 (ix86_expand_epilogue): Modify comments.
16843 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
16844 avx2_runtime, avx512f, and avx512f_runtime.
16846 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
16848 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
16849 (mstackrealign): Do not init to -1.
16850 * config/i386/i386.c (ix86_option_override_internal):
16851 Check opts_set, not opts when setting default value of
16852 opts->x_ix86_force_align_arg_pointer.
16854 2017-08-23 Richard Biener <rguenther@suse.de>
16856 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
16857 lang_hooks.decl_printable_name.
16858 * print-rtl-function.c (print_rtx_function): Likewise.
16859 * tree-pretty-print.c (dump_function_header): Likewise.
16861 2017-08-23 Richard Biener <rguenther@suse.de>
16864 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
16865 -g0 at compile-time.
16867 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16869 PR middle-end/19706
16870 * doc/sourcebuild.texi (Other hardware attributes):
16873 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16875 PR middle-end/19706
16876 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
16877 Fix single-use check.
16879 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
16881 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
16883 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16885 * doc/install.texi: Modify to add more details on running selected
16888 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16890 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
16891 is combined with -mabi=ms.
16892 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
16895 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16897 PR tree-optimization/81488
16898 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
16899 and cached_basis fields.
16900 (MAX_SPREAD): New constant.
16901 (alloc_cand_and_find_basis): Initialize new fields.
16902 (clear_visited): New function.
16903 (create_phi_basis_1): Rename from create_phi_basis, set visited
16904 and cached_basis fields.
16905 (create_phi_basis): New wrapper function.
16906 (phi_add_costs_1): Rename from phi_add_costs, add spread
16907 parameter, set visited field, short-circuit when limits reached.
16908 (phi_add_costs): New wrapper function.
16909 (record_phi_increments_1): Rename from record_phi_increments, set
16911 (record_phi_increments): New wrapper function.
16912 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
16913 (phi_incr_cost): New wrapper function.
16914 (all_phi_incrs_profitable_1): Rename from
16915 all_phi_incrs_profitable, set visited field.
16916 (all_phi_incrs_profitable): New wrapper function.
16918 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16919 Alan Hayward <alan.hayward@arm.com>
16920 David Sherwood <david.sherwood@arm.com>
16922 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
16923 that takes the outer and inner modes.
16924 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
16925 comparison as the canonical test for a paradoxical subreg.
16926 * combine.c (simplify_set): Use paradoxical_subreg_p.
16927 (make_extraction): Likewise.
16928 (force_to_mode): Likewise.
16929 (rtx_equal_for_field_assignment_p): Likewise.
16930 (gen_lowpart_for_combine): Likewise.
16931 (simplify_comparison): Likewise.
16932 * cse.c (equiv_constant): Likewise.
16933 * expmed.c (store_bit_field_1): Likewise.
16934 * final.c (alter_subreg): Likewise.
16935 * fwprop.c (propagate_rtx): Likewise.
16936 (forward_propagate_subreg): Likewise.
16937 * ira-conflicts.c (ira_build_conflicts): Likewise.
16938 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
16939 * lra-constraints.c (curr_insn_transform): Likewise.
16940 (split_reg): Likewise.
16941 * lra-eliminations.c (move_plus_up): Likewise.
16942 (lra_eliminate_regs_1): Likewise.
16943 * recog.c (general_operand): Likewise.
16944 * ree.c (combine_reaching_defs): Likewise.
16945 * reload.c (push_reload): Likewise.
16946 (find_reloads): Likewise.
16947 * reload1.c (elimination_effects): Likewise.
16948 (compute_reload_subreg_offset): Likewise.
16949 (choose_reload_regs): Likewise.
16950 * rtlanal.c (subreg_lsb_1): Likewise.
16951 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16952 (simplify_subreg): Likewise.
16953 * var-tracking.c (track_loc_p): Likewise.
16954 * emit-rtl.c (byte_lowpart_offset): Likewise.
16955 (paradoxical_subreg_p): Delete out-of-line definition.
16957 2017-08-22 Jeff Law <law@redhat.com>
16959 PR tree-optimization/81741
16960 PR tree-optimization/71947
16961 * tree-ssa-dom.c: Include tree-inline.h.
16962 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
16963 equivalences if one is more expensive to compute than the other.
16964 * tree-ssa-scopedtables.h (class const_or_copies): Make
16965 record_const_or_copy_raw method private.
16966 (class avail_exprs_stack): New method simplify_binary_operation.
16967 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
16968 avail_exprs_stack::simplify_binary_operation as needed.
16969 (avail_exprs_stack::simplify_binary_operation): New function.
16971 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
16973 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
16974 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
16975 (DOT_SYMBOLS): Likewise.
16976 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
16977 (RELOCATABLE_NEEDS_FIXUP): Likewise.
16978 (RS6000_ABI_NAME): Likewise.
16979 (TARGET_CMODEL): Likewise.
16980 (TOC_SECTION_ASM_OP): Likewise.
16981 (SET_CMODEL): New macro.
16982 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
16984 2017-08-22 Richard Biener <rguenther@suse.de>
16986 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
16987 to allow for free-lang-data replacements similar to verify_type_variant.
16989 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16990 Alan Hayward <alan.hayward@arm.com>
16991 David Sherwood <david.sherwood@arm.com>
16993 * config/aarch64/aarch64.md (casesi): Use DImode rather than
16994 VOIDmode for the LABEL_REF.
16996 2017-08-22 Richard Biener <rguenther@suse.de>
16998 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17000 2017-08-22 Richard Biener <rguenther@suse.de>
17002 * common.opt (feliminate-dwarf2-dups): Ignore.
17003 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17004 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17005 same_die_p_wrap, compute_section_prefix,
17006 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17007 (comdat_symbol_id, comdat_symbol_number): Likewise.
17008 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17010 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17011 (output_die): Mark unreachable path unreachable.
17012 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17013 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17014 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17015 (dwarf2out_early_finish): Likewise.
17017 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17019 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17021 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17024 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17025 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17026 (avr_attribute_table) <io, io_low, address>: Initialize
17027 .decl_required with true.
17029 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17031 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17032 undocumented debugging options.
17033 (-mvsx-scalar-double): Likewise.
17034 (-mallow-df-permute): Likewise.
17035 (-mvectorize-builtins): Likewise.
17036 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17037 (rs6000_builtin_vectorized_function): Likewise.
17038 (rs6000_builtin_md_vectorized_function): Likewise.
17039 (rs6000_opt_vars): Likewise.
17041 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17044 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17045 (*btrq_imm): Rename from *btrq.
17046 (*btcq_imm): Rename from *btcq.
17047 (btsc): New code attribute.
17048 (*<btsc><mode>): New insn pattern.
17049 (*btr<mode>): Ditto.
17050 (*<btsc><mode>_mask): New insn_and_split pattern.
17051 (*btr<mode>_mask): Ditto.
17053 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17054 Alan Hayward <alan.hayward@arm.com>
17055 David Sherwood <david.sherwood@arm.com>
17057 * function.c (pad_below): Simplify padding calculation.
17059 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17060 Alan Hayward <alan.hayward@arm.com>
17061 David Sherwood <david.sherwood@arm.com>
17063 * target.def (function_prologue): Remove frame size argument.
17064 (function_epilogue): Likewise.
17065 * doc/tm.texi: Regenerate.
17066 * final.c (final_start_function): Update call to function_prologue.
17067 (final_end_function): Update call to function_epilogue.
17068 (default_function_pro_epilogue): Remove frame size argument.
17069 * output.h (default_function_pro_epilogue): Likewise.
17070 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17071 (arm_output_function_prologue): Likewise.
17072 * config/frv/frv.c (frv_function_prologue): Likewise.
17073 (frv_function_epilogue): Likewise.
17074 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17075 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17076 (ia64_output_function_epilogue): Likewise.
17077 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17078 (m32r_output_function_epilogue): Likewise.
17079 * config/microblaze/microblaze.c (microblaze_function_prologue)
17080 (microblaze_function_epilogue): Likewise.
17081 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17082 (mips_output_function_epilogue): Likewise.
17083 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17084 (mmix_target_asm_function_epilogue): Likewise.
17085 * config/msp430/msp430.c (msp430_start_function): Likewise.
17086 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17087 (nds32_asm_function_epilogue): Likewise.
17088 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17089 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17090 (pa_output_function_epilogue): Likewise.
17091 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17092 (rs6000_output_function_epilogue): Likewise.
17093 * config/rl78/rl78.c (rl78_start_function): Likewise.
17094 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17095 (rs6000_output_function_epilogue): Likewise.
17096 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17097 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17098 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17099 (sparc_asm_function_epilogue): Likewise.
17101 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17103 * tree.h (type_has_mode_precision_p): New function.
17104 * convert.c (convert_to_integer_1): Use it.
17105 * expr.c (expand_expr_real_2): Likewise.
17106 (expand_expr_real_1): Likewise.
17107 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17108 * match.pd: Likewise.
17109 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17110 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17111 * tree-tailcall.c (process_assignment): Likewise.
17112 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17113 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17114 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17115 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17116 (vectorizable_assignment): Likewise.
17117 (vectorizable_shift): Likewise.
17118 (vectorizable_operation): Likewise.
17119 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17121 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17123 * match.pd: Add pow (C, x) simplification.
17125 2017-08-21 Richard Biener <rguenther@suse.de>
17127 PR tree-optimization/81900
17128 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17129 for blocks with abnormal predecessors.
17130 (compute_antic): Do not set visited flag prematurely.
17132 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17135 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17137 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17139 * stor-layout.h (vector_type_mode): Move to...
17140 * tree.h (vector_type_mode): ...here.
17141 * stor-layout.c (vector_type_mode): Move to...
17142 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17144 2017-08-21 Richard Biener <rguenther@suse.de>
17146 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17147 register_external_die hooks.
17148 (debug_false_tree_charstarstar_uhwistar): Declare.
17149 (debug_nothing_tree_charstar_uhwi): Likewise.
17150 * debug.c (do_nothing_debug_hooks): Adjust.
17151 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17152 (debug_nothing_tree_charstar_uhwi): Likewise.
17153 * dbxout.c (dbx_debug_hooks): Adjust.
17154 (xcoff_debug_hooks): Likewise.
17155 * sdbout.c (sdb_debug_hooks): Likewise.
17156 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17157 * dwarf2out.c (macinfo_label_base): New global.
17158 (dwarf2out_register_external_die): New function for the
17159 register_external_die hook.
17160 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17161 (dwarf2_debug_hooks): Use them.
17162 (dwarf2_lineno_debug_hooks): Adjust.
17163 (struct die_struct): Add with_offset flag.
17164 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17165 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17166 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17167 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17168 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17169 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17170 defining section names for the early LTO debug variants.
17171 (reset_indirect_string): New helper.
17172 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17173 (print_dw_val): Add support for offsetted symbol references.
17174 (get_ultimate_context): Split out from is_cxx.
17175 (is_cxx): Use get_ultimate_context.
17176 (is_fortran): Add decl overload.
17177 (compute_comp_unit_symbol): Split out worker from
17178 compute_section_prefix.
17179 (compute_section_prefix): Call compute_comp_unit_symbol and
17180 set comdat_type_p here.
17181 (output_die): Skip DIE symbol output for the LTO added one.
17182 Handle DIE symbol references with offset.
17183 (output_comp_unit): Guard section name mangling properly.
17184 For LTO debug sections emit a symbol at the section beginning
17185 which we use to refer to its DIEs.
17186 (add_abstract_origin_attribute): For DIEs registered via
17187 dwarf2out_register_external_die directly refer to the early
17188 DIE rather than indirectly through the shadow one we created.
17189 Remove obsolete call to dwarf2out_abstract_function for
17190 non-function/block origins.
17191 (gen_array_type_die): When generating early LTO debug do
17192 not emit DW_AT_string_length.
17193 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17194 late when in LTO. As suggested place a gcc_unreachable for
17195 the DECL_ABSTRACT_P case.
17196 (gen_subprogram_die): Avoid another specification DIE
17197 for early built declarations/definitions for the late LTO case.
17198 (gen_variable_die): Add type references for late duplicated VLA dies
17200 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17201 we have the abstract instance already.
17202 (process_scope_var): Adjust decl DIE contexts in LTO which
17203 first puts them in limbo.
17204 (gen_decl_die): Do not generate type DIEs late apart from
17205 types for VLAs or for decls we do not yet have a DIE. Do not
17206 call dwarf2out_abstract_function late.
17207 (dwarf2out_early_global_decl): Make sure to create DIEs
17208 for abstract instances of a decl first.
17209 (dwarf2out_late_global_decl): Adjust comment.
17210 (output_macinfo_op): With multiple macro sections use
17211 macinfo_label_base to distinguish labels.
17212 (output_macinfo): Likewise. Update macinfo_label_base.
17213 Pass in the line info label.
17214 (note_variable_value_in_expr): When generating LTO resolve
17215 all variable values here by generating DIEs as needed.
17216 (init_sections_and_labels): Add early LTO debug flag parameter
17217 and generate different sections and names if set. Add generation
17218 counter for the labels so we can have multiple of them.
17219 (reset_dies): Helper to allow DIEs to be output multiple times.
17220 (dwarf2out_finish): When outputting DIEs to the fat part of an
17221 LTO object first reset DIEs.
17222 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17223 (modified_type_die): Check for decl_ultimate_origin being self
17225 (gen_type_die_with_usage): Likewise.
17226 (gen_typedef_die): Allow decl_ultimate_origin being self.
17227 (set_decl_abstract_flags): Remove.
17228 (set_block_abstract_flags): Likewise.
17229 (dwarf2out_abstract_function): Treat the early generated DIEs
17230 as the abstract copy and only add DW_AT_inline and
17231 DW_AT_artificial here and call set_decl_origin_self.
17232 If the DIE has an abstract origin don't do anything.
17233 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17234 if we have none yet (Go fails to build one, PR78628).
17235 (variably_modified_type_p): Prevent endless recursion for Ada
17236 cyclic pointer types.
17237 * lto-streamer-in.c: Include debug.h.
17238 (dref_queue): New global.
17239 (lto_read_tree_1): Stream in DIE references.
17240 (lto_input_tree): Register DIE references.
17241 (input_function): Stream DECL_DEBUG_ARGS.
17242 * lto-streamer-out.c: Include debug.h.
17243 (lto_write_tree_1): Output DIE references.
17244 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17245 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17246 (output_function): Stream DECL_DEBUG_ARGS.
17247 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17248 Stream DECL_ABSTRACT_ORIGIN.
17249 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17250 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17251 DECL_CONTEXT for file-scope decls.
17252 * lto-streamer.h (struct dref_entry): Declare.
17253 (dref_queue): Likewise.
17254 * cfgexpand.c (pass_expand::execute): Do not call the
17255 outlining_inline_function hook here.
17256 * lto-wrapper.c (debug_obj): New global.
17257 (tool_cleanup): Unlink it if required.
17258 (debug_objcopy): New function.
17259 (run_gcc): Handle early debug sections in the IL files by
17260 extracting them to separate files, partially linkin them and
17261 feeding the result back as result to the linker.
17262 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17263 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17264 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17265 sections into a separate segment.
17266 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17268 (darwin_asm_dwarf_section): Likewise.
17269 (darwin_asm_output_dwarf_offset): Likewise.
17270 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17272 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17273 Alan Hayward <alan.hayward@arm.com>
17274 David Sherwood <david.sherwood@arm.com>
17276 * read-md.h (md_reader::record_potential_iterator_use): Replace
17277 pointer argument with an rtx and an index.
17278 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17279 (apply_mode_iterator): Likewise.
17280 (apply_code_iterator): Likewise.
17281 (apply_int_iterator): Likewise.
17282 (apply_subst_iterator): Likewise.
17283 (record_iterator_use): Likewise.
17284 (record_attribute_use): Likewise.
17285 (md_reader::record_potential_iterator_use): Likewise. Update calls
17286 to record_iterator_use and apply_iterator.
17287 (iterator_use): Replace ptr with x and index.
17288 (attribute_use): Likewise.
17289 (apply_attribute_uses): Update calls to apply_iterator.
17290 (apply_iterators): Likewise. Update initialization of iterator_use.
17291 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17292 and record_potential_iterator_use.
17293 (rtx_reader::read_rtx_operand): Likewise.
17295 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17296 Alan Hayward <alan.hayward@arm.com>
17297 David Sherwood <david.sherwood@arm.com>
17299 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17302 2017-08-21 Richard Biener <rguenther@suse.de>
17304 PR middle-end/81884
17305 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17306 at struct end conservatively when comparing common bases.
17308 2017-08-21 Richard Biener <rguenther@suse.de>
17310 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17311 (mem_ref_in_stmt): Remove.
17312 (determine_max_movement): Use ref index to get at the reference.
17313 (invariantness_dom_walker::before_dom_children): Deal with
17314 lim data already initialized.
17315 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17317 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17319 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17320 (TARGET_ISA_ROUND): Ditto.
17321 (TARGET_ROUND): Ditto.
17322 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17323 * config/i386/i386.md: Ditto.
17324 * config/i386/sse.md: Ditto.
17325 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17326 with OPTION_MASK_ISA_SSE4_1.
17328 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17331 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17332 __builtin_ia32_lzcnt_u16.
17334 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17337 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17338 (rs6000_set_current_function): Rewrite function to use it.
17340 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17343 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17344 and TYPE_WARN_IF_NOT_ALIGN.
17345 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17346 (handle_warn_if_not_align): New.
17347 (place_union_field): Call handle_warn_if_not_align.
17348 (place_field): Call handle_warn_if_not_align.
17349 Copy TYPE_WARN_IF_NOT_ALIGN.
17350 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17351 (layout_type): Likewise.
17352 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17354 (tree_decl_common): Add warn_if_not_align.
17355 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17356 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17357 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17358 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17359 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17360 * doc/extend.texi: Document warn_if_not_aligned attribute.
17361 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17363 2017-08-17 Martin Liska <mliska@suse.cz>
17366 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17367 (get_data_dependence): Use it as pointer type.
17368 (distribute_loop): Likewise.
17370 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17372 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17373 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17374 (p8_vmrgow_<mode>_direct): New define_insn.
17375 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17376 handle endianness for vmrgew and vmrgow permute patterns.
17378 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17380 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17381 * config/rs6000/rs6000-cpus.def: Remove comment.
17382 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17383 (POWERPC_MASKS): Likewise.
17384 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17385 use of TARGET_VSX_TIMODE.
17386 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17387 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17388 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17389 (rs6000_option_override_internal): Remove dead code.
17390 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17391 (rs6000_legitimize_reload_address): Likewise.
17392 (rs6000_legitimate_address_p): Likewise.
17393 (rs6000_opt_masks): Delete "vsx-timode".
17394 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17395 from function comment.
17396 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17397 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17398 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17400 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17401 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17402 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17403 (VSX_TI): Likewise.
17405 (define_peephole2): Likewise.
17407 2017-08-17 Martin Sebor <msebor@redhat.com>
17410 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17411 past the end of an array.
17412 (test_pp_format): Add test cases.
17414 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17416 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17417 missing ECF_NOTHROW flags.
17419 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17422 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17423 operands residing in integer registers.
17424 (*vsx_le_perm_load_<mode>): Likewise.
17425 (*vsx_le_perm_store_<mode>): Likewise.
17426 (define_peephole2): Add peepholes to optimize the above.
17428 2017-08-17 Marek Polacek <polacek@redhat.com>
17430 PR middle-end/81814
17431 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17432 to mimic what shorten_compare did. Change the return type to bool.
17433 (fold_cond_expr_with_comparison): Update call to
17434 operand_equal_for_comparison_p.
17435 (fold_ternary_loc): Likewise.
17437 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17439 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17441 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17443 2017-08-17 Richard Biener <rguenther@suse.de>
17445 * tree-ssa-structalias.c (solve_graph): When propagating
17446 to successors update the graphs succ edges and avoid duplicate work.
17448 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17451 * config/i386/i386.c (ix86_option_override_internal): Save target
17452 specific options after ix86_stack_protector_guard_reg was changed.
17454 2017-08-17 Richard Biener <rguenther@suse.de>
17456 PR tree-optimization/81827
17457 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17458 (new_var_info): Initialize it conservatively.
17459 (get_call_vi): Mark register vars.
17460 (new_scalar_tmp_constraint_exp): Likewise.
17461 (handle_rhs_call): Likewise.
17462 (handle_const_call): Likewise.
17463 (create_function_info_for): Likewise.
17464 (solve_constraints): Sort varinfos to separate register from
17465 non-register vars to pack points-to solution bitmaps during
17468 2017-08-17 Marek Polacek <polacek@redhat.com>
17470 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17472 2017-08-17 Richard Biener <rguenther@suse.de>
17474 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17475 to true when overflow is undefined and we saturated the result.
17477 2017-08-17 Alan Modra <amodra@gmail.com>
17480 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17481 Don't use store multiple if only one reg needs saving.
17482 (interesting_frame_related_regno): New function.
17483 (rs6000_frame_related): Don't emit frame info for regs that
17485 (rs6000_emit_epilogue): Likewise.
17487 2017-08-16 Nathan Sidwell <nathan@acm.org>
17489 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17490 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17491 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17492 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17493 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17494 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17495 (verify_type): Adjust for TYPE_BINFO move.
17496 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17497 process TYPE_BINFO directly.
17498 (hash_tree): Likewise.
17499 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17501 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17504 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17506 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17508 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17511 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17512 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17513 Add ix86_binary_operator_ok to insn constraint.
17514 (*iordi_1_bts): Ditto.
17515 (*xordi_1_btc): Ditto.
17516 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17517 Update corresponding peephole2 pattern.
17521 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17523 PR tree-optimization/81832
17524 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17525 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17527 2017-08-16 Marek Polacek <polacek@redhat.com>
17530 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17531 regarding size_zero_node.
17533 2017-08-16 Martin Liska <mliska@suse.cz>
17536 * config.gcc: Respect previously set extra_objs in case
17539 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17541 PR tree-optimization/81835
17542 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17543 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17544 not depend on the phi.
17546 2017-08-16 Alan Modra <amodra@gmail.com>
17548 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17551 2017-08-16 Alan Modra <amodra@gmail.com>
17553 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17554 (save_reg_p): ..into this. Update all callers.
17555 (first_reg_to_save): Simplify.
17557 2017-08-16 Alan Modra <amodra@gmail.com>
17559 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17562 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17566 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17567 constant count if that count is less than 32.
17569 2017-08-15 Nathan Sidwell <nathan@acm.org>
17571 * gcc.c (execute): Emit friendlier message if inferior is killed
17572 by an external cause.
17574 2017-08-15 Richard Biener <rguenther@suse.de>
17576 PR tree-optimization/81790
17577 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17578 CONSTRUCTORs from simplifying and VN.
17580 2017-08-14 Martin Sebor <msebor@redhat.com>
17582 * builtin-attrs.def: Add comments.
17584 2017-08-14 Martin Sebor <msebor@redhat.com>
17587 * doc/extend.texi (attribute nonstring): Document new attribute.
17589 2017-08-14 Martin Sebor <msebor@redhat.com>
17592 * tree-diagnostic.c (default_tree_printer): Handle %G.
17593 * gimple-pretty-print.h (percent_G_format): Declare new function.
17594 * gimple-pretty-print.c (percent_G_format): Define.
17595 * tree-pretty-print.c (percent_K_format): Add argument.
17597 2017-08-14 Martin Sebor <msebor@redhat.com>
17599 PR translation/79998
17600 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17601 Remove a stray space.
17603 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17606 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17607 (*iordi_1_bts): Ditto.
17608 (*xordi_1_btc): Ditto.
17610 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17613 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17614 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17616 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17617 quoted strings, and make more translator-friendly.
17618 (darwin_rs6000_override_options): Likewise.
17619 (rs6000_option_override_internal): Likewise.
17620 (rs6000_return_in_memory): Fix overlong line.
17621 (init_cmulative_args): Use quoted strings, and make more
17622 translator-friendly.
17623 (rs6000_pass_by_reference): Fix overlong line.
17624 (def_builtin): Use quoted strings.
17625 (altivec_expand_predicate_builtin): Use quoted strings, and make
17626 more translator-friendly.
17627 (htm_expand_builtin): Use quoted strings.
17628 (cpu_expand_builtin): Use quoted strings, and make more
17629 translator-friendly.
17630 (altivec_expand_builtin): Likewise.
17631 (paired_expand_predicate_builtin): Likewise.
17632 (rs6000_invalid_builtin): Likewise.
17633 (builtin_function_type): Use quoted strings.
17634 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17635 more translator-friendly.
17636 (rs6000_trampoline_init): Likewise.
17637 (rs6000_handle_altivec_attribute): Likewise.
17638 (rs6000_inner_target_options): Use quoted strings.
17639 (rs6000_disable_incompatible_switches): Likewise.
17640 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17641 strings, and make more translator-friendly.
17642 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17644 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17646 PR tree-optimization/81799
17647 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17648 cond_expr to simple gimple operand.
17650 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17652 PR middle-end/46932
17653 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17655 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17659 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17660 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17661 TARGET_GASISR_PROLOGUES.
17662 * config/avr/avr.c (avr_option_override): Same.
17663 (avr_pass_pre_proep::execute): Same.
17665 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17668 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17669 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17671 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17673 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17674 define_insn_and_split. Split to a memory load from 0 in
17675 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17676 using PTR mode iterator.
17677 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17678 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17679 (*add_tp_<mode>): Redefine as define_insn_and_split.
17680 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17681 address space. Merge with *add_tp_x32 using PTR mode iterator.
17682 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17683 Split to an add with a memory load from 0 in
17684 DEFAULT_TLS_SEG_REG address space.
17686 2017-08-12 Andrew Pinski <apinski@cavium.com>
17688 * config/aarch64/aarch64-option-extensions.def (rdma):
17689 Fix feature string to what Linux prints out in /proc/cpuinfo.
17691 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17694 * dwarf2out.c (modified_type_die): For C typedef types that have
17695 an ultimate origin, process the ultimate origin instead of the
17697 (gen_typedef_die): Assert that input DECLs have no ultimate
17699 (gen_type_die_with_usage): For typedef variants that have an
17700 ultimate origin, just call gen_decl_die on the original DECL.
17701 (process_scope_var): Avoid creating DIEs for local typedefs and
17702 concrete static variables.
17704 2017-08-12 Alan Modra <amodra@gmail.com>
17708 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17709 match gnu-user.h startfile.
17710 (ENDFILE_LINUX_SPEC): Similarly.
17712 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17715 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17717 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17719 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17720 * config/aarch64/aarch64.md (mov<mode>): Change.
17721 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17722 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17723 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17725 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17727 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17728 for storage order barriers.
17730 2017-08-11 Martin Liska <mliska@suse.cz>
17733 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17734 variables of void type.
17736 2017-08-11 Martin Liska <mliska@suse.cz>
17738 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17739 TARGET_SUPPORTS_ALIASES.
17740 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17741 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17742 (optimize_weakref): Likewise.
17743 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17744 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17745 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17747 2017-08-11 Martin Liska <mliska@suse.cz>
17750 * config/i386/i386.c (make_resolver_func): Do complete
17751 refactoring of the function.
17753 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17756 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17757 * config/i386/i386.c (ix86_stack_protect_guard): Use
17758 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17759 * doc/invoke.texi (x86 Options): Document
17760 -mstack-protector-guard-symbol= option.
17762 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17764 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17765 * config/i386/i386.c (ix86_split_stack_guard): New function.
17766 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17767 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17768 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17769 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17770 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17771 (split_stack_space_check): Call ix86_split_stack_guard.
17773 2017-08-10 Martin Sebor <msebor@redhat.com>
17775 * print-tree.c (print_node): Print location using the established
17777 Replace spaces with colons.
17778 (debug_raw, debug): Ditto.
17780 2017-08-10 Martin Sebor <msebor@redhat.com>
17783 * pretty-print.c (pp_format): Correct the handling of %s precision.
17785 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17788 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17790 (ix86_finalize_stack_frame_flags): This. Also clear
17791 frame_pointer_needed if -fno-omit-frame-pointer is used without
17793 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17794 with ix86_finalize_stack_frame_flags.
17795 (ix86_expand_epilogue): Likewise.
17796 (ix86_expand_split_stack_prologue): Likewise.
17797 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17799 2017-08-10 Martin Liska <mliska@suse.cz>
17802 * c-attribs.c (handle_target_attribute):
17803 Report warning for an empty string argument of target attribute.
17805 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17808 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17810 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17811 or DECL_NONLOCAL labels.
17812 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17813 or DECL_NONLOCAL labels here.
17815 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
17817 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
17818 to indicate when early gimple folding has been disabled.
17819 (rs6000_gimple_fold_builtin): Add debug content.
17820 (rs6000_invalid_builtin): Fix whitespace.
17821 (rs6000_expand_builtin): Fix whitespace.
17822 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
17824 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
17827 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
17828 SAVE_MULTIPLE if not all the registers that saves, should be saved.
17830 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
17832 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
17833 (qdf24xx): Likewise.
17834 * config/aarch64/aarch64.md: Include falkor.md.
17835 * config/aarch64/falkor.md: New.
17837 2017-08-09 Marek Polacek <polacek@redhat.com>
17840 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
17841 * diagnostic.c (emit_diagnostic): Add a comment.
17842 (emit_diagnostic_valist): New function.
17844 2017-08-09 Marek Polacek <polacek@redhat.com>
17847 * input.c (make_location): New overload.
17848 * input.h (make_location): Declare.
17850 2017-08-08 Alan Modra <amodra@gmail.com>
17851 H.J. Lu <hongjiu.lu@intel.com>
17854 * gcc.c (NO_PIE_SPEC): Delete.
17855 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
17857 (LINK_PIE_SPEC): ..to here.
17858 (LINK_COMMAND_SPEC): Support -no-pie.
17859 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
17860 chain of crtbegin*.o selection, update for PIE_SPEC changes and
17862 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
17863 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
17864 (ENDFILE_CRTEND_SPEC): Similarly.
17866 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
17869 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
17870 (mstack-protector-guard-offset=): Ditto.
17871 * config/i386/i386.c (ix86_option_override): Handle
17872 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
17874 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
17875 ix86_stack_protect_guard_offset variables.
17876 (TARGET_STACK_PROTECT_GUARD): Always define.
17877 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
17878 and -mstack-protector-guard-offset= options.
17880 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17882 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
17883 boundary case for the last candidate.
17885 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17887 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
17888 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
17890 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17892 PR middle-end/19706
17893 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
17894 * config/aarch64/aarch64-builtins.c
17895 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
17896 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
17897 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
17899 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17900 Andrew Pinski <pinskia@gmail.com>
17902 PR middle-end/19706
17903 * internal-fn.def (XORSIGN): New.
17904 * optabs.def (xorsign_optab): New.
17905 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
17906 (convert_expand_mult_copysign): New.
17907 (pass_optimize_widening_mul::execute): Call
17908 convert_expand_mult_copysign.
17910 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17912 PR tree-optimization/81354
17913 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
17914 Insert on edges rather than explicitly creating landing pads.
17915 (analyze_candidates_and_replace): Commit edge inserts.
17917 2017-08-08 Richard Biener <rguenther@suse.de>
17919 PR middle-end/81719
17920 * tree-ssa-loop-niter.c: Include tree-dfa.h.
17921 (expand_simple_operations): Also look through ADDR_EXPRs with
17922 MEM_REF bases treating them as POINTER_PLUS_EXPR.
17924 2017-08-08 Richard Biener <rguenther@suse.de>
17926 PR tree-optimization/81723
17927 * tree-vect-slp.c (struct bst_traits): New hash traits.
17928 (bst_fail): New global.
17929 (vect_build_slp_tree_2): New worker, split out from ...
17930 (vect_build_slp_tree): ... this now wrapping it with using
17931 bst_fail set to cache SLP tree build fails. Properly handle
17933 (vect_analyze_slp_instance): Allocate and free bst_fail.
17935 2017-08-08 Martin Liska <mliska@suse.cz>
17938 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
17939 LABEL_DECLs that can be from a different function.
17941 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17943 PR tree-optimization/81744
17944 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
17945 loop's number of iterations.
17947 2017-08-08 Martin Liska <mliska@suse.cz>
17949 * asan.c: Include header files.
17950 * attribs.c (build_decl_attribute_variant): New function moved
17952 (build_type_attribute_qual_variant): Likewise.
17953 (cmp_attrib_identifiers): Likewise.
17954 (simple_cst_list_equal): Likewise.
17955 (omp_declare_simd_clauses_equal): Likewise.
17956 (attribute_value_equal): Likewise.
17957 (comp_type_attributes): Likewise.
17958 (build_type_attribute_variant): Likewise.
17959 (lookup_ident_attribute): Likewise.
17960 (remove_attribute): Likewise.
17961 (merge_attributes): Likewise.
17962 (merge_type_attributes): Likewise.
17963 (merge_decl_attributes): Likewise.
17964 (merge_dllimport_decl_attributes): Likewise.
17965 (handle_dll_attribute): Likewise.
17966 (attribute_list_equal): Likewise.
17967 (attribute_list_contained): Likewise.
17968 * attribs.h (lookup_attribute): New function moved from tree.[ch].
17969 (lookup_attribute_by_prefix): Likewise.
17970 * bb-reorder.c: Include header files.
17971 * builtins.c: Likewise.
17972 * calls.c: Likewise.
17973 * cfgexpand.c: Likewise.
17974 * cgraph.c: Likewise.
17975 * cgraphunit.c: Likewise.
17976 * convert.c: Likewise.
17977 * dwarf2out.c: Likewise.
17978 * final.c: Likewise.
17979 * fold-const.c: Likewise.
17980 * function.c: Likewise.
17981 * gimple-expr.c: Likewise.
17982 * gimple-fold.c: Likewise.
17983 * gimple-pretty-print.c: Likewise.
17984 * gimple.c: Likewise.
17985 * gimplify.c: Likewise.
17986 * hsa-common.c: Likewise.
17987 * hsa-gen.c: Likewise.
17988 * internal-fn.c: Likewise.
17989 * ipa-chkp.c: Likewise.
17990 * ipa-cp.c: Likewise.
17991 * ipa-devirt.c: Likewise.
17992 * ipa-fnsummary.c: Likewise.
17993 * ipa-inline.c: Likewise.
17994 * ipa-visibility.c: Likewise.
17996 * lto-cgraph.c: Likewise.
17997 * omp-expand.c: Likewise.
17998 * omp-general.c: Likewise.
17999 * omp-low.c: Likewise.
18000 * omp-offload.c: Likewise.
18001 * omp-simd-clone.c: Likewise.
18002 * opts-global.c: Likewise.
18003 * passes.c: Likewise.
18004 * predict.c: Likewise.
18005 * sancov.c: Likewise.
18006 * sanopt.c: Likewise.
18007 * symtab.c: Likewise.
18008 * toplev.c: Likewise.
18009 * trans-mem.c: Likewise.
18010 * tree-chkp.c: Likewise.
18011 * tree-eh.c: Likewise.
18012 * tree-into-ssa.c: Likewise.
18013 * tree-object-size.c: Likewise.
18014 * tree-parloops.c: Likewise.
18015 * tree-profile.c: Likewise.
18016 * tree-ssa-ccp.c: Likewise.
18017 * tree-ssa-live.c: Likewise.
18018 * tree-ssa-loop.c: Likewise.
18019 * tree-ssa-sccvn.c: Likewise.
18020 * tree-ssa-structalias.c: Likewise.
18021 * tree-ssa.c: Likewise.
18022 * tree-streamer-in.c: Likewise.
18023 * tree-vectorizer.c: Likewise.
18024 * tree-vrp.c: Likewise.
18025 * tsan.c: Likewise.
18026 * ubsan.c: Likewise.
18027 * varasm.c: Likewise.
18028 * varpool.c: Likewise.
18029 * tree.c: Remove functions moved to attribs.[ch].
18030 * tree.h: Likewise.
18031 * config/aarch64/aarch64.c: Add attrs.h header file.
18032 * config/alpha/alpha.c: Likewise.
18033 * config/arc/arc.c: Likewise.
18034 * config/arm/arm.c: Likewise.
18035 * config/avr/avr.c: Likewise.
18036 * config/bfin/bfin.c: Likewise.
18037 * config/c6x/c6x.c: Likewise.
18038 * config/cr16/cr16.c: Likewise.
18039 * config/cris/cris.c: Likewise.
18040 * config/darwin.c: Likewise.
18041 * config/epiphany/epiphany.c: Likewise.
18042 * config/fr30/fr30.c: Likewise.
18043 * config/frv/frv.c: Likewise.
18044 * config/ft32/ft32.c: Likewise.
18045 * config/h8300/h8300.c: Likewise.
18046 * config/i386/winnt.c: Likewise.
18047 * config/ia64/ia64.c: Likewise.
18048 * config/iq2000/iq2000.c: Likewise.
18049 * config/lm32/lm32.c: Likewise.
18050 * config/m32c/m32c.c: Likewise.
18051 * config/m32r/m32r.c: Likewise.
18052 * config/m68k/m68k.c: Likewise.
18053 * config/mcore/mcore.c: Likewise.
18054 * config/microblaze/microblaze.c: Likewise.
18055 * config/mips/mips.c: Likewise.
18056 * config/mmix/mmix.c: Likewise.
18057 * config/mn10300/mn10300.c: Likewise.
18058 * config/moxie/moxie.c: Likewise.
18059 * config/msp430/msp430.c: Likewise.
18060 * config/nds32/nds32-isr.c: Likewise.
18061 * config/nds32/nds32.c: Likewise.
18062 * config/nios2/nios2.c: Likewise.
18063 * config/nvptx/nvptx.c: Likewise.
18064 * config/pa/pa.c: Likewise.
18065 * config/pdp11/pdp11.c: Likewise.
18066 * config/powerpcspe/powerpcspe.c: Likewise.
18067 * config/riscv/riscv.c: Likewise.
18068 * config/rl78/rl78.c: Likewise.
18069 * config/rx/rx.c: Likewise.
18070 * config/s390/s390.c: Likewise.
18071 * config/sh/sh.c: Likewise.
18072 * config/sol2.c: Likewise.
18073 * config/sparc/sparc.c: Likewise.
18074 * config/spu/spu.c: Likewise.
18075 * config/stormy16/stormy16.c: Likewise.
18076 * config/tilegx/tilegx.c: Likewise.
18077 * config/tilepro/tilepro.c: Likewise.
18078 * config/v850/v850.c: Likewise.
18079 * config/vax/vax.c: Likewise.
18080 * config/visium/visium.c: Likewise.
18081 * config/xtensa/xtensa.c: Likewise.
18083 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18086 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18087 constraints since the -mupper-regs-* switches have been
18089 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18090 into a vector from a double word element that was extracted from
18091 another vector, and eliminate extra XXPERMDI instructions.
18092 (vsx_concat_<mode>_2): Likewise.
18093 (vsx_concat_<mode>_3): Likewise.
18094 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18095 concat to allow optimizing inserts from previous extracts.
18097 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18099 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18100 memory reference to a SSP offset in TLS address space.
18101 (ix86_print_operand) <case '@'>: Remove.
18102 (ix86_print_operand_punct_valid_p): Remove '@' code.
18103 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18104 UNSPEC_SP_TLS_TEST.
18105 (stack_tls_protect_set_<mode>): Remove.
18106 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18107 (stack_tls_protect_test_<mode>): Remove.
18108 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18110 2017-08-07 Olivier Hainque <hainque@adacore.com>
18113 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18115 2017-08-07 Douglas Rupp <rupp@adacore.com>
18117 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18118 variable was referenced as multidir in command.
18120 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18123 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18126 2017-08-07 Martin Liska <mliska@suse.cz>
18128 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18129 * config/rl78/rl78.c: Add include of attribs.h.
18130 * config/sh/sh.c: Likewise.
18131 * config/v850/v850.c: Likewise.
18133 2017-08-07 Tom de Vries <tom@codesourcery.com>
18135 PR middle-end/78266
18136 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18138 2017-08-07 Martin Liska <mliska@suse.cz>
18140 * config/mips/mips.c: Include attribs.h.
18142 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18145 * doc/invoke.texi: Document change in behvaior for -Ofast for
18148 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18150 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18152 (aarch64_pop_regs): Likewise.
18153 (aarch64_gen_load_pair): Likewise.
18154 (aarch64_save_callee_saves): Likewise.
18155 (aarch64_restore_callee_saves): Likewise.
18157 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18159 * config/i386/i386.c: Revert the last change.
18161 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18164 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18166 (ix86_finalize_stack_frame_flags): This. Also clear
18167 frame_pointer_needed if -fno-omit-frame-pointer is used without
18169 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18170 with ix86_finalize_stack_frame_flags.
18171 (ix86_expand_epilogue): Likewise.
18172 (ix86_expand_split_stack_prologue): Likewise.
18174 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18177 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18178 to P_AES for Westmere.
18180 2017-08-07 Jonathan Yong <10walls@gmail.com>
18182 * config/i386/mingw.opt (fset-stack-executable): Removed.
18183 * config/i386/cygming.opt (fset-stack-executable): Moved
18185 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18187 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18189 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18191 2017-08-07 Marek Polacek <polacek@redhat.com>
18193 PR middle-end/81737
18194 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18196 2017-08-07 Martin Liska <mliska@suse.cz>
18198 * attribs.h (canonicalize_attr_name): New function.
18199 (cmp_attribs): Move from c-format.c and adjusted.
18200 (is_attribute_p): Moved from tree.h.
18201 * tree-inline.c: Add new includes.
18202 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18203 (private_is_attribute_p): Remove.
18204 (private_lookup_attribute): Likewise.
18205 (private_lookup_attribute_by_prefix): Simplify.
18206 (remove_attribute): Use is_attribute_p.
18207 * tree.h: Remove removed declarations.
18209 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18211 PR middle-end/81698
18212 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18213 instead of computing it in the function. Formatting fix.
18214 (expand_case): Don't rely on default_edge being the first edge,
18215 clear it if removing it, pass default_edge to
18216 emit_case_dispatch_table.
18217 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18220 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18222 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18223 insn in the function, emit NOP after the insn.
18225 2017-08-06 Tom de Vries <tom@codesourcery.com>
18227 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18230 2017-08-06 Tom de Vries <tom@codesourcery.com>
18232 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18235 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18237 PR tree-optimization/57371
18238 * match.pd: New pattern.
18240 2017-08-04 Marek Polacek <polacek@redhat.com>
18242 PR middle-end/81695
18243 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18244 perform the computation in offset_int.
18246 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18248 PR tree-optimization/81136
18249 * tree-vectorizer.h: Include tree-hash-traits.h.
18250 (vec_base_alignments): New typedef.
18251 (vec_info): Add a base_alignments field.
18252 (vect_record_base_alignments): Declare.
18253 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18255 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18256 (create_data_ref): Add an is_conditional_in_stmt argument.
18257 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18258 the is_conditional_in_stmt field.
18259 (data_ref_loc): Add an is_conditional_in_stmt field.
18260 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18261 (find_data_references_in_stmt): Update call to create_data_ref.
18262 (graphite_find_data_references_in_stmt): Likewise.
18263 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18264 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18265 (vect_record_base_alignment): New function.
18266 (vect_record_base_alignments): Likewise.
18267 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18268 for nested statements even if we fail to compute a misalignment.
18269 Use pooled base alignments for unconditional references.
18270 (vect_find_same_alignment_drs): Compare base addresses instead
18272 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18273 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18275 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18277 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18278 Add an explicit name for the enum. Use auto_vec for slp_instances
18279 and grouped_stores.
18280 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18282 (_bb_vec_info): Add a constructor and destructor.
18283 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18284 (destroy_loop_vec_info): Delete.
18285 (vect_destroy_datarefs): Likewise.
18286 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18287 (vec_info::vec_info): New function.
18288 (vec_info::~vec_info): Likewise.
18289 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18290 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18291 destroy_loop_vec_info.
18292 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18293 (_loop_vec_info::_loop_vec_info): ...this.
18294 (destroy_loop_vec_info): Replace with...
18295 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18296 the stmt_vec_infos. Leave handling of vec_info information to its
18297 destructor. Remove explicit vector releases.
18298 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18299 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18300 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18301 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18302 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18303 (destroy_bb_vec_info): Replace with...
18304 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18305 information to its destructor.
18306 (vect_slp_analyze_bb_1): Use new and delete instead of
18307 new_bb_vec_info and destroy_bb_vec_info.
18308 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18311 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18313 * tree-data-ref.h (subscript): Add access_fn field.
18314 (data_dependence_relation): Add could_be_independent_p.
18315 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18316 (same_access_functions): Move to tree-data-ref.c.
18317 * tree-data-ref.c (ref_contains_union_access_p): New function.
18318 (access_fn_component_p): Likewise.
18319 (access_fn_components_comparable_p): Likewise.
18320 (dr_analyze_indices): Add a reference to access_fn_component_p.
18321 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18323 (constant_access_functions): Likewise.
18324 (add_other_self_distances): Likewise.
18325 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18326 (initialize_data_dependence_relation): Use XCNEW and remove
18327 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18328 of access functions that have the same type. Allow the
18329 subsequence to end with different bases in some circumstances.
18330 Record the chosen access functions in SUB_ACCESS_FN.
18331 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18332 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18333 (subscript_dependence_tester_1): Likewise dra and drb.
18334 (build_classic_dist_vector): Update calls accordingly.
18335 (subscript_dependence_tester): Likewise.
18336 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18337 DDR_COULD_BE_INDEPENDENT_P.
18338 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18339 comp_alias_ddrs instead of may_alias_ddrs.
18340 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18342 (vect_analyze_data_ref_dependence): Use it if
18343 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18344 distance vectors if that fails.
18345 (dependence_distance_ge_vf): New function.
18346 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18347 LOOP_VINFO_MAY_ALIAS_DDRS.
18349 2017-08-04 Richard Biener <rguenther@suse.de>
18351 PR middle-end/81705
18352 * fold-const.c (fold_binary_loc): Properly restrict
18353 minus_var0 && minus_var1 case when associating undefined overflow
18356 2017-08-04 Tom de Vries <tom@codesourcery.com>
18358 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18360 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18362 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18363 Don't start diagnostic messages with a capital letter.
18364 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18366 (rs6000_invalid_builtin): Likewise.
18367 (rs6000_trampoline_init): Likewise.
18369 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18372 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18373 after setting changeable df flags.
18375 2017-08-03 Richard Biener <rguenther@suse.de>
18377 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18378 up if the use is in USE - X.
18380 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18382 * toplev.c (dumpfile.h): New include.
18383 (internal_error_reentered): New static function. Use it...
18384 (internal_error_function): ...here to handle reentered internal_error.
18386 2017-08-03 Richard Biener <rguenther@suse.de>
18388 PR middle-end/81148
18389 * fold-const.c (split_tree): Add minus_var and minus_con
18390 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18391 here but always use minus_*.
18392 (associate_trees): Assert we never associate with MINUS_EXPR
18393 and NULL first operand. Do not recurse for PLUS_EXPR operands
18394 when associating as MINUS_EXPR either.
18395 (fold_binary_loc): Track minus_var and minus_con.
18397 2017-08-03 Tom de Vries <tom@codesourcery.com>
18400 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18401 ACCEL_COMPILER, apply finish_options on
18402 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18404 2017-08-03 Tom de Vries <tom@codesourcery.com>
18407 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18408 function_entry_patch_area_size > 0.
18410 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18413 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18414 instead of 10??LU, perform unit multiplication in wide_int,
18415 don't change alloc_object_size_limit if the limit is larger
18418 PR tree-optimization/81655
18419 PR tree-optimization/81588
18420 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18421 the case when ranges[i].low and high are 1 for unsigned type with
18424 PR middle-end/81052
18425 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18426 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18428 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18430 * tree-vrp.h: Add include guard.
18432 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18435 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18436 (ud2): New insn pattern.
18437 * config/i386/i386.c (ix86_expand_epilogue):
18438 For naked functions, generate ud2 instead of trap insn.
18440 2017-08-02 Marek Polacek <polacek@redhat.com>
18443 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18445 2017-08-02 Tom de Vries <tom@codesourcery.com>
18446 Cesar Philippidis <cesar@codesourcery.com>
18448 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18449 Add missing edge probabilities.
18451 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18453 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18454 Correct endianness.
18456 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18458 PR middle-end/79499
18459 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18460 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18461 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18462 prologue_seq sequences - if any.
18464 2017-08-02 Richard Biener <rguenther@suse.de>
18466 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18467 via vectors if supported, integer extracts via punning if supported
18468 or otherwise vector extracts.
18470 2017-08-02 Richard Biener <rguenther@suse.de>
18472 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18474 (bitmap_insert_into_set): ... this.
18476 2017-08-02 Richard Biener <rguenther@suse.de>
18478 PR tree-optimization/81633
18480 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18482 PR tree-optimization/71752
18483 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18485 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18487 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18488 (machine_function::call_ms2sysv_pad_out): Remove field.
18489 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18490 (ix86_compute_frame_layout): Likewise.
18492 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18495 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18496 attribute with interrupt attribute.
18498 2017-08-01 Andrew Pinski <apinski@cavium.com>
18500 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18501 BIT_INSERT_EXPR's operand 1
18502 to see if the types precision matches.
18504 2017-08-01 Martin Liska <mliska@suse.cz>
18506 PR middle-end/70140
18507 * builtins.c (expand_builtin_memcpy_args): Remove.
18508 (expand_builtin_memcpy): Call newly added function
18509 expand_builtin_memory_copy_args.
18510 (expand_builtin_memcpy_with_bounds): Likewise.
18511 (expand_builtin_mempcpy): Remove last argument.
18512 (expand_builtin_mempcpy_with_bounds): Likewise.
18513 (expand_builtin_memory_copy_args): New function created from
18514 expand_builtin_mempcpy_args with small modifications.
18515 (expand_builtin_mempcpy_args): Remove.
18516 (expand_builtin_stpcpy): Remove unused argument.
18517 (expand_builtin): Likewise.
18518 (expand_builtin_with_bounds): Likewise.
18520 2017-08-01 Martin Liska <mliska@suse.cz>
18523 Make mempcpy more optimal (PR middle-end/70140).
18525 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18528 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18529 __builtin_vec_cmpne verify both arguments are compatible vectors
18530 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18531 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18532 move computation of aligned to after checking the argument types.
18536 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18537 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18540 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18541 Doug Rupp <rupp@adacore.com>
18542 Olivier Hainque <hainque@adacore.com>
18544 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18545 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18547 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18548 for TARGET_OS_CPP_BUILTIN.
18549 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18550 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18552 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18553 passing required abi options to the assembler for EABI configurations.
18554 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18555 of .text.hot and .text.unlikely sections for kernel modules when
18556 using ARM style exceptions.
18557 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18558 options. Add EXTRA_CC1_SPEC.
18559 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18561 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18563 (ARM_TARGET2_DWARF_FORMAT): Define.
18564 * config/arm/t-vxworks: Adjust multilib control to removal of the
18565 Diab command line options.
18567 2017-08-01 Martin Liska <mliska@suse.cz>
18569 PR gcov-profile/81561
18570 * gcov.c (unblock): Make unblocking safe as we need to preserve
18571 index correspondence of blocks and block_lists.
18573 2017-08-01 Richard Biener <rguenther@suse.de>
18575 PR tree-optimization/81181
18576 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18577 (compute_antic): ... end of iteration here.
18579 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18581 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18582 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18583 (ftree-slp-vectorize): Likewise.
18584 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18585 can no longer be set independent of flag_tree_loop_vectorize.
18586 * omp-general.c (emp_max_vf): Likewise.
18587 * opts.c (enable_fdo_optimizations): Remove references to
18588 flag_tree_vectorize, these are now implicit.
18589 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18590 and leave it for the options machinery.
18592 2017-08-01 Martin Liska <mliska@suse.cz>
18594 PR middle-end/70140
18595 * builtins.c (expand_builtin_memcpy_args): Remove.
18596 (expand_builtin_memcpy): Call newly added function
18597 expand_builtin_memory_copy_args.
18598 (expand_builtin_memcpy_with_bounds): Likewise.
18599 (expand_builtin_mempcpy): Remove last argument.
18600 (expand_builtin_mempcpy_with_bounds): Likewise.
18601 (expand_builtin_memory_copy_args): New function created from
18602 expand_builtin_mempcpy_args with small modifications.
18603 (expand_builtin_mempcpy_args): Remove.
18604 (expand_builtin_stpcpy): Remove unused argument.
18605 (expand_builtin): Likewise.
18606 (expand_builtin_with_bounds): Likewise.
18608 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18611 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18612 print "ds:" only for immediates in generic address space.
18614 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18617 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18618 (ix86_function_ok_for_sibcall): Return false for naked functions.
18620 2017-08-01 Richard Biener <rguenther@suse.de>
18622 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18623 (compute_antic): Seed worklist with exit block predecessors.
18624 * cfganal.c (dfs_find_deadend): For a cycle return the source
18625 of the edge closing it.
18627 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18629 * config/aarch64/aarch64.c
18630 (aarch64_can_const_movi_rtx_p): Move 0 check.
18632 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18634 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18635 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18637 * match.pd: Ditto in address comparison pattern.
18639 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18641 PR tree-optimization/81627
18642 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18643 closed ssa form for store-store chain.
18645 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18647 PR tree-optimization/81620
18648 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18649 for store-store chain.
18651 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18653 PR tree-optimization/81588
18654 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18655 ranges[i].in_p, invert comparison code ccode. For >/>=,
18656 swap rhs1 and rhs2 and comparison code unconditionally,
18657 for </<= don't do that. Don't swap rhs1/rhs2 again if
18658 ranges[i].in_p, instead invert comparison code ccode if
18659 opcode or oe->rank is BIT_IOR_EXPR.
18662 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18663 a direct optab to conversion optab.
18664 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18665 with GET_MODE_INNER as last argument instead of optab_handler.
18666 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18667 vector extraction if possible and optab is available.
18668 * expr.c (store_constructor): Use convert_optab_handler instead
18669 of optab_handler. Use vector initialization from smaller
18670 vectors if possible and optab is available.
18671 * tree-vect-stmts.c (vectorizable_load): Likewise.
18672 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18673 now have two modes.
18674 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18675 of vec_init from half-sized vectors with the same element mode.
18676 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18677 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18678 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18679 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18680 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18681 after mode in gen_vec_extract* calls.
18682 (vec_extract<mode>): Renamed to ...
18683 (vec_extract<mode><ssescalarmodelower>): ... this.
18684 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18685 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18686 element mode after mode in gen_vec_init* calls.
18687 (VEC_INIT_HALF_MODE): New mode iterator.
18688 (vec_init<mode>): Renamed to ...
18689 (vec_init<mode><ssescalarmodelower>): ... this.
18690 (vec_init<mode><ssehalfvecmodelower>): New expander.
18691 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18692 (vec_extractv2sfsf): ... this.
18693 (vec_initv2sf): Renamed to ...
18694 (vec_initv2sfsf): ... this.
18695 (vec_extractv2si): Renamed to ...
18696 (vec_extractv2sisi): ... this.
18697 (vec_initv2si): Renamed to ...
18698 (vec_initv2sisi): ... this.
18699 (vec_extractv4hi): Renamed to ...
18700 (vec_extractv4hihi): ... this.
18701 (vec_initv4hi): Renamed to ...
18702 (vec_initv4hihi): ... this.
18703 (vec_extractv8qi): Renamed to ...
18704 (vec_extractv8qiqi): ... this.
18705 (vec_initv8qi): Renamed to ...
18706 (vec_initv8qiqi): ... this.
18707 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18708 (vec_init<mode>): Renamed to ...
18709 (vec_init<mode><VEC_base_l>): ... this.
18710 (vec_extract<mode>): Renamed to ...
18711 (vec_extract<mode><VEC_base_l>): ... this.
18712 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18713 (vec_initv2sfsf): ... this.
18714 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18715 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18716 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18717 element mode after mode in gen_vec_init* calls.
18718 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18719 (vec_init<mode><Vel>): ... this.
18720 (vec_extract<mode>): Renamed to ...
18721 (vec_extract<mode><Vel>): ... this.
18722 * config/aarch64/iterators.md (Vel): New mode attribute.
18723 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18724 Add element mode after mode in gen_vec_extract* calls.
18725 * config/s390/vector.md (non_vec_l): New mode attribute.
18726 (vec_extract<mode>): Renamed to ...
18727 (vec_extract<mode><non_vec_l>): ... this.
18728 (vec_init<mode>): Renamed to ...
18729 (vec_init<mode><non_vec_l>): ... this.
18730 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18731 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18733 * config/arm/iterators.md (V_elem_l): New mode attribute.
18734 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18735 (vec_extract<mode><V_elem_l>): ... this.
18736 (vec_extractv2di): Renamed to ...
18737 (vec_extractv2didi): ... this.
18738 (vec_init<mode>): Renamed to ...
18739 (vec_init<mode><V_elem_l>): ... this.
18740 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18741 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18742 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18743 Add element mode after gen_vec_extract* calls.
18744 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18745 (vec_init<mode><unitmode>): ... this.
18746 (vec_extract<mode>): Renamed to ...
18747 (vec_extract<mode><unitmode>): ... this.
18748 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18749 (vec_init<mode><unitmode>): ... this.
18750 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18751 (vec_initv2sfsf): ... this.
18752 (vec_extractv2sf): Renamed to ...
18753 (vec_extractv2sfsf): ... this.
18754 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18755 Add element mode after gen_vec_extract* calls.
18756 * config/mips/mips.md (unitmode): New mode iterator.
18757 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18758 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18759 * config/spu/spu.md (inner_l): New mode attribute.
18760 (vec_init<mode>): Renamed to ...
18761 (vec_init<mode><inner_l>): ... this.
18762 (vec_extract<mode>): Renamed to ...
18763 (vec_extract<mode><inner_l>): ... this.
18764 * config/sparc/sparc.md (veltmode): New mode iterator.
18765 (vec_init<VMALL:mode>): Renamed to ...
18766 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18767 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18768 (vec_initv2sisi): ... this.
18769 (vec_initv2sf): Renamed to ...
18770 (vec_initv2sfsf): ... this.
18771 (vec_extractv2sf): Renamed to ...
18772 (vec_extractv2sfsf): ... this.
18773 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18774 (vec_init<mode>): Renamed to ...
18775 (vec_init<mode><VEC_base_l>): ... this.
18776 (vec_extract<mode>): Renamed to ...
18777 (vec_extract<mode><VEC_base_l>): ... this.
18778 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18779 (vec_initv2sfsf): ... this.
18780 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18781 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18782 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18783 gen_vec_init* calls.
18785 2017-08-01 Richard Biener <rguenther@suse.de>
18787 PR tree-optimization/81297
18788 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18789 TREE_OVERFLOW from INTEGER_CSTs.
18791 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18793 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18795 2017-07-31 Carl Love <cel@us.ibm.com>
18797 * config/rs6000/rs6000-c: Add support for built-in functions
18798 vector signed char vec_xl_be (signed long long, signed char *);
18799 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18800 vector signed int vec_xl_be (signed long long, signed int *);
18801 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18802 vector signed long long vec_xl_be (signed long long, signed long long *);
18803 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18804 vector signed short vec_xl_be (signed long long, signed short *);
18805 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18806 vector double vec_xl_be (signed long long, double *);
18807 vector float vec_xl_be (signed long long, float *);
18808 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18809 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18810 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18812 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18813 (altivec_expand_builtin): Add switch statement to call
18814 altivec_expand_xl_be for each builtin.
18815 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
18816 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
18817 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
18818 __builtin_vsx_le_be_v16qi.
18819 * doc/extend.texi: Update the built-in documentation file for the
18820 new built-in functions.
18822 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18825 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
18827 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18829 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18831 * config.gcc: Add z14.
18832 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
18833 CPU model numbers for z13s and z14.
18834 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
18836 * config/s390/s390-opts.h (enum processor_type): Rename
18837 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18838 * config/s390/s390.c (processor_table): Add field for CPU name to
18839 be passed to Binutils.
18840 (s390_asm_output_machine_for_arch): Use the new field in
18841 processor_table for Binutils.
18842 (s390_expand_builtin): Replace arch12 with z14.
18843 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18844 (s390_get_sched_attrmask): Likewise.
18845 (s390_get_unit_mask): Likewise.
18846 * config/s390/s390.opt: Add z14 to processor_type enum.
18848 2017-07-31 Martin Jambor <mjambor@suse.cz>
18851 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
18852 regardless of optimization level.
18854 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
18855 Martin Liska <mliska@suse.cz>
18857 * predict.def: Remove old comment and adjust probability.
18858 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
18859 PREDICT statements.
18861 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18864 * config/i386/i386.c (ix86_function_naked): New function.
18865 (ix86_can_use_return_insn_p): Return false for naked functions.
18866 (ix86_expand_prologue): Skip prologue for naked functions.
18867 (ix86_expand_epilogue): Skip epilogue for naked functions
18868 and emit trap instruction.
18869 (ix86_warn_func_return): New function.
18870 (ix86_attribute_table): Add "naked" attribute specification.
18871 (TARGET_WARN_FUNC_RETURN): Define.
18872 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
18874 2017-07-31 Martin Liska <mliska@suse.cz>
18876 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
18877 (dump_gimple_bb_header): Always dump BB info.
18878 (pp_cfg_jump): Do not append info about BB when dumping a jump.
18880 2017-07-31 Martin Liska <mliska@suse.cz>
18883 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
18884 also with current_function_decl non-null equality.
18886 2017-07-31 Jakub Jelinek <jakub@redhat.com>
18889 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
18890 change type to the element type, instead add eltype variable and
18891 use it where we are interested in the element type.
18893 PR tree-optimization/81603
18894 * ipa-polymorphic-call.c
18895 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
18896 offset arithmetic in offset_int, bail out if the resulting bit offset
18897 doesn't fit into shwi.
18899 2017-07-31 Martin Liska <mliska@suse.cz>
18901 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
18902 (gimplify_save_expr): Fix comment.
18904 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
18907 * config/i386/i386.c (ix86_function_arg): Update arguments for
18909 (ix86_compute_frame_layout): Set the initial stack offset to
18910 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
18911 INCOMING_FRAME_SP_OFFSET.
18912 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
18913 stack before exception handler returns.
18914 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
18915 the 'ERROR_CODE' for exception handler.
18917 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
18919 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
18920 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
18921 (ASM_OUTPUT_REG_POP): Ditto.
18922 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
18923 instead of asm_fprintf to output pure string.
18925 2017-07-29 Jakub Jelinek <jakub@redhat.com>
18927 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
18928 to imported_module_or_decl hook.
18929 (debug_nothing_tree_tree_tree_bool): Remove.
18930 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
18931 * debug.c (do_nothing_debug_hooks): Use
18932 debug_nothing_tree_tree_tree_bool_bool instead of
18933 debug_nothing_tree_tree_tree_bool.
18934 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18935 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
18936 * sdbout.c (sdb_debug_hooks): Likewise.
18937 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
18938 (gen_namespace_die): Add DW_AT_export_symbols attribute if
18940 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
18941 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
18942 attribute, don't add anything.
18944 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18946 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
18947 (fold_build2_stat_loc): Likewise.
18948 (fold_build3_stat_loc): Likewise.
18949 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
18950 (fold_build1_loc): Remove macro.
18951 (fold_build2_loc): Likewise.
18952 (fold_build3_loc): Likewise.
18954 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18956 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
18957 (gimple_build_debug_bind_source_stat): Likewise.
18958 * gimple.h (gimple_build_debug_bind): Remove macro.
18959 (gimple_build_debug_bind_source): Likewise.
18961 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18963 * bitmap.c (bitmap_alloc): Adjust.
18964 (bitmap_gc_alloc): Likewise.
18965 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
18967 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18969 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
18970 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
18971 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
18972 (bitmap_gc_alloc_stat): Likewise.
18973 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
18975 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18977 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
18978 * rtl.h (shallow_copy_rtx): Remove macro.
18980 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18982 * emit-rtl.c (gen_raw_REG): Adjust.
18983 * gengenrtl.c (gendef): Likewise.
18984 * rtl.c (rtx_alloc_stat): Remove _stat from name.
18985 * rtl.h (rtx_alloc): Remove macro.
18987 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18989 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
18990 (build_tree_list_stat): Likewise.
18991 * tree.h (build_tree_list): Remove macro.
18992 (build_tree_list_vec): Likewise.
18994 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18996 * tree.c (make_vector_stat): Remove _stat from name.
18997 (build_vector_stat): Likewise.
18998 * tree.h (make_vector_stat): Remove macro.
18999 (build_vector_stat): Likewise.
19001 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19003 * tree.h (build_var_debug_value): Remove prototype.
19005 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19007 * tree.c (tree_cons_stat): Remove _stat from name.
19008 * tree.h (tree_cons): Remove macro.
19010 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19012 * tree.c (build_vl_exp_stat): Remove _stat from name.
19013 * tree.h (build_vl_exp): Remove macro.
19015 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19017 * tree.c (build_decl_stat): Remove _stat from name.
19018 * tree.h (build_decl): Remove macro.
19020 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19022 * gimple.c (gimple_build_with_ops_stat): Adjust.
19023 (gimple_alloc_stat): Remove _stat from name.
19024 * gimple.h (gimple_alloc): Remove macro.
19026 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19028 * tree.c (make_tree_vec_stat): Remove _stat from name.
19029 (grow_tree_vec_stat): Likewise.
19030 * tree.h (make_tree_vec_stat): Adjust prototype.
19031 (grow_tree_vec_stat): Likewise.
19032 (make_tree_vec): Remove macro.
19033 (grow_tree_vec): Likewise.
19035 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19037 * fold-const.c (fold_build1_stat_loc): Adjust.
19038 (fold_build2_stat_loc): Likewise.
19039 (fold_build3_stat_loc): Likewise.
19040 * tree.c (build0_stat): Remove _stat from name.
19041 (build1_stat): Likewise.
19042 (build2_stat): Likewise.
19043 (build3_stat): Likewise.
19044 (build4_stat): Likewise.
19045 (build5_stat): Likewise.
19046 * tree.h (build1_loc): Remove macro, and rename _stat function
19048 (build2_loc): Likewise.
19049 (build3_loc): Likewise.
19050 (build4_loc): Likewise.
19051 (build5_loc): Likewise.
19053 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19055 * tree.c (make_int_cst_stat): Remove _stat from name.
19056 * tree.h (make_int_cst_stat): Adjust prototype.
19057 (make_int_cst): Remove macro.
19059 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19061 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19062 * tree.h (make_tree_binfo_stat): Adjust prototype.
19063 (make_tree_binfo): Remove.
19065 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19067 * tree.c (copy_node_stat): Rename to copy_node.
19068 (build_distinct_type_copy): Adjust.
19069 * tree.h (copy_node_stat): Adjust prototype.
19070 (copy_node): Remove macro.
19072 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19074 * tree.c (make_node_stat): rename to make_node.
19075 (build_tree_list_stat): Adjust.
19076 (build0_stat): Likewise.
19077 (build2_stat): Likewise.
19078 (build3_stat): Likewise.
19079 (build4_stat): Likewise.
19080 (build5_stat): Likewise.
19081 (build_decl_stat): Likewise.
19082 * tree.h (make_node_stat): Adjust prototype.
19083 (make_node): remove macro.
19085 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19087 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19088 (PPC_FEATURE2_SCV): Likewise.
19089 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19091 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19093 * config/aarch64/aarch64.c
19094 (aarch64_internal_mov_immediate): Add new special pattern.
19095 * config/aarch64/aarch64.md (*movdi_aarch64):
19096 Add reg/32bit const mov case.
19098 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19099 Richard Sandiford <richard.sandiford@linaro.org>
19101 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19102 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19103 Add integer and movi cases.
19104 (movi-split-hf-df-sf split, fp16): New.
19105 (enabled): Added TARGET_FP_F16INST.
19106 * config/aarch64/iterators.md (GPF_HF): New.
19107 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19109 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19111 * config/aarch64/aarch64.c
19112 (aarch64_simd_container_mode): Add prototype.
19113 (aarch64_expand_mov_immediate): Add HI support.
19114 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19115 (aarch64_can_const_movi_rtx_p): New.
19116 (aarch64_preferred_reload_class):
19117 Remove restrictions of using FP registers for certain SIMD operations.
19118 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19119 (aarch64_valid_floating_const): Add integer move validation.
19120 (aarch64_simd_imm_scalar_p): Remove.
19121 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19122 (aarch64_legitimate_constant_p): Expand list of supported cases.
19123 * config/aarch64/aarch64-protos.h
19124 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19125 (aarch64_reinterpret_float_as_int): New.
19126 (aarch64_simd_imm_scalar_p): Remove.
19127 * config/aarch64/constraints.md (Uvi): New.
19128 (Dd): Split into Ds and new Dd.
19129 * config/aarch64/aarch64.md (*movsi_aarch64):
19131 (*movdi_aarch64): Add SIMD mov case.
19133 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19135 * tree-predcom.c: (struct chain): Handle store-store chain in which
19136 stores for elimination only store loop invariant values.
19137 (execute_pred_commoning_chain): Ditto.
19138 (prepare_initializers_chain_store_elim): Ditto.
19139 (prepare_finalizers): Ditto.
19140 (is_inv_store_elimination_chain): New function.
19141 (initialize_root_vars_store_elim_1): New function.
19143 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19145 * tree-predcom.c: Revise general description of the pass.
19146 (enum chain_type): New enum type for store elimination.
19147 (struct chain): New field supporting store elimination.
19148 (struct component): Ditto.
19149 (dump_chain): Dump store-stores chain.
19150 (release_chain): Release resources.
19151 (split_data_refs_to_components): Compute and create component
19152 contains only stores for elimination.
19153 (get_chain_last_ref_at): New function.
19154 (make_invariant_chain): Initialization.
19155 (make_rooted_chain): Specify chain type in parameter and record it.
19156 (add_looparound_copies): Skip for store-stores chain.
19157 (determine_roots_comp): Compute type of chain and pass it to
19159 (initialize_root_vars_store_elim_2): New function.
19160 (finalize_eliminated_stores): New function.
19161 (remove_stmt): Handle store for elimination.
19162 (execute_pred_commoning_chain): Execute predictive commoning on
19163 store-store chains.
19164 (determine_unroll_factor): Skip unroll for store-stores chain.
19165 (prepare_initializers_chain_store_elim): New function.
19166 (prepare_initializers_chain): Hanlde store-store chain.
19167 (prepare_finalizers_chain, prepare_finalizers): New function.
19168 (tree_predictive_commoning_loop): Return integer value indicating
19169 if loop is unrolled or lcssa form is corrupted.
19170 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19172 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19174 * tree-predcom.c (initialize_root): Delete.
19175 (execute_pred_commoning_chain): Initialize root vars and replace
19176 reference of non-combined chain directly, rather than call above
19179 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19181 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19182 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19184 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19186 * tree-predcom.c (struct chain): New field init_seq.
19187 (release_chain): Release init_seq.
19188 (prepare_initializers_chain): Record intialization stmts in above
19190 (insert_init_seqs): New function.
19191 (tree_predictive_commoning_loop): Call insert_init_seqs.
19193 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19195 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19197 2017-07-28 Richard Biener <rguenther@suse.de>
19199 * match.pd: Remove superfluous :c.
19200 * genmatch.c (simplify::id): Add member.
19201 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19203 (current_id): New global.
19204 (dt_node::parent): Move from ...
19205 (dt_operand::parent): ... here. Add for_id member.
19206 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19207 (decision_tree::find_node): Relax order requirement when
19208 merging DT_TRUE nodes to ones inbetween the current simplify
19209 and the one we try to merge with. Add diagnostic whenever
19210 we need to enforce pattern order by not merging.
19211 (decision_tree::insert): Set current_id.
19212 (decision_tree::print_node): Dump parent node and for_id.
19213 (parser::last_id): Add member.
19214 (parser::push_simplify): Assign unique id.
19215 (parser::parser): Initialize last_id.
19217 2017-07-28 Martin Liska <mliska@suse.cz>
19220 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19221 gimple_build_debug_bind.
19223 2017-07-28 Richard Biener <rguenther@suse.de>
19225 PR tree-optimization/81502
19226 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19228 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19230 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19231 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19232 for BIT_FIELD_REF args.
19233 * fold-const.c (make_bit_field_ref): Likewise.
19234 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19236 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19239 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19240 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19241 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19242 Or it into SANITIZER_UNDEFINED.
19243 * ubsan.c: Include gimple-fold.h and varasm.h.
19244 (ubsan_expand_ptr_ifn): New function.
19245 (instrument_pointer_overflow): New function.
19246 (maybe_instrument_pointer_overflow): New function.
19247 (instrument_object_size): Formatting fix.
19248 (pass_ubsan::execute): Call instrument_pointer_overflow
19249 and maybe_instrument_pointer_overflow.
19250 * internal-fn.c (expand_UBSAN_PTR): New function.
19251 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19252 * sanitizer.def (__ubsan_handle_pointer_overflow,
19253 __ubsan_handle_pointer_overflow_abort): New builtins.
19254 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19255 * internal-fn.def (UBSAN_PTR): New internal function.
19256 * opts.c (sanitizer_opts): Add pointer-overflow.
19257 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19258 * fold-const.c (build_range_check): Compute pointer range check in
19259 integral type if pointer arithmetics would be needed. Formatting
19262 2017-07-28 Martin Liska <mliska@suse.cz>
19265 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19266 parameters that are of a variable-length.
19268 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19270 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19272 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19273 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19274 (CRT_CALL_STATIC_FUNCTION): Likewise.
19275 (ASM_DEFAULT_SPEC): New define.
19276 (ASM_SPEC32): Likewise.
19277 (ASM_SPEC64): Likewise.
19278 (ASM_SPEC_COMMON): Likewise.
19279 (ASM_SPEC): Likewise.
19280 (INVALID_64BIT): Likewise.
19281 (LINK_OS_DEFAULT_SPEC): Likewise.
19282 (LINK_OS_SPEC32): Likewise.
19283 (LINK_OS_SPEC64): Likewise.
19284 (POWERPC_LINUX): Likewise.
19285 (PTRDIFF_TYPE): Likewise.
19286 (RESTORE_FP_PREFIX): Likewise.
19287 (RESTORE_FP_SUFFIX): Likewise.
19288 (SAVE_FP_PREFIX): Likewise.
19289 (SAVE_FP_SUFFIX): Likewise.
19290 (SIZE_TYPE): Likewise.
19291 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19292 (TARGET_64BIT): Likewise.
19293 (TARGET_64BIT): Likewise.
19294 (TARGET_AIX): Likewise.
19295 (WCHAR_TYPE_SIZE): Likewise.
19296 (WCHAR_TYPE): Undefine.
19297 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19298 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19299 (CPP_OS_RTEMS_SPEC): Delete.
19300 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19301 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19303 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19305 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19307 PR tree-optimization/81578
19308 * tree-parloops.c (build_new_reduction): Bail out if
19309 reduction_code isn't one of the standard OpenMP reductions.
19310 Move the details printing after that decision.
19312 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19314 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19315 related to reload_in_progress.
19316 (splat_input_operand): Likewise.
19317 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19319 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19321 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19322 (TARGET_INSTANTIATE_DECLS): Likewise.
19323 (legitimate_indexed_address_p): Delete reload_in_progress code.
19324 (rs6000_debug_legitimate_address_p): Likewise.
19325 (rs6000_eliminate_indexed_memrefs): Likewise.
19326 (rs6000_emit_le_vsx_store): Likewise.
19327 (rs6000_emit_move_si_sf_subreg): Likewise.
19328 (rs6000_emit_move): Likewise.
19329 (register_to_reg_type): Likewise.
19330 (rs6000_pre_atomic_barrier): Likewise.
19331 (rs6000_machopic_legitimize_pic_address): Likewise.
19332 (rs6000_allocate_stack_temp): Likewise.
19333 (rs6000_address_for_fpconvert): Likewise.
19334 (rs6000_address_for_altivec): Likewise.
19335 (rs6000_secondary_memory_needed_rtx): Delete function.
19336 (rs6000_check_sdmode): Likewise.
19337 (rs6000_alloc_sdmode_stack_slot): Likewise.
19338 (rs6000_instantiate_decls): Likewise.
19339 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19340 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19341 Delete reload_in_progress.
19342 (*vec_reload_and_plus_<mptrsize>): Likewise.
19343 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19344 (vsx_div_v2di): Likewise.
19345 (vsx_udiv_v2di): Likewise.
19347 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19349 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19350 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19351 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19352 (rs6000_debug_reg_global): Delete print of LRA status.
19353 (rs6000_option_override_internal): Delete dead LRA related code.
19354 (rs6000_lra_p): Delete function.
19355 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19357 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19359 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19360 * config/riscv/rtems.h: New file.
19362 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19363 Sudakshina Das <sudi.das@arm.com>
19365 * config/aarch64/aarch64.md
19366 (define_split for and<mode>3nr_compare): Move
19367 non aarch64_logical_operand to a register.
19368 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19369 register immediate operand to a register.
19370 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19372 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19374 PR middle-end/81564
19375 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19377 2017-07-27 Richard Biener <rguenther@suse.de>
19379 PR tree-optimization/81573
19380 PR tree-optimization/81494
19381 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19382 multi defuse cycle case.
19384 2017-07-27 Richard Biener <rguenther@suse.de>
19386 PR tree-optimization/81571
19387 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19390 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19392 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19393 earlier and only if MASK_FPU is set. Adjust formatting.
19395 2017-07-27 Martin Liska <mliska@suse.cz>
19397 * opt-functions.awk: Add validation of value of Init.
19398 * optc-gen.awk: Pass new argument.
19400 2017-07-27 Martin Liska <mliska@suse.cz>
19402 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19403 Fix wrong condition.
19405 2017-07-27 Martin Liska <mliska@suse.cz>
19407 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19408 BBs and edges seen by autoFDO.
19410 2017-07-27 Richard Biener <rguenther@suse.de>
19412 PR tree-optimization/81502
19413 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19414 with incompatible but same sized type.
19415 (execute_update_addresses_taken): Likewise.
19417 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19419 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19420 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19422 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19425 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19426 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19427 Change s_operand to memory_operand.
19429 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19431 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19432 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19433 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19434 Emit instructions rather than returning an expression. Handle TFmode
19435 and KFmode by casting to TImode.
19436 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19437 (rs6000_emit_le_vsx_store): Likewise.
19438 * config/rs6000/vsx.md (VSX_TI): New iterator.
19439 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19440 (*vsx_le_undo_permute_<mode>): Likewise.
19441 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19442 emit the split sequence.
19443 (*vsx_le_perm_store_<mode>): Likewise.
19445 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19447 PR tree-optimization/81555
19448 PR tree-optimization/81556
19449 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19450 if true, force CHANGED for the recursive invocation.
19451 (reassociate_bb): Remember original length of ops array, pass
19452 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19454 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19455 attributes for noipa attribute. For naked attribute use
19456 lookup_attribute first before lookup_attribute_spec.
19457 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19459 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19460 for functions with noipa attribute.
19461 (cgraph_externally_visible_p): Return true for functions with noipa
19463 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19464 for functions with noipa attribute.
19465 * doc/extend.texi: Document noipa function attribute.
19466 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19467 also for functions with noipa attribute.
19468 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19470 2017-07-26 Andrew Pinski <apinski@cavium.com>
19472 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19473 vec_unalign_load_cost and vec_unalign_store_cost.
19475 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19477 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19478 -mvsx-small-integer option.
19479 (ISA_3_0_MASKS_IEEE): Likewise.
19480 (OTHER_VSX_VECTOR_MASKS): Likewise.
19481 (POWERPC_MASKS): Likewise.
19482 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19483 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19484 code, only testing for DImode being allowed in non-VSX floating
19486 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19487 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19489 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19490 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19491 TARGET_P8_VECTOR test.
19492 (rs6000_secondary_reload_simple_move): Likewise.
19493 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19494 since TARGET_P9_VECTOR was already tested.
19495 (rs6000_opt_masks): Remove -mvsx-small-integer.
19496 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19497 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19499 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19500 test for TARGET_VEXTRACTUB was used, and that uses
19502 (p9 extract splitter): Likewise.
19503 (vsx_extract_<mode>_di_p9): Likewise.
19504 (vsx_extract_<mode>_store_p9): Likewise.
19505 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19506 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19507 the elimination of TARGET_VSX_SMALL_INTEGER.
19508 (vsx_extract_<mode>_p8): Likewise.
19509 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19510 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19511 (vsx_set_<mode>_p9): Likewise.
19512 (vsx_set_v4sf_p9): Likewise.
19513 (vsx_set_v4sf_p9_zero): Likewise.
19514 (vsx_insert_extract_v4sf_p9): Likewise.
19515 (vsx_insert_extract_v4sf_p9_2): Likewise.
19516 * config/rs6000/rs6000.md (sign extend splitter): Change
19517 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19518 (floatsi<mode>2_lfiwax_mem): Likewise.
19519 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19520 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19521 since a test for TARGET_P9_VECTOR was used.
19522 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19523 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19524 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19525 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19526 TARGET_P8_VECTOR test.
19527 (fix_trunc<mode>si2_stfiwx): Likewise.
19528 (fix_trunc<mode>si2_internal): Likewise.
19529 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19530 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19532 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19533 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19534 TARGET_P8_VECTOR test.
19535 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19536 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19537 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19539 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19540 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19541 since a test for TARGET_P9_VECTOR was used.
19542 (splitter for loading small constants): Likewise.
19544 2017-07-26 Andrew Pinski <apinski@cavium.com>
19546 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19549 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19552 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19553 (fp_valid_at): Likewise.
19555 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19557 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19558 (qdf24xx_addrcost_table): Likewise.
19559 (cortexa57_tunings): Update to use generic_branch_cost.
19560 (cortexa72_tunings): Likewise.
19561 (cortexa73_tunings): Likewise.
19562 (qdf24xx_tunings): Likewise.
19564 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19566 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19567 (thunderx2t99_branch_cost): Likewise.
19568 (cortexa35_tunings): Update to use generic_branch_cost.
19569 (cortexa53_tunings): Likewise.
19570 (cortexa57_tunings): Likewise.
19571 (cortexa72_tunings): Likewise.
19572 (cortexa73_tunings): Likewise.
19573 (thunderx2t99_tunings): Likewise.
19575 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19577 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19578 (sparc_option_override): Honour MASK_FSMULD.
19579 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19580 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19581 * config/sparc/sparc.opt (mfsmuld): New option.
19582 * doc/invoke.texi (mfsmuld): Document option.
19584 2017-07-26 Marek Polacek <polacek@redhat.com>
19586 PR middle-end/70992
19587 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19589 2017-07-26 Richard Biener <rguenther@suse.de>
19591 * gimple-match-head.c (do_valueize): Return OP if valueize
19593 (get_def): New helper to get at the def stmt of a SSA name
19594 if valueize allows.
19595 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19596 do_valueize to get at the def stmt.
19597 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19599 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19601 PR middle-end/46932
19602 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19604 2017-07-26 Martin Liska <mliska@suse.cz>
19607 * function.c (expand_function_start): Make expansion of
19608 nonlocal_goto_save_area after parm_birth_insn.
19610 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19612 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19613 from all CPU target flags enable members.
19615 2017-07-26 Richard Biener <rguenther@suse.de>
19617 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19618 (decision_tree::gen): Make 'type' const.
19619 (write_predicate): Likewise.
19621 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19623 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19624 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19625 (rs6000_option_override_internal): Likewise.
19626 (rs6000_expand_vector_set): Likewise.
19627 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19628 (TARGET_UPPER_REGS_SF): Likewise.
19629 (TARGET_UPPER_REGS_DI): Likewise.
19630 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19631 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19632 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19633 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19634 (Splitters for DI constants in Altivec registers): Likewise.
19635 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19636 (vsx_set_v4sf_p9): Likewise.
19637 (vsx_set_v4sf_p9_zero): Likewise.
19638 (vsx_insert_extract_v4sf_p9): Likewise.
19639 (vsx_insert_extract_v4sf_p9_2): Likewise.
19641 2017-07-25 Carl Love <cel@us.ibm.com>
19643 * doc/extend.texi: Update the built-in documentation file for the
19644 existing built-in functions
19645 vector signed char vec_cnttz (vector signed char);
19646 vector unsigned char vec_cnttz (vector unsigned char);
19647 vector signed short vec_cnttz (vector signed short);
19648 vector unsigned short vec_cnttz (vector unsigned short);
19649 vector signed int vec_cnttz (vector signed int);
19650 vector unsigned int vec_cnttz (vector unsigned int);
19651 vector signed long long vec_cnttz (vector signed long long);
19652 vector unsigned long long vec_cnttz (vector unsigned long long);
19654 2017-07-25 Andrew Pinski <apinski@cavium.com>
19656 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19657 accesses where the use is for the first operand of a BIT_INSERT.
19659 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19662 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19663 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19665 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19667 * config/i386/gstabs.h: Delete.
19668 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19670 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19672 * config/i386/i386.c (ix86_decompose_address): Do not check for
19673 register RTX when looking at index_reg or base_reg.
19674 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19676 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19678 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19679 to update EH info here.
19681 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19683 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19685 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19687 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19689 2017-07-25 Torsten Duwe <duwe@suse.de>
19691 * common.opt: Introduce -fpatchable-function-entry
19692 command line option, and its variables function_entry_patch_area_size
19693 and function_entry_patch_area_start.
19694 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19695 including a two-value parser.
19696 * target.def (print_patchable_function_entry): New target hook.
19697 * targhooks.h (default_print_patchable_function_entry): New function.
19698 * targhooks.c (default_print_patchable_function_entry): Likewise.
19699 * toplev.c (process_options): Switch off IPA-RA if
19700 patchable function entries are being generated.
19701 * varasm.c (assemble_start_function): Look at the
19702 patchable-function-entry command line switch and current
19703 function attributes and maybe generate NOP instructions by
19704 calling the print_patchable_function_entry hook.
19705 * doc/extend.texi: Document patchable_function_entry attribute.
19706 * doc/invoke.texi: Document -fpatchable_function_entry
19707 command line option.
19708 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19710 * doc/tm.texi: Re-generate.
19712 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19715 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19716 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19718 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19720 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19722 2017-07-25 Richard Biener <rguenther@suse.de>
19724 PR tree-optimization/81455
19725 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19726 not walk in cycles when looking for guards.
19728 2017-07-25 Richard Biener <rguenther@suse.de>
19730 PR tree-optimization/81529
19731 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19732 when optimizing backedge uses.
19734 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19736 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19738 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19739 to dl_section_ref. On AIX, append an expression to subtract
19740 the size of the section length to dl_section_ref.
19742 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19744 * configure.ac: If any of the config.* scripts fail, exit 1.
19745 * configure: Regenerate.
19747 2017-07-25 Richard Biener <rguenther@suse.de>
19749 PR middle-end/81546
19750 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19751 of immediate uses, be more verbose on errors.
19753 2017-07-25 Richard Biener <rguenther@suse.de>
19755 PR tree-optimization/81510
19756 * tree-vect-loop.c (vect_is_simple_reduction): When the
19757 reduction stmt is not inside the loop bail out.
19759 2017-07-25 Richard Biener <rguenther@suse.de>
19761 PR tree-optimization/81303
19762 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19763 profitability check against LOOP_VINFO_NITERSM1.
19765 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19767 * domwalk.c (cmp_bb_postorder): Simplify.
19768 (sort_bbs_postorder): New function. Use it...
19769 (dom_walker::walk): ...here to optimize common cases.
19771 2017-07-25 Martin Liska <mliska@suse.cz>
19774 * ipa-visibility.c (function_and_variable_visibility): Make the
19775 redirection just on target that supports aliasing.
19776 Fix GNU coding style.
19778 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19781 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19783 * config/arm/rtems.h: Likewise.
19784 * config/bfin/rtems.h: Likewise.
19785 * config/i386/rtemself.h: Likewise.
19786 * config/lm32/rtems.h: Likewise.
19787 * config/m32c/rtems.h: Likewise.
19788 * config/m68k/rtemself.h: Likewise.
19789 * config/microblaze/rtems.h: Likewise.
19790 * config/mips/rtems.h: Likewise.
19791 * config/moxie/rtems.h: Likewise.
19792 * config/nios2/rtems.h: Likewise.
19793 * config/powerpcspe/rtems.h: Likewise.
19794 * config/rs6000/rtems.h: Likewise.
19795 * config/rtems.h: Likewise.
19796 * config/sh/rtems.h: Likewise.
19797 * config/sh/rtemself.h: Likewise.
19798 * config/sparc/rtemself.h: Likewise.
19800 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19803 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19804 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19805 * tree-ssa-structalias.c (alias_get_name): Same.
19807 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19810 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19811 instructions if no du chain is found.
19813 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19815 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
19817 2017-07-25 Richard Biener <rguenther@suse.de>
19819 PR middle-end/81505
19820 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
19823 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19825 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
19826 upper-regs options.
19827 (ISA_2_7_MASKS_SERVER): Likewise.
19828 (ISA_3_0_MASKS_IEEE): Likewise.
19829 (OTHER_P8_VECTOR_MASKS): Likewise.
19830 (OTHER_VSX_VECTOR_MASKS): Likewise.
19831 (POWERPC_MASKS): Likewise.
19832 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
19833 duplicate list of options.
19834 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
19835 explicit -mupper-regs options.
19836 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
19837 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
19838 alias for -mupper-regs-df.
19839 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19840 (rs6000_init_hard_regno_mode_ok): Likewise.
19841 (rs6000_option_override_internal): Likewise.
19842 (rs6000_opt_masks): Likewise.
19843 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
19844 options in terms of whether -mvsx or -mpower8-vector was used.
19845 (TARGET_UPPER_REGS_DI): Likewise.
19846 (TARGET_UPPER_REGS_SF): Likewise.
19847 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
19848 -mupper-regs-* options.
19850 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19852 * passes.c (emergency_dump_function): Print some empty lines and a
19853 header before the RTL dump.
19855 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19857 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
19859 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
19862 * config/aarch64/aarch64.c (aarch64_classify_symbol):
19863 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
19865 2017-07-24 Carl Love <cel@us.ibm.com>
19867 * config/rs6000/rs6000-c.c: Add support for built-in functions
19868 vector float vec_extract_fp32_from_shorth (vector unsigned short);
19869 vector float vec_extract_fp32_from_shortl (vector unsigned short);
19870 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
19871 vec_extract_fp_from_shortl): Add defines for the two builtins.
19872 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
19873 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
19875 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
19876 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
19877 * doc/extend.texi: Update the built-in documentation file for the
19878 new built-in function.
19880 2017-07-24 Jakub Jelinek <jakub@redhat.com>
19883 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
19885 * doc/generic.texi: Likewise.
19886 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19887 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19889 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
19891 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
19892 (aarch64_mls_elt_merge<mode>): Likewise.
19894 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
19896 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19897 having __cxa_atexit.
19899 2017-07-23 Michael Collison <michael.collison@arm.com>
19901 * config/arm/arm.c (arm_option_override): Deprecate
19902 use of -mstructure-size-boundary.
19903 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
19904 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
19906 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19909 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19910 Reduce cost estimate for direct moves.
19912 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
19915 * config/i386/i386.c (ix86_option_override_internal): Disable
19916 BMI, BMI2 and TBM instructions for -m16.
19918 2017-07-21 Carl Love <cel@us.ibm.com>
19920 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19921 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19922 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19923 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19924 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19925 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19926 VMULOSW): New enum "unspec" values.
19927 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19928 altivec_vmulosw): New patterns.
19929 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19930 VMULOSW): Add definitions.
19932 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
19934 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
19935 (qdf24xx): Likewise.
19936 * config/aarch64/aarch64-options-extensions.def (rdma); New.
19937 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
19938 (AARCH64_FL_V8_1): Renumber.
19939 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
19940 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
19941 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
19942 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
19943 rdma to feature modifiers list.
19945 2017-07-21 Yury Gribov <tetra2005@gmail.com>
19947 PR middle-end/56727
19948 * ipa-visibility (function_and_variable_visibility): Convert
19949 recursive PLT call to direct call if appropriate.
19951 2017-07-21 Andrew Pinski <apinski@cavium.com>
19953 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
19954 operand 1 to see if the types precision matches.
19955 * fold-const.c (operand_equal_p): Likewise.
19957 2017-07-21 Richard Biener <rguenther@suse.de>
19959 PR tree-optimization/81303
19960 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
19961 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
19962 (vect_peeling_hash_get_lowest_cost): Adjust.
19963 (vect_enhance_data_refs_alignment): Likewise. Use
19964 vect_get_peeling_costs_all_drs to compute the penalty for no
19965 peeling to match up costs.
19967 2017-07-21 Richard Biener <rguenther@suse.de>
19969 PR tree-optimization/81500
19970 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
19971 we didn't identify a reduction path.
19973 2017-07-21 Tom de Vries <tom@codesourcery.com>
19974 Cesar Philippidis <cesar@codesourcery.com>
19976 PR gcov-profile/81442
19977 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
19980 2017-07-21 Tom de Vries <tom@codesourcery.com>
19983 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
19985 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
19986 nvptx_override_options_after_change.
19988 2017-07-21 Ulrich Drepper <drepper@redhat.com>
19990 * dwarf2out.c (output_file_names): Avoid double testing for
19991 dwarf_version >= 5.
19993 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
19995 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
19997 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19999 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20001 (try_crossjump_to_edge): Do not punt on partitioned functions.
20003 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20005 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20006 Put all BBs reachable only via paths crossing cold region to cold
20008 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20010 2016-07-21 Richard Biener <rguenther@suse.de>
20012 PR tree-optimization/81303
20013 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20014 into account prologue and epilogue iterations when raising
20015 min_profitable_iters to sth at least covering one vector iteration.
20017 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20019 * config/arm/arm.c (arm_test_cpu_arch_dat):
20022 2017-07-20 Nathan Sidwell <nathan@acm.org>
20024 Remove TYPE_METHODS.
20025 * tree.h (TYPE_METHODS): Delete.
20026 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20027 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20028 (dbxout_type_methods): Scan TYPE_FIELDS.
20029 (dbxout_type): Don't check TYPE_METHODS here.
20030 * function.c (use_register_for_decl): Always ignore register for
20031 class types when not optimizing.
20032 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20033 * tree.c (free_lang_data_in_type): Stitch out member functions and
20034 templates from TYPE_FIELDS.
20035 (build_distinct_type_copy, verify_type_variant,
20036 verify_type): Member fns are on TYPE_FIELDS.
20037 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20038 * tree-pretty-print.c (dump_generic_node): Likewise.
20040 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20043 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20044 V2TImode and V4TImode.
20045 (ix86_expand_vector_extract): Likewise.
20046 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20047 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20048 (ssescalarmode): Handle V4TImode and V2TImode.
20049 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20050 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20051 (VEXTRACTI128_MODE): New mode iterator.
20052 (splitter for *vec_extractv?ti first element): New.
20053 (VEC_INIT_MODE): New mode iterator.
20054 (vec_init<mode>): Consolidate 3 expanders into one using
20055 VEC_INIT_MODE mode iterator.
20057 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20059 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20060 non_spilled_static_chain_regno_p.
20062 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20064 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20066 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20068 * bb-reorder.c (connect_traces): Allow copying of blocks within
20071 2017-07-20 Richard Biener <rguenther@suse.de>
20073 * gimple.h (gimple_phi_result): Add gphi * overload.
20074 (gimple_phi_result_ptr): Likewise.
20075 (gimple_phi_arg): Likewise. Adjust index assert to only
20076 allow actual argument accesses rather than all slots available
20078 (gimple_phi_arg_def): Add gphi * overload.
20079 * tree-phinodes.c (make_phi_node): Initialize only actual
20081 (resize_phi_node): Clear memory not covered by old node,
20082 do not initialize excess argument slots.
20083 (reserve_phi_args_for_new_edge): Initialize new argument slot
20086 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20088 PR tree-optimization/81388
20090 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20092 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20093 by removing computation of may_be_zero.
20095 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20096 Tom de Vries <tom@codesourcery.com>
20098 PR middle-end/81030
20099 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20100 when gimple level profile disagrees with what RTL expander did.
20102 2017-07-20 Richard Biener <rguenther@suse.de>
20104 PR tree-optimization/61171
20105 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20106 (vect_analyze_stmt): Add slp instance parameter.
20107 (vectorizable_reduction): Likewise.
20108 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20109 (vect_is_simple_reduction): Deal with chains not detected
20110 as SLP reduction chain, specifically not properly associated
20111 chains containing a mix of plus/minus.
20112 (get_reduction_op): Remove.
20113 (get_initial_defs_for_reduction): Simplify, pass in whether
20114 this is a reduction chain, pass in the SLP node for the PHIs.
20115 (vect_create_epilog_for_reduction): Get the SLP instance as
20117 (vectorizable_reduction): Get the SLP instance as arg.
20118 During analysis remember the SLP node with the PHIs in the
20119 instance. Simplify getting at the vectorized reduction PHIs.
20120 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20121 through SLP instance.
20122 (vect_slp_analyze_operations): Likewise.
20123 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20124 (vect_transform_stmt): Likewise.
20126 2017-07-20 Tom de Vries <tom@codesourcery.com>
20128 PR tree-optimization/81489
20129 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20130 read of phi arg location to before loop that modifies phi.
20132 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20134 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20137 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20139 PR middle-end/81331
20140 * except.c (execute): Fix ordering issue.
20142 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20144 PR rtl-optimization/81423
20145 * combine.c (make_compound_operation_int): Don't try to optimize
20146 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20148 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20150 PR rtl-optimization/81423
20151 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20152 with a constant that is -1 in the truncated to mode.
20154 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20156 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20157 (determine_unlikely_bbs): ... here.
20158 * predict.h (propagate_unlikely_bbs_forward): Declare.
20159 * cfgexpand.c (pass_expand::execute): Use it.
20160 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20162 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20163 propagate_unlikely_bbs_forward.
20165 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20167 PR middle-end/81331
20168 * except.c (maybe_add_nop_after_section_switch): New function.
20171 2017-07-19 Tom de Vries <tom@codesourcery.com>
20173 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20175 2017-07-19 Tom de Vries <tom@codesourcery.com>
20177 * gimple.h (gimple_phi_arg): Make assert more strict.
20179 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20181 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20182 * config/rs6000/mmintrin.h: New file.
20183 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20185 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20187 PR tree-optimization/81346
20188 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20190 2017-07-19 Tom de Vries <tom@codesourcery.com>
20192 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20194 2017-07-19 Tom de Vries <tom@codesourcery.com>
20196 * config/nvptx/nvptx-modes.def: Add V2DImode.
20197 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20198 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20199 (nvptx_output_mov_insn): Handle lack of mov.b128.
20200 (nvptx_print_operand): Handle 'H' and 'L' codes.
20201 (nvptx_vector_mode_supported): Allow V2DImode.
20202 (nvptx_preferred_simd_mode): New function.
20203 (nvptx_data_alignment): New function.
20204 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20205 nvptx_preferred_simd_mode.
20206 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20208 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20210 2017-07-19 Tom de Vries <tom@codesourcery.com>
20212 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20213 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20214 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20215 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20216 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20217 (mov<VECIM>_insn): New define_insn.
20218 (define_expand "mov<VECIM>): New define_expand.
20220 2017-07-19 Tom de Vries <tom@codesourcery.com>
20222 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20224 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20226 PR tree-optimization/81346
20227 * fold-const.h (fold_div_compare, range_check_type): Declare.
20228 * fold-const.c (range_check_type): New function.
20229 (build_range_check): Use range_check_type.
20230 (fold_div_compare): No longer static, rewritten into
20231 a match.pd helper function.
20232 (fold_comparison): Don't call fold_div_compare here.
20233 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20234 as helper function.
20236 2017-07-19 Nathan Sidwell <nathan@acm.org>
20238 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20239 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20240 * tree.c (find_decls_types_r, verify_type): Use
20241 TYPE_{MIN,MAX}_VALUE_RAW.
20242 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20243 (hash_tree): Likewise.
20244 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20246 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20249 2017-07-18 Tom de Vries <tom@codesourcery.com>
20251 PR middle-end/81464
20252 * omp-expand.c (expand_omp_for_static_chunk): Handle
20253 equal-argument loop exit phi.
20255 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20258 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20259 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20260 operand 2 predicate.
20261 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20262 operand 2 predicate.
20263 (ror,rol -> rorx splitters): Use const_int_operand as
20264 operand 2 predicate.
20266 2017-06-18 Richard Biener <rguenther@suse.de>
20268 PR tree-optimization/81410
20269 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20270 the gap in the ! slp_perm SLP case after each group.
20272 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20274 PR middle-end/81463
20275 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20278 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20280 PR middle-end/81462
20281 * predict.c (set_even_probabilities): Cleanup; do not affect
20282 probabilities that are already known.
20283 (combine_predictions_for_bb): Call even when count is set.
20285 2017-07-18 Nathan Sidwell <nathan@acm.org>
20287 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20290 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20293 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20294 optimization for loop niter analysis.
20296 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20299 * config/avr/avr.c (avr_optimize_casesi): Don't use
20300 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20302 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20304 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20305 body_cost_vec from _vect_peel_extended_info.
20306 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20307 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20310 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20312 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20314 2017-07-18 Richard Biener <rguenther@suse.de>
20316 PR tree-optimization/80620
20317 PR tree-optimization/81403
20318 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20319 info when re-using a VN table entry.
20321 2017-07-18 Richard Biener <rguenther@suse.de>
20323 PR tree-optimization/81418
20324 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20325 vectype_in. Verify that with lane-reducing reduction operations
20326 we have a single def-use cycle.
20328 2017-07-17 Carl Love <cel@us.ibm.com>
20330 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20332 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20333 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20334 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20335 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20336 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20337 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20338 VMULOSW): New enum "unspec" values.
20339 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20340 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20341 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20342 altivec_vmulosw): New patterns.
20343 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20344 VMULOSW): Add definitions.
20346 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20348 * config/alpha/alpha.c: Include predict.h.
20350 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20352 * tree-vrp.c (compare_assert_loc): Fix comparison function
20353 to return predictable results.
20355 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20357 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20359 (subdi3): Likewise.
20360 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20361 * doc/invoke.texi (mexpand-adddi): Update text.
20363 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20365 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20366 that also clobbers the CC register. The old expand code is moved
20368 (*arc_clzsi2): ... here.
20369 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20370 the CC register. The old expand code is moved to ...
20371 (arc_ctzsi2): ... here.
20373 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20375 * config/arc/arc.opt (mindexed-loads): Use initial value
20376 TARGET_INDEXED_LOADS_DEFAULT.
20377 (mauto-modify-reg): Use initial value
20378 TARGET_AUTO_MODIFY_REG_DEFAULT.
20379 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20380 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20381 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20382 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20384 2017-07-17 Martin Liska <mliska@suse.cz>
20387 * opts.c (finish_options): Do not allow -fgnu-tm
20388 w/ -fsanitize={kernel-,}address. Say sorry.
20390 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20393 * tree-loop-distribution.c (classify_partition): Only assert on
20394 numer of iterations.
20395 (merge_dep_scc_partitions): Delete prameter. Update function call.
20396 (distribute_loop): Remove code handling loop with unknown niters.
20397 (pass_loop_distribution::execute): Skip loop with unknown niters.
20399 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20402 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20403 function sort_partitions_by_post_order.
20405 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20407 PR tree-optimization/81374
20408 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20409 the max index of basic blocks, rather than number of basic blocks.
20411 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20413 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20415 (arc_legitimate_pic_operand_p): Likewise.
20416 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20418 (arc_needs_pcl_p): Likewise.
20419 (arc_legitimate_pc_offset_p): Likewise.
20420 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20421 function is also used in constrains.md.
20422 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20423 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20424 PLUS. Only return true/false in known cases, otherwise assert.
20425 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20426 is already called in arc_legitimate_constant_p.
20427 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20429 (LEGITIMATE_PIC_OPERAND_P): Use
20430 arc_raw_symbolic_reference_mentioned_p function.
20431 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20436 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20437 Andrew Burgess <andrew.burgess@embecosm.com>
20439 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20440 (arc_return_address_register): New function.
20441 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20442 (arc_handle_fndecl_attribute): Add naked attribute.
20443 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20444 (TARGET_WARN_FUNC_RETURN): Likewise.
20445 (arc_allocate_stack_slots_for_args): New function.
20446 (arc_warn_func_return): Likewise.
20447 (machine_function): Change type fn_type.
20448 (arc_compute_function_type): Consider new naked function type,
20449 change function return type.
20450 (arc_must_save_register): Adapt to handle new
20451 arc_compute_function_type's return type.
20452 (arc_expand_prologue): Likewise.
20453 (arc_expand_epilogue): Likewise.
20454 (arc_return_address_regs): Delete.
20455 (arc_return_address_register): New function.
20456 (arc_epilogue_uses): Use above function.
20457 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20458 (arc_function_type): Change encoding, add naked type.
20459 (ARC_INTERRUPT_P): Change to handle the new encoding.
20460 (ARC_FAST_INTERRUPT_P): Likewise.
20461 (ARC_NORMAL_P): Define.
20462 (ARC_NAKED_P): Likewise.
20463 (arc_compute_function_type): Delete prototype.
20464 * config/arc/arc.md (in_ret_delay_slot): Use
20465 arc_return_address_register function.
20466 (simple_return): Likewise.
20467 (p_return_i): Likewise.
20469 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20471 PR tree-optimization/81428
20472 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20473 can't be built for those types.
20475 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20477 Remove stuff dead since r239246.
20479 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20480 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20481 (avr_inform_devices): Remove dead stuff.
20483 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20485 * config/arm/arm_neon.h: Fix softp typo.
20487 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20489 PR tree-optimization/81365
20490 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20491 aggregate moves onto bb predecessor edges, make sure there are no
20492 loads that could alias the lhs in between the start of bb and the
20495 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20498 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20499 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20500 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20502 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20504 PR tree-optimization/81396
20505 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20506 (init_symbolic_number): Initialize it to 1.
20507 (perform_symbolic_merge): Add n_ops from both operands into the new
20509 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20510 without base_addr as useless if they need more than one operation.
20511 (bswap_replace): Handle !bswap case for NULL base_addr.
20513 2017-07-17 Tom de Vries <tom@codesourcery.com>
20516 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20519 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20521 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20522 conditional builtin define __FIX_LEON3FT_B2BST.
20524 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20526 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20527 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20530 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20532 PR rtl-optimization/81424
20533 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20534 to remove potential trapping from operands if -fnon-call-exceptions.
20536 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20538 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20539 profile_proability for scalling.
20540 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20542 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20544 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20546 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20548 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20549 fixpoint arithmetics.
20551 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20553 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20554 fixpoint arithmetics.
20556 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20558 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20559 fixpoint arithmetics.
20561 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20563 * profile-count.h (profile_probability::from_reg_br_prob_note,
20564 profile_probability::to_reg_br_prob_note): New functions.
20565 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20566 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20567 * predict.c (probability_reliable_p): Update.
20568 (edge_probability_reliable_p): Update.
20569 (br_prob_note_reliable_p): Update.
20570 (invert_br_probabilities): Update.
20571 (add_reg_br_prob_note): New function.
20572 (combine_predictions_for_insn): Update.
20573 * asan.c (asan_clear_shadow): Update.
20574 * cfgbuild.c (compute_outgoing_frequencies): Update.
20575 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20576 (update_br_prob_note): Update.
20577 (rtl_verify_edges): Update.
20578 (purge_dead_edges): Update.
20579 (fixup_reorder_chain): Update.
20580 * emit-rtl.c (try_split): Update.
20581 * ifcvt.c (cond_exec_process_insns): Update.
20582 (cond_exec_process_if_block): Update.
20583 (dead_or_predicable): Update.
20584 * internal-fn.c (expand_addsub_overflow): Update.
20585 (expand_neg_overflow): Update.
20586 (expand_mul_overflow): Update.
20587 * loop-doloop.c (doloop_modify): Update.
20588 * loop-unroll.c (compare_and_jump_seq): Update.
20589 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20590 * predict.h: Update.
20591 * reorg.c (mostly_true_jump): Update.
20593 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20594 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20595 * config/arc/arc.c: (emit_unlikely_jump): Update.
20596 * config/arm/arm.c: (emit_unlikely_jump): Update.
20597 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20598 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20599 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20600 (ix86_print_operand): Update.
20601 (ix86_split_fp_branch): Update.
20602 (predict_jump): Update.
20603 * config/ia64/ia64.c (ia64_print_operand): Update.
20604 * config/mmix/mmix.c (mmix_print_operand): Update.
20605 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20606 (rs6000_expand_split_stack_prologue): Update.
20607 * config/rs6000/rs6000.c: Update.
20608 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20609 (s390_expand_vec_movstr): Update.
20610 (s390_expand_cs_tdsi): Update.
20611 (s390_expand_split_stack_prologue): Update.
20612 * config/sh/sh.c (sh_print_operand): Update.
20613 (expand_cbranchsi4): Update.
20614 (expand_cbranchdi4): Update.
20615 * config/sparc/sparc.c (output_v9branch): Update.
20616 * config/spu/spu.c (get_branch_target): Update.
20617 (ea_load_store_inline): Update.
20618 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20619 * config/tilepro/tilepro.c: Update.
20621 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20623 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20624 (gimplify_save_expr): Likewise.
20626 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20628 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20630 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20632 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20634 * timevar.def (TV_IPA_FNSUMMARY): Define.
20636 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20638 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20639 to back store errata sensitive sequence from being generated.
20640 (sqrtdf2_fix): Likewise.
20642 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20644 * tree-ssa-threadupdate.c (compute_path_counts,
20645 update_joiner_offpath_counts): Use profile_probability.
20647 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20650 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20652 * config/arm/arm-c.c (arm_cpu_builtins): Define
20653 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20655 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20657 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20658 array entries to represent __ieee128 versions of the
20659 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20660 scalar_extract_sig, and scalar_insert_exp built-in functions.
20661 (altivec_resolve_overloaded_builtin): Add special case handling
20662 for the __builtin_scalar_insert_exp function, as represented by
20663 the P9V_BUILTIN_VEC_VSIEDP constant.
20664 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20665 exponent support for __ieee128 argument.
20666 (VSESQP): Add scalar extract signature support for __ieee128
20668 (VSTDCNQP): Add scalar test negative support for __ieee128
20670 (VSIEQP): Add scalar insert exponent support for __int128 argument
20671 with __ieee128 result.
20672 (VSIEQPF): Add scalar insert exponent support for __ieee128
20673 argument with __ieee128 result.
20674 (VSTDCQP): Add scalar test data class support for __ieee128
20676 (VSTDCNQP): Add overload support for scalar test negative with
20677 __ieee128 argument.
20678 (VSTDCQP): Add overload support for scalar test data class
20679 __ieee128 argument.
20680 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20681 UNSPEC_VSX_SXSIGDP.
20682 (UNSPEC_VSX_SIEXPQP): New constant.
20683 (xsxexpqp): New insn for VSX scalar extract exponent quad
20685 (xsxsigqp): New insn for VSX scalar extract significand quad
20687 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20688 precision with floating point argument.
20689 (xststdcqp): New expand for VSX scalar test data class quad
20691 (xststdcnegqp): New expand for VSX scalar test negative quad
20693 (xststdcqp): New insn to match expansions for VSX scalar test data
20694 class quad precision and VSX scalar test negative quad precision.
20695 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20696 special case operand checking to enforce that second operand of
20697 VSX scalar test data class with quad precision argument is a 7-bit
20699 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20700 prototypes and descriptions of __ieee128 versions of
20701 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20702 scalar_test_data_class, and scalar_test_neg built-in functions.
20704 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20706 PR tree-optimization/81162
20707 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20708 replace a negate with an add.
20710 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20712 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20714 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20716 * config/arm/arm-c.c (arm_cpu_builtins): Define
20717 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20719 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20721 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20722 (armv8-r): Set ARM Cortex-R52 as default CPU.
20723 * config/arm/arm-tables.opt: Regenerate.
20724 * config/arm/arm-tune.md: Regenerate.
20725 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20727 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20728 extension for -mcpu=cortex-r52.
20730 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20732 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20733 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20734 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20735 (fp-armv8): Define it as FP_ARMv8 only.
20736 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20737 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20739 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20740 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20741 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20742 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20743 than TARGET_FPU_ARMV8.
20744 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20745 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20746 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20748 * config/arm/neon.md (neon_vrint): Likewise.
20749 (neon_vcvt): Likewise.
20750 (neon_<fmaxmin_op><mode>): Likewise.
20751 (<fmaxmin><mode>3): Likewise.
20752 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20753 * config/arm/predicates.md (arm_cond_move_operator): Check against
20754 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20756 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20758 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20759 to top of function.
20761 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20763 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20764 loop in comment with memset.
20766 2017-07-14 Martin Liska <mliska@suse.cz>
20768 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20769 * dwarf2out.c (is_java): Remove the function.
20770 (output_pubname): Remove usage of the function.
20771 (lower_bound_default): Remove usage of DW_LANG_Java.
20772 (gen_compile_unit_die): Likewise.
20773 * gcc.c: Remove compiler defaults for .java and .zip files.
20774 * gimple-expr.c (remove_suffix): Change as there's no longer
20775 extension than 4-letter one.
20776 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20777 (gimplify_save_expr): Likewise.
20778 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20779 as it's possible even for other languages than Java.
20780 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20781 * lto-opts.c (lto_write_options): Remove reference to Java.
20782 * opts.c (strip_off_ending): Update file extension handling.
20783 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20784 * tree-eh.c (lower_resx): Likewise.
20785 * tree.c (free_lang_data_in_type): Remove dead code.
20786 (find_decls_types_r): Likewise.
20787 (build_common_builtin_nodes): Remove Java from a comment.
20788 (verify_type): Remove dead code.
20789 * varasm.c (assemble_external): Remove Java from a comment.
20791 2017-07-14 Martin Liska <mliska@suse.cz>
20793 * opts.c (finish_options): Add quotes.
20794 (common_handle_option): Likewise.
20796 2017-07-14 Martin Liska <mliska@suse.cz>
20798 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20799 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20800 Remove N_SO_PASCAL.
20801 * dwarf2out.c (lower_bound_default): Do not handle
20803 (gen_compile_unit_die): Likewise.
20804 * gcc.c: Remove default extension binding for GNU Pascal.
20805 * stmt.c: Remove Pascal language from a comment.
20806 * xcoffout.c: Likewise.
20808 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20811 * diagnostic-show-locus.c (fixit_cmp): New function.
20812 (layout::layout): Sort m_fixit_hints.
20813 (column_range::column_range): Assert that the values are valid.
20814 (struct char_span): New struct.
20815 (correction::overwrite): New method.
20816 (struct source_line): New struct.
20817 (line_corrections::add_hint): Add assertions. Reimplement memcpy
20818 calls in terms of classes source_line and char_span, and
20819 correction::overwrite.
20820 (selftest::test_overlapped_fixit_printing_2): New function.
20821 (selftest::diagnostic_show_locus_c_tests): Call it.
20823 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
20825 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
20826 early if there is no lhs.
20828 2017-07-13 Martin Liska <mliska@suse.cz>
20830 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
20831 (gen_reference_type_die): Likewise.
20832 * stor-layout.c: Remove Pascal-related comment.
20834 2017-07-13 Martin Liska <mliska@suse.cz>
20836 * opts.c (finish_options): Add quotes to error messages.
20837 (parse_sanitizer_options): Likewise.
20839 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
20841 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
20843 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
20845 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
20847 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
20849 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
20851 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
20853 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
20856 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
20857 provides the hardware capability bits, define the macro
20858 __BUILTIN_CPU_SUPPORTS__.
20859 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
20860 if GLIBC does not provide the hardware capability bits. Add a
20861 gcc_unreachable call if the built-in cpu function is neither
20862 __builtin_cpu_is nor __builtin_cpu_supports.
20863 (rs6000_get_function_versions_dispatcher): Change the warning
20864 that an old GLIBC is used which does not export the capability
20865 bits to be an error.
20866 * doc/extend.texi (target_clones attribute): Document the
20867 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
20868 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
20869 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
20870 the macros defined by GCC if the newer GLIBC is available.
20872 2017-07-12 Jeff Law <law@redhat.com>
20874 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
20875 remaining includes slightly.
20876 * config/riscv/riscv-builtins.c: Include profile-count.h.
20878 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20881 * config/avr/avr.c (avr_set_current_function): In diagnostic
20882 messages: Quote keywords and (parts of) identifiers.
20883 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
20886 2017-07-12 Carl Love <cel@us.ibm.com>
20888 * config/rs6000/rs6000-c.c: Add support for built-in functions
20889 vector bool char vec_revb (vector bool char);
20890 vector bool short vec_revb (vector short char);
20891 vector bool int vec_revb (vector bool int);
20892 vector bool long long vec_revb (vector bool long long);
20893 * doc/extend.texi: Update the built-in documentation file for the
20894 new built-in functions.
20896 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20898 * config/s390/s390.md: Remove movcc splitter.
20900 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20902 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
20903 load/store on condition.
20905 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20908 * config/avr/avr.c (avr_encode_section_info)
20909 [progmem && !TREE_READONLY]: Error if progmem object needs
20912 2017-07-11 Michael Collison <michael.collison@arm.com>
20914 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
20917 2017-07-11 Carl Love <cel@us.ibm.com>
20919 * config/rs6000/rs6000-c.c: Add support for builtins
20920 vector unsigned int vec_parity_lsbb (vector signed int);
20921 vector unsigned int vec_parity_lsbb (vector unsigned int);
20922 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
20923 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
20924 vector unsigned long long vec_parity_lsbb (vector signed long long);
20925 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
20926 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
20927 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
20928 * doc/extend.texi: Update the built-in documentation file for the
20929 new built-in functions.
20931 2017-07-11 David Malcolm <dmalcolm@redhat.com>
20933 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
20934 (layout::m_primary_loc): New field.
20935 (layout::layout): Initialize new field. Move location filtering
20936 logic from here to...
20937 (layout::maybe_add_location_range): ...this new method. Add
20938 support for filtering to just the lines already specified by other
20940 (layout::will_show_line_p): New method.
20941 (gcc_rich_location::add_location_if_nearby): New method.
20942 (selftest::test_add_location_if_nearby): New test function.
20943 (selftest::diagnostic_show_locus_c_tests): Call it.
20944 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
20947 2017-07-11 Tom de Vries <tom@codesourcery.com>
20949 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
20950 (bb_first_real_insn): New function.
20951 (nvptx_single): Add extra initialization of broadcasted condition
20954 2017-07-11 Nathan Sidwell <nathan@acm.org>
20956 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
20958 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
20960 * doc/extend.texi (AVR Function Attributes): Remove weblink to
20961 Binutils doc as TEXI will mess them up.
20962 * doc/invoke.texi (AVR Options): Same here.
20964 2017-07-11 Daniel Cederman <cederman@gaisler.com>
20966 * config/sparc/sparc.opt (mfix-ut700): New option.
20967 (mfix-gr712rc): Likewise.
20968 (sparc_fix_b2bst): New variable.
20969 * doc/invoke.texi (SPARC options): Document them.
20970 (ARM options): Fix warnings.
20971 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
20972 instructions to prevent sequences that can trigger the store-store
20973 errata for certain LEON3FT processors.
20974 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
20975 (sparc_option_override): Set sparc_fix_b2bst appropriately.
20976 * config/sparc/sparc.md (fix_b2bst): New attribute.
20977 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
20979 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
20982 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
20984 (*rsqrtsf2_sse): Ditto.
20986 (div<mode>3): Macroize insn from divdf3 and divsf3
20987 using MODEF mode iterator.
20989 2017-07-10 Martin Sebor <msebor@redhat.com>
20991 PR tree-optimization/80397
20992 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
20993 instead of testing for equality to INTEGER_TYPE.
20995 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
20997 * config.gcc: Remove uclibc from arc target spec.
20999 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21001 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21003 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21006 * lto-wrapper.c (remove_option): New function.
21007 (merge_and_complain): Merge PIC/PIE options more realistically.
21009 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21011 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21015 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21016 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21017 * config.in: Regenerate.
21018 * configure: Regenerate.
21019 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21020 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21021 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21022 (TARGET_GASISR_PROLOGUES): ...target mask.
21023 * common/config/avr/avr-common.c
21024 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21025 Set -mgas-isr-prologues.
21026 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21027 INSERT_PASS_BEFORE for it.
21028 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21029 * config/avr/avr.c (avr_option_override)
21030 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21031 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21032 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21033 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21034 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21036 (make_avr_pass_pre_proep): New function.
21037 (emit_push_sfr) <treg>: Add argument to function and use it
21038 instead of TMP_REG.
21039 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21040 and set machine->gasisr.yes.
21041 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21042 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21043 __gcc_isr.n_pushed to .L__stack_usage.
21044 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21045 (avr_asm_final_postscan_insn): ...this new static function.
21046 * config/avr/avr.h (machine_function)
21047 <is_no_gccisr, use_L__stack_usage>: New fields.
21048 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21049 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21050 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21051 (gasisr, *gasisr): New expander and insn.
21052 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21053 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21054 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21056 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21058 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21061 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21063 Move jump-tables out of .text again.
21066 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21067 (ASM_OUTPUT_ADDR_VEC): New function.
21068 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21069 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21070 INSN_ADDRESSes as asm comment.
21071 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21072 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21073 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21074 * config/avr/avr.md (*tablejump): Adjust comment.
21075 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21076 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21078 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21079 (avr_output_addr_vec): New proto.
21080 (avr_log_t) <insn_addresses>: New field.
21082 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21085 * config/i386/i386.c (ix86_function_arg_advance): Set
21086 outgoing_args_on_stack to true if there are outgoing arguments
21088 (ix86_function_arg): Likewise.
21089 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21090 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21091 * config/i386/i386.h (machine_function): Add
21092 outgoing_args_on_stack.
21094 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21096 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21097 supporting pthreds.
21098 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21100 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21102 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21103 (REAL_H): Remove $(MACHMODE_H).
21104 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21106 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21107 $(MACHMODE_H) and double-int.h.
21108 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21110 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21113 2017-07-07 Andrew Pinski <apinski@cavium.com>
21115 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21116 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21118 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21120 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21121 Add warning if GCC was not configured to link against a GLIBC that
21122 exports the hardware capability bits.
21123 (make_resolver_func): Make resolver function private and not a
21124 COMDAT function. Create the name with clone_function_name instead
21125 of make_unique_name.
21128 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21129 correct operand in doing the split.
21131 2017-07-07 Carl Love <cel@us.ibm.com>
21133 * config/rs6000/rs6000-c: Add support for built-in function
21134 vector unsigned short vec_pack_to_short_fp32 (vector float,
21136 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21137 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21138 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21139 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21140 (convert_4f32_8i16): Add define_expand.
21141 * doc/extend.texi: Update the built-in documentation file for the
21142 new built-in function.
21144 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21146 * config/sparc/m8.md: New file.
21147 * config/sparc/sparc.md: Include m8.md.
21149 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21151 * config/sparc/sparc.opt: New option -mvis4b.
21152 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21153 (sparc_option_override): Handle VIS4B.
21154 (enum sparc_builtins): Define
21155 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21156 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21157 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21158 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21159 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21160 (check_constant_argument): New function.
21161 (sparc_vis_init_builtins): Define builtins
21162 __builtin_vis_dictunpack{8,16,32},
21163 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21164 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21165 __builtin_vis_fpcmpde{8,16,32}shl and
21166 __builtin_vis_fpcmpur{8,16,32}shl.
21167 (sparc_expand_builtin): Check that the constant operands to
21168 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21169 constant and in range.
21170 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21172 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21173 (SPARC_IMM5_P): Likewise.
21174 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21175 (enabled): Handle vis4b.
21176 (UNSPEC_DICTUNPACK): New unspec.
21177 (UNSPEC_FPCMPSHL): Likewise.
21178 (UNSPEC_FPUCMPSHL): Likewise.
21179 (UNSPEC_FPCMPDESHL): Likewise.
21180 (UNSPEC_FPCMPURSHL): Likewise.
21181 (cpu_feature): New CPU feature `vis4b'.
21182 (dictunpack{8,16,32}): New insns.
21183 (FPCSMODE): New mode iterator.
21184 (fpcscond): New code iterator.
21185 (fpcsucond): Likewise.
21186 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21187 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21188 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21189 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21190 * config/sparc/constraints.md: Define constraints `q' for unsigned
21191 2-bit integer constants and `t' for unsigned 5-bit integer
21193 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21195 (imm5_operand_dictunpack16): Likewise.
21196 (imm5_operand_dictunpack32): Likewise.
21197 (imm2_operand): Likewise.
21198 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21199 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21200 ditunpack* and fpcmp*shl builtins.
21202 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21204 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21205 * config.in: Add HAVE_AS_SPARC6 define.
21206 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21208 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21210 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21211 (CPP_CPU_SPEC): Handle m8.
21212 (ASM_CPU_SPEC): Likewise.
21213 * config/sparc/sparc-opts.h (enum processor_type): Add
21215 * config/sparc/sparc.c (m8_costs): New struct.
21216 (sparc_option_override): Handle TARGET_CPU_m8.
21217 (sparc32_initialize_trampoline): Likewise.
21218 (sparc64_initialize_trampoline): Likewise.
21219 (sparc_issue_rate): Likewise.
21220 (sparc_register_move_cost): Likewise.
21221 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21222 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21223 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21224 (CPP_CPU_SPEC): Handle M8.
21225 (ASM_CPU_SPEC): Likewise.
21226 (AS_M8_FLAG): Define.
21227 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21228 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21229 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21231 * configure: Regenerate.
21232 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21235 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21237 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21239 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21240 ("*movdi_insn_sp32"): Do not set v3pipe.
21241 ("*movsi_insn"): Likewise.
21242 ("*movdi_insn_sp64"): Likewise.
21243 ("*movsf_insn"): Likewise.
21244 ("*movdf_insn_sp32"): Likewise.
21245 ("*movdf_insn_sp64"): Likewise.
21246 ("*zero_extendsidi2_insn_sp64"): Likewise.
21247 ("*sign_extendsidi2_insn"): Likewise.
21248 ("*mov<VM32:mode>_insn"): Likewise.
21249 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21250 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21251 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21252 ("<vlop:code><VL:mode>3"): Likewise.
21253 ("*not_<vlop:code><VL:mode>3"): Likewise.
21254 ("*nand<VL:mode>_vis"): Likewise.
21255 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21256 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21257 ("one_cmpl<VL:mode>2"): Likewise.
21258 ("faligndata<VM64:mode>_vis"): Likewise.
21259 ("alignaddrsi_vis"): Likewise.
21260 ("alignaddrdi_vis"): Likweise.
21261 ("alignaddrlsi_vis"): Likewise.
21262 ("alignaddrldi_vis"): Likewise.
21263 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21264 ("bmaskdi_vis"): Likewise.
21265 ("bmasksi_vis"): Likewise.
21266 ("bshuffle<VM64:mode>_vis"): Likewise.
21267 ("cmask8<P:mode>_vis"): Likewise.
21268 ("cmask16<P:mode>_vis"): Likewise.
21269 ("cmask32<P:mode>_vis"): Likewise.
21270 ("pdistn<P:mode>_vis"): Likewise.
21271 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21273 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21275 * config/sparc/sparc.md ("subtype"): New insn attribute.
21276 ("*wrgsr_sp64"): Set insn subtype.
21277 ("*rdgsr_sp64"): Likewise.
21278 ("alignaddrsi_vis"): Likewise.
21279 ("alignaddrdi_vis"): Likewise.
21280 ("alignaddrlsi_vis"): Likewise.
21281 ("alignaddrldi_vis"): Likewise.
21282 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21283 ("fexpand_vis"): Likewise.
21284 ("fpmerge_vis"): Likewise.
21285 ("faligndata<VM64:mode>_vis"): Likewise.
21286 ("bshuffle<VM64:mode>_vis"): Likewise.
21287 ("cmask8<P:mode>_vis"): Likewise.
21288 ("cmask16<P:mode>_vis"): Likewise.
21289 ("cmask32<P:mode>_vis"): Likewise.
21290 ("fchksm16_vis"): Likewise.
21291 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21292 ("fmean16_vis"): Likewise.
21293 ("fp<plusminus_insn>64_vis"): Likewise.
21294 ("<plusminus_insn>v8qi3"): Likewise.
21295 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21296 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21297 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21298 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21299 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21300 ("*movqi_insn"): Likewise.
21301 ("*movhi_insn"): Likewise.
21302 ("*movsi_insn"): Likewise.
21303 ("movsi_pic_gotdata_op"): Likewise.
21304 ("*movdi_insn_sp32"): Likewise.
21305 ("*movdi_insn_sp64"): Likewise.
21306 ("movdi_pic_gotdata_op"): Likewise.
21307 ("*movsf_insn"): Likewise.
21308 ("*movdf_insn_sp32"): Likewise.
21309 ("*movdf_insn_sp64"): Likewise.
21310 ("*zero_extendhisi2_insn"): Likewise.
21311 ("*zero_extendqihi2_insn"): Likewise.
21312 ("*zero_extendqisi2_insn"): Likewise.
21313 ("*zero_extendqidi2_insn"): Likewise.
21314 ("*zero_extendhidi2_insn"): Likewise.
21315 ("*zero_extendsidi2_insn_sp64"): Likewise.
21316 ("ldfsr"): Likewise.
21317 ("prefetch_64"): Likewise.
21318 ("prefetch_32"): Likewise.
21319 ("tie_ld32"): Likewise.
21320 ("tie_ld64"): Likewise.
21321 ("*tldo_ldub_sp32"): Likewise.
21322 ("*tldo_ldub1_sp32"): Likewise.
21323 ("*tldo_ldub2_sp32"): Likewise.
21324 ("*tldo_ldub_sp64"): Likewise.
21325 ("*tldo_ldub1_sp64"): Likewise.
21326 ("*tldo_ldub2_sp64"): Likewise.
21327 ("*tldo_ldub3_sp64"): Likewise.
21328 ("*tldo_lduh_sp32"): Likewise.
21329 ("*tldo_lduh1_sp32"): Likewise.
21330 ("*tldo_lduh_sp64"): Likewise.
21331 ("*tldo_lduh1_sp64"): Likewise.
21332 ("*tldo_lduh2_sp64"): Likewise.
21333 ("*tldo_lduw_sp32"): Likewise.
21334 ("*tldo_lduw_sp64"): Likewise.
21335 ("*tldo_lduw1_sp64"): Likewise.
21336 ("*tldo_ldx_sp64"): Likewise.
21337 ("*mov<VM32:mode>_insn"): Likewise.
21338 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21339 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21341 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21343 * config/sparc/sparc.md ("type"): New insn type viscmp.
21344 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21346 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21347 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21348 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21349 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21351 ("n7_vis_logical_11cycle"): Likewise.
21352 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21353 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21354 * config/sparc/niagara.md ("niag_vis"): Likewise.
21355 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21356 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21358 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21360 * config/sparc/sparc.md: New instruction type `bmask'.
21361 (bmaskdi_vis): Use the `bmask' type.
21362 (bmasksi_vis): Likewise.
21363 * config/sparc/ultra3.md (us3_array): Likewise.
21364 * config/sparc/niagara7.md (n7_array): Likewise.
21365 * config/sparc/niagara4.md (n4_array): Likewise.
21366 * config/sparc/niagara2.md (niag2_vis): Likewise.
21367 (niag3_vis): Likewise.
21368 * config/sparc/niagara.md (niag_vis): Likewise.
21370 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21372 * ipa-comdats.c: Remove optimize check from gate.
21373 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21374 for functions not optimized.
21375 (ipa_fn_summary_read): Skip optimize check.
21376 (ipa_fn_summary_write): Likewise.
21377 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21379 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21381 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21382 (check_callers): Give up on caller which is not optimized.
21383 (inline_small_functions): Likewise.
21384 (ipa_inline): Do not give up when not optimizing.
21385 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21386 away unoptimizes cdtors.
21387 (whole_program_function_and_variable_visibility): Do
21388 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21389 * ipa.c (process_references): Do not check optimize.
21390 (symbol_table::remove_unreachable_nodes): Update optimize check.
21391 (set_writeonly_bit): Update optimize check.
21392 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21393 (pass_ipa_single_use::gate): Remove.
21395 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21397 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21398 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21399 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21400 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21401 permute_load, permute_store, adjust_extract, adjust_splat,
21402 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21403 replace_swap_with_copy, dump_swap_insn_table,
21404 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21405 recombine_lvx_pattern, recombine_stvx_pattern,
21406 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21407 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21408 to file rs6000-p8swap.c.
21409 * config/rs6000/rs6000-p8swap.c: New file.
21410 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21411 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21412 and rs6000*-*-* targets.
21414 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21416 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21418 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21420 * lto-wrapper.c (merge_and_complain): Do not merge
21421 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21422 fsigned_zeros, ftrapping_math, fwrapv.
21423 (append_compiler_options): Do not track these options.
21424 (append_linker_options): Likewie
21426 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21428 * cgraphunit.c (cgraph_node::finalize_function): When
21429 !flag_toplevel_reorde set no_reorder flag.
21430 (varpool_node::finalize_decl): Likewise.
21431 (symbol_table::compile): Drop no toplevel reorder path.
21433 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21435 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21436 edges; zero probability is not better than uninitialized.
21438 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21440 * asan.h (asan_sanitize_allocas_p): Declare.
21441 * asan.c (asan_sanitize_allocas_p): New function.
21442 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21443 (handle_builtin_alloca): Likewise.
21444 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21445 if !asan_sanitize_allocas_p.
21446 * params.def (asan-instrument-allocas): Add new option.
21447 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21448 * opts.c (common_handle_option): Disable allocas sanitization for
21451 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21453 * asan.c: Include gimple-fold.h.
21454 (get_last_alloca_addr): New function.
21455 (handle_builtin_stackrestore): Likewise.
21456 (handle_builtin_alloca): Likewise.
21457 (asan_emit_allocas_unpoison): Likewise.
21458 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21459 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21460 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21461 (instrument_builtin_call): Pass gimple iterator to
21462 get_mem_refs_of_builtin_call.
21463 (last_alloca_addr): New global.
21464 * asan.h (asan_emit_allocas_unpoison): Declare.
21465 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21466 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21467 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21468 if function calls alloca.
21469 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21470 * gimple-fold.h (replace_call_with_value): Declare.
21471 * internal-fn.c: Include asan.h.
21472 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21473 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21475 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21477 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21478 (C_SELFTEST_FLAGS): New.
21479 (CPP_SELFTEST_FLAGS): New.
21480 (SELFTEST_DEPS): New, from deps of s-selftest.
21481 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21482 (CPP_SELFTEST_DEPS): New.
21483 (selftest): Add dependency on s-selftest-c++.
21484 (s-selftest): Rename to...
21485 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21486 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21487 than SELFTEST_FLAGS.
21488 (selftest-gdb): Rename to...
21489 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21491 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21492 (selftest-valgrind): Rename to...
21493 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21495 (selftest-valgrind): Reintroduce as an alias for
21496 selftest-c-valgrind.
21497 (s-selftest-c++): New.
21498 (selftest-c++-gdb): New.
21499 (selftest-c++-valgrind): New.
21501 2017-07-06 Olivier Hainque <hainque@adacore.com>
21503 * gcc.c (process_command): When deciding if undefined variables
21504 should be ignored when processing specs, accept "gcc -v" as well.
21506 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21508 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21509 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21511 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21513 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21514 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21515 * config/arm/arm-tables.opt: Regenerate.
21516 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21518 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21520 2017-07-06 Carl Love <cel@us.ibm.com>
21522 * ChangeLog: Clean up from mid air collision
21524 2017-07-06 Carl Love <cel@us.ibm.com>
21526 * config/rs6000/rs6000-c.c: Add support for built-in functions
21527 vector signed int vec_subc (vector signed int, vector signed int);
21528 vector signed __int128 vec_subc (vector signed __int128,
21529 vector signed __int128);
21530 vector unsigned __int128 vec_subc (vector unsigned __int128,
21531 vector unsigned __int128);
21532 vector signed int vec_sube (vector signed int, vector signed int,
21533 vector signed int);
21534 vector unsigned int vec_sube (vector unsigned int,
21535 vector unsigned int,
21536 vector unsigned int);
21537 vector signed __int128 vec_sube (vector signed __int128,
21538 vector signed __int128,
21539 vector signed__int128);
21540 vector unsigned __int128 vec_sube (vector unsigned __int128,
21541 vector unsigned __int128,
21542 vector unsigned __int128);
21543 vector signed int vec_subec (vector signed int, vector signed int,
21544 vector signed int);
21545 vector unsigned int vec_subec (vector unsigned int,
21546 vector unsigned int,
21547 vector unsigned int);
21548 vector signed __int128 vec_subec (vector signed __int128,
21549 vector signed __int128,
21550 vector signed__int128);
21551 vector unsigned __int128 vec_subec (vector unsigned __int128,
21552 vector unsigned __int128,
21553 vector unsigned __int128);
21554 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21555 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21556 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21557 BU_ALTIVEC_OVERLOAD_X definitions.
21558 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21559 * doc/extend.texi: Update the built-in documentation file for the new
21560 built-in functions.
21562 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21565 * diagnostic-show-locus.c (layout::layout): Use start and finish
21566 spelling location for the start and finish of each range.
21567 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21568 Add unused aspect param.
21569 * input.c (expand_location_1): Add "aspect" param, and use it
21570 to access the correct part of the location.
21571 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21573 (expand_location_to_spelling_point): Likewise.
21574 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21575 param, and pass it to expand_location_1.
21577 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21579 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21580 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21581 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21582 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21583 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21584 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21585 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21586 _mm_maskz_getmant_ss): New intrinsics.
21587 (__builtin_ia32_getexpss128_mask): Changed to ...
21588 __builtin_ia32_getexpss128_round ... this.
21589 (__builtin_ia32_getexpsd128_mask): Changed to ...
21590 __builtin_ia32_getexpsd128_round ... this.
21591 * config/i386/i386-builtin-types.def
21592 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21593 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21594 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21595 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21596 __builtin_ia32_getmantss_mask_round): New builtins.
21597 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21598 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21599 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21600 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21601 * config/i386/sse.md
21602 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21603 avx512f_sgetexp<mode><mask_scalar_name>
21604 <round_saeonly_scalar_name> ... this.
21605 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21606 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21607 vgetexp<ssescalarmodesuffix>
21608 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21609 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21610 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21611 avx512f_vgetmant<mode><mask_scalar_name>
21612 <round_saeonly_scalar_name> ... this.
21613 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21614 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21615 vgetmant<ssescalarmodesuffix>
21616 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21617 %0<mask_scalar_operand4>, %1, %2
21618 <round_saeonly_scalar_mask_op4>, %3} ... this.
21619 * config/i386/subst.md (mask_scalar_operand4,
21620 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21621 round_saeonly_scalar_nimm_predicate): New subst attributes.
21623 2017-07-06 Julia Koval <julia.koval@intel.com>
21625 * config/i386/i386.c (ix86_erase_embedded_rounding):
21626 Remove code for old rounding pattern.
21628 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21630 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21632 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21634 * doc/sourcebuild.texi (Test Directives, Variants of
21635 dg-require-support): Add documentation for dg-require-stack-check.
21637 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21639 * config/i386/subst.md (mask_scalar, round_scalar,
21640 round_saeonly_scalar): New meta-templates.
21641 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21642 round_scalar_mask_operand3, round_scalar_mask_op3,
21643 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21644 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21645 round_saeonly_scalar_constraint,
21646 round_saeonly_scalar_prefix): New subst attribute.
21647 * config/i386/sse.md
21648 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21649 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21650 <round_scalar_name> ... this.
21651 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21652 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21653 <round_scalar_name> ... this.
21654 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21655 <sse>_vm<code><mode>3<mask_scalar_name>
21656 <round_saeonly_scalar_name> ... this.
21657 (v<plusminus_mnemonic><ssescalarmodesuffix>
21658 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21659 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21660 v<plusminus_mnemonic><ssescalarmodesuffix>
21661 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21662 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21663 (v<multdiv_mnemonic><ssescalarmodesuffix>
21664 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21665 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21666 v<multdiv_mnemonic><ssescalarmodesuffix>
21667 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21668 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21669 (v<maxmin_float><ssescalarmodesuffix>
21670 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21671 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21672 v<maxmin_float><ssescalarmodesuffix>
21673 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21674 %0<mask_scalar_operand3>, %1, %<iptr>2
21675 <round_saeonly_scalar_mask_op3>} ... this.
21677 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21679 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21680 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21682 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21683 Alan Hayward <alan.hayward@arm.com>
21684 David Sherwood <david.sherwood@arm.com>
21686 * combine.c (simplify_if_then_else): Remove "enum" before
21688 * compare-elim.c (can_eliminate_compare): Likewise.
21689 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21691 (aarch64_lookup_simd_builtin_type): Likewise.
21692 (aarch64_simd_builtin_type): Likewise.
21693 (aarch64_init_simd_builtin_types): Likewise.
21694 (aarch64_simd_expand_args): Likewise.
21695 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21697 (aarch64_reverse_mask): Likewise.
21698 (aarch64_simd_emit_reg_reg_move): Likewise.
21699 (aarch64_gen_adjusted_ldpstp): Likewise.
21700 (aarch64_ccmp_mode_to_code): Likewise.
21701 (aarch64_operands_ok_for_ldpstp): Likewise.
21702 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21703 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21705 (aarch64_min_divisions_for_recip_mul): Likewise.
21706 (aarch64_reassociation_width): Likewise.
21707 (aarch64_get_condition_code_1): Likewise.
21708 (aarch64_simd_emit_reg_reg_move): Likewise.
21709 (aarch64_simd_attr_length_rglist): Likewise.
21710 (aarch64_reverse_mask): Likewise.
21711 (aarch64_operands_ok_for_ldpstp): Likewise.
21712 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21713 (aarch64_gen_adjusted_ldpstp): Likewise.
21714 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21716 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21717 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21718 (arm_lookup_simd_builtin_type): Likewise.
21719 (arm_simd_builtin_type): Likewise.
21720 (arm_init_simd_builtin_types): Likewise.
21721 (arm_expand_builtin_args): Likewise.
21722 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21723 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21724 (ft32_setup_incoming_varargs): Likewise.
21725 (ft32_function_arg): Likewise.
21726 (ft32_function_arg_advance): Likewise.
21727 (ft32_pass_by_reference): Likewise.
21728 (ft32_arg_partial_bytes): Likewise.
21729 (ft32_valid_pointer_mode): Likewise.
21730 (ft32_addr_space_pointer_mode): Likewise.
21731 (ft32_addr_space_legitimate_address_p): Likewise.
21732 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21734 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21735 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21736 (iamcu_alignment): Likewise.
21737 (canonicalize_vector_int_perm): Likewise.
21738 (ix86_noce_conversion_profitable_p): Likewise.
21739 (ix86_mpx_bound_mode): Likewise.
21740 (ix86_operands_ok_for_move_multiple): Likewise.
21741 * config/microblaze/microblaze-protos.h
21742 (microblaze_expand_conditional_branch_reg): Likewise.
21743 * config/microblaze/microblaze.c
21744 (microblaze_expand_conditional_branch_reg): Likewise.
21745 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21747 (rs6000_reassociation_width): Likewise.
21748 (rs6000_invalid_binary_op): Likewise.
21749 (fusion_p9_p): Likewise.
21750 (emit_fusion_p9_load): Likewise.
21751 (emit_fusion_p9_store): Likewise.
21752 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21754 (riscv_hard_regno_mode_ok_p): Likewise.
21755 (riscv_address_insns): Likewise.
21756 (riscv_split_symbol): Likewise.
21757 (riscv_legitimize_move): Likewise.
21758 (riscv_function_value): Likewise.
21759 (riscv_hard_regno_nregs): Likewise.
21760 (riscv_expand_builtin): Likewise.
21761 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21762 (riscv_build_integer): Likewise.
21763 (riscv_split_integer): Likewise.
21764 (riscv_legitimate_constant_p): Likewise.
21765 (riscv_cannot_force_const_mem): Likewise.
21766 (riscv_regno_mode_ok_for_base_p): Likewise.
21767 (riscv_valid_base_register_p): Likewise.
21768 (riscv_valid_offset_p): Likewise.
21769 (riscv_valid_lo_sum_p): Likewise.
21770 (riscv_classify_address): Likewise.
21771 (riscv_legitimate_address_p): Likewise.
21772 (riscv_address_insns): Likewise.
21773 (riscv_load_store_insns): Likewise.
21774 (riscv_force_binary): Likewise.
21775 (riscv_split_symbol): Likewise.
21776 (riscv_force_address): Likewise.
21777 (riscv_legitimize_address): Likewise.
21778 (riscv_move_integer): Likewise.
21779 (riscv_legitimize_const_move): Likewise.
21780 (riscv_legitimize_move): Likewise.
21781 (riscv_address_cost): Likewise.
21782 (riscv_subword): Likewise.
21783 (riscv_output_move): Likewise.
21784 (riscv_canonicalize_int_order_test): Likewise.
21785 (riscv_emit_int_order_test): Likewise.
21786 (riscv_function_arg_boundary): Likewise.
21787 (riscv_pass_mode_in_fpr_p): Likewise.
21788 (riscv_pass_fpr_single): Likewise.
21789 (riscv_pass_fpr_pair): Likewise.
21790 (riscv_get_arg_info): Likewise.
21791 (riscv_function_arg): Likewise.
21792 (riscv_function_arg_advance): Likewise.
21793 (riscv_arg_partial_bytes): Likewise.
21794 (riscv_function_value): Likewise.
21795 (riscv_pass_by_reference): Likewise.
21796 (riscv_setup_incoming_varargs): Likewise.
21797 (riscv_print_operand): Likewise.
21798 (riscv_elf_select_rtx_section): Likewise.
21799 (riscv_save_restore_reg): Likewise.
21800 (riscv_for_each_saved_reg): Likewise.
21801 (riscv_register_move_cost): Likewise.
21802 (riscv_hard_regno_mode_ok_p): Likewise.
21803 (riscv_hard_regno_nregs): Likewise.
21804 (riscv_class_max_nregs): Likewise.
21805 (riscv_memory_move_cost): Likewise.
21806 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21807 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21808 (rl78_addr_space_address_mode): Likewise.
21809 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21811 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21812 (rs6000_reassociation_width): Likewise.
21813 (rs6000_invalid_binary_op): Likewise.
21814 (fusion_p9_p): Likewise.
21815 (emit_fusion_p9_load): Likewise.
21816 (emit_fusion_p9_store): Likewise.
21817 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
21818 (ok_for_simple_move_operands): Likewise.
21819 (ok_for_simple_move_strict_operands): Likewise.
21820 (ok_for_simple_arith_logic_operands): Likewise.
21821 (visium_legitimize_reload_address): Likewise.
21822 (visium_select_cc_mode): Likewise.
21823 (output_cbranch): Likewise.
21824 (visium_split_double_move): Likewise.
21825 (visium_expand_copysign): Likewise.
21826 (visium_expand_int_cstore): Likewise.
21827 (visium_expand_fp_cstore): Likewise.
21828 * config/visium/visium.c (visium_pass_by_reference): Likewise.
21829 (visium_function_arg): Likewise.
21830 (visium_function_arg_advance): Likewise.
21831 (visium_libcall_value): Likewise.
21832 (visium_setup_incoming_varargs): Likewise.
21833 (visium_legitimate_constant_p): Likewise.
21834 (visium_legitimate_address_p): Likewise.
21835 (visium_legitimize_address): Likewise.
21836 (visium_secondary_reload): Likewise.
21837 (visium_register_move_cost): Likewise.
21838 (visium_memory_move_cost): Likewise.
21839 (prepare_move_operands): Likewise.
21840 (ok_for_simple_move_operands): Likewise.
21841 (ok_for_simple_move_strict_operands): Likewise.
21842 (ok_for_simple_arith_logic_operands): Likewise.
21843 (visium_function_value_1): Likewise.
21844 (rtx_ok_for_offset_p): Likewise.
21845 (visium_legitimize_reload_address): Likewise.
21846 (visium_split_double_move): Likewise.
21847 (visium_expand_copysign): Likewise.
21848 (visium_expand_int_cstore): Likewise.
21849 (visium_expand_fp_cstore): Likewise.
21850 (visium_split_cstore): Likewise.
21851 (visium_select_cc_mode): Likewise.
21852 (visium_split_cbranch): Likewise.
21853 (output_cbranch): Likewise.
21854 (visium_print_operand_address): Likewise.
21855 * expmed.c (flip_storage_order): Likewise.
21856 * expmed.h (emit_cstore): Likewise.
21857 (flip_storage_order): Likewise.
21858 * genrecog.c (validate_pattern): Likewise.
21859 * hsa-gen.c (gen_hsa_addr): Likewise.
21860 * internal-fn.c (expand_arith_overflow): Likewise.
21861 * ira-color.c (allocno_copy_cost_saving): Likewise.
21862 * lra-assigns.c (find_hard_regno_for_1): Likewise.
21863 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
21864 (process_invariant_for_inheritance): Likewise.
21865 * lra-eliminations.c (move_plus_up): Likewise.
21866 * omp-low.c (lower_oacc_reductions): Likewise.
21867 * simplify-rtx.c (simplify_subreg): Likewise.
21868 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
21869 (TARGET_CHKP_BOUND_MODE): Likewise..
21870 * targhooks.c (default_chkp_bound_mode): Likewise.
21871 (default_setup_incoming_vararg_bounds): Likewise.
21872 * targhooks.h (default_chkp_bound_mode): Likewise.
21873 (default_setup_incoming_vararg_bounds): Likewise.
21874 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
21875 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
21876 (have_whole_vector_shift): Likewise.
21877 * tree-vect-stmts.c (vectorizable_load): Likewise.
21878 * doc/tm.texi: Regenerate.
21880 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21882 Graceful degrade if Binutils PR21472 is not available.
21885 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
21886 .rodata in flash test fails.
21887 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
21888 * confgure: Regenerate.
21889 * config.in: Regenerate.
21890 * config/avr/avr.c (avr_asm_named_section)
21891 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
21892 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
21893 (avr_asm_init_sections): Same.
21895 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21897 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
21898 (fma<VH:mode>4_intrinsic): Likewise.
21899 (*fmsub<VCVTF:mode>4): Likewise.
21900 (*fmsub<VH:mode>4_intrinsic): Likewise.
21902 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21905 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
21906 Don't depend on "optimize > 0".
21907 (out_movhi_r_mr, out_movqi_mr_r): Same.
21908 (out_movhi_mr_r, out_movqi_r_mr): Same.
21909 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
21910 io_address_operand on "optimize > 0".
21912 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21914 * tree-loop-distribution.c: Add general explanantion on the pass.
21915 (generate_loops_for_partition): Mark distributed loop.
21916 (pg_add_dependence_edges): New parameter. Handle alias data
21917 dependence specially and record it in the parameter if asked.
21918 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
21919 (init_partition_graph_vertices, add_partition_graph_edge): New.
21920 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
21921 (free_partition_graph_vdata, build_partition_graph): New.
21922 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
21923 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
21924 (data_ref_segment_size, latch_dominated_by_data_ref): New.
21925 (compute_alias_check_pairs, version_loop_by_alias_check): New.
21926 (version_for_distribution_p, finalize_partitions): New.
21927 (distribute_loop): Handle alias data dependence specially. Factor
21928 out loop fusion code as functions and call these functions.
21930 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21932 * tree-loop-distribution.c (classify_partition): New parameter and
21933 better handle reduction statement.
21934 (rdg_build_partitions): Revise comment.
21935 (distribute_loop): Compute statements in all partitions and pass it
21936 to classify_partition.
21938 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21940 * tree-loop-distribution.c (enum partition_type): New.
21941 (struct partition): New field type.
21942 (partition_merge_into): Add parameter. Update partition type.
21943 (data_dep_in_cycle_p, update_type_for_merge): New functions.
21944 (build_rdg_partition_for_vertex): Compute partition type.
21945 (rdg_build_partitions): Dump partition type.
21946 (distribute_loop): Update calls to partition_merge_into.
21948 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21950 * tree-loop-distribution.c (struct ddr_hasher): New.
21951 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
21953 (classify_partition): Call get_data_dependence.
21954 (pg_add_dependence_edges): Ditto.
21955 (distribute_loop): Release data dependence hash table.
21957 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21959 * tree-loop-distribution.c (ref_base_address): Delete.
21960 (similar_memory_accesses): Rename ...
21961 (share_memory_accesses): ... to this. Check if partitions access
21962 the same memory reference.
21963 (distribute_loop): Call share_memory_accesses.
21965 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21967 * tree-loop-distribution.c (struct partition): New field recording
21968 its data reference.
21969 (partition_alloc, partition_free): Init and release data refs.
21970 (partition_merge_into): Merge data refs.
21971 (build_rdg_partition_for_vertex): Collect data refs for partition.
21972 (pg_add_dependence_edges): Change parameters from vector to bitmap.
21974 (distribute_loop): Remve data refs from vertice data of partition
21977 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21979 * tree-loop-distribution.c (params.h): Include header file.
21980 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
21981 (datarefs_vec): New global var.
21982 (create_rdg_vertices): Use datarefs_vec directly.
21983 (free_rdg): Don't free data references.
21984 (build_rdg): Update use. Don't free data references.
21985 (distribute_loop): Compute global variable for data references.
21986 Bail out if there are too many data references.
21988 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21990 * tree-loop-distribution.c (loop_nest): New global var.
21991 (build_rdg): Use loop directly, rather than loop nest.
21992 (pg_add_dependence_edges): Remove loop nest parameter. Use global
21994 (distribute_loop): Compute global variable loop nest. Update use.
21996 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21998 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
21999 (partition_merge_into): New parameter. Dump reason for fusion.
22000 (distribute_loop): Update use of partition_merge_into.
22002 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22004 * tree-loop-distribution.c (bb_top_order_index): New.
22005 (bb_top_order_index_size, bb_top_order_cmp): New.
22006 (stmts_from_loop): Use topological order.
22007 (pass_loop_distribution::execute): Compute and release topological
22008 order for basic blocks.
22010 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22012 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22015 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22017 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22018 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22019 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22020 * internal-fn.def (LOOP_DIST_ALIAS): New.
22021 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22022 (fold_loop_internal_call): ... this.
22023 (vect_loop_dist_alias_call): New function.
22024 (set_uid_loop_bbs): Call fold_loop_internal_call.
22025 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22028 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22031 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22032 Require dead FLAGS_REG at the beginning of a peephole.
22034 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22037 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22038 arguments in the call to __builtin_ia32_sbb_u32.
22039 (_subborrow_u64): Swap _X and _Y arguments in the call to
22040 __builtin_ia32_sbb_u64.
22042 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22045 * tree-vrp.c (compare_assert_loc): Turn into a function template
22046 with stable template parameter. Only test if a->e is NULL,
22047 !a->e == !b->e has been verified already. Use e == NULL or
22048 e != NULL instead of e or ! e tests. If stable is true, don't use
22049 iterative_hash_expr, on the other side allow a or b or both NULL
22050 and sort the NULLs last.
22051 (process_assert_insertions): Sort using compare_assert_loc<false>
22052 instead of compare_assert_loc, later sort using
22053 compare_assert_loc<true> before calling process_assert_insertions_for
22054 in a loop. Use break instead of continue once seen NULL pointer.
22056 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22058 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22059 Cortex-R7 and Cortex-R8 processors.
22061 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22063 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22064 uninitialized while src is not.
22066 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22068 * common/config/arm/arm-common.c: Adjust include path for
22070 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22071 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22072 (arm-cpu-data.h): Likewise.
22073 (arm-cpu-cdata.h): Likewise.
22074 * config/arm/arm-cpu.h: Delete.
22075 * config/arm/arm-cpu-cdata.h: Delete.
22076 * config/arm/arm-cpu-data.h: Delete.
22078 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22080 * config/arm/arm-cpus.in (cortex-a55): New.
22081 (cortex-a75): Likewise.
22082 (cortex-a75.cortex-a55): Likewise.
22083 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22085 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22086 * config/arm/arm-cpu-cdata.h: Regenerate.
22087 * config/arm/arm-cpu-data.h: Regenerate.
22088 * config/arm/arm-cpu.h: Regenerate.
22089 * config/arm/arm-tables.opt: Regenerate.
22090 * config/arm/arm-tune.md: Regenerate.
22092 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22094 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22096 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22098 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22101 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22103 PR tree-optimization/81292
22104 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22105 full_string_p, also call adjust_related_strinfos if the adjustment
22106 is simple, otherwise invalidate related strinfos.
22108 2017-07-04 Martin Liska <mliska@suse.cz>
22111 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22112 newly created variable as DECL_IGNORED_P.
22114 2017-07-04 Martin Liska <mliska@suse.cz>
22117 * ipa-inline.c (inline_small_functions):
22118 Use xstrdup_for_dump.
22120 2017-07-04 Tom de Vries <tom@codesourcery.com>
22122 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22124 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22127 * config/darwin.c (darwin_function_switched_text_sections):
22130 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22132 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22134 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22136 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22138 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22140 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22141 min_profitable_iters, and th as inclusive lower bounds.
22142 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22143 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22144 for min_profitable_iters and min_profitable_estimate.
22145 (vect_transform_loop): Treat th as an inclusive lower bound.
22146 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22148 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22151 * config/darwin.c (darwin_function_switched_text_sections):
22152 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22153 in two pieces, and suppress the use of buf.
22155 2017-07-03 Nathan Sidwell <nathan@acm.org>
22157 * hash-table.h (hash_table_mod1): Fix indentation.
22159 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22161 PR middle-end/81290
22162 * predict.c (force_edge_cold): Be more careful about propagation
22164 * profile-count.h (profile_probability::guessed,
22165 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22167 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22169 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22171 * doc/invoke.texi (rcpc architecture extension): Document it.
22173 2017-07-03 Richard Biener <rguenther@suse.de>
22175 PR tree-optimization/60510
22176 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22177 the scalar reduction PHI and use it.
22178 (vectorizable_reduction): Properly guard the single_defuse_cycle
22179 path for non-SLP reduction chains where we cannot use it.
22180 Rework reduc_def/index and vector type deduction. Rework
22181 vector operand gathering during reduction op code-gen.
22182 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22183 chains dissolve the chain and leave it to non-SLP reduction
22186 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22188 * tree-data-ref.h (dr_alignment): Declare.
22189 * tree-data-ref.c (dr_alignment): New function.
22190 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22191 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22193 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22195 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22197 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22198 and base_misalignment fields.
22199 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22200 * tree-data-ref.c: Include builtins.h.
22201 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22202 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22203 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22204 * tree-vect-data-refs.c: Include tree-cfg.h.
22205 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22206 fields instead of calculating an alignment here.
22207 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22208 innermost_loop_behavior fields.
22210 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22212 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22214 (DR_STEP_ALIGNMENT): New macro.
22215 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22216 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22217 (create_data_ref): Print it.
22218 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22219 to tell whether the step preserves vector (mis)alignment.
22220 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22221 Move the check for an integer step and generalise to all INTEGER_CST.
22222 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22223 Print the outer step alignment.
22225 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22227 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22228 with offset_alignment.
22229 (DR_ALIGNED_TO): Delete.
22230 (DR_OFFSET_ALIGNMENT): New macro.
22231 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22232 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22233 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22234 (create_data_ref): Likewise.
22235 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22236 (vect_analyze_data_refs): Likewise.
22237 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22238 creating dummy innermost behavior.
22240 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22242 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22243 with a "innermost_loop_behavior *" and refeence tree.
22244 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22245 (create_data_ref): Update call accordingly.
22246 * tree-predcom.c (find_looparound_phi): Likewise.
22248 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22250 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22251 fields with dr_wrt_vec_loop.
22252 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22253 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22254 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22255 (vect_dr_behavior): New function.
22256 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22257 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22258 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22259 track whether the step preserves the misalignment.
22260 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22261 Use vect_dr_behavior.
22262 (vect_setup_realignment): Update call accordingly.
22263 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22264 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22265 call to vect_create_addr_base_for_vector_ref.
22266 (vect_create_cond_for_align_checks): Likewise.
22267 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22268 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22269 (vect_recog_mask_conversion_pattern): Likewise.
22270 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22271 (new_stmt_vec_info): Remove redundant zeroing.
22273 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22275 * common/config/arm/arm-common.c (arm_be8_option): New function.
22276 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22277 (ISA_ARMv6): Add isa_bit_be8.
22278 * config/arm/arm.h (arm_be8_option): Add prototype.
22279 (BE8_SPEC_FUNCTION): New define.
22280 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22281 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22282 (mlittle-endian): Similarly.
22283 (mbe8, mbe32): New options.
22284 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22285 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22287 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22289 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22291 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22293 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22294 (cleanup_tree_cfg_bb): Use it.
22295 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22297 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22299 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22302 * loop-doloop.c (add_test): Update profile.
22304 2017-07-03 Martin Liska <mliska@suse.cz>
22307 * sanopt.c (rewrite_usage_of_param): New function.
22308 (sanitize_rewrite_addressable_params): Likewise.
22309 (pass_sanopt::execute): Call rewrite_usage_of_param.
22311 2017-07-03 Richard Biener <rguenther@suse.de>
22313 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22314 back to using VIEW_CONVERT_EXPR.
22316 2017-07-03 Martin Liska <mliska@suse.cz>
22319 * doc/extend.texi: Document when a resolver function is
22320 generated for target_clones.
22322 2017-07-03 Martin Liska <mliska@suse.cz>
22324 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22325 and shadow memory of auto variables which are subject of
22326 use-after-scope sanitization.
22327 (asan_expand_mark_ifn): Add do set only when is_poison.
22329 2016-07-03 Richard Biener <rguenther@suse.de>
22331 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22333 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22334 (vectorizable_reduction): Perform reduction PHI creation when
22335 visiting a reduction PHI and adjust and simplify code generation
22336 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22337 (vect_transform_loop): Visit reduction PHIs.
22338 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22339 defs into the SLP tree.
22340 (vect_build_slp_tree): Reduction defs terminate the recursion.
22341 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22343 (vect_get_vec_defs_for_stmt_copy): Export.
22344 (vect_get_vec_defs): Likewise.
22345 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22347 (vect_get_vec_defs_for_stmt_copy): Declare.
22348 (vect_get_vec_defs): Likewise.
22350 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22352 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22353 parameter with a "loop" parameter and use it instead of the
22354 loop containing DR_STMT. Don't check simple_iv when doing
22355 BB analysis. Describe the two analysis modes in the comment.
22357 2017-07-03 Tom de Vries <tom@codesourcery.com>
22359 PR tree-optimization/69468
22360 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22361 (find_same_succ_bb): Handle ignore_edge_flags.
22363 2017-07-03 Tom de Vries <tom@codesourcery.com>
22365 PR tree-optimization/81192
22366 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22368 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22370 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22372 2017-07-03 Tom de Vries <tom@codesourcery.com>
22374 PR tree-optimization/81192
22375 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22376 BB_SAME_SUCC (bb) == NULL.
22378 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22380 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22383 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22385 * dumpfile.c: Include profile-count.h
22386 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22388 (insert_cond_bb): Update profile.
22389 * tree-cfg.h (insert_cond_bb): Update prototype.
22390 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22391 * tree-dump.c: Do not include tree-cfg.
22393 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22395 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22397 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22399 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22402 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22404 * tree-complex.c (expand_complex_div_wide): update profile.
22406 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22407 Alan Hayward <alan.hayward@arm.com>
22408 David Sherwood <david.sherwood@arm.com>
22410 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22411 (CORETYPES_H): New define.
22412 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22413 (insn-modes-inline.h, s-modes-inline-h): New rules.
22414 (generated_files): Add insn-modes-inline.h.
22415 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22416 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22417 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22418 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22419 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22420 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22421 (build/gencodes.o, build/genconditions.o): Likewise.
22422 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22423 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22424 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22425 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22426 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22427 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22428 * coretypes.h: Include everything up to real.h for generators.
22429 Include insn-modes.h first. Include wide-int-print.h after
22430 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22431 * machmode.h: Don't include insn-modes.h here.
22432 * function-tests.c: Remove includes of signop.h, machmode.h,
22433 double-int.h and wide-int.h.
22435 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22437 * optc-save-gen.awk: Likewise.
22438 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22439 * godump.c: Remove include of wide-int-print.h.
22440 * pretty-print.h: Likewise.
22441 * wide-int-print.cc: Likewise.
22442 * wide-int.cc: Likewise.
22443 * hash-map-tests.c: Remove include of signop.h.
22444 * hash-set-tests.c: Likewise.
22445 * rtl-tests.c: Likewise.
22446 * mkconfig.sh: Remove include of machmode.h.
22447 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22449 (emit_insn_modes_inline_h): ...this new function. Emit the code
22450 into an insn-modes-inline.h header file, adding appropriate
22451 include guards and end comments.
22452 (emit_insn_modes_c_header): Remove include of machmode.h.
22453 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22455 (main): Handle -i flag and call emit_insn_modes_inline_h when
22458 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22460 * tree-ssa-strlen.c (strinfo): Rename the length field to
22461 nonzero_chars. Add a full_string_p field.
22462 (compare_nonzero_chars, zero_length_string_p): New functions.
22463 (get_addr_stridx): Add an offset_out parameter.
22464 Use compare_nonzero_chars.
22465 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22466 (new_strinfo): Update after above changes to strinfo.
22467 (set_endptr_and_length): Set full_string_p.
22468 (get_string_length): Update after above changes to strinfo.
22469 (unshare_strinfo): Update call to new_strinfo.
22470 (maybe_invalidate): Likewise.
22471 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22472 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22473 as a uhwi instead of an shwi. Update after above changes to
22474 strinfo and new_strinfo.
22475 (zero_length_string): Assert that chainsi contains full strings.
22476 Use zero_length_string_p. Update call to new_strinfo.
22477 (adjust_related_strinfos): Update after above changes to strinfo.
22478 Copy full_string_p from origsi.
22479 (adjust_last_stmt): Use zero_length_string_p.
22480 (handle_builtin_strlen): Update after above changes to strinfo and
22481 new_strinfo. Install the lhs as the string length if the previous
22482 entry didn't describe a full string.
22483 (handle_builtin_strchr): Update after above changes to strinfo
22485 (handle_builtin_strcpy): Likewise.
22486 (handle_builtin_strcat): Likewise.
22487 (handle_builtin_malloc): Likewise.
22488 (handle_pointer_plus): Likewise.
22489 (handle_builtin_memcpy): Likewise. Track nonzero characters
22490 that aren't necessarily followed by a nul terminator.
22491 (handle_char_store): Likewise.
22493 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22495 PR tree-optimization/80769
22496 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22497 for malloc and calloc. Document the new invariant that all related
22498 strinfos have delayed lengths or none do.
22499 (verify_related_strinfos): Move earlier in file.
22500 (set_endptr_and_length): New function, split out from...
22501 (get_string_length): ...here. Also set the lengths of related
22503 (zero_length_string): Assert that chainsi has known (rather than
22505 (adjust_related_strinfos): Likewise.
22507 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22509 PR tree-optimization/81136
22510 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22511 assert that two references with the same misalignment have the same
22512 compile-time misalignment if those compile-time misalignments
22515 2017-07-01 Andi Kleen <ak@linux.intel.com>
22517 * print-tree.c (print_node): Print all attributes.
22519 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22521 * cfg.c (scale_bbs_frequencies): New function.
22522 * cfg.h (scale_bbs_frequencies): Declare it.
22523 * cfgloopanal.c (single_likely_exit): Cleanup.
22524 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22526 (scale_loop_profile): Likewise.
22527 (loop_version): Likewise.
22528 (create_empty_loop_on_edge): Update.
22529 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22530 scale_loop_frequencies, scale_loop_profile, loopify,
22531 loop_version): Update prototypes.
22532 * modulo-sched.c (sms_schedule): Update.
22533 * predict.c (unlikely_executed_edge_p): Also check probability.
22534 (probably_never_executed_edge_p): Fix typo.
22535 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22536 * tree-parloops.c (gen_parallel_loop): Update.
22537 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22538 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22539 * tree-ssa-loop-split.c (split_loop): Update.
22540 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22541 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22542 (vect_loop_versioning): Update.
22543 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22545 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22547 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22549 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22551 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22552 to keep profile consistent.
22554 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22556 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22557 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22558 * profile-count.h (max_safe_multiplier): Make unsigned.
22559 (profile_count::guessed_zero): New.
22561 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22563 * bb-reorder.c (fix_up_crossing_landing_pad,
22564 fix_crossing_conditional_branches): Use make_single_succ_edge
22565 to keep profile consistent.
22567 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22569 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22572 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22575 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22576 the right scopes, make sure cond_jump isn't preserved between multiple
22577 iterations. Search for fallthru edge whenever there are 3+ edges and
22578 use find_fallthru_edge for it.
22580 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22582 Patch by Alexander Monakov <amonakov@ispras.ru>
22583 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22584 probabilities consistently.
22586 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22588 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22589 emit_cmp_and_jump_insns.
22591 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22594 * tree-inline.c (expand_call_inline): Combine profile statuses.
22596 2017-06-30 Andrew Pinski <apinski@cavium.com>
22598 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22599 fold_stmt returned true.
22601 2017-06-30 Nathan Sidwell <nathan@acm.org>
22603 * ggc.h (empty_string): Delete.
22604 * cfgexpand.c (expand_asm_stmt): Use plain "".
22605 * optabs.c (expand_asm_memory_barrier): Likewise.
22606 * stringpool.c (empty_string): Delete.
22607 (digit_vector, digit_string): Delete.
22608 (ggc_alloc_string): Use plain "", don't optimize single digit
22609 strings. Use ggc_alloc_atomic.
22611 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22613 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22614 comparison set and one other set, use the cost of the non-comparison
22617 2017-06-30 Nathan Sidwell <nathan@acm.org>
22619 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22622 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22624 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22625 loops. Remove now unneeded calls to gimple_switch_set_label() that
22626 just set removed labels to NULL_TREE.
22628 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22630 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22631 (set_range_info): ...here. Only call set_range_info_raw if domain
22633 (set_nonzero_bits): Call set_range_info_raw.
22634 * tree-ssanames.h (set_range_info_raw): New.
22636 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22639 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22640 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22641 of nonimmediate_operand and <store_mask_constraint> instead of m
22642 for the input operand. For V8FI iterator, always split if input
22643 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22644 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22645 <store_mask_predicate> instead of register_operand and
22646 <store_mask_constraint> instead of v for the input operand. Make
22647 sure both operands aren't MEMs for if not <mask_applied>.
22649 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22651 * lto-wrapper.c (copy_file) Close both file descriptors before
22654 2017-06-30 Martin Liska <mliska@suse.cz>
22657 * multiple_target.c (create_dispatcher_calls): Make ifunc
22658 also for function that don't have calls or are not referenced.
22660 2017-06-30 Richard Biener <rguenther@suse.de>
22662 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22663 analyze the first scalar stmt. Move vector type computation
22664 for the BB case here from ...
22665 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22666 live operation processing in the SLP case properly.
22668 2017-06-30 Richard Biener <rguenther@suse.de>
22670 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22672 2017-06-30 Martin Liska <mliska@suse.cz>
22675 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22676 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22678 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22680 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22683 2017-06-30 Richard Biener <rguenther@suse.de>
22685 PR tree-optimization/81249
22686 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22687 condition reduction result to original scalar type.
22689 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22691 * profile-count.h (enum profile_quality): Fix typos and whitespace
22694 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22696 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22697 type for branch probabilities.
22699 2017-06-29 Julian Brown <julian@codesourcery.com>
22700 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22702 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22703 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22704 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22705 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22707 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22709 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22710 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22711 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22712 CC usage from generic code to here.
22713 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22714 CC usage into the target macros.
22716 2017-06-29 Maya Rashish <coypu@sdf.org>
22718 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22721 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22723 * arm/arm-builtins.c: Include profile-count.h
22724 * except.c (sjlj_emit_function_enter): Use
22725 profile_probability::unlikely.
22727 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22729 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22730 and tocrel_offset be pointer args rather than implicitly using
22732 (legitimate_constant_pool_address_p, rs6000_emit_move,
22733 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22734 tocrel_offset and use in toc_relative_expr_p call.
22735 (print_operand, print_operand_address): Use static tocrel_base_oac
22736 and tocrel_offset_oac.
22737 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22740 2017-06-29 Maya Rashish <coypu@sdf.org>
22742 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22744 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22746 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22747 objects, take into account only the alignment of 'op0' and 'mode1' if
22750 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22752 * ccmp.c (ccmp_tree_comparison_p): New function.
22753 (ccmp_candidate_p): Update to use above function.
22754 (get_compare_parts): New function.
22755 (expand_ccmp_next): Update to use new functions.
22756 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22758 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22759 take mode as argument.
22760 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22761 * expr.c (expand_expr_real_1): Pass mode as argument.
22763 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22765 * combine.c (combine_instructions): Print insns to dump_file, together
22768 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22770 * asan.c (asan_emit_stack_protection): Update.
22771 (create_cond_insert_point): Update.
22772 * auto-profile.c (afdo_propagate_circuit): Update.
22773 * basic-block.h (struct edge_def): Turn probability to
22774 profile_probability.
22775 (EDGE_FREQUENCY): Update.
22776 * bb-reorder.c (find_traces_1_round): Update.
22777 (better_edge_p): Update.
22778 (sanitize_hot_paths): Update.
22779 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22780 (make_single_succ_edge): Update.
22781 (check_bb_profile): Update.
22782 (dump_edge_info): Update.
22783 (update_bb_profile_for_threading): Update.
22784 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22785 probabilitycount to 0.
22786 * cfgbuild.c (compute_outgoing_frequencies): Update.
22787 * cfgcleanup.c (try_forward_edges): Update.
22788 (outgoing_edges_match): Update.
22789 (try_crossjump_to_edge): Update.
22790 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22791 (expand_gimple_tailcall): Update.
22792 (construct_init_block): Use make_single_succ_edge.
22793 (construct_exit_block): Use make_single_succ_edge.
22794 * cfghooks.c (verify_flow_info): Update.
22795 (redirect_edge_succ_nodup): Update.
22796 (split_edge): Update.
22797 (account_profile_record): Update.
22798 * cfgloopanal.c (single_likely_exit): Update.
22799 * cfgloopmanip.c (scale_loop_profile): Update.
22800 (set_zero_probability): Remove.
22801 (duplicate_loop_to_header_edge): Update.
22802 * cfgloopmanip.h (loop_version): Update prototype.
22803 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22804 (force_nonfallthru_and_redirect): Update.
22805 (update_br_prob_note): Update.
22806 (rtl_verify_edges): Update.
22807 (purge_dead_edges): Update.
22808 (rtl_lv_add_condition_to_bb): Update.
22809 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22810 * cgraphunit.c (init_lowered_empty_function): Update.
22811 (cgraph_node::expand_thunk): Update.
22812 * cilk-common.c: Include profile-count.h
22813 * dojump.c (inv): Remove.
22814 (jumpifnot): Update.
22815 (jumpifnot_1): Update.
22816 (do_jump_1): Update.
22818 (do_jump_by_parts_greater_rtx): Update.
22819 (do_compare_rtx_and_jump): Update.
22820 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
22821 do_jump_1. do_compare_rtx_and_jump): Update prototype.
22822 * dwarf2cfi.c: Include profile-count.h
22823 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
22824 (sjlj_emit_dispatch_table): Likewise.
22825 * explow.c: Include profile-count.h
22826 * expmed.c (emit_store_flag_force): Update.
22827 (do_cmp_and_jump): Update.
22828 * expr.c (compare_by_pieces_d::generate): Update.
22829 (compare_by_pieces_d::finish_mode): Update.
22830 (emit_block_move_via_loop): Update.
22831 (store_expr_with_bounds): Update.
22832 (store_constructor): Update.
22833 (expand_expr_real_2): Update.
22834 (expand_expr_real_1): Update.
22835 * expr.h (try_casesi, try_tablejump): Update prototypes.
22836 * gimple-pretty-print.c (dump_probability): Update.
22837 (dump_profile): New.
22838 (dump_gimple_label): Update.
22839 (dump_gimple_bb_header): Update.
22840 * graph.c (draw_cfg_node_succ_edges): Update.
22841 * hsa-gen.c (convert_switch_statements): Update.
22842 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
22843 (find_if_case_1): Update.
22844 (find_if_case_2): Update.
22845 * internal-fn.c (expand_arith_overflow_result_store): Update.
22846 (expand_addsub_overflow): Update.
22847 (expand_neg_overflow): Update.
22848 (expand_mul_overflow): Update.
22849 (expand_vector_ubsan_overflow): Update.
22850 * ipa-cp.c (good_cloning_opportunity_p): Update.
22851 * ipa-split.c (split_function): Use make_single_succ_edge.
22852 * ipa-utils.c (ipa_merge_profiles): Update.
22853 * loop-doloop.c (add_test): Update.
22854 (doloop_modify): Update.
22855 * loop-unroll.c (compare_and_jump_seq): Update.
22856 (unroll_loop_runtime_iterations): Update.
22857 * lra-constraints.c (lra_inheritance): Update.
22858 * lto-streamer-in.c (input_cfg): Update.
22859 * lto-streamer-out.c (output_cfg): Update.
22860 * mcf.c (adjust_cfg_counts): Update.
22861 * modulo-sched.c (sms_schedule): Update.
22862 * omp-expand.c (expand_omp_for_init_counts): Update.
22863 (extract_omp_for_update_vars): Update.
22864 (expand_omp_ordered_sink): Update.
22865 (expand_omp_for_ordered_loops): Update.
22866 (expand_omp_for_generic): Update.
22867 (expand_omp_for_static_nochunk): Update.
22868 (expand_omp_for_static_chunk): Update.
22869 (expand_cilk_for): Update.
22870 (expand_omp_simd): Update.
22871 (expand_omp_taskloop_for_outer): Update.
22872 (expand_omp_taskloop_for_inner): Update.
22873 * omp-simd-clone.c (simd_clone_adjust): Update.
22874 * optabs.c (expand_doubleword_shift): Update.
22875 (expand_abs): Update.
22876 (emit_cmp_and_jump_insn_1): Update.
22877 (expand_compare_and_swap_loop): Update.
22878 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
22879 * predict.c (predictable_edge_p): Update.
22880 (edge_probability_reliable_p): Update.
22881 (set_even_probabilities): Update.
22882 (combine_predictions_for_insn): Update.
22883 (combine_predictions_for_bb): Update.
22884 (propagate_freq): Update.
22885 (estimate_bb_frequencies): Update.
22886 (force_edge_cold): Update.
22887 * profile-count.c (profile_count::dump): Add missing space into dump.
22888 (profile_count::debug): Add newline.
22889 (profile_count::differs_from_p): Explicitly convert to unsigned.
22890 (profile_count::stream_in): Update.
22891 (profile_probability::dump): New member function.
22892 (profile_probability::debug): New member function.
22893 (profile_probability::differs_from_p): New member function.
22894 (profile_probability::differs_lot_from_p): New member function.
22895 (profile_probability::stream_in): New member function.
22896 (profile_probability::stream_out): New member function.
22897 * profile-count.h (profile_count_quality): Rename to ...
22898 (profile_quality): ... this one.
22899 (profile_probability): New.
22900 (profile_count): Update.
22901 * profile.c (compute_branch_probabilities): Update.
22902 * recog.c (peep2_attempt): Update.
22903 * sched-ebb.c (schedule_ebbs): Update.
22904 * sched-rgn.c (find_single_block_region): Update.
22905 (compute_dom_prob_ps): Update.
22906 (schedule_region): Update.
22907 * sel-sched-ir.c (compute_succs_info): Update.
22908 * stmt.c (struct case_node): Update.
22909 (do_jump_if_equal): Update.
22910 (get_outgoing_edge_probs): Update.
22911 (conditional_probability): Update.
22912 (emit_case_dispatch_table): Update.
22913 (expand_case): Update.
22914 (expand_sjlj_dispatch_table): Update.
22915 (emit_case_nodes): Update.
22916 * targhooks.c: Update.
22917 * tracer.c (better_p): Update.
22918 (find_best_successor): Update.
22919 * trans-mem.c (expand_transaction): Update.
22920 * tree-call-cdce.c: Update.
22921 * tree-cfg.c (gimple_split_edge): Upate.
22922 (move_sese_region_to_fn): Upate.
22923 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
22924 * tree-eh.c (lower_resx): Upate.
22925 (cleanup_empty_eh_move_lp): Upate.
22926 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22927 * tree-inline.c (copy_edges_for_bb): Update.
22928 (copy_cfg_body): Update.
22929 * tree-parloops.c (gen_parallel_loop): Update.
22930 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
22931 (gimple_gen_time_profiler): Update.
22932 * tree-ssa-dce.c (remove_dead_stmt): Update.
22933 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22934 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
22935 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
22936 (unloop_loops): Update.
22937 (try_peel_loop): Update.
22938 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22939 * tree-ssa-loop-split.c (connect_loops): Update.
22940 (split_loop): Update.
22941 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22942 (hoist_guard): Update.
22943 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
22944 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
22945 (value_replacement): Update.
22946 * tree-ssa-reassoc.c (branch_fixup): Update.
22947 * tree-ssa-tail-merge.c (replace_block_by): Update.
22948 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
22949 (create_edge_and_update_destination_phis): Update.
22950 (compute_path_counts): Update.
22951 (recompute_probabilities): Update.
22952 (update_joiner_offpath_counts): Update.
22953 (freqs_to_counts_path): Update.
22954 (duplicate_thread_path): Update.
22955 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
22956 (struct switch_conv_info): Update.
22957 (gen_inbound_check): Update.
22958 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22959 (vect_do_peeling): Update.
22960 (vect_loop_versioning): Update.
22961 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22962 (optimize_mask_stores): Update.
22963 * ubsan.c (ubsan_expand_null_ifn): Update.
22964 * value-prof.c (gimple_divmod_fixed_value): Update.
22965 (gimple_divmod_fixed_value_transform): Update.
22966 (gimple_mod_pow2): Update.
22967 (gimple_mod_pow2_value_transform): Update.
22968 (gimple_mod_subtract): Update.
22969 (gimple_mod_subtract_transform): Update.
22970 (gimple_ic): Update.
22971 (gimple_stringop_fixed_value): Update.
22972 (gimple_stringops_transform): Update.
22973 * value-prof.h: Update.
22975 2017-06-29 Carl Love <cel@us.ibm.com>
22977 * config/rs6000/rs6000-c.c: Add support for built-in functions
22978 vector signed int vec_signed (vector float);
22979 vector signed long long vec_signed (vector double);
22980 vector signed int vec_signed2 (vector double, vector double);
22981 vector signed int vec_signede (vector double);
22982 vector signed int vec_signedo (vector double);
22983 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
22984 instruction generator.
22985 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22986 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
22987 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
22989 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
22990 vunsignede_v2df): Add define_expands.
22991 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
22992 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
22993 VEC_UNSIGNEDO): Add definitions.
22994 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22995 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
22996 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
22997 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
22998 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
22999 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23000 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23001 vec_unsignede, vec_unsignedo): Add builtin defines.
23002 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23004 * doc/extend.texi: Update the built-in documentation file for the
23005 new built-in functions.
23007 2017-06-29 Richard Biener <rguenther@suse.de>
23009 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23010 reduction chains to LOOP_VINFO_REDUCTIONS.
23011 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23012 SLP reductions after processing reduction chains.
23014 2017-06-29 Nathan Sidwell <nathan@acm.org>
23016 * builtins.c (fold_builtin_FUNCTION): Use
23017 lang_hooks.decl_printable_name.
23019 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23021 PR middle-end/81194
23022 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23023 with only one label.
23024 * stmt.c (expand_case): Assert NCASES is greater than one.
23026 2017-06-29 Richard Biener <rguenther@suse.de>
23028 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23030 (group_case_labels): Likewise.
23031 (find_taken_edge): Push sanity checking on val to workers...
23032 (find_taken_edge_cond_expr): ... here
23033 (find_taken_edge_switch_expr): ... and here, handle cases
23034 with just a default label.
23035 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23036 (group_case_labels): Likewise.
23037 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23038 group_case_labels does anything cleanup the CFG again.
23040 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23042 PR tree-optimization/81196
23043 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23044 exit condition comparing two IVs.
23046 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23048 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23049 profile to the dummy entry at the end of the list of architectures.
23050 * config/arm/arm-cpu-cdata.h: Regenerated.
23052 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23053 Michael Collison <michael.collison@arm.com>
23056 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23058 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23059 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23060 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23061 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23062 of shift when the shift amount is masked with constant equal to
23063 the size of the mode.
23064 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23067 2017-06-29 Martin Liska <mliska@suse.cz>
23069 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23071 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23074 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23077 (edge_growth_cache_entry::edge_growth_cache_entry): New
23079 (reset_edge_growth_cache): Update to use constructor.
23081 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23083 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23084 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23085 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23087 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23089 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23090 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23092 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23094 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23095 (*-linux-uclibc*): Add t-uclibc tmake_file.
23096 * config/t-musl: New.
23097 * config/t-uclibc: New.
23099 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23101 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23103 (gen_comm_data): Emit architectural setting of arch_prof.
23104 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23106 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23107 (armv8-m.base, armv8-m.main): Likewise.
23108 * arm-protos.h (arm_build_target): Add profile field.
23109 (arch_option): Likewise.
23110 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23112 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23113 arm_active_target.profile.
23115 2017-06-28 Richard Biener <rguenther@suse.de>
23117 PR middle-end/81227
23118 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23119 TYPE_OVERFLOW_WRAPS.
23120 * match.pd (negate_expr_p): Likewise.
23121 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23122 fold_build2, not fold_binary.
23124 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23126 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23127 Convert memory address to Pmode.
23128 (aarch64_print_operand): Assert MEM operands are always Pmode.
23130 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23133 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23134 Remove redundant if.
23135 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23136 * config/arm/aarch-common-protos.h
23137 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23138 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23140 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23143 * multiple_target.c (create_dispatcher_calls): Set the default
23144 clone to be static, not public.
23146 2017-06-28 Richard Biener <rguenther@suse.de>
23148 * tree-vect-loop.c (vectorizable_reduction): Move special
23149 cond reduction IV var creation ...
23150 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23151 parameter. Use STMT_VINFO_VECTYPE.
23152 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23155 2017-06-28 Martin Liska <mliska@suse.cz>
23158 * ipa-visibility.c (non_local_p): Handle visibility.
23160 2017-06-28 Martin Liska <mliska@suse.cz>
23163 * common.opt: Add IntegerRange to various options.
23164 * opt-functions.awk (integer_range_info): New function.
23165 * optc-gen.awk: Add integer_range_info to cl_options struct.
23166 * opts-common.c (decode_cmdline_option): Handle
23167 CL_ERR_INT_RANGE_ARG.
23168 (cmdline_handle_error): Likewise.
23169 * opts.c (print_filtered_help): Show valid interval in
23170 when --help is provided.
23171 * opts.h (struct cl_option): Add range_min and range_max fields.
23172 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23174 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23176 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23177 (x * C EQ/NE y * C): New transformation.
23179 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23181 * genmultilib (combination_space): Accept '+' in option names.
23183 2017-06-28 Martin Liska <mliska@suse.cz>
23186 * asan.c (instrument_derefs): Bail out inner references
23187 that are hard register variables.
23189 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23192 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23193 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23195 2017-06-28 Richard Biener <rguenther@suse.de>
23197 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23198 (vect_get_slp_defs): Adjust.
23199 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23201 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23203 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23204 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23205 (vectorizable_reduction): Adjust.
23206 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23208 (vect_get_slp_defs): Likewise.
23209 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23210 (vectorizable_bswap): Adjust.
23211 (vectorizable_call): Likewise.
23212 (vectorizable_conversion): Likewise.
23213 (vectorizable_assignment): Likewise.
23214 (vectorizable_shift): Likewise.
23215 (vectorizable_operation): Likewise.
23216 (vectorizable_store): Likewise.
23217 (vectorizable_condition): Likewise.
23218 (vectorizable_comparison): Likewise.
23220 2017-06-28 Michael Collison <michael.collison@arm.com>
23223 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23225 (arm_gen_movmemqi): Removed unused variable 'i'.
23226 Convert 'for' loop into 'while' loop.
23227 (arm_expand_prologue): Remove last unnecessary set of insn.
23228 (thumb_pop): Remove unused variable 'pushed_words'.
23229 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23231 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23233 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23234 * config/s390/s390-protos.h: Add prototype of
23235 s390_rel_address_ok_p.
23236 * config/s390/s390.c (s390_got_symbol): New function.
23237 (s390_rel_address_ok_p): New function.
23238 (legitimize_pic_address): Use s390_rel_address_ok_p.
23239 (s390_load_got): Use s390_got_symbol.
23240 (s390_option_override): Issue error if
23241 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23242 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23244 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23246 2017-06-27 Andrew Pinski <apinski@cavium.com>
23248 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23249 (X * copysign (1.0, X)): New pattern.
23250 (X * copysign (1.0, -X)): New pattern.
23251 (copysign (-1.0, CST)): New pattern.
23253 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23255 * genmultilib (combination_space): Remove variable.
23256 Validate reuse rules against regular expression for any sequence
23257 of multilib options in any order.
23259 2017-06-27 Michael Collison <michael.collison@arm.com>
23261 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23262 call aarch64_split_simd_combine.
23263 * (aarch64_combine_internal<mode>): Delete pattern.
23264 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23265 Allow register and subreg operands.
23267 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23269 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23270 specific need, just fallback on defaults.
23271 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23273 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23274 Olivier Hainque <hainque@adacore.com>
23276 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23278 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23279 targets. Pick a default if no particular attempt applied.
23280 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23283 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23285 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23286 (x86_64-wrs-vxworks7): Likewise.
23288 2017-06-27 Marek Polacek <polacek@redhat.com>
23291 * ubsan.c (instrument_null): Check get_base_address's result for null.
23293 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23295 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23297 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23299 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23300 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23301 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23302 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23303 New function types.
23304 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23305 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23306 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23307 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23308 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23309 BUILT_IN_FEUPDATEENV): New builtins.
23310 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23311 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23312 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23313 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23315 (builtin_structptr_types): Adjust size.
23316 * tree.c (builtin_structptr_types): Add four entries.
23318 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23319 Olivier Hainque <hainque@adacore.com>
23321 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23322 (TLS_SYM): New local macro, forcing reference to __tls__ on
23323 link command lines for VxWorks 7 RTPs, triggering initialization
23325 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23326 OS features TLS support, true for RTPs on VxWorks 7.
23327 * config/vxworks.c (vxworks_override_options): Setup emutls
23330 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23332 * predict.c (test_prediction_value_range): Use -1U instead of -1
23333 to avoid narrowing conversion warning.
23334 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23335 to avoid narrowing conversion warning.
23336 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23338 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23340 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23342 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23343 64bit configurations.
23344 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23345 (SIZE_TYPE): Likewise.
23346 * config/vxworks.c (vxworks_emutls_var_fields): Use
23347 long_unsigned_type_node instead of unsigned_type_node as the offset
23348 field type, which is "pointer" mode in emutls.c.
23350 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23353 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23355 PR middle-end/81207
23356 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23357 gimple_vuse copying separately from gimple_vdef copying.
23359 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23361 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23364 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23365 Olivier Hainque <hainque@adacore.com>
23367 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23368 for all vxworks7 targets.
23369 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23370 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23371 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23372 variations for VX6/VX7 and 32/64bits later on in ...
23373 (VXWORKS_LIB_SPEC): Leverage new macros.
23374 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23375 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23377 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23379 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23380 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23382 2017-06-26 Carl Love <cel@us.ibm.com>
23384 * config/rs6000/rs6000-c.c: Add support for built-in functions
23385 vector bool char vec_reve (vector bool char);
23386 vector signed char vec_reve (vector signed char);
23387 vector unsigned char vec_reve (vector unsigned char);
23388 vector bool int vec_reve (vector bool int);
23389 vector signed int vec_reve (vector signed int);
23390 vector unsigned int vec_reve (vector unsigned int);
23391 vector bool long long vec_reve (vector bool long long);
23392 vector signed long long vec_reve (vector signed long long);
23393 vector unsigned long long vec_reve (vector unsigned long long);
23394 vector bool short vec_reve (vector bool short);
23395 vector signed short vec_reve (vector signed short);
23396 vector double vec_reve (vector double);
23397 vector float vec_reve (vector float);
23398 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23399 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23400 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23401 (altivec_vreve): New pattern.
23402 * config/rs6000/altivec.h (vec_reve): New define.
23403 * doc/extend.texi (vec_rev): Update the built-in documentation file
23404 for the new built-in functions.
23406 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23408 PR tree-optimization/71815
23409 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23411 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23413 (slsr_process_phi): Likewise.
23414 (replace_uncond_cands_and_profitable_phis): Don't replace a
23415 multiply candidate with a stride of 1 (copy or cast).
23416 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23418 (lowest_cost_path): Likewise.
23419 (total_savings): Likewise.
23421 2017-06-26 Richard Biener <rguenther@suse.de>
23424 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23425 Use def_builtin_pure for all gather builtins.
23427 2017-06-26 Richard Biener <rguenther@suse.de>
23429 PR tree-optimization/81203
23430 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23431 non-dominating BBs.
23433 2017-06-26 Marek Polacek <polacek@redhat.com>
23436 * doc/invoke.texi: Document -Wmultistatement-macros.
23438 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23440 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23441 arm_neon_ok_no_float_abi effective target.
23443 2017-06-26 Richard Biener <rguenther@suse.de>
23445 PR tree-optimization/80928
23446 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23447 (copy_bbs): Set BB_DUPLICATED flag early.
23448 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23450 (execute_on_shrinking_pred): Likewise.
23451 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23452 BB_DUPLICATED blocks.
23453 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23454 iterate over all PHIs considering removal of *gsi.
23456 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23458 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23461 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23463 * config/rs6000/rs6000-string.c: (expand_block_clear,
23464 do_load_for_compare, select_block_compare_mode,
23465 compute_current_alignment, expand_block_compare,
23466 expand_strncmp_align_check, expand_strn_compare,
23467 expand_block_move, rs6000_output_load_multiple)
23468 Move functions related to string/block move/compare
23469 to a separate file.
23470 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23471 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23472 for this function which is now used in two files.
23473 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23474 * config.gcc: Add rs6000-string.o to extra_objs for
23475 targets powerpc*-*-* and rs6000*-*-*.
23477 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23480 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23481 32-bit, since indexed is not valid for DImode.
23482 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23483 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23484 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23485 (define_peephole2 for Altivec d-form store): Likewise.
23488 * multiple_target.c (create_dispatcher_calls): Only create the
23489 dispatcher call if the function is the default clone of a
23490 versioned function.
23492 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23494 PR middle-end/80902
23495 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23496 a call, force the call to not be a tail call.
23498 2017-06-23 Jeff Law <law@redhat.com>
23500 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23503 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23505 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23506 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23507 handling for early expansion of vector shifts (sl,sr,sra,rl).
23508 (builtin_function_type): Add vector shift right instructions
23509 to the unsigned argument list.
23511 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23513 rtl-optimizatoin/79286
23514 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23515 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23516 trap. PIC register plus a const unspec without offset can never trap.
23518 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23520 * tree.h (builtin_structptr_type): New type.
23521 (builtin_structptr_types): Declare new array.
23522 * tree.c (builtin_structptr_types): New array.
23523 (free_lang_data, build_common_tree_nodes): Use it.
23525 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23528 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23531 2017-06-22 Matt Turner <mattst88@gmail.com>
23533 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23534 Lake models to skylake case. Assume skylake for unknown
23535 models with clflushopt.
23537 2017-06-22 Jeff Law <law@redhat.com>
23539 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23540 frame sizes that do not satisfy aarch64_uimm12_shift.
23542 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23544 * profile-count.h (apply_probability,
23545 apply_scale, probability_in): Fix checks for zero.
23547 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23549 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23550 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23552 2016-06-22 Richard Biener <rguenther@suse.de>
23554 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23555 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23556 REDUC_MAX_EXPR support.
23557 (vectorizable_reduction): Likewise.
23558 (vect_create_epilog_for_reduction): Likewise.
23560 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23562 * match.pd (A / (1 << B) -> A >> B): New.
23563 * generic-match-head.c: Include optabs-tree.h.
23564 * gimple-match-head.c: Likewise.
23565 * optabs-tree.h (target_supports_op_p): New.
23566 * optabs-tree.c (target_supports_op_p): New.
23568 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23570 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23571 $gcc_cv_ld --help output.
23572 (gcc_cv_ld_demangle): Likewise.
23573 (gcc_cv_ld_eh_frame_hdr): Likewise.
23574 (gcc_cv_ld_pie): Likewise.
23575 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23576 (gcc_cv_ld_buildid): Likewise.
23577 (gcc_cv_ld_sysroot): Likewise.
23578 (ld_bndplt_support): Likewise.
23579 (ld_pushpopstate_support): Likewise.
23580 * configure: Regenerate.
23581 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23583 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23586 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23587 operands indexes to avoid gap between operands and match_dups.
23589 2017-06-21 Andrew Pinski <apinski@cavium.com>
23591 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23592 Increment Arith_shift and Arith_shift_reg by 1.
23593 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23595 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23596 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23597 (aarch64_strip_extend): Add new argument and test for it.
23598 (aarch64_cheap_mult_shift_p): New function.
23599 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23600 add a cost if it is true.
23601 Update calls to aarch64_strip_extend.
23602 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23604 2017-06-21 Andrew Pinski <apinski@cavium.com>
23606 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23608 (thunderxt88): Likewise.
23609 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23610 (thunderx_prefetch_tune): New variable.
23611 (thunderx2t99_prefetch_tune): Update for the correct values.
23612 (thunderxt88_tunings): New variable.
23613 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23614 generic_prefetch_tune.
23615 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23617 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23619 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23620 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23621 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23622 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23623 (aarch64_atomic_cas<mode>, GPI): Likewise.
23625 2017-06-21 Martin Liska <mliska@suse.cz>
23627 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23628 statements on cold and hot labels.
23629 * predict.c (tree_estimate_probability_bb): Remove the
23630 prediction from this place.
23632 2017-06-21 Martin Liska <mliska@suse.cz>
23634 PR tree-optimization/79489
23635 * gimplify.c (maybe_add_early_return_predict_stmt): New
23637 (gimplify_return_expr): Call the function.
23638 * predict.c (tree_estimate_probability_bb): Remove handling
23640 * predict.def: Update comment about early return predictor.
23641 * gimple-predict.h (is_gimple_predict): New function.
23642 * predict.def: Change default value of early return to 66.
23643 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23645 * passes.def: Put pass_strip_predict_hints to the beginning of
23648 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23650 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23651 FUNCTION_DECL declarations.
23652 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23654 (dwaf2out_decl): Likewise.
23655 * godump.c (go_early_global_decl): Skip call to the real debug hook
23656 for FUNCTION_DECL declarations.
23657 * passes.c (rest_of_decl_compilation): Skip call to the
23658 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23659 -fdump-go-spec is passed.
23661 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23663 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23664 Reorder for compactness.
23665 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23666 (def_builtin_pure, def_builtin_pure2): New functions.
23667 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23669 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23671 * match.pd (nop_convert): New predicate.
23672 ((A +- CST1) +- CST2): Allow some NOP conversions.
23674 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23677 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23678 with ctors/dtors if GOVD_SHARED is set.
23680 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23682 * config/aarch64/aarch64.md (movti_aarch64):
23683 Emit mov rather than orr.
23684 (movtf_aarch64): Likewise.
23685 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23686 Emit mov rather than orr.
23688 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23690 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23691 Swap alternatives, make integer dup more expensive.
23693 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23695 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23696 Return true for non-tls symbols.
23698 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23700 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23701 (cortex-a75): Likewise.
23702 (cortex-a75.cortex-a55): Likewise.
23703 * config/aarch64/aarch64-tune.md: Regenerate.
23704 * doc/invoke.texi (-mtune): Document new values for -mtune.
23706 2017-06-21 Tom de Vries <tom@codesourcery.com>
23708 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23709 stack_size feature.
23710 (Effective-Target Keywords, Other attributes): Suggest using
23711 dg-add-options stack_size feature to get stack limit in stack_size
23712 effective target documentation.
23714 2017-06-21 Julian Brown <julian@codesourcery.com>
23715 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23717 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23718 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23719 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23721 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23722 attribute type list for neon_multiply.
23723 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23724 attribute type list for neon_multiply.
23725 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23726 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23727 attribute type list for neon_multiply.
23728 * config/arm/types.md (crypto_pmull): Add.
23729 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23730 attribute type list.
23732 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23734 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23737 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23739 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23740 to make sure not to dereference a NULL cost_classes_ptr pointer.
23742 2017-06-20 Carl Love <cel@us.ibm.com>
23744 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23745 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23746 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23747 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23748 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23749 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23750 VMULOSW): New enum "unspec" values.
23751 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23752 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23753 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23754 altivec_vmulosw): New patterns.
23755 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23756 VMULOSW): Add definitions.
23758 2017-06-20 Julia Koval <julia.koval@intel.com>
23760 * config/i386/i386.c: Fix rounding expand for new pattern.
23761 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23763 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23765 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23766 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23768 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23770 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23773 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23775 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23776 architecture, then by implementer ID.
23777 * config/aarch64/aarch64-tune.md: Regenerate.
23779 2017-06-20 Richard Biener <rguenther@suse.de>
23781 PR middle-end/81097
23782 * fold-const.c (split_tree): Fold to type before negating.
23784 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23786 * diagnostic-show-locus.c
23787 (selftest::test_fixit_deletion_affecting_newline): New function.
23788 (selftest::diagnostic_show_locus_c_tests): Call it.
23790 2017-06-20 Andreas Schwab <schwab@suse.de>
23793 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23796 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23798 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23799 __ARM_FEATURE_COPROC according to support.
23801 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23803 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23804 Rewritten to avoid overflow for > 32-bit pointers.
23807 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23808 by removing enum keyword.
23809 (ubsan_type_descriptor): Likewise. Formatting fix.
23812 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23813 splitter): Require TARGET_SSE2 in the condition.
23815 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
23818 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
23819 for doing vector set of SFmode on ISA 3.0.
23820 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
23821 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
23823 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
23824 SFmode value into a V4SF variable that was extracted from another
23825 V4SF variable without converting the element to double precision
23826 and back to single precision vector format.
23827 (vsx_insert_extract_v4sf_p9_2): Likewise.
23829 2017-06-19 Jakub Jelinek <jakub@redhat.com>
23831 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
23832 in UWHI to avoid undefined overflow.
23835 * ubsan.h (enum ubsan_encode_value_phase): New.
23836 (ubsan_encode_value): Change second argument to
23837 enum ubsan_encode_value_phase with default value of
23838 UBSAN_ENCODE_VALUE_GENERIC.
23839 * ubsan.c (ubsan_encode_value): Change second argument to
23840 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
23841 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
23842 create_tmp_var_raw instead of create_tmp_var and use a
23844 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
23845 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
23846 ubsan_encode_value callers.
23849 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
23850 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
23851 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
23853 2017-06-19 Richard Biener <rguenther@suse.de>
23855 PR middle-end/81118
23856 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
23857 estimates if we changed anything.
23859 2017-06-19 Richard Biener <rguenther@suse.de>
23861 PR tree-optimization/80887
23862 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
23863 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
23864 simplified lookups, then reset mprts_hook.
23865 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
23867 (try_to_simplify): Likewise.
23869 2017-06-19 Martin Liska <mliska@suse.cz>
23872 * gimplify.c (gimplify_switch_expr):
23873 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
23875 2017-06-19 Martin Liska <mliska@suse.cz>
23877 * doc/install.texi: Document that PGO runs in 4 stages.
23879 2017-06-19 Martin Liska <mliska@suse.cz>
23882 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
23883 to dispatcher function name.
23884 * multiple_target.c (replace_function_decl): New function.
23885 (create_dispatcher_calls): Redirect both edges and references.
23887 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
23889 * profile-count.c (profile_count::dump): Dump quality.
23890 (profile_count::differs_from_p): Update for unsigned val.
23891 * profile-count.h (profile_count_quality): New enum.
23892 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
23894 2017-06-19 Richard Biener <rguenther@suse.de>
23896 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
23897 struct function as arg.
23898 (estimate_numbers_of_iterations): Export overload with loop arg.
23899 (free_numbers_of_iterations_estimates_loop): Use an overload of
23900 free_numbers_of_iterations_estimates instead.
23901 * tree-cfg.c (remove_bb): Adjust.
23902 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
23903 * tree-parloops.c (gen_parallel_loop): Likewise.
23904 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
23906 (tree_unroll_loops_completely): Likewise.
23907 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
23908 Use an overload instead and export.
23909 (estimated_loop_iterations): Adjust.
23910 (max_loop_iterations): Likewise.
23911 (likely_max_loop_iterations): Likewise.
23912 (estimate_numbers_of_iterations): Take struct function as arg
23914 (loop_exits_before_overflow): Adjust.
23915 (free_numbers_of_iterations_estimates_loop): Use an overload.
23916 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
23917 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
23919 2017-06-19 Richard Biener <rguenther@suse.de>
23922 * ipa-prop.c (find_constructor_constant_at_offset): Handle
23923 RANGE_EXPR conservatively.
23925 2017-06-16 Carl Love <cel@us.ibm.com>
23927 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23928 definitions for vec_float, vec_float2, vec_floato,
23929 vec_floate built-ins.
23930 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
23931 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
23933 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
23934 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
23935 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
23936 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
23937 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
23938 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
23939 vec_floato): Add builtin defines.
23940 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
23941 Update the built-in documentation file for the new built-in
23944 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23946 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
23947 (mthumb): Mark as the negative of -marm.
23949 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23951 * doc/invoke.texi (ARM Options, -mcpu): Document supported
23953 (ARM Options, -mtune): Document that this accepts the same
23954 extension options as -mcpu.
23955 (ARM Options, -mfpu): Document addition of -mfpu=auto.
23957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23959 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
23960 permitted extensions.
23962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23964 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
23965 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
23966 (armv8-m.main): Add option +nodsp.
23967 * config/arm/arm-cpu-cdata.h: Regenerated.
23969 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23971 * config/arm/t-fuchsia: New file.
23972 * config.gcc (arm*-*-fuchsia*): Use it.
23974 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23976 * config/arm/t-symbian: Rewrite for new option infrastructure.
23978 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23980 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
23981 (MULTILIB_REQUIRED): Likewise.
23983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23985 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
23986 (MULTILIB_RESUE): Likewise.
23987 (MULTILIB_MATCHES): Likewise.
23988 (MULTLIB_REQUIRED): Likewise.
23990 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23992 * config/arm/t-rtems: Rewrite for new option framework.
23994 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23996 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
23997 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
23998 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
23999 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24000 * config/arm/t-multilib: ... here.
24001 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24002 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24003 armv7-a and armv8*-a when A-profile libraries have not been built.
24004 * config/arm/t-rmprofile: Rewrite.
24006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24008 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24009 with a backslash. Remove the backslash after substituting unescaped
24011 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24013 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24015 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24016 the driver to pass through the default setting of -mfloat-abi.
24017 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24019 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24020 (all_feat_combs): New rule.
24021 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24023 * config/arm/t-aprofile: Rewrite.
24025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24027 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24028 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24029 fpu is not specified by the user/command-line.
24030 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24031 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24032 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24033 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24034 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24035 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24037 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24039 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24040 * config/arm/t-arm-elf: Rewritten.
24042 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24044 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24045 have some floating-point instructions.
24046 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24047 (TARGET_MAYBE_HARD_FLOAT): New macro.
24048 * config/arm/arm-builtins.c (arm_init_builtins): Use
24049 TARGET_MAYBE_HARD_FLOAT.
24050 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24054 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24055 (configargs.h): Include it.
24056 (arm_print_hint_for_fpu_option): New function.
24057 (arm_parse_fpu_option): New function.
24058 (candidate_extension): New class.
24059 (arm_canon_for_multilib): New function.
24060 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24061 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24062 (ARCH_CANONICAL_SPECS): New macro.
24063 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24065 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24067 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24068 are set after handling multilib fragments. Set target_cpu_default2
24071 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24073 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24075 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24076 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24078 (check_arch): Likewise.
24079 * config/arm/arm.c (arm_configure_build_target): Handle
24080 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24081 options in the default.
24083 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24085 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24086 when an option is an alias of another.
24087 * config/arm/parsecpu.awk (optalias): New parser token.
24088 (gen_comm_data): Mark non-alias options as such. Emit entries
24089 for extension aliases.
24090 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24091 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24092 (armv6kz, armv6zk, armv6t2): Likewise.
24093 (armv7): Make vfpv3-d16 an alias.
24094 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24096 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24097 Sort in canonical order.
24098 (armv8-a): Sort in canonical order.
24099 (armv8.1-a, armv8.2-a): Likewise.
24100 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24102 (cortex-a9): Sort in canonical order.
24103 * config/arm/arm.c (selftests.h): Include it.
24104 (arm_test_cpu_arch_data): New function.
24105 (arm_run_self_tests): New function.
24106 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24107 (targetm): Move declaration to the end of the file.
24108 * arm-cpu-cdata.h: Regenerated.
24110 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24112 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24113 call to target_mode_check describing the type of option passed.
24114 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24115 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24116 arm_parse_cpu_option_name to match parameters against list of
24118 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24119 arm_arch_core_flags data structure.
24120 * config/arm/arm-cpu_cdata.h: Regenerated.
24122 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24124 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24126 (arm_print_hint_for_cpu_option): Likewise.
24127 (arm_print_hint_for_arch_option): Likewise.
24128 (arm_parse_cpu_option_name): Likewise.
24129 (arm_parse_arch_option_name): Likewise.
24130 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24131 of entries in the all_fpus list.
24132 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24133 (arm_parse_cpu_option_name): Declare.
24134 (arm_parse_arch_option_name): Declare.
24135 (arm_parse_option_features): Declare.
24136 (arm_intialize_isa): Declare.
24137 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24139 (gen_comm_data): ... here. Make definitions non-static.
24140 * config/arm/arm-cpu-data.h: Regenerated.
24141 * config/arm/arm-cpu-cdata.h: Regenerated.
24143 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24145 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24146 (cpu_arch_extension): New structure.
24147 (cpu_arch_option, arch_option, cpu_option): New structures.
24148 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24149 architecture types.
24150 (gen_data): Generate new format data tables.
24151 * config/arm/arm.c (cpu_tune): New structure.
24152 (cpu_option, processors): Delete.
24153 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24154 (arm_print_hint_for_cpu_option): ... this and ...
24155 (arm_print_hint_for_arch_option): ... this.
24156 (arm_parse_arch_cpu_name): Delete. Replace with ...
24157 (arm_parse_cpu_option_name): ... this and ...
24158 (arm_parse_arch_option_name): ... this.
24159 (arm_unrecognized_feature): Change type of target parameter to
24161 (arm_parse_arch_cpu_features): Delete. Replace with ...
24162 (arm_parse_option_features): ... this.
24163 (arm_configure_build_target): Rework to use new configuration data
24165 (arm_print_tune_info): Rework for new configuration data tables.
24166 * config/arm/arm-cpu-data.h: Regenerated.
24167 * config/arm/arm-cpu.h: Regenerated.
24169 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24171 * Makefile.in (OBJS): Move sbitmap.o from here ...
24172 (OBJS-libcommon): ... to here.
24174 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24176 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24177 (ISA_ALL_CRYPTO): New macro.
24178 (ISA_ALL_SIMD): New macro
24179 (ISA_ALL_FP): New macro.
24180 * config/arm/arm.c (fpu_bitlist): Update initializer.
24181 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24183 (arm9e): Add fpu. Add option for nofp
24184 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24185 (arm926ej-s, arm1026ej-s): Likewise.
24186 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24187 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24188 neon-fp16, neon-vfpv4, nofp and nosimd.
24189 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24190 (cortex-a8): Add fpu. Add option for nofp.
24191 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24192 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24193 (cortex-r4f): Add fpu.
24194 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24195 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24197 (cortex-r8): Likewise.
24198 (cortex-m4): Add fpu. Add option for nofp.
24199 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24200 (cortex-a17.cortex-a7): Likewise.
24201 (cortex-a32): Add fpu. Add options for crypto and nofp.
24202 (cortex-a35, cortex-a53): Likewise.
24203 (cortex-a57): Add fpu. Add option for crypto.
24204 (cortex-a72, cortex-a73): Likewise.
24205 (exynos-m1): Likewise.
24206 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24207 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24208 (cortex-m33): Add fpu. Add option for nofp.
24209 * config/arm/arm-cpu-cdata.h: Regenerated
24210 * config/arm/arm-cpu-data.h: Regenerated.
24212 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24214 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24215 (armv5te, armv5tej): Likewise.
24216 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24217 (armv7): Add options fp and vfpv3-d16.
24218 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24219 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24221 (armv7ve): Likewise.
24222 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24223 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24224 (armv8-a): Add nocrypto option.
24225 (armv8.1-a, armv8.2-a): Likewise.
24226 (armv8-m.main): add options fp, fp.dp and nofp.
24228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24230 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24232 (armv8-a+crc): Delete.
24233 (armv8.1-a): Add options simd, crypto and nofp.
24234 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24235 (armv8.2-a+fp16): Delete.
24236 (armv8-m.main): Add option dsp.
24237 (armv8-m.main+dsp): Delete.
24238 (cortex-a8): Add fpu. Add nofp option.
24239 (cortex-a9): Add fpu. Add nofp and nosimd options.
24240 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24241 link to main cpu and architecture data structures.
24242 (gen_comm_data): Only put isa attributes from the main architecture
24244 (option): New statement for architecture and CPU entries.
24245 * arm.c (struct cpu_option): New structure.
24246 (struct processors): Add entry for options.
24247 (arm_unrecognized_feature): New function.
24248 (arm_parse_arch_cpu_name): Ignore any characters after the first
24250 (arm_parse_arch_cpu_feature): New function.
24251 (arm_configure_build_target): Separate out any CPU and architecture
24252 features and parse separately. Don't error out if -mfpu=auto is
24253 used with only an architecture string.
24254 (arm_print_asm_arch_directives): New function.
24255 (arm_file_start): Call it.
24256 * config/arm/arm-cpu-cdata.h: Regenerated.
24257 * config/arm/arm-cpu-data.h: Likewise.
24258 * config/arm/arm-tables.opt: Likewise.
24260 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24262 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24263 assembler when it is not -mfpu=auto.
24265 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24267 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24268 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24269 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24270 (ASM_CPU_SPEC): Rewrite.
24271 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24272 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24273 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24274 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24275 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24276 copied string is NUL-terminated. Also strip any characters prefixed
24278 (arm_rewrite_selected_arch): New function.
24279 (arm_rewrite_march): New function.
24281 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24283 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24284 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24285 (march, mcpu, mtune): Convert to string-based options.
24286 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24287 (arm_parse_arch_cpu_name): New function.
24288 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24289 identify selected architecture or CPU.
24290 (arm_option_save): New function.
24291 (TARGET_OPTION_SAVE): Redefine.
24292 (arm_option_restore): Restore string options.
24293 (arm_option_print): Print string options.
24295 2017-06-16 Martin Sebor <msebor@redhat.com>
24297 PR tree-optimization/80933
24298 PR tree-optimization/80934
24299 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24300 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24301 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24302 (gimple_fold_builtin): Call them.
24304 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24306 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24307 as unlikely; update profile.
24309 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24311 * predict.c (force_edge_cold): Handle declaring edges impossible
24314 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24316 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24318 (try_unroll_loop_completely): Fix reporting.
24320 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24322 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24324 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24327 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24328 if given a non-constant argument for an intrinsic which requires a
24331 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24333 * profile.c (compare_freqs): New function.
24334 (branch_prob): Sort edge list.
24335 (find_spanning_tree): Assume that the list is priority sorted.
24337 2017-06-16 Richard Biener <rguenther@suse.de>
24339 PR tree-optimization/81090
24340 * passes.def (pass_record_bounds): Remove.
24341 * tree-pass.h (make_pass_record_bounds): Likewise.
24342 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24343 make_pass_record_bounds): Likewise.
24344 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24345 not free niter estimates at the beginning but at the end.
24346 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24348 2017-06-16 Richard Biener <rguenther@suse.de>
24350 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24351 initializer to workaround ICE in host GCC 4.8.
24353 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24355 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24357 (clone_inlined_nodes): Update.
24359 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24361 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24362 prefetch settings, and enable prefetching by default at -O3.
24364 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24366 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24367 Set flag_prefetch_loop_arrays according to tuning data.
24369 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24371 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24372 New tune structure.
24373 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24374 [Unrelated to main purpose of the patch] Place the pointer field last
24375 to enable type checking errors when tune structure are wrongly merged.
24376 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24377 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24378 (thunderx2t99_prefetch_tune): New tune constants.
24379 (tune_params *_tunings): Update all tunings (no functional change).
24380 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24381 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24382 from tunings structures.
24384 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24387 * ubsan.c (instrument_null): Add T argument, use it instead
24388 of computing it based on IS_LHS.
24389 (instrument_object_size): Likewise.
24390 (pass_ubsan::execute): Adjust instrument_null and
24391 instrument_object_size callers to pass gimple_get_lhs or
24392 gimple_assign_rhs1 result to it. Use instrument_null instead of
24393 calling get_base_address and instrument_mem_ref. Handle
24394 aggregate call arguments for object-size sanitization.
24396 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24398 PR tree-optimization/81089
24399 * tree-vrp.c (is_masked_range_test): Validate operands of
24402 2017-06-15 Martin Sebor <msebor@redhat.com>
24405 * dumpfile.c (dump_register): Avoid calling memset to initialize
24406 a class with a default ctor.
24407 * gcc.c (struct compiler): Remove const qualification.
24408 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24409 * hash-table.h: Ditto.
24410 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24412 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24413 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24415 * params.h (struct param_info): Make struct members non-const.
24416 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24417 with default initialization.
24418 * vec.h (vec_copy_construct, vec_default_construct): New helper
24420 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24421 with vec_copy_construct.
24422 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24423 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24424 * doc/invoke.texi (-Wclass-memaccess): Document.
24426 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24428 * emit-rtl.h (is_leaf): Update comment about local
24429 register allocator.
24431 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24434 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24435 for a variable to have a section before checking if the section has a
24437 Set section to.persistent if persistent attribute is set.
24438 Warn if .persistent attribute is used on an automatic variable.
24440 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24442 PR rtl-optimization/80474
24443 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24445 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24447 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24450 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24452 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24453 (casesi): Emit bounds checking as RTL.
24454 (casesi_internal_mips16_<mode>): Remove bounds checking.
24456 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24458 * config/xtensa/xtensa.c (xtensa_option_override): Append
24459 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24460 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24461 xtensa_doloop_hooks): Define unconditionally.
24462 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24464 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24465 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24466 for it in xtensa_option_override.
24467 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24468 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24470 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24472 * doc/cppopts.texi: Document '-' special value to -MF.
24474 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24476 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24477 (cortex_a53_fconst): Likewise.
24478 (cortex_a53_fpmul): Likewise.
24479 (cortex_a53_f_load_64): Likewise.
24480 (cortex_a53_f_load_many): Likewise.
24481 (cortex_a53_advsimd_alu): Likewise.
24482 (cortex_a53_advsimd_alu_q): Likewise.
24483 (cortex_a53_advsimd_mul): Likewise.
24484 (cortex_a53_advsimd_mul_q): Likewise.
24485 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24486 Add missing fmul, r2f_cvt and fconst cases.
24488 2017-06-14 Richard Biener <rguenther@suse.de>
24490 PR middle-end/81088
24491 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24493 (fold_binary_loc): When associating do not treat pre-existing
24494 TREE_OVERFLOW on literal constants as a reason to allow
24495 TREE_OVERFLOW on associated literal constants.
24497 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24499 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24500 (MASK_FEATURES): New macro.
24501 * config/sparc/sparc.c (sparc_option_override): Remove the special
24502 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24504 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24506 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24507 a division of 0 if non-call exceptions are enabled.
24509 2017-06-14 Andrew Pinski <apinski@cavium.com>
24510 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24513 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24514 Improve vector initialization code gen for only variable case.
24516 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24518 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24520 2017-06-14 Richard Biener <rguenther@suse.de>
24522 PR tree-optimization/81083
24523 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24526 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24528 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24529 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24530 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24531 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24532 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24533 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24535 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24537 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24538 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24540 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24542 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24544 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24546 * config/rs6000/t-rtems: Don't handle SPE.
24548 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24550 * config/rs6000/t-linux: Don't handle SPE.
24552 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24554 * config/rs6000/eabispe.h: Delete file.
24556 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24558 * config/rs6000/t-spe: Delete file.
24560 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24562 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24563 (rs6000_legitimate_offset_address_p): Return false for anything in
24564 V2SImode or V2SFmode.
24566 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24568 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24569 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24570 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24572 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24573 (rs6000_legitimate_offset_address_p): Ditto.
24574 (rs6000_emit_move): Ditto.
24575 (rs6000_init_builtins): Remove V4HI_type_node.
24577 2017-06-13 Martin Liska <mliska@suse.cz>
24580 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24581 (gate_asan): Likewise.
24582 * asan.h (asan_no_sanitize_address_p): Remove the function.
24583 (sanitize_flags_p): New function.
24584 * builtins.def: Fix coding style.
24585 * common.opt: Use renamed enum value.
24586 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24587 * doc/extend.texi: Document no_sanitize attribute.
24588 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24589 to SANITIZE_UNDEFINED_NONDEFAULT.
24590 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24591 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24592 Use sanitize_flags_p.
24593 * gimplify.c (gimplify_function_tree): Likewise.
24594 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24595 * opts.c (parse_no_sanitize_attribute): New function.
24596 (common_handle_option): Use renamed enum value.
24597 * opts.h (parse_no_sanitize_attribute): Declare.
24598 * tree.c (sanitize_flags_p): New function.
24599 * tree.h: Declared here.
24600 * tsan.c: Use sanitize_flags_p.
24601 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24602 (instrument_mem_ref): Likewise.
24603 (instrument_bool_enum_load): Likewise.
24604 (do_ubsan_in_current_function): Remove the function.
24605 (pass_ubsan::execute): Use sanitize_flags_p.
24606 * ubsan.h: Remove do_ubsan_in_current_function
24607 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24608 (dump_function_to_file): Use it here.
24610 2017-06-13 Martin Jambor <mjambor@suse.cz>
24612 PR tree-optimization/80803
24613 PR tree-optimization/81063
24614 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24615 (propagate_subaccesses_across_link): Enqueue subtree whenever
24616 necessary instead of relying on the caller.
24618 2017-06-13 Martin Jambor <mjambor@suse.cz>
24620 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24621 that have a first_link.
24622 (sort_and_splice_var_accesses): Do not check first_link before
24624 (subtree_mark_written_and_enqueue): Likewise.
24625 (propagate_all_subaccesses): Likewise and do not stop at first
24626 parent with a first_link.
24628 2017-06-13 Martin Jambor <mjambor@suse.cz>
24630 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24633 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24635 * match.pd: New pattern.
24637 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24639 * tree-vrp.c (is_masked_range_test): New function.
24640 (register_edge_assert_for): Determine ranges for
24643 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24645 PR tree-optimization/67328
24646 * fold-const.c (maskable_range_p): New function.
24647 (build_range_check): Generate bittests if possible.
24649 2017-06-13 Martin Liska <mliska@suse.cz>
24651 * gimple-pretty-print.c (dump_probability): Add new argument.
24652 (dump_edge_probability): Dump both probability and count.
24653 (dump_gimple_label): Likewise.
24654 (dump_gimple_bb_header): Likewise.
24656 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24659 * config/avr/avr-devices.c: Fix indentation.
24660 * config/avr/gen-avr-mmcu-specs.c: Dito.
24662 2017-06-13 Richard Biener <rguenther@suse.de>
24664 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24665 instead get vector type from stmt_info.
24666 (vectorizable_reduction): Adjust. Remove dead code.
24668 2017-06-13 Richard Biener <rguenther@suse.de>
24670 PR middle-end/81065
24671 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24672 case of C * (x * C2 + C3).
24673 (fold_addr_of_array_ref_difference): Properly fold index difference.
24675 2017-06-12 David S. Miller <davem@davemloft.net>
24678 * config/sparc/sparc.md (return expander): Emit frame blockage if
24679 function uses alloca.
24681 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24683 * combine.c (make_field_assignment): Check len rather than the mode
24684 precision when calling force_to_mode.
24686 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24688 Support multilibs and devices that see flash in RAM address range.
24691 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24692 (avr_mcu_t) <flash_pm_offset>: New field.
24693 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24694 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24695 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24696 (AVR_TINY_PM_OFFSET): Remove macro.
24697 * config/avr/avr.opt (-mshort-calls): New option.
24698 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24699 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24700 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24701 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24702 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24703 instead of avr_arch->have_jmp_call.
24704 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24705 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24706 avr_arch->flash_pm_offset to define.
24707 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24708 new field flash_pm_offset. Add entry for avrxmega3.
24709 (avr_texinfo): Add entry for avrxmega3.
24710 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24711 attiny212, attiny214,
24712 attiny412, attiny414, attiny416, attiny417,
24713 attiny814, attiny816, attiny817,
24714 attiny1614, attiny1616, attiny1617,
24715 attiny3214, attiny3216, attiny3217.
24716 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24717 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24718 (avr_print_operand_address) [AVR_TINY]: Same.
24719 (avr_asm_init_sections) <readonly_data_section>: Only patch
24720 callback if avr_arch->flash_pm_offset = 0.
24721 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24722 for rodata if avr_arch->flash_pm_offset != 0.
24723 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24724 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24725 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24726 (m_options): Append opt_rcall.
24727 (m_dirnames): Append dir_rcall.
24728 * config/avr/t-multilib: Regenerate.
24730 * configure.ac [target=avr]: Check whether avrxmega3 default
24731 linker description file works as needed.
24732 * configure: Regenerate.
24733 * doc/avr-mmcu.texi: Regenerate.
24734 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24735 <__AVR_ARCH__>: Document avrxmega3 and 103.
24736 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24737 <__AVR_SHORT_CALLS__>: Document it.
24738 <__AVR_PM_BASE_ADDRESS__>: Document it.
24739 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24740 (AVR Variable Attributes) <progmem>: Document this is
24741 not needed for avrxmega3.
24742 (AVR Named Address Spaces) <__flash>: Dito.
24744 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24746 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24748 2017-06-12 Doug Rupp <rupp@adacore.com>
24750 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24751 Append vxworks-stdint.h to the tm_file list.
24752 * config/vxworks-stdint.h: New file.
24754 2017-06-12 Martin Liska <mliska@suse.cz>
24756 PR tree-optimization/81041
24757 * tree-profile.c (gimple_gen_ic_func_profiler):
24758 Create an extra BB in profile-generate
24759 (gimple_gen_time_profiler): Likewise.
24761 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24763 PR tree-optimization/81003
24764 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24765 (update_range_test): Use it instead of force_gimple_operand_gsi.
24767 2017-06-12 Richard Biener <rguenther@suse.de>
24769 PR tree-optimization/81053
24770 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24771 with backedge value not defined in loop. Simplify def stmt
24774 2017-06-11 Tom de Vries <tom@codesourcery.com>
24777 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24779 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24780 nvptx_cannot_force_const_mem.
24782 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24784 * opts.c (finish_options): Move test for flag_split_stack after
24785 it has been initialized.
24787 2017-06-11 Jason Merrill <jason@redhat.com>
24789 * tree.h (id_equal): New.
24790 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24791 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24792 instead of strcmp of IDENTIFIER_POINTER.
24794 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24796 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24797 (mark_all_inlined_calls_cdtor): Fix formating.
24798 (inline_transform): Rescale profile before inlining.
24800 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24802 * cgraph.h (cgraph_edge::clone): Update prototype.
24803 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24804 (cgraph_node::create_clone): Update.
24805 (cgraph_node::create_version_clone): Update.
24806 * tree-inline.c (copy_bb): Update.
24807 (expand_call_inline): Update.
24809 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24811 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24812 factored out from ...
24813 (rs6000_emit_prologue): ... here.
24815 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24817 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
24818 factored out from ...
24819 (rs6000_emit_prologue): ... here.
24821 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24823 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
24825 (handle_missing_profiles): Fix computation of tp_first_run.
24826 (counts_to_freqs): Do not touch freqs when count is 0.
24828 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24830 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
24833 2017-06-10 Tom de Vries <tom@codesourcery.com>
24835 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24836 attributes): Document signal effective target.
24838 2017-06-10 Tom de Vries <tom@codesourcery.com>
24840 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24841 Document effective target stack_size.
24843 2017-06-09 David Malcolm <dmalcolm@redhat.com>
24845 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
24846 to the edit_context if they can be auto-applied.
24848 2017-06-9 Ian Lance Taylor <iant@golang.org>
24850 * opts.c (finish_options): If -fsplit-stack, disable implicit
24851 -forder-blocks-and-partition.
24852 * doc/invoke.texi (Optimize Options): Document that when using
24853 -fsplit-stack -forder-blocks-and-partition is not implicitly
24856 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24858 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
24859 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
24860 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
24861 * builtins.def (abort, trap, unreachable): Declare cold.
24862 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
24863 * tree-core.h (ECF_COLD): New.
24864 * tree.c (set_call_expr_flags): Handle ECF_COLD.
24865 (build_common_builtin_nodes): Mark unreachable and abort as cold.
24867 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24869 * predict.c (unlikely_executed_stmt_p): Cleanup.
24871 2017-06-09 Richard Biener <rguenther@suse.de>
24873 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
24874 model if the ref is always written to.
24876 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24878 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
24880 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24882 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
24885 2017-06-09 Tom de Vries <tom@codesourcery.com>
24888 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
24889 "target cannot support label values" when encountering LABEL_REF.
24891 2017-06-09 Martin Liska <mliska@suse.cz>
24893 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
24894 (gimple_gen_ic_func_profiler): Emit direct comparison
24895 of __gcov_indirect_call_callee with NULL.
24896 (gimple_gen_time_profiler): Change probability from
24897 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
24899 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24901 * profile.c (edge_gcov_counts): Turn to pointer.
24902 (compute_branch_probabilities, compute_branch_probabilities): Update.
24903 (branch_prob): Do not clear edge_gcov_count.
24904 * profile.h (edge_gcov_counts): Turn to pointer.
24905 (edge_gcov_count): Update.
24907 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24909 * gimple.h (gimple_check_failed): Mark cold.
24911 2017-06-09 Richard Biener <rguenther@suse.de>
24913 PR tree-optimization/66623
24914 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
24915 refactor check_reduction into two parts, properly computing
24916 whether we have to check reduction validity for outer loop
24919 2017-06-09 Richard Biener <rguenther@suse.de>
24921 PR tree-optimization/79483
24922 * graphite-scop-detection.c (order): New global.
24923 (get_order): Compute bb to order mapping that satisfies code
24924 generation constraints.
24925 (cmp_pbbs): New helper.
24926 (build_scops): Start domwalk at entry block, sort generated
24929 2017-06-09 Richard Biener <rguenther@suse.de>
24931 PR middle-end/81007
24932 * ipa-polymorphic-call.c
24933 (ipa_polymorphic_call_context::restrict_to_inner_class):
24934 Skip FIELD_DECLs with error_mark_node type.
24935 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
24938 2017-06-09 Martin Liska <mliska@suse.cz>
24940 * predict.c (struct branch_predictor): New struct.
24941 (test_prediction_value_range): New test.
24942 (predict_c_tests): New function.
24943 * selftest-run-tests.c (selftest::run_tests): Run the function.
24944 * selftest.h: Declare new tests.
24946 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
24949 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
24950 gen_add3_insn did not fail.
24951 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
24952 r0, construct that number in a temporary reg and add that reg to r0.
24953 If asked to put the result in r0 as well, fail.
24955 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
24957 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24958 for early expansion of vec_eqv.
24960 2017-06-08 Jakub Jelinek <jakub@redhat.com>
24962 PR middle-end/81005
24963 * ubsan.c (instrument_null): Avoid pointless code temporary.
24964 (pass_ubsan::execute): Instrument aggregate arguments of calls.
24966 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
24970 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
24973 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
24976 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24978 * predict.c (unlikely_executed_edge_p): Move ahead.
24979 (probably_never_executed_edge_p): Use it.
24981 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24983 PR middle-end/79988
24984 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
24985 gimple_call_builtin_p call.
24987 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24989 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
24990 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
24991 rtl_check_failed_type2, rtl_check_failed_code1,
24992 rtl_check_failed_code2, rtl_check_failed_code_mode,
24993 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
24994 rtvec_check_failed_bounds, rtl_check_failed_flag,
24995 _fatal_insn_not_found, _fatal_insn): Likewise.
24996 * tree.h (tree_contains_struct_check_failed,
24997 tree_check_failed, tree_not_check_failed,
24998 tree_class_check_failed, tree_range_check_failed,
24999 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25000 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25001 tree_operand_check_failed, omp_clause_check_failed,
25002 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25005 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25007 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25008 flag_branch_probabilities.
25009 * ipa-inline.c (edge_badness): Likewise.
25010 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25011 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25012 * predict.c (maybe_hot_frequency_p): Likewise.
25013 (probably_never_executed): Likewise.
25014 * sched-ebb.c (schedule_ebbs): Likewise.
25015 * sched-rgn.c (find_single_block_region): Likewise.
25016 * tracer.c (tail_duplicate): Likewise.
25018 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25020 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25021 longer requires x_flag_profile_use.
25023 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25025 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25026 instead of flag_reorder_blocks_and_partition.
25027 * dbxout.c (dbxout_function_end): Likewise.
25028 * dwarf2out.c (gen_subprogram_die): Likewise.
25029 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25030 * hw-doloop.c (reorg_loops): Likewise.
25031 * varasm.c (assemble_start_function,
25032 assemble_end_function): Likewise.
25033 (decide_function_section): Do not check for
25034 flag_reorder_blocks_and_partition.
25036 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25038 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25040 (chkp_get_hard_register_fake_addr_expr): Ditto.
25041 (chkp_build_addr_expr): Add check for hard reg case.
25042 (chkp_parse_array_and_component_ref): Ditto.
25043 (chkp_find_bounds_1): Ditto.
25044 (chkp_process_stmt): Don't generate bounds store for
25047 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25049 * predict.c (maybe_hot_bb_p): Do not check profile status.
25050 (maybe_hot_edge_p): Likewise.
25051 (probably_never_executed): Check for zero counts even if profile
25053 (unlikely_executed_edge_p): New function.
25054 (unlikely_executed_stmt_p): New function.
25055 (unlikely_executed_bb_p): New function.
25056 (set_even_probabilities): Use unlikely predicates.
25057 (combine_predictions_for_bb): Likewise.
25058 (predict_paths_for_bb): Likewise.
25059 (predict_paths_leading_to_edge): Likewise.
25060 (determine_unlikely_bbs): New function.
25061 (estimate_bb_frequencies): Use it.
25062 (compute_function_frequency): Use zero counts even if profile is
25064 * profile-count.h: Fix typo.
25066 2017-08-08 Julia Koval <julia.koval@intel.com>
25068 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25069 _mm512_mask_cvtsepi16_storeu_epi8,
25070 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25071 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25072 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25073 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25074 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25075 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25076 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25077 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25078 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25079 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25080 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25081 __builtin_ia32_pmovuswb256mem_mask,
25082 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25083 __builtin_ia32_pmovwb512mem_mask): New builtins.
25085 2017-08-08 Julia Koval <julia.koval@intel.com>
25087 PR target/73350,80862
25088 * config/i386/subst.md (round): Fix round pattern.
25089 * config/i386/i386.c (ix86_erase_embedded_rounding):
25090 Fix erasing rounding for the fixed pattern.
25092 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25094 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25096 2017-06-08 Martin Liska <mliska@suse.cz>
25098 PR gcov-profile/80911
25099 * gcov.c (block_info::block_info): New constructor.
25101 2017-06-07 Carl Love <cel@us.ibm.com>
25103 * config/rs6000/rs6000-c: The return type of the following
25104 built-in functions was implemented as int not long long. Fix sign
25105 of return value for the unsigned version of vec_mulo and vec_mule.
25106 vector unsigned long long vec_bperm (vector unsigned long long,
25107 vector unsigned char)
25108 vector signed long long vec_mule (vector signed int,
25110 vector unsigned long long vec_mule (vector unsigned int,
25111 vector unsigned int)
25112 vector signed long long vec_mulo (vector signed int,
25114 vector unsigned long long vec_mulo (vector unsigned int,
25115 vector unsigned int)
25116 * doc/extend.texi: Fix the documentation for the built-in
25119 2017-06-07 Carl Love <cel@us.ibm.com>
25122 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25125 2017-06-07 Carl Love <cel@us.ibm.com>
25127 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25128 support, Generate doublehv for signed int/float for BE case only.
25130 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25132 * doc/invoke.texi (mcx16): Rewrite.
25134 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25136 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25137 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25138 *mov<mode>_softfloat, and an anonymous splitter): Use
25139 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25141 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25143 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25145 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25146 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25147 (rs6000_debug_reg_global): Adjust.
25148 (rs6000_init_hard_regno_mode_ok): Adjust.
25149 (rs6000_dbx_register_number): Adjust.
25150 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25151 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25152 Remove SPE_ACC and SPEFSCR.
25153 (REG_ALLOC_ORDER): Ditto.
25154 (FRAME_POINTER_REGNUM): Change to 111.
25155 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25156 (REG_CLASS_NAMES): Ditto.
25157 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25158 (REGISTER_NAMES): Ditto.
25159 (ADDITIONAL_REG_NAMES): Ditto.
25160 (rs6000_reg_names): Ditto.
25161 * config/rs6000/rs6000.md: Renumber some register number
25164 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25166 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25168 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25169 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25171 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25172 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25173 Delete the SPE high registers.
25174 (REG_ALLOC_ORDER): Ditto.
25175 (enum reg_class): Remove SPE_HIGH_REGS.
25176 (REG_CLASS_NAMES): Ditto.
25177 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25178 (REGISTER_NAMES): Ditto.
25179 (rs6000_reg_names): Ditto.
25180 * doc/tm.texi.in: Remove SPE as example.
25181 * doc/tm.texi: Regenerate.
25183 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25185 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25186 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25187 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25188 * config/rs6000/rs6000.md (type): Remove "brinc".
25190 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25192 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25193 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25194 * config/rs6000/linuxspe.h: Delete file.
25195 * config/rs6000/rs6000.md: Don't include spe.md.
25196 * config/rs6000/spe.h: Delete file.
25197 * config/rs6000/spe.md: Delete file.
25198 * config/rs6000/t-rs6000: Remove spe.md.
25200 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25202 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25203 (reg_or_none500mem_operand): Delete.
25204 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25205 instead of reg_or_none500mem_operand.
25207 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25209 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25210 handling of SPE flags.
25211 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25213 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25215 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25217 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25218 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25219 paired_divv2sf3): Similar.
25220 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25221 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25222 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25224 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25225 Rename the paired_* instruction patterns.
25226 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25228 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25229 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25230 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25231 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25232 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25233 PAIRED_VECTOR_MODE.
25234 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25235 (spe_func_has_64bit_regs_p): Delete.
25236 (spe_expand_predicate_builtin): Delete.
25237 (spe_expand_evsel_builtin): Delete.
25238 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25239 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25240 (invalid_e500_subreg): Delete.
25241 (rs6000_legitimize_address): Always force_reg op2 as well, for
25242 paired single memory accesses.
25243 (rs6000_member_type_forces_blk): Delete.
25244 (rs6000_spe_function_arg): Delete.
25245 (rs6000_expand_unop_builtin): Delete SPE handling.
25246 (rs6000_expand_binop_builtin): Ditto.
25247 (spe_expand_stv_builtin): Delete.
25248 (bdesc_2arg_spe): Delete.
25249 (spe_expand_builtin): Delete.
25250 (spe_expand_predicate_builtin): Delete.
25251 (spe_expand_evsel_builtin): Delete.
25252 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25253 (spe_init_builtins): Delete.
25254 (spe_func_has_64bit_regs_p): Delete.
25255 (savres_routine_name): Delete "info" parameter. Adjust callers.
25256 (rs6000_emit_stack_reset): Ditto.
25257 (rs6000_dwarf_register_span): Delete.
25258 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25259 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25260 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25262 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25264 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25265 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25266 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25267 mulv2sf3, divv2sf3): Delete expanders.
25269 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25271 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25273 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25275 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25276 * config/rs6000/rs6000.c: Ditto.
25278 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25280 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25281 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25282 comparison_operator.
25284 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25286 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25287 * config/rs6000/rs6000.opt: Ditto.
25288 * config/rs6000/t-rtems: Ditto.
25290 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25292 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25293 TARGET_E500_SINGLE by 0, simplify.
25294 * config/rs6000/rs6000.c: Ditto.
25295 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25296 (spe_build_register_parallel): Delete.
25297 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25298 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25299 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25300 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25301 (E500_CONVERT): Delete.
25302 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25304 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25306 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25307 * config/rs6000/dfp.md: Ditto.
25308 (negdd2, *negdd2_fpr): Merge.
25309 (absdd2, *absdd2_fpr): Merge.
25310 (negtd2, *negtd2_fpr): Merge.
25311 (abstd2, *abstd2_fpr): Merge.
25312 * config/rs6000/e500.h: Delete file.
25313 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25314 TARGET_FPRS by 1 and simplify.
25315 * config/rs6000/rs6000-c.c: Ditto.
25316 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25317 TARGET_DF_SPE by 0.
25318 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25320 * config/rs6000/rs6000.md: Ditto.
25321 (floatdidf2, *floatdidf2_fpr): Merge.
25322 (move_from_CR_gt_bit): Delete.
25323 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25324 (E500_CR_IOR_COMPARE): Delete.
25325 (All patterns that require !TARGET_FPRS): Delete.
25326 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25328 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25330 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25332 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25334 * graphds.c (add_edge): Intitialize edge's attached data.
25335 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25336 pointer parameter. Call pointed function on each edge during
25337 graph traversing. Skip traversing the edge when the function
25339 (graphds_dfs, graphds_scc): Ditto.
25340 (for_each_edge): New parameter. Pass the new parameter to callback
25342 * graphds.h (skip_edge_callback): New function pointer type.
25343 (graphds_dfs, graphds_scc): New function pointer parameter.
25344 (graphds_edge_callback, for_each_edge): New parameter.
25346 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25348 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25349 out code checking if runtime alias check is possible to below ...
25350 Call the new function.
25351 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25352 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25354 2017-06-07 Marek Polacek <polacek@redhat.com>
25357 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25358 TYPE_OVERFLOW_WRAPS check.
25360 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25362 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25363 if versioning is required.
25364 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25365 peeling with the check for versioning.
25367 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25369 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25370 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25371 Set true to new parameter if new ssa variable is defined.
25372 (vect_gen_vector_loop_niters): Refactor. Set range information
25373 for the new vector loop bound variable.
25374 (vect_do_peeling): Ditto.
25376 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25378 * tree-affine.c (ssa.h): Include header file.
25379 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25380 has wrapping overflow behavior.
25382 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25384 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25386 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25388 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25389 (tree_to_aff_combination): ... here.
25391 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25393 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25394 reg_pressure model function.
25395 (ivopts_global_cost_for_size): Delete.
25396 (determine_set_costs, iv_ca_recount_cost): Call new model function
25397 ivopts_estimate_reg_pressure.
25399 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25401 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25402 expensive than udiv. Remove floating point cases from mod.
25404 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25406 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25407 Increase idiv cost.
25409 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25411 * config/aarch64/aarch64.md
25412 (copysignsf3): Fix mask generation.
25414 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25416 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25418 (class dump_manager): Add register_dumps method.
25419 * dumpfile.c: Include langhooks.h.
25420 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25421 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25422 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25423 (dump_manager::dump_register): Start with 512 entries instead of 32.
25424 (dump_manager::register_dumps): New method.
25425 * toplev.c (general_init): Instead of invoking register_dumps
25426 langhook, invoke register_dumps method on the dump manager.
25427 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25430 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25432 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25433 Say that # requires an associated define_split to exist, and that
25434 the define_split must be suitable for use after register allocation.
25436 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25438 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25439 (compute_outgoing_frequencies): Also initialize zero counts.
25440 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25441 around loops; preserve more of profile when nothing changes.
25443 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25445 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25447 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25448 * config/arm/arm-cpu-cdata.h: Regenerate.
25449 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25450 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25451 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25452 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25453 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25455 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25457 * config/arm/t-rmprofile: Likewise.
25458 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25460 2017-06-06 David S. Miller <davem@davemloft.net>
25463 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25464 blockage if function uses alloca.
25466 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25468 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25469 New "uid" fields to hold pretty-print IDs of group and ref.
25470 Memory references are now identified as <group_id>:<ref_id>
25471 instead of using [random] addresses.
25472 (dump_mem_details): Simplify, no functional change.
25473 (dump_mem_ref): Simplify and make output more concise.
25474 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25475 (find_or_create_group): Initialize group uid.
25476 (record_ref): Initialize ref uid. Improve debug output.
25477 (prune_group_by_reuse, should_issue_prefetch_p,)
25478 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25479 (mark_nontemporal_store, determine_loop_nest_reuse):
25480 Improve debug output.
25482 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25484 * dbgcnt.def (prefetch): New debug counter.
25485 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25486 (schedule_prefetches): Stop issueing prefetches if debug counter
25489 2017-06-06 Tom de Vries <tom@codesourcery.com>
25491 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25492 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25494 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25496 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25497 Use aarch64_reg_or_zero predicate for operand 4.
25498 (aarch64_compare_and_swap<mode> define_insn_and_split):
25499 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25500 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25502 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25504 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25505 (arm_compute_save_core_reg_mask): This.
25506 (thumb1_compute_save_reg_mask): Rename into ...
25507 (thumb1_compute_save_core_reg_mask): This.
25508 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25509 (arm_compute_frame_layout): Likewise.
25511 2017-06-06 Richard Biener <rguenther@suse.de>
25513 PR tree-optimization/80974
25514 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25515 keep or clear leaders SSA info.
25517 2017-06-06 Tom de Vries <tom@codesourcery.com>
25519 * config/nvptx/nvptx.c (split_mode_p): New function.
25520 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25522 2017-06-06 Tom de Vries <tom@codesourcery.com>
25524 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25526 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25529 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25532 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25534 * shrink-wrap.c (handle_simple_exit): Update profile.
25535 (try_shrink_wrapping): Upate profile.
25537 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25539 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25540 (tree_guess_outgoing_edge_probabilities): New.
25541 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25542 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25544 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25546 * ipa-split.c (split_function): Initialize return bb profile.
25548 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25550 * profile.c (compute_branch_probabilities): Also initialize
25551 EXIT_BLOCK profile.
25553 2017-06-06 Richard Biener <rguenther@suse.de>
25555 PR tree-optimization/80928
25556 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25557 (vect_analyze_loop_operations): Properly guard analysis for
25559 (vect_transform_loop): Likewise.
25560 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25561 (vect_model_induction_cost): Do not cost for pure SLP.
25562 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25563 of induction in inner loop vectorization.
25564 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25565 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25566 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25568 (vect_analyze_slp_cost_1): Cost induction.
25569 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25570 (vect_get_slp_vect_defs): Likewise.
25571 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25572 (vect_transform_stmt): Handle SLP reductions.
25573 * tree-vectorizer.h (vectorizable_induction): Adjust.
25575 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25577 * config/rs6000/rs6000.c (make_resolver_func): Update
25578 init_lowered_empty_function call.
25580 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25582 * doc/invoke.texi: Document the -fprofile-abs-path option.
25583 * common.opt (fprofile-abs-path): New option.
25584 * gcov-io.h (gcov_write_filename): Declare.
25585 * gcov-io.c (gcov_write_filename): New function.
25586 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25587 * profile.c (output_location): Likewise.
25589 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25591 * shring-wrap.c: Revert accidental commit.
25593 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25595 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25597 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25599 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25601 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25602 profile in callgraph edge.
25603 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25604 (apply_scale): Likewise.
25605 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25606 Also scale profile when inlining function with zero profile.
25607 (initialize_cfun): Update exit block profile even when it is zero.
25608 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25609 when profile is read.
25611 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25613 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25614 (CLONE_*): New constants to define the processors we can generate
25615 code for with the target_clone attribute.
25616 (rs6000_clone_map): New array to identify which clone processors
25617 the current program is running on.
25618 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25619 target_clone attribute.
25620 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25621 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25622 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25623 (cpu_expand_builtin): Add support for target_clone attribute.
25624 (rs6000_valid_attribute_p): Allow "default" attribute.
25625 (get_decl_name): New debug function to simplify printing the
25626 current function name in debugging statements.
25627 (rs6000_clone_priority): New functions to support the target_clone
25628 attribute, and be able to generate code to switch between ISA 2.05
25629 through ISA 3.0 (power6 through power9).
25630 (rs6000_compare_version_priority): Likewise.
25631 (rs6000_get_function_versions_dispatcher): Likewise.
25632 (make_resolver_func): Likewise.
25633 (add_condition_to_bb): Likewise.
25634 (dispatch_function_versions): Likewise.
25635 (rs6000_generate_version_dispatcher_body): Likewise.
25636 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25637 (fusion_gpr_load_p): Fix a spacing issue.
25638 * doc/extend.texi (Common Function Attributes): Document that the
25639 PowerPC supports the target_clone attribute.
25641 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25643 * config/arm/arm.h: explain F symbol found in description of ARM
25644 register allocation in its legend.
25646 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25648 * config/mips/frame-header-opt.c: Include profile-count.h.
25649 * config/riscv/riscv.c: Include profile-count.h
25651 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25653 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25655 (sm_set_flag_if_changed): Add bbs field.
25656 (execute_sm_if_changed_flag_set): Pass BBS.
25657 (execute_sm): Update.
25659 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25661 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25664 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25666 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25667 (peephole2): New peephole2 to emit the above.
25668 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25670 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25672 * config/aarch64/aarch64.c (define_peephole2 above
25673 *sub_<shift>_<mode>): New peephole.
25675 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25677 * config/i386/i386.c (make_resolver_func): Update.
25678 * Makefile.in: Add profile-count.h and profile-count.o
25679 * auto-profile.c (afdo_indirect_call): Update to new API.
25680 (afdo_set_bb_count): Update.
25681 (afdo_propagate_edge): Update.
25682 (afdo_propagate_circuit): Update.
25683 (afdo_calculate_branch_prob): Update.
25684 (afdo_annotate_cfg): Update.
25685 * basic-block.h: Include profile-count.h
25686 (struct edge_def): Turn count to profile_count.
25687 (struct basic_block_def): Likewie.
25688 (REG_BR_PROB_BASE): Move to profile-count.h
25689 (RDIV): Move to profile-count.h
25690 * bb-reorder.c (max_entry_count): Turn to profile_count.
25691 (find_traces): Update.
25692 (rotate_loop):Update.
25693 (connect_traces):Update.
25694 (sanitize_hot_paths):Update.
25695 * bt-load.c (migrate_btr_defs): Update.
25696 * cfg.c (RDIV): Remove.
25697 (init_flow): Use alloc_block.
25698 (alloc_block): Uninitialize count.
25699 (unchecked_make_edge): Uninitialize count.
25700 (check_bb_profile): Update.
25701 (dump_edge_info): Update.
25702 (dump_bb_info): Update.
25703 (update_bb_profile_for_threading): Update.
25704 (scale_bbs_frequencies_int): Update.
25705 (scale_bbs_frequencies_gcov_type): Update.
25706 (scale_bbs_frequencies_profile_count): New.
25707 * cfg.h (update_bb_profile_for_threading): Update.
25708 (scale_bbs_frequencies_profile_count): Declare.
25709 * cfgbuild.c (compute_outgoing_frequencies): Update.
25710 (find_many_sub_basic_blocks): Update.
25711 * cfgcleanup.c (try_forward_edges): Update.
25712 (try_crossjump_to_edge): Update.
25713 * cfgexpand.c (expand_gimple_tailcall): Update.
25714 (construct_exit_block): Update.
25715 * cfghooks.c (verify_flow_info): Update.
25716 (dump_bb_for_graph): Update.
25717 (split_edge): Update.
25718 (make_forwarder_block): Update.
25719 (duplicate_block): Update.
25720 (account_profile_record): Update.
25721 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25722 (get_estimated_loop_iterations): Update.
25723 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25724 (single_likely_exit): Update.
25725 * cfgloopmanip.c (scale_loop_profile): Update.
25727 (set_zero_probability): Update.
25728 (lv_adjust_loop_entry_edge): Update.
25729 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25730 (purge_dead_edges): Update.
25731 (rtl_account_profile_record): Update.
25732 * cgraph.c (cgraph_node::create): Uninitialize count.
25733 (symbol_table::create_edge): Uninitialize count.
25734 (cgraph_update_edges_for_call_stmt_node): Update.
25735 (cgraph_edge::dump_edge_flags): Update.
25736 (cgraph_node::dump): Update.
25737 (cgraph_edge::maybe_hot_p): Update.
25738 * cgraph.h: Include profile-count.h
25739 (create_clone), create_edge, create_indirect_edge): Update.
25740 (cgraph_node): Turn count to profile_count.
25741 (cgraph_edge0: Likewise.
25742 (make_speculative, clone): Update.
25743 (create_edge): Update.
25744 (init_lowered_empty_function): Update.
25745 * cgraphclones.c (cgraph_edge::clone): Update.
25746 (duplicate_thunk_for_node): Update.
25747 (cgraph_node::create_clone): Update.
25748 * cgraphunit.c (cgraph_node::analyze): Update.
25749 (cgraph_node::expand_thunk): Update.
25750 * final.c (dump_basic_block_info): Update.
25751 * gimple-streamer-in.c (input_bb): Update.
25752 * gimple-streamer-out.c (output_bb): Update.
25753 * graphite.c (print_global_statistics): Update.
25754 (print_graphite_scop_statistics): Update.
25755 * hsa-brig.c: Include basic-block.h.
25756 * hsa-dump.c: Include basic-block.h.
25757 * hsa-gen.c (T sum_slice): Update.
25758 (convert_switch_statements):Update.
25759 * hsa-regalloc.c: Include basic-block.h.
25760 * ipa-chkp.c (chkp_produce_thunks): Update.
25761 * ipa-cp.c (struct caller_statistics): Update.
25762 (init_caller_stats): Update.
25763 (gather_caller_stats): Update.
25764 (ipcp_cloning_candidate_p): Update.
25765 (good_cloning_opportunity_p): Update.
25766 (get_info_about_necessary_edges): Update.
25767 (dump_profile_updates): Update.
25768 (update_profiling_info): Update.
25769 (update_specialized_profile): Update.
25770 (perhaps_add_new_callers): Update.
25771 (decide_about_value): Update.
25772 (ipa_cp_c_finalize): Update.
25773 * ipa-devirt.c (struct odr_type_warn_count): Update.
25774 (struct decl_warn_count): Update.
25775 (struct final_warning_record): Update.
25776 (possible_polymorphic_call_targets): Update.
25777 (ipa_devirt): Update.
25778 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25779 * ipa-icf.c (sem_function::merge): Update.
25780 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25781 * ipa-inline.c (compute_uninlined_call_time): Update.
25782 (compute_inlined_call_time): Update.
25783 (want_inline_small_function_p): Update.
25784 (want_inline_self_recursive_call_p): Update.
25785 (edge_badness): Update.
25786 (lookup_recursive_calls): Update.
25787 (recursive_inlining): Update.
25788 (inline_small_functions): Update.
25789 (dump_overall_stats): Update.
25790 (dump_inline_stats): Update.
25791 * ipa-profile.c (ipa_profile_generate_summary): Update.
25792 (ipa_propagate_frequency): Update.
25793 (ipa_profile): Update.
25794 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25795 * ipa-utils.c (ipa_merge_profiles): Update.
25796 * loop-doloop.c (doloop_modify): Update.
25797 * loop-unroll.c (report_unroll): Update.
25798 (unroll_loop_runtime_iterations): Update.
25799 * lto-cgraph.c (lto_output_edge): Update.
25800 (lto_output_node): Update.
25801 (input_node): Update.
25802 (input_edge): Update.
25803 (merge_profile_summaries): Update.
25804 * lto-streamer-in.c (input_cfg): Update.
25805 * lto-streamer-out.c (output_cfg): Update.
25806 * mcf.c (create_fixup_graph): Update.
25807 (adjust_cfg_counts): Update.
25808 (sum_edge_counts): Update.
25809 * modulo-sched.c (sms_schedule): Update.
25810 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25811 * predict.c (maybe_hot_count_p): Update.
25812 (probably_never_executed): Update.
25813 (dump_prediction): Update.
25814 (combine_predictions_for_bb): Update.
25815 (propagate_freq): Update.
25816 (handle_missing_profiles): Update.
25817 (counts_to_freqs): Update.
25818 (rebuild_frequencies): Update.
25819 (force_edge_cold): Update.
25820 * predict.h: Include profile-count.h
25821 (maybe_hot_count_p, counts_to_freqs): UPdate.
25822 * print-rtl-function.c: Do not include cfg.h
25823 * print-rtl.c: Include basic-block.h
25824 * profile-count.c: New file.
25825 * profile-count.h: New file.
25826 * profile.c (is_edge_inconsistent): Update.
25827 (correct_negative_edge_counts): Update.
25828 (is_inconsistent): Update.
25829 (set_bb_counts): Update.
25830 (read_profile_edge_counts): Update.
25831 (compute_frequency_overlap): Update.
25832 (compute_branch_probabilities): Update; Initialize and deinitialize
25834 (branch_prob): Update.
25835 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
25836 (edge_gcov_count): New.
25837 (bb_gcov_count): New.
25838 * shrink-wrap.c (try_shrink_wrapping): Update.
25839 * tracer.c (better_p): Update.
25840 * trans-mem.c (expand_transaction): Update.
25841 (ipa_tm_insert_irr_call): Update.
25842 (ipa_tm_insert_gettmclone_call): Update.
25843 * tree-call-cdce.c: Update.
25844 * tree-cfg.c (gimple_duplicate_sese_region): Update.
25845 (gimple_duplicate_sese_tail): Update.
25846 (gimple_account_profile_record): Update.
25847 (execute_fixup_cfg): Update.
25848 * tree-inline.c (copy_bb): Update.
25849 (copy_edges_for_bb): Update.
25850 (initialize_cfun): Update.
25851 (freqs_to_counts): Update.
25852 (copy_cfg_body): Update.
25853 (expand_call_inline): Update.
25854 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25855 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
25856 (try_unroll_loop_completely): Update.
25857 (try_peel_loop): Update.
25858 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25859 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
25860 * tree-ssa-loop-split.c (connect_loops): Update.
25861 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
25862 * tree-ssa-reassoc.c (branch_fixup): Update.
25863 * tree-ssa-tail-merge.c (replace_block_by): Update.
25864 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
25865 (compute_path_counts): Update.
25866 (update_profile): Update.
25867 (recompute_probabilities): Update.
25868 (update_joiner_offpath_counts): Update.
25869 (estimated_freqs_path): Update.
25870 (freqs_to_counts_path): Update.
25871 (clear_counts_path): Update.
25872 (ssa_fix_duplicate_block_edges): Update.
25873 (duplicate_thread_path): Update.
25874 * tree-switch-conversion.c (case_bit_test_cmp): Update.
25875 (struct switch_conv_info): Update.
25876 * tree-tailcall.c (decrease_profile): Update.
25877 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25878 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25879 * value-prof.c (check_counter): Update.
25880 (gimple_divmod_fixed_value): Update.
25881 (gimple_mod_pow2): Update.
25882 (gimple_mod_subtract): Update.
25883 (gimple_ic_transform): Update.
25884 (gimple_stringop_fixed_value): Update.
25885 * value-prof.h (gimple_ic): Update.
25887 2017-06-02 Carl Love <cel@us.ibm.com>
25889 * config/rs6000/rs6000-c: Add support for built-in functions
25890 vector double vec_doublee (vector signed int);
25891 vector double vec_doublee (vector unsigned int);
25892 vector double vec_doublee (vector float);
25893 vector double vec_doubleh (vector signed int);
25894 vector double vec_doubleh (vector unsigned int);
25895 vector double vec_doubleh (vector float);
25896 vector double vec_doublel (vector signed int);
25897 vector double vec_doublel (vector unsigned int);
25898 vector double vec_doublel (vector float);
25899 vector double vec_doubleo (vector signed int);
25900 vector double vec_doubleo (vector unsigned int);
25901 vector double vec_doubleo (vector float);.
25902 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
25903 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
25905 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
25906 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
25907 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
25909 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
25910 vec_doublel, vec_doubleh.
25911 * doc/extend.texi: Update the built-in documentation file for the
25912 new built-in functions.
25914 2017-06-02 David Malcolm <dmalcolm@redhat.com>
25917 * ipa-inline-analysis.c (free_growth_caches): Set
25918 edge_removal_hook_holder to NULL after removing it.
25920 2017-06-02 Sudakshina Das <sudi.das@arm.com>
25922 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
25923 comparision with zero.
25925 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25926 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25927 for early expansion of vec_min and vec_max builtins.
25928 (builtin_function_type): Add min/max unsigned variants to those
25929 identified as having unsigned arguments.
25931 2017-06-02 Olivier Hainque <hainque@adacore.com>
25933 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
25935 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25937 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
25938 Use VALL_F16 iterator rather than VALL.
25940 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25942 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
25943 Emit CBNZ inside loop when doing a strong exchange and comparing
25944 against zero. Generate the CC flags after the loop.
25946 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
25948 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
25949 (dl_section_ref): New.
25950 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
25951 On AIX, append an expression to subtract the size of the
25952 section length to dl_section_ref.
25954 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25956 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25957 for early expansion of vector absolute builtins.
25959 2017-06-02 Richard Biener <rguenther@suse.de>
25961 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
25962 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
25964 2017-06-02 Richard Biener <rguenther@suse.de>
25966 PR tree-optimization/80948
25967 * tree-tailcall.c (find_tail_calls): Track stmts to move in
25968 stmt order as well.
25970 2017-06-02 Richard Biener <rguenther@suse.de>
25972 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
25974 * tree-vect-stmts.c (process_use): Do not mark backedge defs
25975 for inductions as relevant.
25977 2017-06-02 Richard Biener <rguenther@suse.de>
25979 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
25980 (vectorizable_induction): ... this. Remove dead code.
25982 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
25984 * builtins. (expand_builtin_alloca): Remove second parameter and
25985 infer its value from the first parameter instead.
25986 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
25988 2017-06-02 Jakub Jelinek <jakub@redhat.com>
25990 PR rtl-optimization/80903
25991 * loop-doloop.c (add_test): Unshare sequence.
25993 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
25995 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
25997 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
25999 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26001 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26002 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26003 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26004 (xlogue_layout::get_stub_rtx): Make static.
26005 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26006 (xlogue_layout::compute_stub_managed_regs): Rename to...
26007 (xlogue_layout::count_stub_managed_regs): ...this.
26008 (xlogue_layout::is_stub_managed_reg): New function.
26009 (xlogue_layout::m_stub_names): Rename to...
26010 (xlogue_layout::s_stub_names): ...this, make static.
26011 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26012 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26013 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26014 xlogue_layout::s_stub_names): Instantiate statics.
26015 (stub_managed_regs): Remove.
26016 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26017 (disable_call_ms2sysv_xlogues): Rename to...
26018 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26019 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26021 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26022 change after reload_completed.
26023 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26025 (ix86_expand_prologue): Likewise.
26026 (ix86_expand_epilogue): Likewise.
26027 (ix86_expand_split_stack_prologue): Likewise.
26028 (ix86_compute_frame_layout): Remove frame parameter ...
26029 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26030 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26032 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26033 (ix86_frame): Move from here ...
26034 * config/i386/i386.h (ix86_frame): ... to here.
26035 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26036 complete ix86_frame data structure instead. Remove some_ld_name.
26038 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26040 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26041 symbols that hold a DECL_VALUE_EXPR.
26043 2017-06-01 Martin Jambor <mjambor@suse.cz>
26045 PR tree-optimization/80898
26046 * tree-sra.c (process_subtree_disqualification): Removed.
26047 (disqualify_candidate): Do not acll
26048 process_subtree_disqualification.
26049 (subtree_mark_written_and_enqueue): New function.
26050 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26051 RHS has been disqualified and re-queue LHS if necessary. Apart
26052 from that, ignore disqualified RHS.
26054 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26056 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26057 address fetch for z10 or later.
26059 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26061 * config/arc/arc.md (tst_movb): Add guard when splitting.
26063 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26065 * config/arc/arc.c (arc_can_eliminate): Test against
26066 arc_frame_pointer_needed.
26068 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26070 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26071 to prevent store reordering.
26072 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26073 (type): Add block type.
26074 (stack_tie): Define special instruction to be used in
26077 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26079 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26080 constraint. It is not valid for the pattern.
26081 (noncommutative_binary_comparison): Likewise.
26083 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26085 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26088 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26090 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26091 be used by the reg-alloc.
26093 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26095 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26096 reg-alloc when having mul64 or mul32x16 instructions.
26097 (mulsidi3): Likewise.
26098 (umulsidi3): Likewise.
26099 (mulsi32x16): New pattern.
26100 (mulsi64): Likewise.
26101 (mulsidi64): Likewise.
26102 (umulsidi64): Likewise.
26103 (MUL32x16_REG): Define.
26104 (mul64_600): Use MUL32x16_REG.
26105 (mac64_600): Likewise.
26106 (umul64_600): Likewise.
26107 (umac64_600): Likewise.
26109 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26111 * config/arc/arc.md (mulsi3_700): Make it commutative.
26113 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26115 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26117 (*sign_extendsidi2_insn): Likewise for movstosw.
26119 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26121 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26122 the type of the input discriminant value. Convert the
26123 discriminant value of signedness vary.
26125 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26127 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26128 Add to -Wall section.
26130 2017-06-01 Richard Biener <rguenther@suse.de>
26132 PR middle-end/66313
26133 * fold-const.c (fold_plusminus_mult_expr): If the factored
26134 factor may be zero use a wrapping type for the inner operation.
26135 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26136 and handle moved defs.
26137 (process_assignment): Properly guard the unary op case. Return a
26138 tri-state indicating that moving the stmt before the call may allow
26139 to continue. Pass through to_move.
26140 (find_tail_calls): Handle moving unrelated defs before
26143 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26146 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26147 splitter result in the canonical way.
26149 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26151 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26152 also for 32bit target. Update insn attributes.
26153 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26155 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26157 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26158 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26159 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26160 (_mm_maskz_min_ss): New intrinsics.
26162 2017-05-31 Martin Liska <mliska@suse.cz>
26164 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26165 Change comment style to one we normally use.
26166 (vectorizable_reduction): Likewise.
26167 (vectorizable_induction): Likewise.
26168 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26169 (vectorizable_call): Likewise.
26170 (vectorizable_simd_clone_call): Likewise.
26171 (vectorizable_conversion): Likewise.
26172 (vectorizable_assignment): Likewise.
26173 (vectorizable_shift): Likewise.
26174 (vectorizable_operation): Likewise.
26175 (vectorizable_store): Likewise.
26176 (vectorizable_load): Likewise.
26177 * tree-vectorizer.h: Likewise.
26179 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26181 * passes.c (emergency_dump_function): New.
26182 * tree-pass.h (emergency_dump_function): Declare.
26183 * plugin.c (plugins_internal_error_function): Remove.
26184 * plugin.h (plugins_internal_error_function): Remove declaration.
26185 * toplev.c (internal_error_function): New static function. Use it...
26186 (general_init): ...here.
26188 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26190 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26191 (arc_rtx_costs): Add costs for new patterns.
26192 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26193 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26195 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26197 * tree-ssa-strlen.c (get_next_strinfo): New function.
26198 (get_stridx_plus_constant): Use it.
26199 (zero_length_string): Likewise.
26200 (adjust_related_strinfos): Likewise.
26201 (adjust_last_stmt): Likewise.
26203 2017-05-31 Richard Biener <rguenther@suse.de>
26206 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26207 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26209 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26211 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26212 loop_vinfo argument and use of dependence distance vectors.
26213 Check instead whether the two references differ only in their
26214 initial value and assume that they have the same alignment if the
26215 difference is a multiple of the vector alignment.
26216 (vect_analyze_data_refs_alignment): Update call accordingly.
26218 2017-05-31 Martin Liska <mliska@suse.cz>
26221 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26223 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26225 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26226 (create_intersect_range_checks): Move from ...
26227 * tree-data-ref.c (create_intersect_range_checks_index)
26228 (create_intersect_range_checks): ... to here.
26229 (create_runtime_alias_checks): New function factored from ...
26230 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26231 here. Call above function.
26232 * tree-data-ref.h (create_runtime_alias_checks): New function.
26234 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26236 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26237 segment length for dr_b and compute it in wide_int.
26239 2017-05-31 Richard Biener <rguenther@suse.de>
26241 PR tree-optimization/80906
26242 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26243 and pass through iv_map.
26244 (copy_bb_and_scalar_dependences): Adjust.
26245 (translate_pending_phi_nodes): Likewise.
26246 (copy_loop_close_phi_args): Handle code-generating IVs instead
26249 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26251 * diagnostic-color.c (color_dict): Add "type-diff".
26252 (parse_gcc_colors): Update comment.
26253 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26254 -fdiagnostics-show-template-tree and -fno-elide-type.
26255 (GCC_COLORS): Add type-diff to example.
26257 (-fdiagnostics-show-template-tree): New.
26258 (-fno-elide-type): New.
26259 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26260 the pp_format_decoder callback. Call any m_format_postprocessor's
26262 (pretty_printer::pretty_printer): Initialize
26263 m_format_postprocessor.
26264 (pretty_printer::~pretty_printer): Delete any
26265 m_format_postprocessor.
26266 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26267 (class format_postprocessor): New class.
26268 (struct pretty_printer::format_decoder): Document the new parameters.
26269 (struct pretty_printer::m_format_postprocessor): New field.
26270 * tree-diagnostic.c (default_tree_printer): Update for new
26271 bool and const char ** params.
26272 * tree-diagnostic.h (default_tree_printer): Likewise.
26274 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26276 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26277 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26278 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26279 rs6000_gen_cell_microcode code.
26280 (rs6000_final_prescan_insn): Delete.
26281 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26282 "warn-cell-microcode" entries.
26283 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26284 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26285 throughout. Change cc_reg_not_micro_cr0_operand to
26286 cc_reg_not_cr0_operand throughout.
26287 (*extendhi<mode>2_noload): Delete.
26288 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26289 (mwarn-cell-microcode): Delete.
26290 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26291 -mgen-cell-microcode and -mwarn-cell-microcode.
26293 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26296 * config/i386/constraints.md (Yd): New constraint.
26298 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26299 and (?Yd, r) alternatives. Update insn attributes.
26300 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26301 and (?*Yd, r) alternatives. Update insn attributes.
26302 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26304 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26306 * gimplify.c (gimplify_modify_expr): Don't create a
26307 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26310 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26312 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26314 2017-05-30 Richard Biener <rguenther@suse.de>
26316 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26317 and reduc_def fields.
26318 (STMT_VINFO_REDUC_TYPE): New define.
26319 (STMT_VINFO_REDUC_DEF): Likewise.
26320 (vect_force_simple_reduction): Adjust prototype.
26321 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26322 (vect_is_simple_reduction): Remove check_reduction argument.
26323 (vect_force_simple_reduction): Adjust and set
26324 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26325 (vectorizable_reduction): Do not re-do reduction analysis
26326 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26327 * tree-parloops.c (gather_scalar_reductions): Adjust.
26329 2017-05-30 Richard Biener <rguenther@suse.de>
26331 PR middle-end/80901
26332 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26335 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26337 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26338 Introduce unknown_misalignment parameter and remove vf.
26339 (vect_peeling_hash_get_lowest_cost):
26340 Pass unknown_misalignment parameter.
26341 (vect_enhance_data_refs_alignment):
26342 Fix unsupportable data ref treatment.
26344 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26346 * tree-vect-data-refs.c (vect_get_data_access_cost):
26347 Workaround for SLP handling.
26348 (vect_enhance_data_refs_alignment):
26349 Compute costs for doing no peeling at all, compare to the best
26350 peeling costs so far and avoid peeling if cheaper.
26352 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26354 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26355 Return peeling info and set costs to zero for unlimited cost
26357 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26358 with unknown misalignment. Compute and costs for unknown
26359 misalignment, compare them to the costs for known misalignment
26360 and choose the cheapest for peeling.
26362 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26364 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26365 (vect_get_peeling_costs_all_drs): Create function.
26366 (vect_peeling_hash_get_lowest_cost):
26367 Use vect_get_peeling_costs_all_drs.
26368 (vect_peeling_supportable): Create function.
26369 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26371 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26373 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26374 DR_HAS_NEGATIVE_STEP.
26375 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26376 (vect_enhance_data_refs_alignment): Use.
26377 (vect_duplicate_ssa_name_ptr_info): Use.
26378 * tree-vectorizer.h (dr_misalignment): Use.
26379 (known_alignment_for_access_p): Use.
26381 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26384 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26386 (has_section_name): New function.
26388 2017-05-30 Martin Liska <mliska@suse.cz>
26391 * auto-profile.c (get_function_decl_from_block): Fix
26394 2017-05-30 Richard Biener <rguenther@suse.de>
26396 PR middle-end/80876
26397 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26399 2017-05-30 Martin Liska <mliska@suse.cz>
26401 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26402 * dumpfile.h (struct dump_file_info): Remove ctors.
26404 2017-05-30 Martin Liska <mliska@suse.cz>
26406 * predict.def: Fix GNU coding style.
26408 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26410 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26411 Mark 'to' argument with ATTRIBUTE_UNUSED.
26413 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26415 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26416 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26417 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26420 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26422 * doc/install.texi (Options specification): Restore entry of
26423 --enable-sjlj-exceptions.
26425 2017-05-27 Michael Eager <eager@eagercon.com>
26428 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26430 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26432 * config/microblaze/microblaze.h
26433 (FIXED_REGISTERS): Update in macro.
26434 (CALL_USED_REGISTERS): Update in macro.
26436 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26438 * doc/install.texi: Add links to macOS binary distributions.
26440 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26444 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26446 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26448 2017-05-26 Martin Liska <mliska@suse.cz>
26450 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26452 2017-05-26 Martin Liska <mliska@suse.cz>
26454 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26455 always leading ';; '.
26456 (dump_bb_info): Likewise.
26457 (brief_dump_cfg): Likewise.
26458 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26459 * dumpfile.c: Remove usage of TDF_VERBOSE.
26460 * dumpfile.h (enum dump_kind): Likewise.
26461 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26462 * print-tree.c (debug_verbose): Remove.
26463 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26464 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26465 * tree-diagnostic.c (default_tree_printer): Replace
26466 TDF_DIAGNOSTIC with TDF_SLIM.
26468 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26470 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26471 in parameter loop, rather than loop_vinfo.
26472 (create_intersect_range_checks): Ditto.
26473 (vect_create_cond_for_alias_checks): Update call to above functions.
26475 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26477 PR tree-optimization/80815
26478 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26479 for merging runtime alias checks. Handle negative DR_STEPs.
26481 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26483 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26485 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26486 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26487 out code pruning runtime alias checks.
26488 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26489 factored out from above.
26490 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26492 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26494 (prune_runtime_alias_test_list): New decalaration.
26496 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26498 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26499 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26500 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26501 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26502 (operator==, comp_dr_with_seg_len_pair): Ditto.
26503 (vect_prune_runtime_alias_test_list): Ditto.
26505 2017-05-26 Martin Liska <mliska@suse.cz>
26508 * params.def: Bound partial-inlining-entry-probability param.
26510 2017-05-26 Marek Polacek <polacek@redhat.com>
26513 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26516 2017-05-26 Richard Biener <rguenther@suse.de>
26518 PR tree-optimization/80842
26519 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26522 2017-05-26 Richard Biener <rguenther@suse.de>
26524 PR tree-optimization/80844
26525 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26527 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26529 * doc/md.texi (Machine Constraints): Update x86 family
26530 machine constraints section to match 'config/i386/constraints.md'.
26532 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26534 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26536 2017-05-25 Nathan Sidwell <nathan@acm.org>
26538 * doc/invoke.texi (--enable-languages): Update documentation.
26540 2017-05-25 Martin Liska <mliska@suse.cz>
26542 * dumpfile.c: Add TDF_FOLDING.
26543 * dumpfile.h (enum dump_kind): Likewise.
26544 * genmatch.c (dt_simplify::gen_1): Use it.
26546 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26548 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26550 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26552 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26553 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26555 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26557 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26558 * match.pd (X == C): Rewrite it here.
26559 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26560 with_certain_nonzero_bits2): New predicates.
26561 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26563 2017-05-24 Nathan Sidwell <nathan@acm.org>
26565 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26568 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26571 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26573 * config/powerpcspe: New port. Files are copied from the rs6000
26574 port, with "rs6000" in filenames replaced by "powerpcspe".
26576 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26578 PR rtl-optimization/80754
26579 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26581 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26583 * config/sparc/sparc.md (length): Return the correct value for -mflat
26584 sibcalls to match output_sibcall.
26586 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26590 * config/rs6000/rs6000.c (struct machine_function): Add new field
26592 (rs6000_get_separate_components): Init that field, use it.
26593 (rs6000_components_for_bb): Use the field.
26595 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26597 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26599 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26601 PR middle-end/80823
26602 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26604 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26607 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26608 address against address_operand predicate.
26609 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26611 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26613 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26614 parameters passed indirectly.
26616 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26618 * config/i386/i386.md (*movdi_internal): Remove SSE4
26619 alternative 18 (?r, *v). Update insn attributes.
26620 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26621 Update insn attributes.
26622 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26623 Update insn attributes.
26624 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26625 alternative 1 (r, v). Remove isa attribute.
26626 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26627 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26628 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26630 2017-05-23 Tom de Vries <tom@codesourcery.com>
26632 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26635 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26637 * cgraphunit.c (symbol_table::process_new_functions): Update.
26638 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26639 (inline_generate_summary): Rename to ...
26640 (ipa_fn_summary_generate): ... this one.
26641 (inline_read_summary): Rename to ...
26642 (ipa_fn_summary_read): ... this one.
26643 (inline_write_summary): Rename to ...
26644 (ipa_fn_summary_write): ... this one.
26645 (inline_free_summary): Rename to ...
26646 (ipa_free_fn_summary): ... this one.
26647 (pass_data_local_fn_summary, pass_local_fn_summary,
26648 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26649 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26650 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26651 make_pass_ipa_fn_summary): New.
26652 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26653 inline_write_summary, inline_free_summary): Remove.
26654 (ipa_free_fn_summary) : New.
26655 * ipa-inline.c (ipa_inline): Update.
26656 (pass_ipa_inline): Do not generate summaries.
26657 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26659 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26660 and add pass_ipa_fn_summary.
26661 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26663 (make_pass_inline_parameters): Remove.
26665 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26667 * omp-low.c (struct omp_context): Remove "default_kind" member.
26670 * omp-offload.c (execute_oacc_device_lower): Remove the
26671 parallelism dimensions function attributes for unparallelized
26672 OpenACC kernels constructs.
26674 2017-05-23 Martin Liska <mliska@suse.cz>
26676 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26678 (cgraph_edge::make_speculative): Likewise.
26679 (cgraph_edge::resolve_speculation): Likewise.
26680 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26681 (cgraph_node::dump): Likewise.
26682 * cgraph.h: Likewise.
26683 * cgraphunit.c (analyze_functions): Likewise.
26684 (symbol_table::compile): Likewise.
26685 * ipa-cp.c (print_all_lattices): Likewise.
26686 (determine_versionability): Likewise.
26687 (initialize_node_lattices): Likewise.
26688 (ipcp_verify_propagated_values): Likewise.
26689 (estimate_local_effects): Likewise.
26690 (update_profiling_info): Likewise.
26691 (create_specialized_node): Likewise.
26692 (perhaps_add_new_callers): Likewise.
26693 (decide_about_value): Likewise.
26694 (decide_whether_version_node): Likewise.
26695 (identify_dead_nodes): Likewise.
26696 (ipcp_store_bits_results): Likewise.
26697 * ipa-devirt.c (dump_targets): Likewise.
26698 (ipa_devirt): Likewise.
26699 * ipa-icf.c (sem_item::dump): Likewise.
26700 (sem_function::equals): Likewise.
26701 (sem_variable::equals): Likewise.
26702 (sem_item_optimizer::read_section): Likewise.
26703 (sem_item_optimizer::execute): Likewise.
26704 (congruence_class::dump): Likewise.
26705 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26706 (dump_inline_summary): Likewise.
26707 (estimate_node_size_and_time): Likewise.
26708 (inline_analyze_function): Likewise.
26709 * ipa-inline-transform.c (inline_call): Likewise.
26710 * ipa-inline.c (report_inline_failed_reason): Likewise.
26711 (want_early_inline_function_p): Likewise.
26712 (edge_badness): Likewise.
26713 (update_edge_key): Likewise.
26714 (inline_small_functions): Likewise.
26715 * ipa-profile.c (ipa_profile): Likewise.
26716 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26717 (ipa_make_edge_direct_to_target): Likewise.
26718 (remove_described_reference): Likewise.
26719 (ipa_impossible_devirt_target): Likewise.
26720 (propagate_controlled_uses): Likewise.
26721 (ipa_print_node_params): Likewise.
26722 (ipcp_transform_function): Likewise.
26723 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26724 (propagate_pure_const): Likewise.
26725 * ipa-reference.c (generate_summary): Likewise.
26726 (read_write_all_from_decl): Likewise.
26727 (propagate): Likewise.
26728 (ipa_reference_read_optimization_summary): Likewise.
26729 * ipa-utils.c (ipa_merge_profiles): Likewise.
26730 * ipa.c (walk_polymorphic_call_targets): Likewise.
26731 (symbol_table::remove_unreachable_nodes): Likewise.
26732 (ipa_single_use): Likewise.
26733 * passes.c (execute_todo): Likewise.
26734 * predict.c (drop_profile): Likewise.
26735 * symtab.c (symtab_node::get_dump_name): New function.
26736 (symtab_node::dump_name): Likewise.
26737 (symtab_node::dump_asm_name): Likewise.
26738 (symtab_node::dump_references): Likewise.
26739 (symtab_node::dump_referring): Likewise.
26740 (symtab_node::dump_base): Likewise.
26741 (symtab_node::debug_symtab): Likewise.
26742 * tree-sra.c (convert_callers_for_node): Likewise.
26743 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26744 * value-prof.c (init_node_map): Likewise.
26746 2017-05-23 Martin Liska <mliska@suse.cz>
26748 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26749 and symtab_node::debug_symtab to symbol_table::debug.
26750 * cgraphunit.c (analyze_functions): Use the renamed function.
26751 (symbol_table::compile): Likewise.
26752 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26753 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26754 * passes.c (execute_todo): Likewise.
26755 * symtab.c (symbol_table::dump): New function.
26756 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26758 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26760 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26761 that nonconst implies exec.
26763 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26765 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26766 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26767 (inline_edge_summary_vec): Turn into ...
26768 (ipa_call_summaries): ... this one.
26769 (redirect_to_unreachable, edge_set_predicate,
26770 evaluate_properties_for_edge, inline_summary_alloc,
26771 reset_ipa_call_summary, reset_inline_summary,
26772 inline_summary_t::duplicate): Update.
26773 (inline_edge_duplication_hook): Turn to ...
26774 (ipa_call_summary_t::duplicate): ... this one.
26775 (inline_edge_removal_hook): Turn to ...
26776 (ipa_call_summary_t::remove): ... this one.
26777 (dump_inline_edge_summary): Turn to ...
26778 (dump_ipa_call_summary): ... this one.
26779 (estimate_function_body_sizes): Update.
26780 (inline_update_callee_summaries): Update.
26781 (remap_edge_change_prob): Update.
26782 (remap_edge_summaries): Update.
26783 (inline_merge_summary): Update.
26784 (do_estimate_edge_time): Update.
26785 (inline_generate_summary): Update.
26786 (inline_read_section): Update.
26787 (inline_read_summary): Update.
26788 (inline_free_summary): Update.
26789 * ipa-inline.c (can_inline_edge_p): Update.
26790 (compute_inlined_call_time): Update.
26791 (want_inline_small_function_p): Update.
26792 (edge_badness): Update.
26793 (early_inliner): Update.
26794 * ipa-inline.h (inline_edge_summary): Turn to ...
26795 (ipa_call_summary): ... this one.
26796 (ipa_call_summary_t): New class.
26797 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26798 (ipa_call_summaries): New.
26799 (inline_edge_summary): Remove.
26800 (estimate_edge_growth): Update.
26801 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26802 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26803 * ipa-split.c (execute_split_functions): Update.
26804 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26806 2017-05-23 Tom de Vries <tom@codesourcery.com>
26808 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26809 attributes): Document rdrand effective target.
26811 2017-05-23 Tom de Vries <tom@codesourcery.com>
26813 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26814 attributes): Sort alphabetically.
26816 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
26818 * config/avr/genmultilib.awk: Use gsub instead of gensub.
26820 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26823 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
26824 V2DF/V2DI splat into two separate patterns, one that handles
26825 registers, and the other that only handles memory. Drop support
26826 for splatting from a GPR on ISA 2.07 and then splitting the
26827 splat into direct move and splat.
26828 (vsx_splat_<mode>_reg): Likewise.
26829 (vsx_splat_<mode>_mem): Likewise.
26831 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
26833 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
26835 2017-05-22 Jakub Jelinek <jakub@redhat.com>
26837 PR middle-end/80809
26838 * omp-low.c (finish_taskreg_remap): New function.
26839 (finish_taskreg_scan): If unit size of ctx->record_type
26840 is non-constant, unshare the size expression and replace
26841 decls in it with possible outer var refs.
26843 PR middle-end/80809
26844 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
26845 GOVD_SHARED rather than GOVD_PRIVATE with it.
26846 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
26847 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
26849 PR middle-end/80853
26850 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
26851 as last argument to build_outer_var_ref for pointer bases of array
26852 section reductions.
26854 2017-05-19 Martin Sebor <msebor@redhat.com>
26856 * print-tree.c (print_node): Print DECL_READ_P flag.
26858 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26860 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
26861 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
26862 * cgraph.c: Likewise.
26863 * cgraphunit.c: Likewise.
26864 * gengtype.c: Likewise.
26865 * ipa-cp.c: Likewise.
26866 * ipa-devirt.c: Likewise.
26867 * ipa-icf.c: Likewise.
26868 * ipa-predicate.c: Likewise.
26869 * ipa-profile.c: Likewise.
26870 * ipa-prop.c: Likewise.
26871 * ipa-split.c: Likewise.
26873 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
26874 edge_predicate_pool, dump_inline_hints,
26875 inline_summary::account_size_time, redirect_to_unreachable,
26876 edge_set_predicate, set_hint_predicate,
26877 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
26878 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
26879 inline_summary_t::remove, remap_hint_predicate_after_duplication,
26880 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
26881 ipa_call_summary_t::remove, initialize_growth_caches,
26882 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
26883 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
26884 mark_modified, unmodified_parm_1, unmodified_parm,
26885 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
26886 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
26887 compute_bb_predicates, will_be_nonconstant_expr_predicate,
26888 will_be_nonconstant_predicate, record_modified_bb_info,
26889 get_minimal_bb, record_modified, param_change_prob,
26890 phi_result_unknown_predicate, predicate_for_phi_result,
26891 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
26892 estimate_function_body_sizes, compute_inline_parameters,
26893 compute_inline_parameters_for_curren, pass_data_inline_parameters,
26894 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
26895 inline_update_callee_summaries, remap_edge_change_prob,
26896 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
26897 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
26898 inline_analyze_function, inline_summary_t::insert,
26899 inline_generate_summary, read_ipa_call_summary, inline_read_section,
26900 inline_read_summary, write_ipa_call_summary, inline_write_summary,
26901 inline_free_summary): Move to ipa-fnsummary.h
26902 (predicate_t): Remove.
26903 * ipa-fnsummary.c: New file.
26904 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
26905 (enum inline_hints_vals, inline_hints, agg_position_info,
26906 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
26907 inline_summaries, ipa_call_summary, ipa_call_summary_t,
26908 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
26909 dump_inline_summary, dump_inline_hints, inline_generate_summary,
26910 inline_read_summary, inline_write_summary, inline_free_summary,
26911 inline_analyze_function, initialize_inline_failed,
26912 inline_merge_summary, inline_update_overall_summary,
26913 compute_inline_parameters): Move to ipa-fnsummary.h
26914 * ipa-fnsummary.h: New file.
26915 * ipa-inline-transform.h: Include ipa-inline.h.
26916 * ipa-inline.c: LIkewise.
26918 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26920 * ipa-inline.c (edge_badness): Use inlined_time instead of
26921 inline_summaries->get.
26923 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26925 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
26927 2017-05-22 Nathan Sidwell <nathan@acm.org>
26929 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
26930 (fdump-lang): Document 'raw' option.
26931 * dumpfile.h (TDI_tu): Delete.
26932 * dumpfile.c (dump_files): Remove translation-unit.
26933 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
26935 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
26937 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
26938 command option from $(AWK) call.
26939 * config/avr/genmultilib.awk: Simplify and rewrite so that it
26940 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
26941 [FORMAT]: Remove handling of variable.
26942 * config/avr/t-multilib: Regenerate.
26944 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26946 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
26948 (dump_inline_summary): Do not print self_time.
26949 (estimate_function_body_sizes): Do not set self_time.
26950 (compute_inline_parameters): Likewise.
26951 (inline_read_section, inline_write_summary): Do not stream self_time.
26952 * ipa-inline.h (inline_summary): Drop self_time.
26954 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26956 * ipa-inline-analysis.c (account_size_time): Rename to ...
26957 (inline_summary::account_size_time): ... this one.
26958 (reset_ipa_call_summary): Turn to ...
26959 (ipa_call_summary::reset): ... this one.
26960 (reset_inline_summary): Turn to ...
26961 (inline_summary::reset): ... this one.
26962 (inline_summary_t::remove): Update.
26963 (inline_summary_t::duplicate): Update.
26964 (ipa_call_summary_t::remove): Update.
26965 (dump_inline_summary): Update.
26966 (estimate_function_body_sizes): Update.
26967 (compute_inline_parameters): Update.
26968 (estimate_node_size_and_time): Update.
26969 (inline_merge_summary): Update.
26970 (inline_update_overall_summary): Update.
26971 (inline_read_section): Update.
26972 (inline_write_summary): Update.
26973 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
26974 add account_size_time and reset member functions.
26975 (ipa_call_summary): Add reset function.
26976 * ipa-predicate.h (predicate::operator &): Constify.
26978 2017-05-22 Richard Biener <rguenther@suse.de>
26980 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
26982 2017-05-19 Jason Merrill <jason@redhat.com>
26984 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
26986 2017-05-19 Marek Polacek <polacek@redhat.com>
26989 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
26990 TYPE_OVERFLOW_WRAPS checks.
26992 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
26994 * tree-core.h (enum omp_clause_default_kind): Add
26995 "OMP_CLAUSE_DEFAULT_PRESENT".
26996 * tree-pretty-print.c (dump_omp_clause): Handle it.
26997 * gimplify.c (enum gimplify_omp_var_data): Add
26998 "GOVD_MAP_FORCE_PRESENT".
26999 (gimplify_adjust_omp_clauses_1): Map it to
27000 "GOMP_MAP_FORCE_PRESENT".
27001 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27003 * gimplify.c (oacc_default_clause): Clarify.
27005 2017-05-19 Nathan Sidwell <nathan@acm.org>
27007 LANG_HOOK_REGISTER_DUMPS
27008 * toplev.c (general_init): Call register dump lang hook.
27009 * doc/invoke.texi: Document -fdump-lang option family.
27010 * dumpfile.c (dump_files): Remove class dump here.
27011 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27012 * dumpfile.h (tree_dump_index): Remove TDI_class.
27013 * langhooks-def.h (lhd_register_dumps): Declare.
27014 (LANG_HOOKS_REGISTER_DUMPS): Define.
27015 (LANG_HOOKS_INITIALIZER): Add it.
27016 * langhooks.c (lhd_register_dumps): Define.
27017 * langhooks.h (struct lang_hooks): Add register_dumps.
27019 2017-05-19 Nathan Sidwell <nathan@acm.org>
27021 * context.h (context::set_passes): New.
27022 * context.c (context::context): Do not create pass manager.
27023 * toplev.c (general_init): Create pass manager here.
27025 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27027 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27028 use this splitter if two add or or instructions would also work for
27029 the constant we want to generate.
27031 2017-05-19 Richard Biener <rguenther@suse.de>
27034 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27035 predicate evaluation.
27037 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27039 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27041 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27044 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27046 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27047 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27048 (inline_edge_summary_vec): Turn into ...
27049 (ipa_call_summaries): ... this one.
27050 (redirect_to_unreachable, edge_set_predicate,
27051 evaluate_properties_for_edge, inline_summary_alloc,
27052 reset_ipa_call_summary, reset_inline_summary,
27053 inline_summary_t::duplicate): Update.
27054 (inline_edge_duplication_hook): Turn to ...
27055 (ipa_call_summary_t::duplicate): ... this one.
27056 (inline_edge_removal_hook): Turn to ...
27057 (ipa_call_summary_t::remove): ... this one.
27058 (dump_inline_edge_summary): Turn to ...
27059 (dump_ipa_call_summary): ... this one.
27060 (estimate_function_body_sizes): Update.
27061 (inline_update_callee_summaries): Update.
27062 (remap_edge_change_prob): Update.
27063 (remap_edge_summaries): Update.
27064 (inline_merge_summary): Update.
27065 (do_estimate_edge_time): Update.
27066 (inline_generate_summary): Update.
27067 (inline_read_section): Update.
27068 (inline_read_summary): Update.
27069 (inline_free_summary): Update.
27070 * ipa-inline.c (can_inline_edge_p): Update.
27071 (compute_inlined_call_time): Update.
27072 (want_inline_small_function_p): Update.
27073 (edge_badness): Update.
27074 (early_inliner): Update.
27075 * ipa-inline.h (inline_edge_summary): Turn to ...
27076 (ipa_call_summary): ... this one.
27077 (ipa_call_summary_t): New class.
27078 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27079 (ipa_call_summaries): New.
27080 (inline_edge_summary): Remove.
27081 (estimate_edge_growth): Update.
27082 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27083 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27084 * ipa-split.c (execute_split_functions): Update.
27085 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27087 2017-05-19 Richard Biener <rguenther@suse.de>
27089 PR middle-end/80764
27090 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27092 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27094 * config/rs6000/rs6000.c (struct machine_function): Add field
27095 fpr_is_wrapped_separately.
27096 (rs6000_get_separate_components): Use 64 components. Handle the
27097 new FPR components.
27098 (rs6000_components_for_bb): Handle the FPR components.
27099 (rs6000_emit_prologue_components): Handle the FPR components.
27100 (rs6000_emit_epilogue_components): Handle the FPR components.
27101 (rs6000_set_handled_components): Handle the FPR components.
27102 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27103 that are already separately shrink-wrapped.
27104 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27105 that are already separately shrink-wrapped.
27107 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27110 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27113 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27114 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27115 cases where the register allocator uses a move and an offsettable
27116 memory operation to/from a FPR register on ISA 2.06/2.07.
27117 (define_peephole2 for Altivec d-form store): Likewise.
27119 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27122 * config/i386/mmx.md (*mov<mode>_internal): Enable
27123 alternatives 11, 12, 13 and 14 also for 32bit targets.
27124 Remove alternatives 15, 16, 17 and 18.
27125 * config/i386/sse.md (vec_concatv2di): Change
27126 alternative (!x, *y) to (x, ?!*Yn).
27128 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27130 * dumpfile.h (enum dump_kind): Remove stray comma.
27132 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27134 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27135 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27136 predicate::num_conditions
27137 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27138 (CHANGED): turn into predicate::changed.
27139 (agg_position_info): Move to ipa-predicate.h
27140 (add_condition, predicate::add_clause, predicate::operator &=,
27141 predicate::or_with, predicate::evaluate, predicate::probability,
27142 dump_condition, dump_clause, predicate::dump,
27143 predicate::remap_after_duplication, predicate::remap_after_inlining,
27144 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27145 (evaluate_conditions_for_known_args): Update.
27146 (set_cond_stmt_execution_predicate): Update.
27147 * ipa-inline.h: Include ipa-predicate.h
27148 (condition, inline_param_summary, conditions, agg_position_info,
27149 predicate): Move to ipa-predicate.h
27150 * ipa-predicate.c: New file.
27151 * ipa-predicate.h: New file.
27153 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27155 * final.c (leaf_function_p): Check we are not in a sequence.
27157 2017-05-18 Martin Liska <mliska@suse.cz>
27159 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27160 * dumpfile.c (dump_register): Use new enum dump_kind.
27161 (get_dump_file_name): Likewise.
27162 (dump_enable_all): Likewise.
27163 (dump_switch_p_1): Likewise.
27164 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27165 * dumpfile.h (enum dump_kind): New enum type.
27166 (struct dump_file_info): Create constructor and
27167 format fields and comments.
27168 * passes.c (pass_manager::register_one_dump_file):
27170 * statistics.c (statistics_early_init): Likewise.
27171 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27172 TDF_TREE with TDF_SLIM.
27173 (gather_memory_references_ref): Likewise.
27175 2017-05-18 Martin Liska <mliska@suse.cz>
27177 * vec.h (struct vnull): Use it.
27179 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27181 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27182 (true_predicate, false_predicate, true_predicate_p,
27183 false_predicate_p): Remove.
27184 (single_cond_predicate, not_inlined_predicate): Turn to member function
27186 (add_condition): Update.
27187 (add_clause): Turn to...
27188 (predicate::add_clause): ... this one; update; allow passing NULL
27190 (and_predicates): Turn to ...
27191 (predicate::operator &=): ... this one.
27192 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27193 (or_predicates): Turn to ...
27194 (predicate::or_with): ... this one.
27195 (evaluate_predicate): Turn to ...
27196 (predicate::evaluate): ... this one.
27197 (predicate_probability): Turn to ...
27198 (predicate::probability): ... this one.
27199 (dump_condition): Update.
27200 (dump_predicate): Turn to ...
27201 (predicate::dump): ... this one.
27202 (account_size_time): Update.
27203 (edge_set_predicate): Update.
27204 (set_hint_predicate): UPdate.
27205 (evaluate_conditions_for_known_args): Update.
27206 (evaluate_properties_for_edge): Update.
27207 (remap_predicate_after_duplication): Turn to...
27208 (predicate::remap_after_duplication): ... this one.
27209 (remap_hint_predicate_after_duplication): Update.
27210 (inline_summary_t::duplicate): UPdate.
27211 (dump_inline_edge_summary): Update.
27212 (dump_inline_summary): Update.
27213 (set_cond_stmt_execution_predicate): Update.
27214 (set_switch_stmt_execution_predicate): Update.
27215 (compute_bb_predicates): Update.
27216 (will_be_nonconstant_expr_predicate): Update.
27217 (will_be_nonconstant_predicate): Update.
27218 (phi_result_unknown_predicate): Update.
27219 (predicate_for_phi_result): Update.
27220 (array_index_predicate): Update.
27221 (estimate_function_body_sizes): Update.
27222 (estimate_node_size_and_time): Update.
27223 (estimate_ipcp_clone_size_and_time): Update.
27224 (remap_predicate): Rename to ...
27225 (predicate::remap_after_inlining): ... this one.
27226 (remap_hint_predicate): Update.
27227 (inline_merge_summary): Update.
27228 (inline_update_overall_summary): Update.
27229 (estimate_size_after_inlining): Update.
27230 (read_predicate): Rename to ...
27231 (predicate::stream_in): ... this one.
27232 (read_inline_edge_summary): Update.
27233 (write_predicate): Rename to ...
27234 (predicate::stream_out): ... this one.
27235 (write_inline_edge_summary): Update.
27236 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27237 (clause_t): Turn to uint32_t
27238 (predicate): Turn to class; implement constructor and operators
27240 (size_time_entry): Update.
27241 (inline_summary): Update.
27242 (inline_edge_summary): Update.
27244 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27246 * fold-const.c (fold_binary_loc): Move transformation...
27247 * match.pd (C - X CMP X): ... here.
27249 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27251 * config/sparc/sparc.c (sparc_option_override): Set function
27252 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27253 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27255 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27257 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27259 2017-05-18 Marek Polacek <polacek@redhat.com>
27262 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27263 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27264 gimple_assign_load_p.
27266 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27268 PR middle-end/80692
27269 * real.c (do_compare): Give decimal_do_compare preference over
27270 comparing just the signs.
27272 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27274 * doc/md.texi (Canonicalization of Instructions): Describe the
27275 canonical form of instructions that inherently set a condition
27278 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27280 PR middle-end/80775
27281 * tree-cfg.c: Move deletion of unreachable case statements to after
27282 the merging of consecutive case labels.
27284 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27286 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27287 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27288 restoring of callee-saved registers.
27290 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27292 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27293 * config/visium/visium.c (single_set_and_flags): Likewise.
27294 * config/visium/visium.md (Substitutions): Likewise.
27296 2017-05-17 Martin Liska <mliska@suse.cz>
27298 * cfg.c: Introduce dump_flags_t type and
27299 use it instead of int type.
27301 * cfghooks.c: Likewise.
27302 * cfghooks.h (struct cfg_hooks): Likewise.
27303 * cfgrtl.c: Likewise.
27304 * cfgrtl.h: Likewise.
27305 * cgraph.c (cgraph_node::get_body): Likewise.
27306 * coretypes.h: Likewise.
27307 * domwalk.c: Likewise.
27308 * domwalk.h: Likewise.
27309 * dumpfile.c (struct dump_option_value_info): Likewise.
27310 (dump_enable_all): Likewise.
27311 (dump_switch_p_1): Likewise.
27312 (opt_info_switch_p): Likewise.
27313 * dumpfile.h (enum tree_dump_index): Likewise.
27314 (struct dump_file_info): Likewise.
27315 * genemit.c: Likewise.
27316 * generic-match-head.c: Likewise.
27317 * gengtype.c (open_base_files): Likewise.
27318 * gimple-pretty-print.c: Likewise.
27319 * gimple-pretty-print.h: Likewise.
27320 * graph.c (print_graph_cfg): Likewise.
27321 * graphite-scop-detection.c (dot_all_sese): Likewise.
27322 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27323 * loop-unroll.c (report_unroll): Likewise.
27324 * passes.c (pass_manager::register_one_dump_file): Likewise.
27325 * print-tree.c: Likewise.
27326 * statistics.c: Likewise.
27327 * tree-cfg.c: Likewise.
27328 * tree-cfg.h: Likewise.
27329 * tree-dfa.c: Likewise.
27330 * tree-dfa.h: Likewise.
27331 * tree-dump.c (dump_function): Likewise.
27332 * tree-dump.h (struct dump_info): Likewise.
27333 * tree-pretty-print.c: Likewise.
27334 * tree-pretty-print.h: Likewise.
27335 * tree-ssa-live.c: Likewise.
27336 * tree-ssa-live.h: Likewise.
27337 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27338 * tree-vect-loop.c: Likewise.
27339 * tree-vect-slp.c: Likewise.
27341 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27342 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27344 PR tree-optimization/80457
27345 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27346 of all arguments to a statement as scalar_to_vec operations.
27347 (vectorizable_call): Adjust call to vect_model_simple_cost for
27349 (vectorizable_conversion): Likewise.
27350 (vectorizable_assignment): Likewise.
27351 (vectorizable_shift): Likewise.
27352 (vectorizable_operation): Likewise.
27353 (vectorizable_comparison): Likewise.
27354 (vect_is_simple_cond): Record the def types for operands.
27355 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27356 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27357 for statement argument count.
27359 2017-05-16 Carl Love <cel@us.ibm.com>
27361 * config/rs6000/rs6000-c: Add support for built-in functions
27362 vector unsigned long long vec_bperm (vector unsigned long long,
27363 vector unsigned char)
27364 vector signed long long vec_mule (vector signed int,
27366 vector unsigned long long vec_mule (vector unsigned int,
27367 vector unsigned int)
27368 vector signed long long vec_mulo (vector signed int,
27370 vector unsigned long long vec_mulo (vector unsigned int,
27371 vector unsigned int)
27372 vector signed char vec_sldw (vector signed char,
27373 vector signed char,
27375 vector unsigned char vec_sldw (vector unsigned char,
27376 vector unsigned char,
27378 vector signed short vec_sldw (vector signed short,
27379 vector signed short,
27381 vector unsigned short vec_sldw (vector unsigned short,
27382 vector unsigned short,
27384 vector signed int vec_sldw (vector signed int,
27387 vector unsigned int vec_sldw (vector unsigned int,
27388 vector unsigned int,
27390 vector signed long long vec_sldw (vector signed long long,
27391 vector signed long long,
27393 vector unsigned long long vec_sldw (vector unsigned long long,
27394 vector unsigned long long,
27396 * config/rs6000/rs6000-c: Add support for built-in functions
27397 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27398 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27399 * doc/extend.texi: Update the built-in documentation for the
27400 new built-in functions.
27402 2017-05-16 Marek Polacek <polacek@redhat.com>
27406 * tree.c (save_expr): Don't fold the expression.
27408 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27410 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27411 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27412 and (?*y,m). Update insn attributes.
27414 2017-05-16 Martin Liska <mliska@suse.cz>
27416 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27417 flags argument of print_gimple_stmt, print_gimple_expr,
27418 print_generic_stmt and print_generic_expr.
27419 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27420 * coretypes.h: Likewise.
27421 * except.c (dump_eh_tree): Likewise.
27422 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27423 * gimple-pretty-print.h: Likewise.
27424 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27425 (backprop::push_to_worklist): Likewise.
27426 (backprop::pop_from_worklist): Likewise.
27427 (backprop::process_use): Likewise.
27428 (backprop::intersect_uses): Likewise.
27429 (note_replacement): Likewise.
27430 * gimple-ssa-store-merging.c
27431 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27432 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27433 (pass_store_merging::execute): Likewise.
27434 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27435 (ssa_base_cand_dump_callback): Likewise.
27436 (dump_incr_vec): Likewise.
27437 (replace_refs): Likewise.
27438 (replace_mult_candidate): Likewise.
27439 (create_add_on_incoming_edge): Likewise.
27440 (create_phi_basis): Likewise.
27441 (insert_initializers): Likewise.
27442 (all_phi_incrs_profitable): Likewise.
27443 (introduce_cast_before_cand): Likewise.
27444 (replace_one_candidate): Likewise.
27445 * gimplify.c (gimplify_expr): Likewise.
27446 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27447 (set_rename): Likewise.
27448 (rename_uses): Likewise.
27449 (copy_loop_phi_nodes): Likewise.
27450 (add_close_phis_to_merge_points): Likewise.
27451 (copy_loop_close_phi_args): Likewise.
27452 (copy_cond_phi_args): Likewise.
27453 (graphite_copy_stmts_from_block): Likewise.
27454 (translate_pending_phi_nodes): Likewise.
27455 * graphite-poly.c (print_pdr): Likewise.
27456 (dump_gbb_cases): Likewise.
27457 (dump_gbb_conditions): Likewise.
27458 (print_scop_params): Likewise.
27459 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27460 (build_cross_bb_scalars_use): Likewise.
27461 (gather_bbs::before_dom_children): Likewise.
27462 * hsa-dump.c (dump_hsa_immed): Likewise.
27463 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27464 (get_replacement_map): Likewise.
27465 * ipa-inline-analysis.c (dump_condition): Likewise.
27466 (estimate_function_body_sizes): Likewise.
27467 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27468 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27469 * ipa-prop.c (ipa_dump_param): Likewise.
27470 (ipa_print_node_jump_functions_for_edge): Likewise.
27471 (ipa_modify_call_arguments): Likewise.
27472 (ipa_modify_expr): Likewise.
27473 (ipa_dump_param_adjustments): Likewise.
27474 (ipa_dump_agg_replacement_values): Likewise.
27475 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27476 * ipa-pure-const.c (check_stmt): Likewise.
27477 (pass_nothrow::execute): Likewise.
27478 * ipa-split.c (execute_split_functions): Likewise.
27479 * omp-offload.c (dump_oacc_loop_part): Likewise.
27480 (dump_oacc_loop): Likewise.
27481 * trans-mem.c (tm_log_emit): Likewise.
27482 (tm_memopt_accumulate_memops): Likewise.
27483 (dump_tm_memopt_set): Likewise.
27484 (dump_tm_memopt_transform): Likewise.
27485 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27486 (print_loop): Likewise.
27487 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27488 (chkp_gather_checks_info): Likewise.
27489 (chkp_get_check_result): Likewise.
27490 (chkp_remove_check_if_pass): Likewise.
27491 (chkp_use_outer_bounds_if_possible): Likewise.
27492 (chkp_reduce_bounds_lifetime): Likewise.
27493 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27494 (chkp_mark_completed_bounds): Likewise.
27495 (chkp_register_incomplete_bounds): Likewise.
27496 (chkp_mark_invalid_bounds): Likewise.
27497 (chkp_maybe_copy_and_register_bounds): Likewise.
27498 (chkp_build_returned_bound): Likewise.
27499 (chkp_get_bound_for_parm): Likewise.
27500 (chkp_build_bndldx): Likewise.
27501 (chkp_get_bounds_by_definition): Likewise.
27502 (chkp_generate_extern_var_bounds): Likewise.
27503 (chkp_get_bounds_for_decl_addr): Likewise.
27504 * tree-chrec.c (chrec_apply): Likewise.
27505 * tree-data-ref.c (dump_data_reference): Likewise.
27506 (dump_subscript): Likewise.
27507 (dump_data_dependence_relation): Likewise.
27508 (analyze_overlapping_iterations): Likewise.
27509 * tree-inline.c (expand_call_inline): Likewise.
27510 (tree_function_versioning): Likewise.
27511 * tree-into-ssa.c (dump_defs_stack): Likewise.
27512 (dump_currdefs): Likewise.
27513 (dump_names_replaced_by): Likewise.
27514 (dump_update_ssa): Likewise.
27515 (update_ssa): Likewise.
27516 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27517 * tree-parloops.c (build_new_reduction): Likewise.
27518 (try_create_reduction_list): Likewise.
27519 (ref_conflicts_with_region): Likewise.
27520 (oacc_entry_exit_ok_1): Likewise.
27521 (oacc_entry_exit_single_gang): Likewise.
27522 * tree-pretty-print.h: Likewise.
27523 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27524 (get_scalar_evolution): Likewise.
27525 (add_to_evolution): Likewise.
27526 (get_loop_exit_condition): Likewise.
27527 (analyze_evolution_in_loop): Likewise.
27528 (analyze_initial_condition): Likewise.
27529 (analyze_scalar_evolution): Likewise.
27530 (instantiate_scev): Likewise.
27531 (number_of_latch_executions): Likewise.
27532 (gather_chrec_stats): Likewise.
27533 (final_value_replacement_loop): Likewise.
27534 (scev_const_prop): Likewise.
27535 * tree-sra.c (dump_access): Likewise.
27536 (disqualify_candidate): Likewise.
27537 (create_access): Likewise.
27538 (reject): Likewise.
27539 (maybe_add_sra_candidate): Likewise.
27540 (create_access_replacement): Likewise.
27541 (analyze_access_subtree): Likewise.
27542 (analyze_all_variable_accesses): Likewise.
27543 (sra_modify_assign): Likewise.
27544 (initialize_constant_pool_replacements): Likewise.
27545 (find_param_candidates): Likewise.
27546 (decide_one_param_reduction): Likewise.
27547 (replace_removed_params_ssa_names): Likewise.
27548 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27549 * tree-ssa-copy.c (dump_copy_of): Likewise.
27550 (copy_prop_visit_cond_stmt): Likewise.
27551 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27552 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27553 (record_equivalences_from_stmt): Likewise.
27554 * tree-ssa-dse.c (compute_trims): Likewise.
27555 (delete_dead_call): Likewise.
27556 (delete_dead_assignment): Likewise.
27557 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27558 (forward_propagate_into_cond): Likewise.
27559 (pass_forwprop::execute): Likewise.
27560 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27561 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27563 (move_computations_worker): Likewise.
27564 (execute_sm): Likewise.
27565 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27566 (remove_exits_and_undefined_stmts): Likewise.
27567 (remove_redundant_iv_tests): Likewise.
27568 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27569 (adjust_iv_update_pos): Likewise.
27570 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27571 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27572 (value_replacement): Likewise.
27573 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27574 * tree-ssa-pre.c (print_pre_expr): Likewise.
27575 (get_representative_for): Likewise.
27576 (create_expression_by_pieces): Likewise.
27577 (insert_into_preds_of_block): Likewise.
27578 (eliminate_insert): Likewise.
27579 (eliminate_dom_walker::before_dom_children): Likewise.
27580 (eliminate): Likewise.
27581 (remove_dead_inserted_code): Likewise.
27582 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27583 * tree-ssa-reassoc.c (get_rank): Likewise.
27584 (eliminate_duplicate_pair): Likewise.
27585 (eliminate_plus_minus_pair): Likewise.
27586 (eliminate_not_pairs): Likewise.
27587 (undistribute_ops_list): Likewise.
27588 (eliminate_redundant_comparison): Likewise.
27589 (update_range_test): Likewise.
27590 (optimize_range_tests_var_bound): Likewise.
27591 (optimize_vec_cond_expr): Likewise.
27592 (rewrite_expr_tree): Likewise.
27593 (rewrite_expr_tree_parallel): Likewise.
27594 (linearize_expr): Likewise.
27595 (break_up_subtract): Likewise.
27596 (linearize_expr_tree): Likewise.
27597 (attempt_builtin_powi): Likewise.
27598 (attempt_builtin_copysign): Likewise.
27599 (transform_stmt_to_copy): Likewise.
27600 (transform_stmt_to_multiply): Likewise.
27601 (dump_ops_vector): Likewise.
27602 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27603 (print_scc): Likewise.
27604 (set_ssa_val_to): Likewise.
27605 (visit_reference_op_store): Likewise.
27606 (visit_use): Likewise.
27607 (sccvn_dom_walker::before_dom_children): Likewise.
27608 (run_scc_vn): Likewise.
27609 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27611 (expr_hash_elt::print): Likewise.
27612 (const_and_copies::pop_to_marker): Likewise.
27613 (const_and_copies::record_const_or_copy_raw): Likewise.
27614 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27615 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27616 (dump_predicates): Likewise.
27617 (find_uninit_use): Likewise.
27618 (warn_uninitialized_phi): Likewise.
27619 (pass_late_warn_uninitialized::execute): Likewise.
27620 * tree-ssa.c (verify_vssa): Likewise.
27621 (verify_ssa): Likewise.
27622 (maybe_optimize_var): Likewise.
27623 * tree-vrp.c (dump_value_range): Likewise.
27624 (dump_all_value_ranges): Likewise.
27625 (dump_asserts_for): Likewise.
27626 (register_edge_assert_for_2): Likewise.
27627 (vrp_visit_cond_stmt): Likewise.
27628 (vrp_visit_switch_stmt): Likewise.
27629 (vrp_visit_stmt): Likewise.
27630 (vrp_visit_phi_node): Likewise.
27631 (simplify_cond_using_ranges_1): Likewise.
27632 (fold_predicate_in): Likewise.
27633 (evrp_dom_walker::before_dom_children): Likewise.
27634 (evrp_dom_walker::push_value_range): Likewise.
27635 (evrp_dom_walker::pop_value_range): Likewise.
27636 (execute_early_vrp): Likewise.
27638 2017-05-16 Richard Biener <rguenther@suse.de>
27640 * dwarf2out.c (loc_list_from_tree_1): Do not create
27641 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27643 2017-05-16 Richard Biener <rguenther@suse.de>
27645 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27647 (note_variable_value_in_expr): If we resolved the decl ref
27648 do not push to the stack.
27650 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27652 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27653 operations in fast-math mode.
27654 (vaddq_f16): Likewise.
27655 (vmul_f16): Likewise.
27656 (vmulq_f16): Likewise.
27657 (vsub_f16): Likewise.
27658 (vsubq_f16): Likewise.
27659 * config/arm/neon.md (add<mode>3): New.
27661 (fma:<VH:mode>3): New. Also remove outdated comment.
27664 2017-05-16 Martin Liska <mliska@suse.cz>
27668 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27669 (odr_types_equivalent_p): Likewise.
27671 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27673 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27675 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27678 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27679 non-interunit SSE move alternatives with '?'.
27680 (zero-extendsidi peephole2): New peephole to skip intermediate
27681 general register in SSE zero-extend sequence.
27683 2017-05-15 Jeff Law <law@redhat.com>
27685 * reorg.c (relax_delay_slots): Create a new variable to hold
27686 the temporary target rather than clobbering TARGET_LABEL.
27688 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27689 missing argument to extract_bit_field call.
27690 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27692 2017-05-15 Martin Liska <mliska@suse.cz>
27695 * gcc.c (process_command): Do not allow empty argument of -o option.
27697 2017-05-15 Renlin Li <renlin.li@arm.com>
27699 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27700 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27701 * config/aarch64/constraints.md (Usf): Add long call check.
27702 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27703 (call_value): Likewise.
27704 (sibcall): Likewise.
27705 (sibcall_value): Likewise.
27707 (call_value_insn): New.
27708 (sibcall_insn): Update rtx pattern.
27709 (sibcall_value_insn): Likewise.
27710 (call_internal): Remove.
27711 (call_value_internal): Likewise.
27712 (sibcall_internal): Likewise.
27713 (sibcall_value_internal): Likewise.
27714 (call_reg): Likewise.
27715 (call_symbol): Likewise.
27716 (call_value_reg): Likewise.
27717 (call_value_symbol): Likewise.
27719 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27722 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27724 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27726 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27727 compatible with CCGOCmode and with CCZmode.
27729 2017-05-14 Martin Sebor <msebor@redhat.com>
27731 PR middle-end/77671
27732 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27733 (gimple_fold_builtin_snprintf): Same.
27734 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27735 (gimple_fold_builtin_snprintf): Same.
27736 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27737 of character types.
27738 (is_call_safe): New function.
27739 (try_substitute_return_value): Call it.
27740 (try_simplify_call): New function.
27741 (pass_sprintf_length::handle_gimple_call): Call it.
27743 2017-05-14 Martin Sebor <msebor@redhat.com>
27745 PR middle-end/80669
27746 * builtins.c (expand_builtin_stpncpy): Simplify.
27748 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27750 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27751 * config/i386/i386.h
27752 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27753 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27754 (struct machine_function): Add new members call_ms2sysv,
27755 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27756 (struct machine_frame_state): New fields sp_realigned and
27757 sp_realigned_offset.
27758 * config/i386/i386.c
27759 (enum xlogue_stub): New enum.
27760 (enum xlogue_stub_sets): New enum.
27761 (class xlogue_layout): New class.
27762 (struct ix86_frame): New fields stack_realign_allocate_offset,
27763 stack_realign_offset and outlined_save_offset. Modify comments to
27764 detail stack layout when using out-of-line stubs.
27765 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27766 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27767 -mcall-ms2sysv-xlogues.
27768 (stub_managed_regs): New static variable.
27769 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27770 registers managed by out-of-line stub.
27771 (disable_call_ms2sysv_xlogues): New function.
27772 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27773 m->call_ms2sysv when appropriate and compute frame layout for
27775 (sp_valid_at, fp_valid_at): New inline functions.
27776 (choose_basereg): New function.
27777 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27779 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27780 Use align parameter of choose_baseaddr to generated aligned SSE movs
27782 (pro_epilogue_adjust_stack): Modify to track
27783 machine_frame_state::sp_realigned.
27784 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27785 (ix86_nsaved_sseregs): Likewise.
27786 (ix86_emit_save_regs): Likewise.
27787 (ix86_emit_save_regs_using_mov): Likewise.
27788 (ix86_emit_save_sse_regs_using_mov): Likewise.
27789 (get_scratch_register_on_entry): Likewise.
27790 (gen_frame_set): New function.
27791 (gen_frame_load): Likewise.
27792 (gen_frame_store): Likewise.
27793 (emit_outlined_ms2sysv_save): Likewise.
27794 (emit_outlined_ms2sysv_restore): Likewise.
27795 (ix86_expand_prologue): Modify stack re-alignment code and call
27796 emit_outlined_ms2sysv_save when appropriate.
27797 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27798 parameter rtx_insn *insn, which allows the function to be used to only
27799 generate the notes.
27800 (ix86_expand_epilogue): Modify validity checks of frame and stack
27801 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27802 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27803 * config/i386/predicates.md
27804 (save_multiple): New predicate.
27805 (restore_multiple): Likewise.
27806 * config/i386/sse.md
27807 (save_multiple<mode>): New pattern.
27808 (save_multiple_realign<mode>): Likewise.
27809 (restore_multiple<mode>): Likewise.
27810 (restore_multiple_and_return<mode>): Likewise.
27811 (restore_multiple_leave_return<mode>): Likewise.
27812 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27814 2017-05-14 Julia Koval <julia.koval@intel.com>
27816 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
27817 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
27818 (__builtin_ia32_xsetbv): New builtins.
27819 * config/i386/i386.c (ix86_expand_special_args_builtin):
27821 (ix86_expand_builtin): Special expand for new intrinsics.
27822 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
27823 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
27824 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
27826 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27828 * cfganal.c (inverted_post_order_compute): Change argument type
27830 * cfganal.h (inverted_post_order_compute): Adjust prototype.
27831 * df-core.c (rest_of_handle_df_initialize): Adjust.
27832 (rest_of_handle_df_finish): Likewise.
27833 (df_analyze_1): Likewise.
27834 (df_analyze): Likewise.
27835 (loop_inverted_post_order_compute): Change argument to be a vec *.
27836 (df_analyze_loop): Adjust.
27837 (df_get_n_blocks): Likewise.
27838 (df_get_postorder): Likewise.
27839 * df.h (struct df_d): Change field to be a vec.
27840 * lcm.c (compute_laterin): Adjust.
27841 (compute_available): Likewise.
27842 * lra-lives.c (lra_create_live_ranges_1): Likewise.
27843 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
27844 * tree-ssa-pre.c (compute_antic): Likewise.
27846 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27848 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
27849 (depth_first_search::depth_first_search): Change structure init
27850 function to this constructor.
27851 (depth_first_search::add_bb): Rename function to this member.
27852 (depth_first_search::execute): Likewise.
27853 (flow_dfs_compute_reverse_finish): Adjust.
27855 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27857 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
27858 (longest_simple_path): Likewise.
27859 * shrink-wrap.c (spread_components): Likewise.
27860 (disqualify_problematic_components): Likewise.
27861 (emit_common_heads_for_components): Likewise.
27862 (emit_common_tails_for_components): Likewise.
27863 (insert_prologue_epilogue_for_components): Likewise.
27865 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27867 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
27870 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27872 * df-core.c (df_set_blocks): Start using auto_bitmap.
27873 (df_compact_blocks): Likewise.
27874 * df-problems.c (df_rd_confluence_n): Likewise.
27875 * df-scan.c (df_insn_rescan_all): Likewise.
27876 (df_process_deferred_rescans): Likewise.
27877 (df_update_entry_block_defs): Likewise.
27878 (df_update_exit_block_uses): Likewise.
27879 (df_entry_block_bitmap_verify): Likewise.
27880 (df_exit_block_bitmap_verify): Likewise.
27881 (df_scan_verify): Likewise.
27882 * lra-constraints.c (lra_constraints): Likewise.
27883 (undo_optional_reloads): Likewise.
27884 (lra_undo_inheritance): Likewise.
27885 * lra-remat.c (calculate_gen_cands): Likewise.
27886 (do_remat): Likewise.
27887 * lra-spills.c (assign_spill_hard_regs): Likewise.
27888 (spill_pseudos): Likewise.
27889 * tree-ssa-pre.c (bitmap_set_and): Likewise.
27890 (bitmap_set_subtract_values): Likewise.
27892 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27894 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
27895 management with auto_bitmap.
27896 (fix_inter_tick): Likewise.
27897 (fix_recovery_deps): Likewise.
27898 * ira.c (add_store_equivs): Likewise.
27899 (find_moveable_pseudos): Likewise.
27900 (split_live_ranges_for_shrink_wrap): Likewise.
27901 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
27902 (rtx_reuse_manager::seen_def_p): Likewise.
27903 (rtx_reuse_manager::set_seen_def): Likewise.
27904 * print-rtl.h (class rtx_reuse_manager): Likewise.
27906 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27908 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
27910 (migrate_btr_def): Likewise.
27911 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
27912 * df-core.c (loop_post_order_compute): Likewise.
27913 (loop_inverted_post_order_compute): Likewise.
27914 * hsa-common.h: Likewise.
27915 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
27916 * init-regs.c (initialize_uninitialized_regs): Likewise.
27917 * ipa-inline.c (resolve_noninline_speculation): Likewise.
27918 (inline_small_functions): Likewise.
27919 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
27920 * ira.c (combine_and_move_insns): Likewise.
27921 (build_insn_chain): Likewise.
27922 * loop-invariant.c (find_invariants): Likewise.
27923 * lower-subreg.c (propagate_pseudo_copies): Likewise.
27924 * predict.c (tree_predict_by_opcode): Likewise.
27925 (predict_paths_leading_to): Likewise.
27926 (predict_paths_leading_to_edge): Likewise.
27927 (estimate_loops_at_level): Likewise.
27928 (estimate_loops): Likewise.
27929 * shrink-wrap.c (try_shrink_wrapping): Likewise.
27930 (spread_components): Likewise.
27931 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
27932 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
27933 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
27934 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
27935 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
27936 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
27937 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
27938 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
27939 (mark_threaded_blocks): Likewise.
27940 (thread_through_all_blocks): Likewise.
27941 * tree-ssa.c (verify_ssa): Likewise.
27942 (execute_update_addresses_taken): Likewise.
27943 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
27945 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27947 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
27949 (post_order_compute): Likewise.
27950 (inverted_post_order_compute): Likewise.
27951 (pre_and_rev_post_order_compute_fn): Likewise.
27953 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27955 * genrecog.c (int_set::int_set): Explicitly construct our
27956 auto_vec base class.
27957 * vec.h (auto_vec::auto_vec): New constructor.
27959 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27961 * bitmap.h (class auto_bitmap): New constructor taking
27962 bitmap_obstack * argument.
27964 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27966 * bitmap.h (class auto_bitmap): Change type of m_bits to
27967 bitmap_head, and adjust ctor / dtor and member operators.
27969 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
27971 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
27972 when returned register mode doesn't match original mode.
27974 2017-05-12 Jeff Law <law@redhat.com>
27975 Jakub Jelinek <jakub@redhat.com>
27977 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
27978 we look for cc setter after the compare-elim changes.
27979 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
27980 within the vector to match what compare-elim now expects.
27981 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
27982 (xorsi3_flags, one_cmplsi2_flags): Likewise.
27984 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
27985 after the compare-elim changes.
27986 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
27987 the vector to match what compare-elim now expects.
27988 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
27989 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
27990 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
27991 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
27992 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
27994 * config/visium/visium.c (single_set_and_flags): Fix where
27995 we look for cc setter after the compare-elim changes.
27996 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
27997 with the vector to match what compare-elim now expects.
27998 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
27999 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28000 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28001 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28002 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28003 (neg<mode>2_insn_set_overflow): Likewise.
28005 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28007 PR middle-end/79794
28008 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28009 maybe_expand_insn call, set ops[0].target. If still set after call,
28010 set alt_rtl. Add extra arg to recursive calls.
28011 (extract_bit_field): Add alt_rtl argument. Pass to
28013 * expmed.h (extract_bit_field): Fix prototype.
28014 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28015 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28016 to extract_bit_field_calls.
28017 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28018 Pass alt_rtl to extract_bit_field calls.
28019 * calls.c (store_unaligned_arguments_into_psuedos)
28020 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28021 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28023 * optabs.h (struct expand_operand): Add target bitfield.
28025 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28027 * compare-elim.c (try_eliminate_compare): Canonicalize
28028 operation with embedded compare to
28029 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28030 (set (reg) (operation)].
28032 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28034 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28037 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28038 cost of adding a carry flag for ADC instruction.
28039 [case MINUS]: Ignore the cost of subtracting a carry flag
28040 for SBB instruction.
28042 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28044 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28046 * config/rs6000/bmiintrin.h: New file.
28047 * config/rs6000/bmi2intrin.h: New file.
28048 * config/rs6000/x86intrin.h: New file.
28050 2017-05-12 Jeff Law <law@redhat.com>
28052 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28055 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28057 PR middle-end/80707
28058 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28060 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28062 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28063 early expansion of vector divide builtins.
28064 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28065 builtins identified as having unsigned arguments.
28067 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28069 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28070 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28071 expansion of vector logical operations (and, andc, or, xor,
28074 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28076 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28077 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28079 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28081 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28082 early GIMPLE expansion of vector multiplies.
28084 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28086 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28087 TARGET_HAVE_MOVT conditional.
28088 (movt splitter): Likewise.
28090 2017-05-12 Richard Biener <rguenther@suse.de>
28092 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28093 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28094 Fold all stmts not inplace.
28096 2017-05-12 Richard Biener <rguenther@suse.de>
28098 PR tree-optimization/80713
28099 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28100 inserted_exprs bit for not removed stmts.
28102 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28104 PR middle-end/69921
28105 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28106 parallelized" attribute for parallelized OpenACC kernels.
28107 * omp-offload.c (execute_oacc_device_lower): Use it.
28109 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28110 Set "oacc kernels" attribute.
28111 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28112 parameter. Adjust all users.
28113 (oacc_fn_attrib_kernels_p): Remove function.
28114 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28115 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28116 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28117 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28118 assert "oacc kernels" attribute is set.
28120 2017-05-11 Carl Love <cel@us.ibm.com>
28122 * config/rs6000/rs6000-c: Add support for built-in functions
28123 vector unsigned char vec_popcnt (vector signed char)
28124 vector unsigned char vec_popcnt (vector unsigned char)
28125 vector unsigned short vec_popcnt (vector signed short)
28126 vector unsigned short vec_popcnt (vector unsigned short)
28127 vector unsigned int vec_popcnt (vector signed int)
28128 vector unsigned int vec_popcnt (vector unsigned int)
28129 vector unsigned long long vec_popcnt (vector signed long long)
28130 vector unsigned long long vec_popcnt (vector unsigned long long)
28131 vector signed long long vec_slo (vector signed long long,
28132 vector signed char)
28133 vector signed long long vec_slo (vector signed long long,
28134 vector unsigned char)
28135 vector unsigned long long vec_slo (vector unsigned long long,
28136 vector signed char)
28137 vector unsigned long long vec_slo (vector unsigned long long,
28138 vector unsigned char)
28139 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28140 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28141 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28142 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28143 * doc/extend.texi: Update the built-in documentation file for the
28144 new built-in functions.
28146 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28148 * attribs.h (sorted_attr_string): Move machine independent
28149 functions for target clone support from the i386 port to common
28150 code. Rename ix86_function_versions to common_function_versions.
28151 Rename make_name to make_unique_name.
28152 (common_function_versions): Likewise.
28153 (make_unique_name): Likewise.
28154 (make_dispatcher_decl): Likewise.
28155 (is_function_default_version): Likewise.
28156 * attribs.c (attr_strcmp): Likewise.
28157 (sorted_attr_string): Likewise.
28158 (common_function_versions): Likewise.
28159 (make_unique_name): Likewise.
28160 (make_dispatcher_decl): Likewise.
28161 (is_function_default_version): Likewise.
28162 * config/i386/i386.c (attr_strcmp): Likewise.
28163 (sorted_attr_string): Likewise.
28164 (ix86_function_versions): Likewise.
28165 (make_name): Likewise.
28166 (make_dispatcher_decl): Likewise.
28167 (is_function_default_version): Likewise.
28168 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28170 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28173 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28174 Account for direct move costs for vec_construct of integer
28177 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28180 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28181 (UNSPEC_STX_ATOMIC): Ditto.
28182 (loaddi_via_sse): New insn.
28183 (storedi_via_sse): Ditto.
28184 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28185 Update corresponding peephole2 patterns.
28186 (atomic_storedi_fpu): Ditto.
28188 2017-05-11 Julia Koval <julia.koval@intel.com>
28190 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28191 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28193 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28194 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28195 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28197 2017-05-11 Nathan Sidwell <nathan@acm.org>
28199 * graphite-poly.c: Include dumpfile.h.
28201 * dumpfle.h (dump_function): Declare here ...
28202 * tree-dump.h (dump_function): ... not here.
28203 * dumpfile.c: #include tree-cfg.h.
28204 (dump_function): Move here from ...
28205 * tree-dump.c (dump_function): ... here.
28206 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28207 * graphite-poly.c: Don't include tree-dump.h.
28208 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28209 * print-tree.c: Likewise.
28210 * stor-layout.c: Likewise.
28211 * tree-nested.c: Likewise.
28213 * dumpfile.c (dump_start): Use TDF_FLAGS.
28214 (dump_enable_all): Fix TDF_KIND check thinko.
28216 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28218 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28219 array entries to represent two legal parameterizations of the
28220 overloaded __builtin_cmpb function, as represented by the
28221 P6_OV_BUILTIN_CMPB constant.
28222 (altivec_resolve_overloaded_builtin): Add special case handling
28223 for the __builtin_cmpb function, as represented by the
28224 P6_OV_BUILTIN_CMPB constant.
28225 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28226 (BU_P6_64BIT_2): New macro.
28227 (BU_P6_OVERLOAD_2): New macro
28228 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28229 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28230 (CMPB): Add overload support to represent both 32-bit and 64-bit
28231 compare-bytes function.
28232 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28233 support for TARGET_CMPB.
28234 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28235 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28236 documentation of the __builtin_cmpb overloaded built-in function.
28238 2017-05-11 Richard Biener <rguenther@suse.de>
28240 PR tree-optimization/80705
28241 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28242 bases are not vectorizable.
28244 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28246 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28247 when counting register pressure.
28249 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28251 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28252 (iv_ca_more_deps): Renamed to ...
28253 (iv_ca_compare_deps): ... this.
28254 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28256 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28258 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28260 (determine_group_iv_costs): ... here.
28261 (find_inv_vars_cb): Record inv var if it's not recorded before.
28263 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28265 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28266 (get_shiftadd_cost): Ditto.
28268 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28270 * tree-ssa-address.c: Include header file.
28271 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28273 (add_to_parts): Refactor.
28274 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28275 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28278 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28280 PR tree-optimization/53090
28281 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28283 (extract_cond_operands): Detect condition with IV on both sides
28284 and return COMP_IV_EXPR_2.
28285 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28286 (rewrite_use_compare): Simplify by removing call to function
28287 extract_cond_operands.
28289 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28291 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28292 (extract_cond_operands): Detect condition comparing against non-
28293 invariant bound and return appropriate enum value.
28294 (find_interesting_uses_cond): Update use of extract_cond_operands.
28295 Handle its return value accordingly.
28296 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28298 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28300 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28301 nonlinear iv_use computation in loop invariant sensitive way.
28303 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28305 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28306 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28308 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28310 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28311 (dump_cand): Support iv_cand.inv_exprs.
28312 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28314 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28317 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28319 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28321 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28322 as local function. Include necessary header files.
28323 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28325 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28327 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28329 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28331 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28332 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28333 RSHIFT_EXPR and BIT_NOT_EXPR.
28335 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28337 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28338 (adjust_setup_cost): New parameter supporting round up adjustment.
28339 (struct address_cost_data): Delete.
28340 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28341 (split_address_cost, ptr_difference_cost): Delete.
28342 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28343 (struct ainc_cost_data): New struct.
28344 (get_address_cost_ainc): New function.
28345 (get_address_cost, get_computation_cost): Reimplement.
28346 (determine_group_iv_cost_address): Record inv_expr for all uses of
28348 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28349 (iv_ca_has_deps): Reimplemented to ...
28350 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28352 (iv_ca_extend): Call iv_ca_more_deps.
28354 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28356 * tree-ssa-address.c (struct mem_address): Move to header file.
28357 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28358 * tree-ssa-address.h (struct mem_address): Move from C file.
28359 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28361 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28363 * tree-affine.h (aff_combination_type): New interface.
28364 (aff_combination_zero_p): Remove static.
28365 (aff_combination_const_p): New interface.
28366 (aff_combination_singleton_var_p): New interfaces.
28368 2017-05-11 Richard Biener <rguenther@suse.de>
28370 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28371 Skip unreachable blocks and destinations.
28372 (eliminate): Move stmt removal and fixup ...
28373 (fini_eliminate): ... here. Skip inserted exprs.
28374 (pass_pre::execute): Move fini_pre after fini_eliminate.
28375 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28376 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28377 PRE to get rid of dead code that has invalid SSA form and
28378 split critical edges again.
28380 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28382 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28384 2017-05-11 Richard Biener <rguenther@suse.de>
28386 * passes.c (execute_function_todo): Verify loops if they are
28387 said to be up-to-date.
28388 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28389 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28391 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28394 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28395 handle calling assemble_external ourself.
28398 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28399 modes with zero size. Enhance comment.
28401 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28403 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28404 built-ins for vec_xl and vec_xst with short and char pointer
28407 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28409 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28410 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28411 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28412 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28413 (_mm_maskz_min_round_ss): New intrinsics.
28414 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28415 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28416 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28417 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28418 (__builtin_ia32_minss_mask_round): New builtins.
28419 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28420 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28421 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28423 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28424 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28426 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28429 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28431 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28432 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28433 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28434 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28435 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28436 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28437 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28438 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28439 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28440 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28441 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28442 (__builtin_ia32_mulss_mask_round): New builtins.
28443 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28444 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28445 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28447 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28448 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28450 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28453 2017-05-10 Julia Koval <julia.koval@intel.com>
28455 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28456 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28457 (_mm256_setr_m128i): New intrinsics.
28459 2017-05-10 Julia Koval <julia.koval@intel.com>
28461 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28462 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28463 (_mm_maskz_rcp14_ss): New intrinsics.
28464 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28465 (__builtin_ia32_rcp14ss_mask): New builtins.
28466 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28468 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28470 PR tree-optimization/51513
28471 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28472 (assert_unreachable_fallthru_edge_p): Use it.
28473 (group_case_labels_stmt): Likewise.
28474 * tree-cfg.h: Prototype it.
28475 * stmt.c: Include cfghooks.h and tree-cfg.h.
28476 (emit_case_dispatch_table) <gap_label>: New local variable.
28477 Use it to fill dispatch table gaps.
28478 Test for default_label before updating probabilities.
28479 (expand_case) <default_label>: Remove unneeded initialization.
28480 Test for unreachable default case statement and remove its edge.
28481 Set default_label accordingly.
28482 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28484 2017-05-10 Carl Love <cel@us.ibm.com>
28486 * config/rs6000/rs6000-c: Add support for built-in functions
28487 vector signed char vec_neg (vector signed char)
28488 vector signed short int vec_neg (vector short int)
28489 vector signed int vec_neg (vector signed int)
28490 vector signed long long vec_neg (vector signed long long)
28491 vector float vec_neg (vector float)
28492 vector double vec_neg (vector double)
28493 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28495 * config/rs6000/altivec.h: Add define for vec_neg
28496 * doc/extend.texi: Update the built-in documentation for the
28497 new built-in functions.
28499 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28501 PR tree-optimization/77644
28502 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28504 2017-05-10 Nathan Sidwell <nathan@acm.org>
28506 * dumpfile.h (TDI_lang_all): New.
28507 (TDF_KIND): New. Renumber others
28508 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28510 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28512 (get_dump_file_name): Adjust suffix generation.
28513 (dump_enable_all): Use TDF_KIND.
28514 * doc/invoke.texi (-fdump-lang-all): Document.
28516 * dumpfile.h: Tabify.
28518 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28521 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28522 Move member access before delete.
28524 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28526 * tree-inline.c (expand_call_inline): Split block at stmt
28529 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28532 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28533 are now unused after splitting mov{sf,sd}_hardfloat.
28534 (f32_lr2): Likewise.
28535 (f32_lm): Likewise.
28536 (f32_lm2): Likewise.
28537 (f32_li): Likewise.
28538 (f32_li2): Likewise.
28539 (f32_lv): Likewise.
28540 (f32_sr): Likewise.
28541 (f32_sr2): Likewise.
28542 (f32_sm): Likewise.
28543 (f32_sm2): Likewise.
28544 (f32_si): Likewise.
28545 (f32_si2): Likewise.
28546 (f32_sv): Likewise.
28547 (f32_dm): Likewise.
28548 (f32_vsx): Likewise.
28549 (f32_av): Likewise.
28550 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28551 For movsf, order stores so the VSX stores occur before the GPR
28552 store which encourages the register allocator to use a traditional
28553 FPR instead of a GPR. For movsd, order the stores so that the GPR
28554 store comes before the VSX stores to allow the power6 to work.
28555 This is due to the power6 not having a 32-bit integer store
28556 instruction from a FPR.
28557 (movsf_hardfloat): Likewise.
28558 (movsd_hardfloat): Likewise.
28560 2017-05-09 Martin Sebor <msebor@redhat.com>
28562 PR translation/80280
28563 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28566 PR translation/80280
28567 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28568 data member added in r247778.
28569 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28571 2017-05-09 Nathan Sidwell <nathan@acm.org>
28573 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28575 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28578 2017-05-09 Marek Polacek <polacek@redhat.com>
28580 * doc/invoke.texi: Fix typo.
28582 2017-05-09 Richard Biener <rguenther@suse.de>
28584 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28585 (vrp_val_is_min): Likewise.
28586 (set_value_range_to_value): Likewise.
28587 (set_value_range_to_nonnegative): Likewise.
28588 (gimple_assign_nonzero_p): Likewise.
28589 (gimple_stmt_nonzero_p): Likewise.
28590 (vrp_int_const_binop): Likewise. Remove unreachable case.
28591 (adjust_range_with_scev): Adjust comments.
28592 (compare_range_with_value): Likewise.
28593 (extract_range_from_phi_node): Likewise.
28594 (test_for_singularity): Likewise.
28596 2017-05-09 Richard Biener <rguenther@suse.de>
28598 * tree-vrp.c (get_single_symbol): Add assert that we don't
28599 get overflowed constants as invariant part.
28600 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28601 checks. Use wi::cmp instead of recursing for integer constants.
28602 (compare_values): Just ignore whether we assumed undefined
28603 overflow instead of failing the compare.
28604 (extract_range_for_var_from_comparison_expr): Add comment before the
28605 TREE_NO_WARNING sets.
28606 (test_for_singularity): Likewise.
28607 (extract_range_from_comparison): Do not disable optimization
28608 when we assumed undefined overflow.
28609 (extract_range_basic): Remove init of unused var.
28611 2017-05-09 Richard Biener <rguenther@suse.de>
28613 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28614 (extract_range_from_multiplicative_op_1): Adjust.
28615 (extract_range_from_binary_expr_1): Use int_const_binop.
28617 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28620 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28621 rs6000_store_data_bypass_p in seven define_bypass directives and
28622 in several comments.
28623 * config/rs6000/rs6000-protos.h: Add prototype for
28624 rs6000_store_data_bypass_p function.
28625 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28626 function implements slightly different (rs6000-specific) semantics
28627 than store_data_bypass_p, returning false rather than aborting
28628 with assertion error when arguments do not satisfy the
28629 requirements of store data bypass.
28630 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28631 rs6000_store_data_bypass_p.
28633 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28635 * config/xtensa/xtensa-protos.h
28636 (xtensa_initial_elimination_offset): New declaration.
28637 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28638 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28639 macro definition, add case for FRAME_POINTER_REGNUM when
28640 FRAME_GROWS_DOWNWARD.
28641 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28642 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28643 xtensa_initial_elimination_offset.
28645 2017-05-08 Nathan Sidwell <nathan@acm.org>
28647 * doc/invoke.texi: Alphabetize -fdump options.
28649 2017-05-08 Martin Sebor <msebor@redhat.com>
28651 PR translation/80280
28652 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28654 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28656 * target.def (compute_frame_layout): New optional target hook.
28657 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28658 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28659 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28661 * reload1.c (verify_initial_elim_offsets): Likewise.
28662 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28663 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28664 (arm_get_frame_offsets): Split up into this ...
28665 (arm_compute_frame_layout): ... and this function.
28667 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28669 * config/aarch64/constraints.md (Usa): New constraint.
28670 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28672 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28674 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28675 with_multilib_list after it has been checked.
28677 2017-05-08 Richard Biener <rguenther@suse.de>
28679 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28680 (bitmap_set_subtract_values): Likewise.
28682 2017-05-08 Richard Biener <rguenther@suse.de>
28684 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28685 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28687 (gimple_stmt_nonzero_warnv_p): Rename to ...
28688 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28690 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28691 (extract_range_basic): Adjust, do not disable propagation on
28692 strict overflow sensitive simplification.
28693 (vrp_visit_cond_stmt): Likewise.
28695 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28697 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28698 body size unconditionally.
28700 2017-05-07 Jeff Law <law@redhat.com>
28703 2017-05-06 Jeff Law <law@redhat.com>
28704 PR tree-optimization/78496
28705 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28708 PR tree-optimization/78496
28709 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28710 (simplify_stmt_using_ranges): Call it.
28711 (vrp_dom_walker::before_dom_children): Extract equivalences
28712 from an ASSERT_EXPR with an equality comparison against a
28715 2017-05-06 Jeff Law <law@redhat.com>
28717 PR tree-optimization/78496
28718 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28721 PR tree-optimization/78496
28722 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28723 (simplify_stmt_using_ranges): Call it.
28724 (vrp_dom_walker::before_dom_children): Extract equivalences
28725 from an ASSERT_EXPR with an equality comparison against a
28728 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28730 * lra-constraints.c (lra_copy_reg_equiv): New function.
28731 (split_reg): Use it to copy equivalence information from the
28732 original register to the spill register.
28734 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28736 PR rtl-optimization/75964
28737 * simplify-rtx.c (simplify_const_relational_operation): Remove
28738 invalid handling of comparisons of integer ABS.
28740 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28742 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28743 initialize to zero.
28744 (init_regs): Remove declaration.
28745 (function_arg_advance_32): Initialize error_p as boolean variable.
28747 2017-05-05 Nathan Sidwell <nathan@acm.org>
28749 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28750 lines. Use for (;;).
28752 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28754 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28755 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28756 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28757 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28758 VF=2 that require versioning.
28760 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28762 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28765 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28767 * diagnostic.h (diagnostic_override_option_index): Convert from
28768 macro to inline function.
28770 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28772 * diagnostic.c (last_module_changed_p): New function.
28773 (set_last_module): New function.
28774 (diagnostic_report_current_module): Convert macro usage to
28775 the above functions.
28776 * diagnostic.h (diagnostic_context::last_module): Strengthen
28777 from const line_map * to const line_map_ordinary *.
28778 (diagnostic_last_module_changed): Delete macro.
28779 (diagnostic_set_last_module): Delete macro.
28781 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28783 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28784 with diagnostic_report_diagnostic.
28785 (diagnostic_n_impl_richloc): Likewise.
28786 * diagnostic.h (report_diagnostic): Delete macro.
28787 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28788 with diagnostic_report_diagnostic.
28789 * substring-locations.c (format_warning_va): Likewise.
28791 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28793 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28794 save/restor of format_spec. Move option-printing code to...
28795 (print_option_information): ...this new function, and
28796 reimplement by simply printing to the pretty_printer,
28797 rather than appending to the format string.
28799 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28801 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28802 handling logic into...
28803 (update_effective_level_from_pragmas): ...this new function.
28805 2017-05-04 Andrew Waterman <andrew@sifive.com>
28807 * config/riscv/riscv.opt (mstrict-align): New option.
28808 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28809 (SLOW_UNALIGNED_ACCESS): Define.
28810 (riscv_slow_unaligned_access): Declare.
28811 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28813 (riscv_slow_unaligned_access): New variable.
28814 (rocket_tune_info): Set slow_unaligned_access to true.
28815 (optimize_size_tune_info): Set slow_unaligned_access to false.
28816 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
28817 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
28818 (riscv_option_override): Set riscv_slow_unaligned_access.
28819 * doc/invoke.texi: Add -mstrict-align to RISC-V.
28821 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
28823 * config/riscv/riscv.md: Unify indentation.
28825 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28830 * config/rs6000/rs6000.md (u code attribute): Add FIX and
28832 (extendsi<mode>2): Add support for doing sign extension via
28833 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
28834 don't have ISA 3.0 instructions.
28835 (extendsi<mode>2 splitter): Likewise.
28836 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
28837 generate the normal insns since SImode can now go in vector
28838 registers. Disallow the special UNSPECs needed for previous
28839 machines to hide SImode being used. Add new insns
28840 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
28841 (fix_trunc<mode>si2_stfiwx): Likewise.
28842 (fix_trunc<mode>si2_internal): Likewise.
28843 (fixuns_trunc<mode>si2): Likewise.
28844 (fixuns_trunc<mode>si2_stfiwx): Likewise.
28845 (fctiw<u>z_<mode>_smallint): Likewise.
28846 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
28847 of floating point to 32-bit integer from doing a direct move to
28848 the GPR registers to do a store.
28849 (fctiwz_<mode>): Break long line.
28851 2017-05-05 Bin Cheng <bin.cheng@arm.com>
28853 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
28854 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
28855 (addr_list, addr_offset_valid_p): New.
28856 (split_address_groups): Check offset validity with above function.
28857 (gt-tree-ssa-loop-ivopts.h): Include header file.
28859 2017-05-05 Nathan Sidwell <nathan@acm.org>
28861 * config.gcc (arm*-*-*): Add missing 'fi'.
28863 2017-05-05 Steve Ellcey <sellcey@cavium.com>
28865 * doc/invoke.texi (-fopt-info): Explicitly say order of options
28866 included in -fopt-info does not matter.
28867 * doc/optinfo.texi (-fopt-info): Fix description of default
28868 behavour. Explicitly say order of options included in -fopt-info
28871 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28873 * config.gcc: Allow combinations of aprofile and rmprofile values for
28874 --with-multilib-list.
28875 * config/arm/t-multilib: New file.
28876 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
28877 variables. Remove setting of ISA and floating-point ABI in
28878 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
28879 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
28880 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
28881 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
28883 * config/arm/t-rmprofile: Likewise except for the matches changes.
28884 * doc/install.texi (--with-multilib-list): Document the combination of
28885 aprofile and rmprofile values and warn about pitfalls in doing that.
28887 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28889 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
28890 (movdi_aarch64): Likewise.
28892 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28894 PR tree-optimization/80632
28895 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
28897 (build_arrays): Initialize it for virtual phis.
28898 (fix_phi_nodes): Use it for virtual phis.
28900 PR tree-optimization/80558
28901 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
28902 [x, y] op z into [x op, y op z] for op & or | if conditions
28905 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
28906 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28909 * config/arm/arm.md (use_literal_pool): Remove.
28910 (64-bit immediate split): No longer takes cost into consideration
28911 if arm_disable_literal_pool is enabled.
28912 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
28913 used when arm_disable_literal_pool is enabled.
28914 (arm_max_const_double_inline_cost): Remove use of
28915 arm_disable_literal_pool.
28916 (push_minipool_fix): Add assert.
28917 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
28918 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
28919 (no_literal_pool_sf_immediate): New.
28921 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28923 PR tree-optimization/80613
28924 * tree-ssa-dce.c (propagate_necessity): Remove cases for
28925 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
28927 2017-05-05 Richard Biener <rguenther@suse.de>
28929 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
28931 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
28933 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
28934 of this flag from insn conditions due to removal from r247495.
28936 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28938 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
28940 (arm_early_store_addr_dep_ptr): Likewise.
28941 * config/arm/aarch-common-protos.h
28942 (arm_early_load_addr_dep_ptr): Add prototype.
28943 (arm_early_store_addr_dep_ptr): Likewise.
28944 * config/arm/cortex-a53.md: Add new bypasses.
28946 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28948 * tree.c (next_type_uid): Change type to unsigned.
28949 (type_hash_canon): Decrement back next_type_uid if
28950 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
28951 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
28954 2017-05-04 Martin Sebor <msebor@redhat.com>
28956 * builtins.c: Fix a trivial typo in a comment.
28958 PR middle-end/79234
28959 * builtins.c (check_sizes): Adjust to handle reading past the end.
28960 Avoid printing excessive upper bound of ranges. Use %E to print
28961 tree nodes instead of converting them to %wu.
28962 (expand_builtin_memchr): New function.
28963 (compute_dest_size): Rename...
28964 (compute_objsize): ...to this.
28965 (expand_builtin_memcpy): Adjust.
28966 (expand_builtin_mempcpy): Adjust.
28967 (expand_builtin_strcat): Adjust.
28968 (expand_builtin_strcpy): Adjust.
28969 (check_strncat_sizes): Adjust.
28970 (expand_builtin_strncat): Adjust.
28971 (expand_builtin_strncpy): Adjust and simplify.
28972 (expand_builtin_memset): Adjust.
28973 (expand_builtin_bzero): Adjust.
28974 (expand_builtin_memcmp): Adjust.
28975 (expand_builtin): Handle memcmp.
28976 (maybe_emit_chk_warning): Check strncat just once.
28978 2017-05-04 Martin Sebor <msebor@redhat.com>
28980 PR preprocessor/79214
28981 PR middle-end/79222
28982 PR middle-end/79223
28983 * builtins.c (check_sizes): Add inlining context and issue
28984 warnings even when -Wno-system-headers is set.
28985 (check_strncat_sizes): Same.
28986 (expand_builtin_strncat): Same.
28987 (expand_builtin_memmove): New function.
28988 (expand_builtin_stpncpy): Same.
28989 (expand_builtin): Handle memmove and stpncpy.
28991 2017-05-04 Bin Cheng <bin.cheng@arm.com>
28993 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
28994 which is not used any more.
28996 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28998 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29000 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29002 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29003 (cortexa53_tunings): Likewise.
29004 (cortexa57_tunings): Likewise.
29005 (cortexa72_tunings): Likewise.
29006 (cortexa73_tunings): Likewise.
29008 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29010 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29011 Set loop alignment to 8.
29013 2017-05-04 Martin Sebor <msebor@redhat.com>
29015 PR translation/80280
29016 * builtins.c (expand_builtin_object_size): Add missing quoting to
29017 %D and like directives.
29018 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29019 (hsa_type_for_tree_type): Same.
29020 (verify_function_arguments): Same.
29021 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29022 * varasm.c (get_section): Same.
29025 2017-05-04 Martin Sebor <msebor@redhat.com>
29027 PR translation/80280
29028 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29030 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29032 * config/aarch64/aarch64.c (generic_addrcost_table):
29033 Change HI/TI mode setting.
29035 2017-05-04 Martin Jambor <mjambor@suse.cz>
29037 PR tree-optimization/80622
29038 * tree-sra.c (comes_initialized_p): New function.
29039 (build_accesses_from_assign): Only set write lazily when
29040 comes_initialized_p is false.
29041 (analyze_access_subtree): Use comes_initialized_p.
29042 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29043 instead of testing for PARM_DECL.
29045 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29047 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29048 constraint on operand 0 to allow more general addressing modes.
29049 Adjust output template.
29050 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29052 * config/aarch64/aarch64-protos.h
29053 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29054 * config/aarch64/constraints.md (Dp): New address constraint.
29055 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29058 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29060 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29061 update use of estimate_ipcp_clone_size_and_time.
29062 (estimate_local_effects): Update use of
29063 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29064 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29065 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29066 Return nonspecialized time.
29068 2017-05-04 Richard Biener <rguenther@suse.de>
29070 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29071 for the last VUSE which def dominates the PHI. Directly call
29073 (get_continuation_for_phi_1): Remove.
29075 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29077 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29078 to explain the use of truncating division. Cap the number of
29079 iterations to the maximum given by nb_iterations_upper_bound,
29082 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29084 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29085 * configure: Regenerate.
29086 * config.in: Regenerate.
29087 * config/i386/driver-mingw32.c: new file.
29088 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29089 * config.host: Link driver-mingw32.o on MinGW host.
29090 * doc/install.texi: Document new --enable-mingw-wildcard configure
29093 2017-05-04 Marek Polacek <polacek@redhat.com>
29095 PR tree-optimization/80612
29096 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29098 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29099 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29101 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29102 (movt splitter): Likewise.
29103 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29104 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29105 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29106 block for Thumb-1 with MOVT.
29107 (thumb2_legitimate_address_p): Move code block ...
29108 (can_avoid_literal_pool_for_label_p): ... into this new function.
29109 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29111 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29112 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29113 "M-profile targets with the MOVT instruction".
29115 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29117 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29118 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29119 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29121 2017-05-04 Martin Liska <mliska@suse.cz>
29123 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29124 variable cond_code.
29126 2017-05-04 Richard Biener <rguenther@suse.de>
29128 * tree.c (array_at_struct_end_p): Handle arrays at struct
29129 end with flexarrays more conservatively. Refactor and treat
29130 arrays of arrays or aggregates more strict. Fix
29131 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29132 * tree.c (array_at_struct_end_p): Adjust prototype.
29133 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29134 * gimple-fold.c (get_range_strlen): Likewise.
29135 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29137 2017-05-04 Richard Biener <rguenther@suse.de>
29139 PR tree-optimization/31130
29140 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29142 (supports_overflow_infinity): Likewise.
29143 (is_negative_overflow_infinity): Likewise.
29144 (is_positive_overflow_infinity): Likewise.
29145 (is_overflow_infinity): Likewise.
29146 (stmt_overflow_infinity): Likewise.
29147 (overflow_infinity_range_p): Likewise.
29148 (usable_range_p): Remove as always returning true.
29149 (make_overflow_infinity): Remove.
29150 (negative_overflow_infinity): Likewise.
29151 (positive_overflow_infinity): Likewise.
29152 (avoid_overflow_infinity): Likewise.
29153 (set_value_range): Adjust accordingly.
29154 (set_value_range_to_nonnegative): Likewise, remove now unused
29155 overflow_infinity arg.
29156 (vrp_operand_equal_p): Adjust.
29157 (update_value_range): Likewise.
29158 (range_int_cst_singleton_p): Likewise.
29159 (operand_less_p): Likewise.
29160 (compare_values_warnv): Likewise.
29161 (extract_range_for_var_from_comparison_expr): Likewise.
29162 (vrp_int_const_binop): Likewise.
29163 (zero_nonzero_bits_from_vr): Likewise.
29164 (extract_range_from_multiplicative_op_1): Likewise.
29165 (extract_range_from_binary_expr_1): Likewise.
29166 (extract_range_from_unary_expr): Likewise.
29167 (extract_range_from_comparison): Likewise.
29168 (extract_range_basic): Likewise.
29169 (adjust_range_with_scev): Likewise.
29170 (compare_ranges): Likewise.
29171 (compare_range_with_value): Likewise.
29172 (dump_value_range): Likewise.
29173 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29175 (simplify_cond_using_ranges): Adjust.
29177 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29179 * brig-builtins.def: Added a builtin for class_f64.
29180 * builtin-types.def: Added a builtin type needed by class_f64.
29182 2017-05-03 Jason Merrill <jason@redhat.com>
29184 * timevar.def: Add TV_CONSTEXPR.
29186 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29188 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29190 2017-05-03 Martin Jambor <mjambor@suse.cz>
29192 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29193 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29194 * ipa-cp.c (ipcp_propagate_stage): Do not call
29195 ipa_update_after_lto_read.
29196 * ipa-inline.c (ipa_inline): Likewise.
29198 2017-05-03 Martin Jambor <mjambor@suse.cz>
29200 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29201 tag. Added a default constructor and a destructor.
29202 (ipa_edge_args_sum_t): New class;
29203 (ipa_edge_args_sum): Declare.
29204 (ipa_edge_args_vector): Remove declaration.
29205 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29206 (ipa_free_edge_args_substructures): Remove declaration.
29207 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29208 (ipa_edge_args_info_available_for_edge_p): Likewise.
29209 * ipa-prop.c (ipa_edge_args_vector): Removed.
29210 (edge_removal_hook_holder): Likewise.
29211 (edge_duplication_hook_holder): Likewise.
29212 (ipa_edge_args_sum): New variable.
29213 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29214 ipa_edge_args_vector.
29215 (ipa_free_edge_args_substructures): Likewise.
29216 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29217 ipa_edge_args_vector.
29218 (ipa_edge_removal_hook): Turned into method
29219 ipa_edge_args_sum_t::remove.
29220 (ipa_edge_duplication_hook): Turned into method
29221 ipa_edge_args_sum_t::duplicate.
29222 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29223 registering edge hooks.
29224 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29225 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29226 ipa_edge_args_sum instead of ipa_edge_args_vector.
29227 * ipa-profile.c (ipa_profile): Likewise.
29229 2017-05-03 Martin Jambor <mjambor@suse.cz>
29231 * symbol-summary.h (function_summary): New method exists.
29232 (function_summary::symtab_removal): Deallocate through release.
29233 (call_summary): New class.
29234 (gt_ggc_mx): New overload.
29235 (gt_pch_nx): Likewise.
29236 (gt_pch_nx): Likewise.
29238 2017-05-03 Jeff Law <law@redhat.com>
29240 PR tree-optimization/78496
29241 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29242 from simplify_cond_using_ranges. Split off code to walk
29243 backwards through casts into ...
29244 (simplify_cond_using_ranges_2): New function.
29245 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29246 (execute_vrp): After identifying jump threads, call
29247 simplify_cond_using_ranges_2.
29249 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29252 * ipa-inline.h (inline_summary): Add ctor.
29253 (create_ggc): Do not use ggc_cleared_alloc.
29255 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29256 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29258 * gcc.c (handle_braces): Support escaping in switch matching
29260 * doc/invoke.texi (Spec Files): Document it.
29261 Remove superfluous @code markup in items.
29263 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29265 * diagnostic-show-locus.c (struct column_range): New struct.
29266 (get_affected_columns): New function.
29267 (get_printed_columns): New function.
29268 (struct correction): New struct.
29269 (correction::ensure_capacity): New function.
29270 (correction::ensure_terminated): New function.
29271 (struct line_corrections): New struct.
29272 (line_corrections::~line_corrections): New dtor.
29273 (line_corrections::add_hint): New function.
29274 (layout::print_trailing_fixits): Reimplement in terms of the new
29276 (selftest::test_overlapped_fixit_printing): New function.
29277 (selftest::diagnostic_show_locus_c_tests): Call it.
29279 2017-05-03 Nathan Sidwell <nathan@acm.org>
29281 Canonicalize canonical type hashing
29282 * tree.h (type_hash_canon_hash): Declare.
29283 * tree.c (type_hash_list, attribute_hash_list): Move into
29284 type_hash_canon_hash.
29285 (build_type_attribute_qual_variant): Break out hash code calc into
29286 type_hash_canon_hash.
29287 (type_hash_canon_hash): New. Generic type hash computation.
29288 (build_range_type_1, build_array_type_1, build_function_type,
29289 build_method_type_directly, build_offset_type, build_complex_type,
29290 make_vector_type): Call it.
29292 2017-05-03 Richard Biener <rguenther@suse.de>
29294 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29295 When all DRs have unknown misaligned do not always peel
29296 when there is a store but apply the same costing model as if
29297 there were only loads.
29299 2017-05-03 Richard Biener <rguenther@suse.de>
29302 PR tree-optimization/80492
29303 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29304 compare_base_decls returning dont-know properly.
29306 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29308 * config/arm/iterators.md (CCSI): New mode iterator.
29309 (arch): New mode attribute.
29310 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29311 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29312 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29313 code iterator for success result mode.
29314 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29315 the corresponding new insn generators.
29317 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29320 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29321 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29323 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29325 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29326 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29327 (DDR_A): Wrap DDR argument in brackets.
29328 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29329 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29330 (DDR_REVERSED_P): Likewise.
29332 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29334 PR tree-optimization/79472
29335 * tree-switch-conversion.c (struct switch_conv_info): Add
29336 contiguous_range and default_case_nonstandard fields.
29337 (collect_switch_conv_info): Compute contiguous_range and
29338 default_case_nonstandard fields, don't clear final_bb if
29339 contiguous_range and only the default case doesn't have the required
29341 (check_all_empty_except_final): Set default_case_nonstandard instead
29342 of failing if contiguous_range and the default case doesn't have empty
29344 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29345 and only the default case doesn't have the required constants. Skip
29347 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29348 if default_case_nonstandard.
29349 (build_constructors): Build constant 1 just once. Assert that default
29350 values aren't inserted in between cases if contiguous_range. Skip
29352 (build_arrays): Skip virtual phis.
29353 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29354 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29355 Handle virtual phis.
29356 (gen_inbound_check): Handle default_case_nonstandard case.
29357 (process_switch): Adjust check_final_bb caller. Call
29358 gather_default_values with the first non-default case instead of
29359 default case if default_case_nonstandard.
29361 2017-05-02 Nathan Sidwell <nathan@acm.org>
29363 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29364 check. Fix formatting.
29366 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29368 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29369 errors when comparing specialized and unspecialized times.
29371 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29373 * diagnostic-show-locus.c
29374 (layout::should_print_annotation_line_p): Make private.
29375 (layout::print_annotation_line): Make private.
29376 (layout::annotation_line_showed_range_p): Make private.
29377 (layout::show_ruler): Make private.
29378 (layout::print_source_line): Make private. Pass in line and
29379 line_width, rather than calling location_get_source_line. Drop
29381 (layout::print_leading_fixits): New method.
29382 (layout::print_any_fixits): Rename to...
29383 (layout::print_trailing_fixits): ...this, and make private.
29384 Don't print newline fixits.
29385 (diagnostic_show_locus): Move logic for printing one row into...
29386 (layout::print_line): ...this new function. Move the
29387 location_get_source_line call and error-handling from
29388 print_source_line to here. Call print_leading_fixits, and rename
29389 print_any_fixits to print_trailing_fixits.
29390 (selftest::test_fixit_insert_containing_newline): Update now that
29391 newlines are partially supported.
29392 (selftest::test_fixit_insert_containing_newline_2): New test.
29393 (selftest::test_fixit_replace_containing_newline): Update comments.
29394 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29395 * edit-context.c (class added_line): New class.
29396 (class edited_line): Describe newline handling in comment.
29397 (edited_line::actually_edited_p): New method.
29398 (edited_line::print_content): Delete redundant decl.
29399 (edited_line::m_predecessors): New field.
29400 (edited_file::print_content): Call edited_line::print_content.
29401 (edited_file::print_diff): Update to support newlines.
29402 (edited_file::print_diff_hunk): Likewise.
29403 (edited_file::print_run_of_changed_lines): New function.
29404 (edited_file::print_diff_line): Convert to...
29405 (print_diff_line): ...this.
29406 (edited_file::get_effective_line_count): New function.
29407 (edited_line::edited_line): Initialize new field m_predecessors.
29408 (edited_line::~edited_line): Clean up m_predecessors.
29409 (edited_line::apply_fixit): Handle newlines.
29410 (edited_line::get_effective_line_count): New function.
29411 (edited_line::print_content): New function.
29412 (edited_line::print_diff_lines): New function.
29413 (selftest::test_applying_fixits_insert_containing_newline): New
29415 (selftest::test_applying_fixits_replace_containing_newline): New
29417 (selftest::insert_line): New function.
29418 (selftest::test_applying_fixits_multiple_lines): Add example of
29420 (selftest::edit_context_c_tests): Call the new tests.
29422 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29424 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29425 parameter cand. Update dump information.
29426 (get_computation_cost): Update uses.
29428 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29430 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29431 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29432 (get_computation_at, rewrite_use_address): Update use of
29433 get_computation_aff.
29435 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29437 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29438 (get_computation): Delete.
29439 (get_computation_cost): Implement like get_computation_cost_at.
29440 Use get_computation_at.
29441 (get_computation_cost_at): Delete.
29442 (rewrite_use_nonlinear_expr): Use get_computation_at.
29443 (rewrite_use_compare, remove_unused_ivs): Ditto.
29445 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29447 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29449 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29451 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29452 (ivopts_global_cost_for_size): Rename parameter and update uses.
29453 (iv_ca_recount_cost): Update uses.
29454 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29455 candidates seperately in n_invs and n_cands.
29456 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29458 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29460 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29461 (find_inv_vars_cb): New.
29462 (find_depends): Renamed to ...
29463 (find_inv_vars): ... this.
29464 (add_candidate_1, force_var_cost): Call find_inv_vars.
29465 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29467 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29469 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29470 inv_vars. Add inv_exprs.
29471 (struct iv_cand): Rename depends_on to inv_vars.
29472 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29473 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29474 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29475 (dump_cand): Dump inv_vars.
29476 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29477 (record_invariant, find_depends, add_candidate_1): Ditto.
29478 (set_group_iv_cost, force_var_cost): Ditto.
29479 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29480 (get_computation_cost_at, get_computation_cost): Ditto.
29481 (determine_group_iv_cost_generic): Ditto.
29482 (determine_group_iv_cost_address): Ditto.
29483 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29484 (determine_group_iv_costs): Ditto.
29485 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29486 (iv_ca_set_remove_invariants): Renamed to ...
29487 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29488 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29489 (iv_ca_set_add_invariants): Renamed to ...
29490 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29491 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29492 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29493 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29494 (create_new_ivs): Remove useless dump.
29496 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29498 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29500 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29501 (iv_ca_set_no_cp, create_new_iv): Ditto.
29503 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29505 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29507 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29509 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29510 function tree_check2.
29512 2017-05-02 Martin Liska <mliska@suse.cz>
29514 * doc/gcov.texi: Add missing preposition.
29515 * gcov.c (function_info::function_info): Properly fill up
29516 all member variables.
29518 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29520 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29522 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29524 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29526 2017-05-02 Martin Liska <mliska@suse.cz>
29529 * lto-streamer-in.c (lto_read_tree_1): Remove
29530 LTO_STREAMER_DEBUG.
29531 * lto-streamer.c (struct tree_hash_entry): Likewise.
29532 (struct tree_entry_hasher): Likewise.
29533 (tree_entry_hasher::hash): Likewise.
29534 (tree_entry_hasher::equal): Likewise.
29535 (lto_streamer_init): Likewise.
29536 (lto_orig_address_map): Likewise.
29537 (lto_orig_address_get): Likewise.
29538 (lto_orig_address_remove): Likewise.
29539 * lto-streamer.h: Likewise.
29540 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29541 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29543 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29545 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29546 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29547 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29548 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29549 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29550 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29551 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29552 (mm_maskz_sub_ss): New intrinsics.
29553 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29554 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29555 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29556 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29557 (__builtin_ia32_subss_mask_round): New builtins.
29558 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29559 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29560 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29562 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29563 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29565 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29568 2017-05-02 Martin Jambor <mjambor@suse.cz>
29570 PR tree-optimization/78687
29571 * tree-sra.c (access): New field parent.
29572 (process_subtree_disqualification): New function.
29573 (disqualify_candidate): Call it.
29574 (build_accesses_from_assign): Reset write flag if creating an
29576 (build_access_subtree): Fill in parent field and also prpagate
29577 down grp_write flag.
29578 (create_artificial_child_access): New parameter set_grp_write, set
29579 grp_write to its value.
29580 (propagate_subaccesses_across_link): Also propagate grp_write flag
29582 (propagate_all_subaccesses): Push the closest parent back to work
29583 queue if add_access_to_work_queue returned true.
29585 2017-05-02 Richard Biener <rguenther@suse.de>
29587 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29588 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29589 -fstrict-overflow documentation.
29590 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29591 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29592 flag_strict_overflow.
29593 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29594 * lto-opts.c (lto_write_options): Do not stream it.
29595 * lto-wrapper.c (merge_and_complain): Do not handle it.
29596 * opts.c (default_options_table): Do not set -fstrict-overflow.
29597 (finish_options): Likewise do not clear it when sanitizing.
29598 * simplify-rtx.c (simplify_const_relational_operation): Do not
29599 test flag_strict_overflow.
29601 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29603 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29604 using enabled attribute.
29605 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29606 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29607 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29608 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29609 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29610 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29611 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29612 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29613 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29614 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29616 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29618 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29620 2017-05-02 Richard Biener <rguenther@suse.de>
29622 PR tree-optimization/80591
29624 2017-04-10 Richard Biener <rguenther@suse.de>
29626 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29629 2017-05-02 Richard Biener <rguenther@suse.de>
29631 PR tree-optimization/80549
29632 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29633 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29634 headers if they do not have a preheader.
29636 2017-05-02 Martin Liska <mliska@suse.cz>
29639 * common.opt: Fix typo.
29640 * doc/invoke.texi: Likewise.
29642 2017-05-01 Jan Beulich <jbeulich@suse.com>
29644 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29645 swapping, add (x,x,m,x,n) alternative.
29647 2017-05-01 Nathan Sidwell <nathan@acm.org>
29649 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29650 unnecessary unadjusted_alignment check.
29652 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29655 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29657 * gimplify.c (gimplify_cilk_detach): New function.
29658 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29659 * tree-core.h: Document EXPR_CILK_SPAWN.
29660 * tree.h (EXPR_CILK_SPAWN): Define.
29662 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29664 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29665 to use new fixit_hint representation, using the "replace" logic.
29666 (get_line_span_for_fixit_hint): Likewise.
29667 (layout::print_any_fixits): Likewise.
29668 (selftest::test_one_liner_many_fixits): Rename to...
29669 (selftest::test_one_liner_many_fixits_1): ...this, and update
29670 comment and expected output to reflect that the multiple fix-it
29671 hints are now consolidated into one insertion.
29672 (selftest::test_one_liner_many_fixits_2): New test.
29673 (selftest::test_diagnostic_show_locus_one_liner): Update for
29675 (selftest::test_fixit_consolidation): Update for fix-it API
29677 * diagnostic.c (print_parseable_fixits): Likewise.
29678 * edit-context.c (edited_line::m_line_events): Convert from
29679 auto_vec <line_event *> to auto_vec <line_event>.
29680 (class line_event): Convert from abstract base class to a concrete
29681 class, taking over the role of replace_event.
29682 (class insert_event): Delete.
29683 (class replace_event): Rename to class line_event. Convert to
29685 (edit_context::add_fixits): Reimplement.
29686 (edit_context::apply_insert): Delete.
29687 (edit_context::apply_replace): Rename to...
29688 (edit_context::apply_fixit): ...this. Convert to half-open range.
29689 (edited_file::apply_insert): Delete.
29690 (edited_file::apply_replace): Rename to...
29691 (edited_file::apply_fixit): ...this.
29692 (edited_line::~edited_line): Drop deletion of events.
29693 (edited_line::apply_insert): Delete.
29694 (edited_line::apply_replace): Rename to...
29695 (edited_line::apply_fixit): ...this. Convert to half-open range.
29696 Update for change to type of m_line_events.
29697 * edit-context.h (edit_context::apply_insert): Delete.
29698 (edit_context::apply_replace): Rename to...
29699 (edit_context::apply_fixit): ...this.
29701 2017-05-01 Martin Sebor <msebor@redhat.com>
29703 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29706 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29709 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29710 __get_cpuid_max returns 0.
29711 (__get_cpuid_count): Ditto.
29713 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29715 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29716 replacement expression is another instance of one of its arguments.
29718 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29721 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29722 check for stack push/pop autoinc.
29723 * config/i386/i386.c (ix86_agi_dependent): Return false
29724 if the only reason why modified_in_p returned true is that
29725 addr is SP based and set_insn is a push or pop.
29727 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29729 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29732 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29735 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29736 (account_size_time): Use two predicates - exec_pred and
29738 (evaluate_conditions_for_known_args): Compute both clause and
29740 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29741 (inline_summary_t::duplicate): Update.
29742 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29744 (compute_inline_parameters): Likewise.
29745 (estimate_edge_size_and_time): Update caluclation of time.
29746 (estimate_node_size_and_time): Compute both time and nonspecialized
29748 (estimate_ipcp_clone_size_and_time): Update.
29749 (inline_merge_summary): Update.
29750 (do_estimate_edge_time): Update.
29751 (do_estimate_edge_size): Update.
29752 (do_estimate_edge_hints): Update.
29753 (inline_read_section, inline_write_summary): Stream both new predicates.
29754 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29756 (compute_inlined_call_time): Cleanup.
29757 (big_speedup_p): Update.
29758 (edge_badness): Update.
29759 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29760 (size_time_entry): Replace predicate by exec_predicate and
29761 nonconst_predicate.
29762 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29763 (estimate_edge_time): Return also nonspec_time.
29764 (reset_edge_growth_cache): Update.
29766 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29768 PR rtl-optimization/80491
29769 * ifcvt.c (noce_process_if_block): When looking for x setter
29770 with missing else_bb, don't check only the insn right before
29771 cond_earliest, but look for the last insn that x is modified in
29772 within the same bb.
29774 PR rtl-optimization/80491
29775 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29777 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29779 PR tree-optimization/80487
29780 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29782 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29784 PR tree-optimization/79697
29785 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29786 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29787 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29789 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29790 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29792 2017-04-28 Martin Sebor <msebor@redhat.com>
29794 PR tree-optimization/80523
29795 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29796 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29798 (maybe_warn, format_directive, parse_directive): Use new functions.
29799 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29801 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29803 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29805 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29807 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29808 target_header_dir): Set correctly.
29809 * configure: Regenerated.
29810 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29811 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29812 instead of SYSTEM_HEADER_DIR.
29814 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
29816 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
29817 (estimate_local_effects): Likewise.
29818 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
29819 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
29820 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
29821 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
29822 do_estimate_edge_time, estimate_edge_time): Likewise.
29823 * ipa-inline-analysis.c (estimate_node_size_and_time,
29824 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
29825 (estimate_time_after_inlining): Remove.
29827 2017-04-28 Martin Liska <mliska@suse.cz>
29829 * doc/gcov.texi: Enhance documentation of gcov.
29831 2017-04-28 Martin Liska <mliska@suse.cz>
29833 * doc/gcov.texi: Sort options in alphabetic order.
29834 * doc/gcov-dump.texi: Likewise.
29835 * doc/gcov-tool.texi: Likewise.
29836 * gcov.c (print_usage): Likewise.
29837 * gcov-dump.c (print_usage): Likewise.
29838 * gcov-tool.c (print_merge_usage_message): Likewise.
29839 (print_rewrite_usage_message): Likewise.
29840 (print_overlap_usage_message): Likewise.
29842 2017-04-28 Martin Liska <mliska@suse.cz>
29844 PR gcov-profile/53915
29845 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
29847 2017-04-28 Martin Liska <mliska@suse.cz>
29849 PR gcov-profile/79891
29850 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
29851 is marked by compiler as living on a line.
29852 (get_cycles_count): Remove usage of the union.
29853 (output_intermediate_file): Likewise.
29854 (find_source): Fix GNU coding style.
29855 (accumulate_line_counts): Remove old non-all block mode.
29856 (output_lines): Remove usage of the union.
29857 * profile.c (output_location): Include all BBs, even if
29858 belonging to a same line (and file) as a previous BB.
29860 2017-04-28 Martin Liska <mliska@suse.cz>
29862 * gcov.c (process_args): Handle new argument 'w'.
29863 (read_graph_file): Assign ID to BBs.
29864 (output_branch_count): Display BB # if verbose flag is set.
29865 (output_lines): Likewise for arcs.
29866 (print_usage): Add '--verbose' option help.
29867 * doc/gcov.texi: Document --verbose (-w) option.
29869 2017-04-28 Martin Liska <mliska@suse.cz>
29871 * gcov.c (struct block_location_info): New struct.
29872 (process_file): Fill up the new structure.
29873 (read_graph_file): Replace usage of encoding by the newly added
29875 (add_line_counts): Likewise.
29876 (accumulate_line_counts): Remove usage of the union.
29877 (function_info::function_info): New function.
29878 (function_info::~function_info): Likewise.
29879 (process_file): Call delete instead of release_function.
29880 (release_function): Release the function.
29881 (release_structures): Call delete instead of release_function.
29882 (solve_flow_graph): Replace usage of num_blocks.
29883 (find_exception_blocks): Likewise.
29884 (output_lines): Fix GNU coding style.
29886 2017-04-28 Martin Liska <mliska@suse.cz>
29889 * coverage.c (coverage_remove_note_file): New function.
29890 * coverage.h: Declare the function.
29891 * toplev.c (finalize): Clean if an error has been seen.
29893 2017-04-28 Martin Liska <mliska@suse.cz>
29895 PR gcov-profile/80031
29896 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
29897 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
29898 * gcov.c (read_graph_file): Read just number of blocks.
29899 * profile.c (branch_prob): Do not stream 0 flags per a basic
29902 2017-04-28 Martin Liska <mliska@suse.cz>
29904 * gcov-dump.c (tag_*): Add new argument to declarations.
29905 (dump_gcov_file): Likewise.
29906 (tag_blocks): Add and use new argument depth.
29907 (tag_arcs): Likewise.
29908 (tag_lines): Likewise.
29909 (tag_counters): Likewise.
29910 (tag_summary): Likewise.
29911 (dump_working_sets): Use depth to do a proper indentation.
29913 2017-04-28 Jakub Jelinek <jakub@redhat.com>
29916 * cgraph.h (symtab_node::debug_symtab): No longer inline.
29917 * symtab.c (symtab_node::debug_symtab): Move definition here.
29919 2017-04-28 Richard Biener <rguenther@suse.de>
29921 * lto-streamer.h (LTO_major_version): Bump to 7.
29923 2017-04-28 Richard Biener <rguenther@suse.de>
29925 * tree-vrp.c (assert_info): New struct.
29926 (add_assert_info): New helper.
29927 (register_edge_assert_for_2): Refactor to add asserts to a vector
29929 (register_edge_assert_for_1): Likewise.
29930 (register_edge_assert_for): Likewise.
29931 (finish_register_edge_assert_for): New helper actually registering
29932 asserts where live on edge.
29933 (find_conditional_asserts): Adjust.
29934 (find_switch_asserts): Likewise.
29935 (evrp_dom_walker::try_find_new_range): Generalize.
29936 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
29938 2017-04-27 Marek Polacek <polacek@redhat.com>
29941 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
29942 arg10 and arg11 to itype.
29944 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
29946 * doc/extend.texi (Object Size Checking): Improve grammar.
29948 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
29951 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
29952 that the logic for permitting reciprocal estimates matches that
29955 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29958 * tree.c (type_cache_hasher::equal): Only compare
29959 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
29960 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
29961 non-aggregate element types.
29962 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
29963 about the flag on ARRAY_TYPEs in the comment, formatting fix.
29965 2017-04-27 Richard Biener <rguenther@suse.de>
29967 PR middle-end/80533
29968 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
29969 stripping ARRAY_REFs from MEM_EXPR make sure we're not
29970 keeping a reference to a trailing array.
29972 2017-04-27 Richard Biener <rguenther@suse.de>
29974 PR middle-end/80539
29975 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
29976 being in loop-closed SSA form conservatively.
29977 (chrec_fold_multiply_poly_poly): Likewise.
29979 2017-04-27 Tamar Christina <tamar.christina@arm.com>
29981 PR middle-end/79665
29982 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
29983 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
29985 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29988 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
29989 (aarch64_function_arg_alignment): Return unsigned int again, but still
29990 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
29991 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29992 Don't emit -Wpsabi note.
29993 (aarch64_function_arg_boundary): Likewise.
29994 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29997 2017-04-26 Nathan Sidwell <nathan@acm.org>
29999 * tree.h (crc32_unsigned_n): Declare.
30000 (crc32_unsigned, crc32_unsigned): Make inline.
30001 * tree.c (crc32_unsigned_bits): Replace with ...
30002 (crc32_unsigned_n): ... this.
30003 (crc32_unsigned, crc32_byte): Remove.
30004 (crc32_string): Remove unnecessary braces.
30006 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30008 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30009 * ipa-inline-analysis.c (MAX_TIME): Remove.
30010 (account_size_time): Use sreal for time.
30011 (dump_inline_summary): Update.
30012 (estimate_function_body_sizes): Update.
30013 (estimate_edge_size_and_time): Update.
30014 (estimate_calls_size_and_time): Update.
30015 (estimate_node_size_and_time): Update.
30016 (inline_merge_summary): Update.
30017 (inline_update_overall_summary): Update.
30018 (estimate_time_after_inlining): Update.
30019 (inline_read_section): Update.
30020 (inline_write_summary): Update.
30021 * ipa-inline.c (compute_uninlined_call_time): Update.
30022 (compute_inlined_call_time): Update.
30023 (recursive_inlining): Update.
30024 (inline_small_functions): Update.
30025 (dump_overall_stats): Update.
30026 * ipa-inline.h: Include sreal.h.
30027 (size_time_entry): Turn time to sreal.
30028 (inline_summary): Turn self_time nad time to sreal.
30030 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30032 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30034 (sreal::stream_out, sreal::stream_in): New.
30035 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30037 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30039 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30042 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30045 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30047 (dimode_scalar_chain::compute_convert_gain): Ditto.
30048 (dimode_scalar_chain::make_vector_copies): Ditto.
30049 (dimode_scalar_chain::convert_reg): Ditto.
30050 (dimode_scalar_chain::convert_insn): Ditto.
30051 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30052 (VI248_AVX512BW_1): New mode iterator.
30053 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30054 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30057 2017-04-25 Martin Sebor <msebor@redhat.com>
30059 PR tree-optimization/80497
30060 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30061 constants are representable in HOST_WIDE_INT.
30062 (parse_directive): Ditto.
30064 2017-04-25 Martin Sebor <msebor@redhat.com>
30067 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30068 (new_zero_array): Adjust signature.
30069 (dom_info::dom_init): Used unsigned rather that size_t.
30070 (dom_info::dom_info): Same.
30072 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30073 Jakub Jelinek <jakub@redhat.com>
30076 * config/arm/arm.c: Include gimple.h.
30077 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30078 returns negative, increment ncrn only if it returned positive.
30079 (arm_needs_doubleword_align): Return int instead of bool,
30080 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30081 members, but if there is any such non-FIELD_DECL
30082 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30083 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30084 returns negative, increment nregs only if it returned positive.
30085 (arm_setup_incoming_varargs): Likewise.
30086 (arm_function_arg_boundary): Emit -Wpsabi note if
30087 arm_needs_doubleword_align returns negative, return
30088 DOUBLEWORD_ALIGNMENT only if it returned positive.
30090 2017-04-25 Marek Polacek <polacek@redhat.com>
30093 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30094 first argument to type.
30096 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30099 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30100 type checks to test for compatibility instead of equality.
30102 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30103 Jakub Jelinek <jakub@redhat.com>
30106 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30108 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30109 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30110 the alignment computation, but return their maximum in warn_alignment.
30111 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30112 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30114 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30115 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30118 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30120 * config/arc/simdext.md (dmpyh): Fix typo.
30122 2017-04-25 Richard Biener <rguenther@suse.de>
30124 PR tree-optimization/80492
30125 * alias.c (compare_base_decls): Handle registers with asm
30126 specification conservatively.
30127 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30128 compare_base_decls returning dont-know properly.
30130 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30132 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30133 (legitimate_offset_address_p): New function.
30134 (arc_legitimate_address_p): Use above function.
30136 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30138 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30140 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30142 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30143 ACCH registers whenever they are available.
30145 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30147 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30148 double regs fix when not used.
30150 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30152 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30154 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30155 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30157 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30159 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30160 when emitting short ADD instructions.
30162 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30164 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30166 (cmpsi_cc_c_insn): Likewise.
30167 (cbranchsi4_scratch): Compute proper instruction length using
30168 compact_hreg_operand.
30169 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30171 2017-04-25 Richard Biener <rguenther@suse.de>
30173 PR middle-end/80509
30174 * passes.c (pass_manager::pass_manager): Initialize
30175 m_name_to_pass_map.
30177 2017-04-25 Richard Biener <rguenther@suse.de>
30179 PR tree-optimization/79201
30180 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30182 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30185 * config/s390/vector.md: Split MEM->GPR vector moves for
30186 non-s_operand addresses.
30188 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30191 * config/s390/predicates.md (reload_const_wide_int_operand): New
30193 * config/s390/s390.md ("movti"): Remove d/P alternative.
30194 ("movti_bigconst"): New pattern definition.
30196 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30199 * s390-protos.h (s390_expand_cs_hqi): Removed.
30200 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30201 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30202 modes as well as CCZ1mode and CCZmode.
30203 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30204 signature of s390_emit_compare_and_swap.
30205 (s390_expand_cs_hqi): Likewise, make static.
30206 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30207 compare-and-swap, in some cases.
30208 (s390_expand_cs): Wrapper function.
30209 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30211 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30212 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30213 patterns for small and large integers. Forbid symref memory operands.
30214 Move expander to s390.c. Require cc register.
30215 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30216 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30217 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30218 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30219 symref memory operands. Remove CC mode and call s390_match_ccmode
30221 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30223 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30225 * config/s390/s390.md (define_peephole2): New peephole to help
30226 combining the load-and-test pattern with volatile memory.
30228 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30230 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30231 with CCZmode for TARGET_Z196.
30233 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30235 PR rtl-optimization/80501
30236 * combine.c (make_compound_operation_int): Set subreg_code to SET
30237 even for AND with mask of the sign bit of mode.
30239 PR rtl-optimization/80500
30240 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30241 sum's initial value.
30243 2017-04-25 Julian Brown <julian@codesourcery.com>
30244 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30246 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30248 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30250 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30252 2017-04-25 Julian Brown <julian@codesourcery.com>
30253 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30255 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30256 (thunderx2t99_sha): New Reservation.
30258 2017-04-25 Julian Brown <julian@codesourcery.com>
30259 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30261 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30262 type for 1-element load.
30264 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30266 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30268 2017-04-24 Martin Jambor <mjambor@suse.cz>
30270 PR tree-optimization/80293
30271 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30272 char arrays not totally scalarizable if it is false.
30273 (analyze_all_variable_accesses): Pass correct value in the new
30274 parameter. Add a statistics counter.
30276 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30278 PR middle-end/79931
30279 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30281 2017-04-24 Richard Biener <rguenther@suse.de>
30283 PR tree-optimization/80494
30284 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30285 out for complex types.
30287 2017-04-24 Richard Biener <rguenther@suse.de>
30289 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30290 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30291 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30292 (DFS): Adjust and never fail.
30293 (sccvn_dom_walker::fail): Remove.
30294 (sccvn_dom_walker::before_dom_children): Adjust.
30295 (run_scc_vn): Likewise and never fail.
30296 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30297 (pass_fre::execute): Likewise.
30299 2017-04-24 Richard Biener <rguenther@suse.de>
30301 PR tree-optimization/79725
30302 * tree-ssa-sink.c (statement_sink_location): Return whether
30303 failure reason was zero uses. Move that check later.
30304 (sink_code_in_bb): Deal with zero uses by removing the stmt
30307 2017-04-24 Richard Biener <rguenther@suse.de>
30310 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30311 pointer-based references.
30313 2017-04-24 Richard Biener <rguenther@suse.de>
30316 * pass_manager.h (pass_manager::operator new): Remove.
30317 (pass_manager::operator delete): Likewise.
30318 * passes.c (pass_manager::operator new): Remove.
30319 (pass_manager::operator delete): Likewise.
30320 (pass_manager::pass_manager): Zero individual pass members.
30322 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30325 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30326 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30327 Check "XEXP (src, 1)" operand here.
30328 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30329 Check "XEXP (src, 1)" operand here.
30330 (dimode_scalar_chain::make_vector_copies): Detect count register
30331 of a shift instruction. Zero extend count register from QImode
30332 to DImode to satisfy vector shift pattern count operand predicate.
30333 Substitute vector shift count operand with a DImode copy.
30334 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30337 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30339 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30340 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30341 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30342 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30343 (UNSPEC_NOREX_MEM): Remove definition.
30345 2017-04-21 Richard Biener <rguenther@suse.de>
30347 PR tree-optimization/79547
30348 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30349 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30350 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30351 without any constraints.
30353 2017-04-21 Richard Biener <rguenther@suse.de>
30355 PR tree-optimization/78847
30356 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30358 2017-04-21 Richard Biener <rguenther@suse.de>
30360 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30361 (build_distinct_type_copy): Likewise.
30362 (build_variant_type_copy): Likewise.
30363 * tree.c (build_qualified_type): Pass down mem-stat info.
30364 (build_distinct_type_copy): Likewise.
30365 (build_variant_type_copy): Likewise.
30367 2017-04-21 Richard Biener <rguenther@suse.de>
30369 PR tree-optimization/80237
30370 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30372 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30373 for a simplified result.
30375 2016-04-21 Richard Biener <rguenther@suse.de>
30377 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30378 sth as strict as a simple_iv but a chrec without symbols and an
30379 operand defined in the loop we are peeling (and not some subloop).
30380 (propagate_constants_for_unrolling): Propagate all constants.
30382 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30385 * config/i386/i386.c (print_reg): Remove assert for disalowed
30386 regno values, call output_operand_lossage instead.
30388 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30391 * config/i386/constraints.md (Yc): New register constraint.
30392 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30393 Use Yc constraint for alternative 2 of operand 0. Remove
30394 preferred_for_speed attribute.
30396 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30398 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30399 lastprivate clauses in SIMT case.
30401 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30403 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30405 2017-04-20 Richard Biener <rguenther@suse.de>
30407 PR tree-optimization/57796
30408 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30409 as N scalar stores.
30410 (vect_model_load_cost): Cost gathers as N scalar loads.
30412 2017-04-20 Richard Biener <rguenther@suse.de>
30414 * ggc-page.c (ggc_allocated_p): Rename to ...
30415 (safe_lookup_page_table_entry): ... this and return the lookup
30417 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30419 2017-04-20 Richard Biener <rguenther@suse.de>
30421 PR tree-optimization/80453
30422 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30423 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30424 from the conditions.
30425 (vn_phi_eq): Pass them down.
30426 (vn_phi_lookup): Record them.
30427 (vn_phi_insert): Likewise.
30429 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30431 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30432 uninitialized variable warning to avoid buffer overrun.
30434 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30437 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30438 is suppressed for '{ 0 }' in C.
30440 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30442 * BASE-VER: Set to 8.0.0.
30444 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30446 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30447 priority .init_array and .fini_array section with SECTION_NOTYPE
30450 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30452 PR middle-end/80423
30453 * tree.h (build_array_type): Add typeless_storage default argument.
30454 * tree.c (type_cache_hasher::equal): Also compare
30455 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30456 (build_array_type): Add typeless_storage argument, set
30457 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30459 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30460 (build_array_type): Likewise. Add typeless_storage argument.
30462 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30463 Jakub Jelinek <jakub@redhat.com>
30465 PR tree-optimization/80426
30466 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30467 operation on symbolic operands, also compute the overflow for the
30468 invariant part when the operation degenerates into a negation.
30470 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30473 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30474 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30477 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30479 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30482 * config/avr/avr.c (tree.h): Include it.
30483 (cgraph.h): Include it.
30484 (avr_encode_section_info): Don't warn for uninitialized progmem
30485 variable if it's just an alias.
30487 2017-04-19 Richard Biener <rguenther@suse.de>
30490 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30491 when needed by AutoPGO.
30493 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30496 * doc/lto.texi: Remove an extra 'that'.
30498 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30500 PR rtl-optimization/80429
30501 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30502 are only used in debug insns.
30504 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30505 Vladimir Makarov <vmakarov@redhat.com>
30507 * config/sparc/predicates.md (input_operand): Add comment. Return
30508 true for any memory operand when LRA is in progress.
30509 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30511 2017-04-18 Jeff Law <law@redhat.com>
30514 * mips.md ({return,simple_return}_internal): Do not overwrite
30517 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30519 PR tree-optimization/80443
30520 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30521 instead of adding 1, subtract -1 and similarly instead of subtracting
30524 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30526 PR rtl-optimization/80357
30527 * haifa-sched.c (tmp_bitmap): New variable.
30528 (model_recompute): Handle duplicate use records.
30529 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30530 (free_global_sched_pressure_data): Free it.
30532 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30535 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30536 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30537 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30538 instead of SYSTEM_HEADER_DIR.
30540 2017-04-18 Jeff Law <law@redhat.com>
30542 PR middle-end/80422
30543 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30544 predecessors after walking up the insn chain.
30546 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30549 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30550 sizetype type into debug info.
30552 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30555 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30556 unneeded test for TARGET_UPPER_REGS_SF.
30557 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30559 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30562 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30563 instead of gsi_after_labels.
30565 2017-04-18 Jeff Law <law@redhat.com>
30567 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30571 2017-04-13 Jeff Law <law@redhat.com>
30572 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30573 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30575 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30578 * config/avr/avr.c (intl.h): Include it.
30579 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30581 2017-04-18 Martin Liska <mliska@suse.cz>
30583 PR gcov-profile/78783
30584 * gcov-tool.c (gcov_output_files): Validate that destination
30585 file is either removed by the tool or by a user.
30587 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30588 Guy Benyei <guybe@mellanox.com>
30590 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30591 block, and do not negate it, the stored id is already negative.
30593 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30595 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30597 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30600 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30601 masks of options that should be turned off if the VSX vector
30602 options are turned off.
30603 (OTHER_P8_VECTOR_MASKS): Likewise.
30604 (OTHER_VSX_VECTOR_MASKS): Likewise.
30605 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30606 rs6000_disable_incompatible_switches to validate no type switches
30608 (rs6000_incompatible_switch): New function to disallow turning on
30609 other vector options if -mno-vsx, -mno-power8-vector, or
30610 -mno-power9-vector are specified.
30612 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30614 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30616 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30618 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30619 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30620 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30621 (ARG_POINTER_CFA_OFFSET): Likewise.
30623 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30625 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30626 conditions to take advantage of various optimizations.
30628 2017-04-13 Jeff Law <law@redhat.com>
30630 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30631 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30632 (zero_extendsidi2_dext): Likewise.
30634 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30637 * fold-const.c (fold_ternary_loc): Revert
30638 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30641 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30643 PR rtl-optimization/80343
30644 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30645 new scratch pseudo.
30647 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30650 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30651 to ubsan_encode_value.
30653 2017-04-13 Jeff Law <law@redhat.com>
30655 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30656 appearing in DEBUG_INSNs.
30658 2017-04-13 Martin Liska <mliska@suse.cz>
30660 PR gcov-profile/80413
30661 * gcov-io.c (gcov_write_string): Copy to buffer just when
30662 allocated size is greater than zero.
30664 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30667 * dwarf2out.c (decls_for_scope): Ignore declarations of
30668 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30670 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30673 * ipa-visibility.c (non_local_p): Fix typos.
30674 (localize_node): When localizing symbol in same comdat group,
30675 dissolve the group only when we know external symbols are going
30677 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30679 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30681 PR tree-optimization/79390
30682 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30683 order does not result in usable sequence, retry with reversed operand
30689 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30690 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30691 op0 instead of fold_convert_loc (loc, type, arg0).
30693 2017-04-12 Jeff Law <law@redhat.com>
30695 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30696 has a delay slot in the generated code.
30698 * config/cris/cris.md (cris_preferred_reload_class): Return
30699 GENNONACR_REGS rather than GENERAL_REGS.
30701 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30704 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30705 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30706 signedness of the result type.
30708 2017-04-12 Richard Biener <rguenther@suse.de>
30709 Jeff Law <law@redhat.com>
30711 PR tree-optimization/80359
30712 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30713 trim stores to TARGET_MEM_REFs.
30715 2017-04-12 Richard Biener <rguenther@suse.de>
30717 PR tree-optimization/79390
30718 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30719 threading case even more.
30721 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30724 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30725 for quad_address_p for TImode, instead of just not indexed_address.
30727 2017-04-12 Richard Biener <rguenther@suse.de>
30728 Bernd Edlinger <bernd.edlinger@hotmail.de>
30730 PR middle-end/79671
30731 * alias.c (component_uses_parent_alias_set_from): Handle
30732 TYPE_TYPELESS_STORAGE.
30733 (get_alias_set): Likewise.
30734 * tree-core.h (tree_type_common): Add typeless_storage flag.
30735 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30736 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30737 for types containing members with TYPE_TYPELESS_STORAGE.
30738 (place_field): Likewise.
30739 (layout_type): Likewise for ARRAY_TYPE.
30740 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30741 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30742 TYPE_TYPELESS_STORAGE.
30743 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30745 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30748 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30749 first argument to type.
30751 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30755 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30756 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30757 (rs6000_expand_binop_builtin): Likewise.
30758 (rs6000_expand_ternop_builtin): Likewise; also add missing
30759 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30760 vshasigma built-ins.
30761 * doc/extend.texi: Document that vec_xxpermdi's third argument
30762 must be a constant.
30764 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30766 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30767 Use shift_const cost parameter when calculating gain of STV shifts.
30769 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30771 PR rtl-optimization/70478
30772 * lra-constraints.c (process_alt_operands): Check memory for
30773 disfavoring memory insn operand.
30775 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30777 PR middle-end/80100
30778 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30779 left shift in unsigned HOST_WIDE_INT type.
30781 PR rtl-optimization/80385
30782 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30783 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30786 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30787 if they have any depend clauses.
30789 2017-04-11 Martin Liska <mliska@suse.cz>
30792 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30793 * ipa-split.c (split_function): Create a local comdat symbol
30794 if caller is in a comdat group.
30796 2017-04-11 Martin Liska <mliska@suse.cz>
30799 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30802 2017-04-11 Martin Sebor <msebor@redhat.com>
30804 PR middle-end/80364
30805 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30806 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30808 (directive::set_width, directive::set_precision, format_character):
30810 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30813 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
30816 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
30817 conflict, set target->arch_name instead of target->cpu_name.
30819 2017-04-11 Richard Biener <rguenther@suse.de>
30821 PR tree-optimization/80374
30822 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
30823 build_zero_cst, remove fold_convertible_p check again.
30825 2017-04-11 Martin Liska <mliska@suse.cz>
30828 * ubsan.c (instrument_object_size): Do not instrument register
30831 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30834 * config/i386/i386-builtin-types.def
30835 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
30836 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
30837 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
30838 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
30839 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
30840 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
30841 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
30842 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
30843 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
30844 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
30845 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
30846 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
30847 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
30848 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
30849 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
30850 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
30851 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
30852 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
30853 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
30854 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
30855 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
30856 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
30857 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
30858 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
30859 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
30860 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
30861 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
30862 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
30863 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
30864 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
30865 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
30866 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
30867 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
30868 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
30869 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
30870 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
30871 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
30872 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
30873 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
30874 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
30875 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
30876 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
30877 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
30878 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
30879 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
30880 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
30882 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
30883 flag to second_arg_count, handle 4 argument function type _COUNT
30884 aliases, handle second_arg_count on second argument rather than last.
30886 2017-04-10 Jeff Law <law@redhat.com>
30888 PR tree-optimization/80374
30889 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
30890 record anything if we can not convert integer_zero_node to the
30893 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30896 * config/rs6000/rs6000.c (rs6000_option_override_internal):
30897 Enhance special handling given to the TARGET_P9_MINMAX option in
30898 relation to certain other options.
30900 2017-04-10 Bin Cheng <bin.cheng@arm.com>
30902 PR tree-optimization/80153
30903 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
30904 remove POINTER_PLUS_EXPR's base part directly, rather than through
30907 2017-04-10 Richard Biener <rguenther@suse.de>
30908 Bin Cheng <bin.cheng@arm.com>
30910 PR tree-optimization/80153
30911 * tree-affine.c (aff_combination_to_tree): Get base pointer from
30912 the first element of pointer type aff_tree. Build result expr in
30914 (add_elt_to_tree): Convert to type unconditionally. Remove other
30915 fold_convert calls.
30916 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
30917 (rewrite_use_nonlinear_expr): Check invariant using iv information.
30919 2017-04-10 Richard Biener <rguenther@suse.de>
30921 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30924 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
30926 PR rtl-optimization/70478
30927 * lra-constraints.c (curr_small_class_check): New.
30928 (update_and_check_small_class_inputs): New.
30929 (process_alt_operands): Update curr_small_class_check. Disfavor
30930 alternative insn memory operands. Check available regs for small
30933 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30936 * config/mips/mips.opt (-mvirt): Update description.
30937 * doc/invoke.texi (-mvirt): Likewise.
30939 2017-04-10 Richard Biener <rguenther@suse.de>
30941 PR middle-end/80362
30942 * fold-const.c (fold_binary_loc): Look at unstripped ops when
30943 looking for NEGATE_EXPR in -A / -B to A / B folding.
30945 2017-04-10 Martin Liska <mliska@suse.cz>
30947 PR gcov-profile/80224
30948 * gcov.c (print_usage): Fix usage string.
30949 (get_gcov_intermediate_filename): Remove.
30950 (output_gcov_file): Use both for normal and intermediate format.
30951 (generate_results): Do not initialize special file for
30952 intermediate format.
30954 2017-04-10 Richard Biener <rguenther@suse.de>
30956 PR tree-optimization/80304
30957 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
30960 2017-04-10 Nathan Sidwell <nathan@acm.org>
30963 * config/rs6000/rs6000.c (rs6000_vector_type): New.
30964 (rs6000_init_builtins): Use it.
30966 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30968 * config/arm/arm.md (<mrc>): Add mode to SET source.
30969 (<mrrc>): Likewise.
30971 2017-04-10 Richard Biener <rguenther@suse.de>
30973 PR middle-end/80344
30974 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
30976 2017-04-10 Jakub Jelinek <jakub@redhat.com>
30979 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
30980 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
30981 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
30982 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
30983 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
30984 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
30985 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
30986 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
30987 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
30988 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
30989 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
30990 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
30991 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
30992 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
30993 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
30994 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
30995 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
30996 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
30997 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
30998 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
30999 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31000 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31001 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31003 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31005 PR rtl-optimization/70478
31006 * lra-constraints.c: Reverse the last patch.
31008 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31010 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31011 Add comment for WCHAR_T.
31013 2017-04-08 Martin Liska <mliska@suse.cz>
31016 2017-04-07 Martin Liska <mliska@suse.cz>
31019 * ipa-split.c (split_function): Add function part to a same comdat
31022 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31025 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31027 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31029 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31030 (vec_pairing): ...to this.
31031 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31032 (rs6000_sched_init): Adjust for name change.
31033 (struct rs6000_sched_context): Likewise.
31034 (rs6000_init_sched_context): Likewise.
31035 (rs6000_set_sched_context): Likewise.
31037 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31043 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31045 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31046 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31047 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31049 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31051 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31053 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31055 PR rtl-optimization/70703
31056 * ira-color.c (update_conflict_hard_regno_costs): Use
31057 int64_t instead of HOST_WIDE_INT.
31059 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31061 PR rtl-optimization/70478
31062 * lra-constraints.c (process_alt_operands): Disfavor alternative
31063 insn memory operands.
31065 2017-04-07 Jeff Law <law@redhat.com>
31067 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31068 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31070 2017-04-07 Martin Liska <mliska@suse.cz>
31073 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31074 Show error message instead of an ICE.
31076 2017-04-07 Martin Liska <mliska@suse.cz>
31079 * ipa-split.c (split_function): Add function part to a same comdat
31082 2017-04-07 Richard Biener <rguenther@suse.de>
31084 PR middle-end/80341
31085 * tree.c (get_unwidened): Also handle ! for_type case for
31087 * convert.c (do_narrow): Split out from ...
31088 (convert_to_integer_1): ... here. Do not pass final truncation
31089 type to get_unwidened for TRUNC_DIV_EXPR.
31091 2017-04-07 Richard Biener <rguenther@suse.de>
31093 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31095 (aff_combination_const): Adjust.
31096 (aff_combination_scale): Likewise.
31097 (aff_combination_add_elt): Likewise.
31098 (aff_combination_add_cst): Likewise.
31099 (aff_combination_convert): Likewise.
31100 (add_elt_to_tree): Likewise. Remove unused argument.
31101 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31103 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31105 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31107 * config/arm/arm.c (arm_default_short_enums): Use
31108 ARM_DEFAULT_SHORT_ENUMS.
31109 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31111 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31114 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31115 members with redundant out-of-class redeclaration.
31117 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31120 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31121 * config/i386/i386.md (*zero_extendsidi2):
31122 Add (?*x,*x) and (?*v,*v) alternatives.
31124 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31127 * config/i386/i386.c (ix86_expand_builtin)
31128 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31129 mode from insn data. Convert operands to insn operand mode.
31130 Copy operands that don't satisfy insn predicate to a register.
31132 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31134 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31137 2017-04-06 Richard Biener <rguenther@suse.de>
31139 PR tree-optimization/80334
31140 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31141 preserve alignment of accesses.
31143 2017-04-06 Richard Biener <rguenther@suse.de>
31145 PR tree-optimization/80262
31146 * tree-sra.c (build_ref_for_offset): Preserve address-space
31148 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31149 Drop useless address-space information on MEM_REF offsets.
31151 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31153 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31155 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31157 PR rtl-optimization/70703
31158 * ira-color.c (update_conflict_hard_regno_costs): Use
31159 HOST_WIDE_INT instead of long.
31161 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31164 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31165 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31167 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31168 for x86_64 target. Handle -m3dnowa option.
31170 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31172 PR rtl-optimization/70703
31173 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31174 (update_conflict_hard_regno_costs): Use long instead of unsigned
31175 arithmetic for cost calculation.
31177 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31178 Bernd Edlinger <bernd.edlinger@hotmail.de>
31181 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31184 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31187 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31188 ptr_mode with Pmode throughout.
31189 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31190 into probe_stack_range and use DImode.
31192 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31195 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31196 call_eh_return is true.
31198 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31200 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31201 Initialize last_match_fntype_index.
31203 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31206 * tree-nvr.c: Include internal-fn.h.
31207 (pass_return_slot::execute): Ignore internal calls without
31210 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31211 Richard Biener <rguenther@suse.de>
31214 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31215 captures used multiple times, except for the last use.
31216 * generic-match-head.c: Include gimplify.h.
31218 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31220 PR tree-optimization/79390
31221 * target.h (struct noce_if_info): Declare.
31222 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31223 * target.def (noce_conversion_profitable_p): New target hook.
31224 * ifcvt.h (struct noce_if_info): New type, moved from ...
31225 * ifcvt.c (struct noce_if_info): ... here.
31226 (noce_conversion_profitable_p): Renamed to ...
31227 (default_noce_conversion_profitable_p): ... this. No longer
31229 (noce_try_store_flag_constants, noce_try_addcc,
31230 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31231 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31232 instead of noce_conversion_profitable_p.
31233 * config/i386/i386.c: Include ifcvt.h.
31234 (ix86_option_override_internal): Don't override
31235 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31236 (ix86_noce_conversion_profitable_p): New function.
31237 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31238 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31239 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31240 * doc/tm.texi: Regenerated.
31242 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31244 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31247 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31250 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31251 instructions for small multiply cores.
31253 2017-04-04 Jeff Law <law@redhat.com>
31255 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31257 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31258 that are not set by the loop over the elements.
31260 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31263 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31264 int mode, convert_modes it to mode as unsigned, otherwise use
31265 lowpart_subreg to mode rather than SImode.
31266 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31267 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31268 Use DImode instead of SImode for the shift count operand.
31269 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31272 2017-04-04 Richard Biener <rguenther@suse.de>
31274 PR middle-end/80281
31275 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31276 arithmetic done for the negate or the plus. Simplify.
31277 (A - (-B) -> A + B): Likewise.
31278 * fold-const.c (split_tree): Make sure to not negate pointers.
31280 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31282 PR rtl-optimization/60818
31283 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31284 a compare of comparisons with the thing compared if this results
31285 in a different machine mode.
31287 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31289 * alias.c (base_alias_check): Fix typo in comment.
31290 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31291 * cgraphunit.c (symbol_table::compile): Likewise.
31292 * collect2.c (maybe_run_lto_and_relink): Likewise.
31293 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31294 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31295 * config/avr/avr.c (avr_map_op_t): Likewise.
31296 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31297 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31298 * config/epiphany/epiphany.md (movcc): Likewise.
31299 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31300 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31302 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31303 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31304 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31305 * config/sh/sh.c (sh_rtx_costs): Likewise.
31306 * fold-const.c (fold_truth_andor): Likewise.
31307 * genautomata.c (collapse_flag): Likewise.
31308 * gengtype.h (struct type::u::s): Likewise.
31309 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31310 * input.c (FORMAT_AMOUNT): Likewise.
31311 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31312 (known_aggs_to_agg_replacement_list): Likewise.
31313 * ipa-inline-analysis.c: Likewise.
31314 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31315 * ipa-polymorphic-call.c
31316 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31317 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31318 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31320 * modulo-sched.c (apply_reg_moves): Likewise.
31321 * omp-expand.c (build_omp_regions_1): Likewise.
31322 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31323 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31324 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31325 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31326 * value-prof.c: Likewise.
31327 * var-tracking.c (val_reset): Likewise.
31329 2017-04-03 Richard Biener <rguenther@suse.de>
31331 PR tree-optimization/80275
31332 * fold-const.c (split_address_to_core_and_offset): Handle
31335 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31337 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31338 descriptors is at least equal to that of functions.
31340 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31342 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31344 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31347 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31348 (mov<IMOD4:mode>): New expander.
31349 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31351 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31353 PR rtl-optimization/79405
31354 * fwprop.c (propagations_left): New variable.
31355 (forward_propagate_into): Decrement it.
31356 (fwprop_init): Initialize it.
31357 (fw_prop): If the variable has reached zero, stop propagating.
31358 (fwprop_addr): Ditto.
31360 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31363 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31364 a FUNCTION_DECL, pass it as decl instead of origin to
31367 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31369 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31372 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31375 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31376 TARGET_VSX_SMALL_INTEGER.
31378 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31380 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31381 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31383 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31385 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31386 extraction from odd-numbered MSA register.
31388 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31390 PR middle-end/80173
31391 * expmed.c (store_bit_field_1): Don't attempt to create
31392 a word subreg out of hard registers wider than word if they
31393 have HARD_REGNO_NREGS of 1 for their mode.
31395 PR middle-end/80163
31396 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31397 conversions to integer types wider than word and pointer.
31400 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31401 (rtx_equal_for_cselib_p): Pass 0 to it.
31402 * cselib.c (cselib_hasher::equal): Likewise.
31403 (rtx_equal_for_cselib_1): Add depth argument. If depth
31404 is 128, don't look up VALUE locs and punt. Increment
31405 depth in recursive calls when walking VALUE locs.
31407 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31409 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31410 (make_gcov_file_name): Use the canonical path name for generating
31412 (read_line): Fix handling of files with ascii null bytes.
31414 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31416 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31417 to initialise a vector register instead
31418 of using a const_int.
31420 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31422 PR translation/80189
31423 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31424 diagnostic messages.
31426 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31429 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31430 (dfp_diex_<mode>): Update mode of operand 1.
31431 * doc/extend.texi (dxex, dxexq): Document change to return type.
31432 (diex, diexq): Document change to argument type.
31434 2017-03-30 Martin Jambor <mjambor@suse.cz>
31437 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31438 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31439 it reflects the signature changes performed at the callee side.
31440 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31441 to cgraph_build_function_type_skip_args.
31442 (build_function_decl_skip_args): Adjust call to the above function.
31444 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31447 * config/i386/sse.md
31448 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31449 register as dest whenever it is a MEM not rtx_equal_p to the
31450 corresponding dup operand, and when forcing into reg move the
31451 reg into the memory afterwards.
31452 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31453 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31454 for the force_reg mode.
31455 (avx512vl_vextractf128<mode>): Use register as dest either
31456 always when a MEM, or when it is a MEM not rtx_equal_p to the
31457 corresponding dup operand, or even not when it is a CONST_VECTOR
31458 depending on the mode and lo vs. hi.
31459 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31461 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31462 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31463 Likewise. Require that operands[2] is even.
31464 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31465 Remove extraneous parens. Require that operands[2] is a multiple
31467 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31468 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31469 disallow memory then.
31471 2017-03-30 Richard Biener <rguenther@suse.de>
31473 PR tree-optimization/77498
31474 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31475 to non-constants over backedges.
31477 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31479 PR rtl-optimization/80233
31480 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31481 as last_combined_insn. Do not test for BARRIER_P separately.
31483 2017-03-29 Andreas Schwab <schwab@suse.de>
31486 * calls.c (prepare_call_address): Convert funexp to Pmode before
31487 copying to temp reg.
31489 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31491 PR tree-optimization/80158
31492 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31493 Handle possible future case of more than one alternate
31495 (replace_rhs_if_not_dup): Likewise.
31496 (replace_one_candidate): Likewise.
31498 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31500 PR rtl-optimization/80193
31501 * ira.c (ira): Do not check allocation for LRA.
31503 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31505 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31506 (nvptx_output_simt_exit): Declare.
31507 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31508 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31509 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31510 (nvptx_declare_function_name): ...here. Emit declaration of local
31511 memory space buffer for omp_simt_enter insn.
31512 (nvptx_output_unisimt_switch): New.
31513 (nvptx_output_softstack_switch): New.
31514 (nvptx_output_simt_enter): New.
31515 (nvptx_output_simt_exit): New.
31516 * config/nvptx/nvptx.h (struct machine_function): New fields
31517 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31518 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31519 (UNSPECV_SIMT_EXIT): Ditto.
31520 (omp_simt_enter_insn): New insn.
31521 (omp_simt_enter): New expansion.
31522 (omp_simt_exit): New insn.
31523 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31525 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31526 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31527 (expand_GOMP_SIMT_EXIT): New.
31528 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31529 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31530 (GOMP_SIMT_EXIT): Ditto.
31531 * target-insns.def (omp_simt_enter): New insn.
31532 (omp_simt_exit): Ditto.
31533 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31535 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31536 (lower_rec_input_clauses): Likewise.
31537 (lower_lastprivate_clauses): Handle SIMT privatization.
31539 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31540 (ompdevlow_adjust_simt_enter): New.
31541 (find_simtpriv_var_op): New.
31542 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31543 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31545 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31546 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31547 (copy_decl_for_dup_finish): Ditto.
31549 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31551 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31554 * config/i386/i386.c (ix86_option_override_internal): Always
31555 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31557 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31559 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31561 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31563 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31564 mark new edge's irreducible flag accordign to it.
31565 (vect_do_peeling): Check loop preheader edge's irreducible flag
31566 and pass it to function slpeel_add_loop_guard.
31568 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31570 PR tree-optimization/80218
31571 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31572 Update block frequencies and counts.
31574 2017-03-28 Richard Biener <rguenther@suse.de>
31576 PR tree-optimization/78644
31577 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31578 of a simplification result we may not use it at all.
31580 2017-03-28 Richard Biener <rguenther@suse.de>
31583 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31584 without arguments, generate default definition of a SSA name.
31586 2017-03-28 Richard Biener <rguenther@suse.de>
31588 PR middle-end/80222
31589 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31590 TYPE_REF_CAN_ALIAS_ALL references.
31591 * fold-const.c (fold_indirect_ref_1): Likewise.
31593 2017-03-28 Martin Liska <mliska@suse.cz>
31596 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31597 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31599 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31600 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31602 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31603 (EXTRA_SPECS): Define.
31604 (SUBTARGET_EXTRA_SPECS): Likewise.
31605 (SUBTARGET_CPP_SPEC): Likewise.
31606 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31607 SUBTARGET_EXTRA_SPECS.
31608 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31610 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31612 * config/arc/simdext.md (vst64_insn): Update pattern.
31613 (vld32wh_insn): Likewise.
31614 (vld32wl_insn): Likewise.
31615 (vld64_insn): Likewise.
31616 (vld32_insn): Likewise.
31618 2017-03-28 Marek Polacek <polacek@redhat.com>
31621 * fold-const.c (fold_comparison): Use protected_set_expr_location
31622 instead of SET_EXPR_LOCATION.
31624 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31626 * tree.c (add_expr): Avoid name lookup warning.
31628 2017-03-27 Jeff Law <law@redhat.com>
31630 PR tree-optimization/80216
31631 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31632 function name. Limit recursion depth.
31633 (record_temporary_equivalences): Corresponding changes.
31635 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31637 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31640 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31643 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31645 * cfgcleanup.c (reg_note_cfa_p): New array.
31646 (insns_have_identical_cfa_notes): New function.
31647 (old_insns_match_p): Don't cross-jump in between /f
31648 and non-/f instructions. If both i1 and i2 are frame related,
31649 verify all CFA notes, their order and content.
31651 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31654 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31655 HImode and SImode with zero extend to DImode to one insn.
31656 (bswap<mode>2_extenddi): Likewise.
31657 (bswapsi2_extenddi): Likewise.
31658 (bswaphi2_extendsi): Likewise.
31659 (bswaphi2): Combine bswap HImode and SImode into one insn.
31660 Separate memory insns from swapping register.
31661 (bswapsi2): Likewise.
31662 (bswap<mode>2): Likewise.
31663 (bswaphi2_internal): Delete, no longer used.
31664 (bswapsi2_internal): Likewise.
31665 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31666 store, and gpr<-gpr swap insns.
31667 (bswap<mode>2_store): Likewise.
31668 (bswaphi2_reg): Register only splitter, combine with the splitter.
31669 (bswaphi2 splitter): Likewise.
31670 (bswapsi2_reg): Likewise.
31671 (bswapsi2 splitter): Likewise.
31672 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31673 the insns into load, store, and register/register insns.
31674 (bswapdi2_ldbrx): Likewise.
31675 (bswapdi2_load): Likewise.
31676 (bswapdi2_store): Likewise.
31677 (bswapdi2_reg): Likewise.
31679 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31681 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31682 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31684 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31687 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31689 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31690 special handling for target option conflicts between dform
31691 options (-mpower9-dform, -mpower9-dform-vector,
31692 -mpower9-dform-scalar) and -mno-direct-move.
31694 2017-03-27 Richard Biener <rguenther@suse.de>
31696 PR tree-optimization/80181
31697 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31699 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31701 * config/arc/predicates.md (move_double_src_operand): Replace the
31702 call to move_double_src_operand with a call to address_operand.
31704 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31706 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31707 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31708 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31710 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31712 * config/arc/predicates.md (long_immediate_loadstore_operand):
31713 Consider scaled addresses cases.
31715 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31717 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31718 restored when in interrupt.
31719 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31720 doesn't have delay slot.
31722 2017-03-27 Richard Biener <rguenther@suse.de>
31725 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31726 inlined thunk clones.
31728 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31731 * asan.c (instrument_derefs): Copy over last operand from
31732 original COMPONENT_REF to the new COMPONENT_REF with
31733 DECL_BIT_FIELD_REPRESENTATIVE.
31734 * ubsan.c (instrument_object_size): Likewise.
31736 2017-03-27 Richard Biener <rguenther@suse.de>
31738 PR tree-optimization/80170
31739 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31740 sure DR/SCEV didnt fold in constants we do not see when looking
31741 at the reference base alignment.
31743 2017-03-27 Richard Biener <rguenther@suse.de>
31745 PR middle-end/80171
31746 * gimple-fold.c (fold_ctor_reference): Properly guard against
31747 NULL return value from canonicalize_constructor_val.
31749 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31752 * config/i386/i386.c (ix86_expand_builtin)
31753 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31754 flags reg setting and flags reg using instructions.
31755 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31756 clobbering instructions to zero extend op2.
31758 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31760 * doc/install.texi (Configuration) <--with-aix-soname>:
31761 Update link to AIX ld.
31763 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31765 PR rtl-optimization/80160
31766 PR rtl-optimization/80159
31767 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31768 reg_alternate_class into account.
31770 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31773 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31774 to consider in curr_insn_transform.
31776 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31778 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31779 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31780 and emit_mode_inner.
31782 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31784 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31785 argument to the overloaded builtin variants. Use the new flag to
31786 deprecate certain builtin variants.
31787 * config/s390/s390-builtin-types.def: Add new builtin types.
31788 * config/s390/s390-builtins.h: Support new flags field for
31789 overloaded builtins.
31790 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31791 (s390_macro_to_expand): Enable vector float data type.
31792 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31793 builtins by incrementing the __VEC__ version number.
31794 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31796 (s390_resolve_overloaded_builtin): Emit error messages depending
31797 on the builtin flags.
31798 * config/s390/s390.c (s390_expand_builtin): Support additional
31799 flags argument. Change error message to match the messages
31800 emitted in s390-c.c.
31801 * config/s390/s390.md: New UNSPEC_* constants.
31802 (op_type): Add new instruction types.
31803 * config/s390/vecintrin.h: Add new builtins and test data class
31805 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31806 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31807 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31808 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31809 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31810 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31812 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31813 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31814 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
31815 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
31817 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
31818 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
31819 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
31820 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
31822 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
31823 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
31824 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
31825 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
31826 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
31827 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
31828 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
31830 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
31831 ("vec_scatter_element<V_HW_4:mode>_DI")
31832 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
31833 ("vec_fpint<mode>", "vflls")
31834 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
31835 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
31836 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
31837 ("*vec_cmphe<mode>_cc"): ... these.
31839 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
31840 mode constant instead of magic value.
31842 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31844 * config/s390/s390.c (s390_expand_vec_compare): Support other
31845 vector floating point modes than just V2DF.
31846 (s390_expand_vcond): Likewise.
31847 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
31848 (s390_cannot_change_mode_class): Prevent mode changes between TF
31849 and V1TF in vector registers.
31850 * config/s390/s390.md (DF, SF): New mode attributes.
31851 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
31852 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
31853 SFmode support for VRs.
31854 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
31856 (VFT, VF_HW): New mode iterators.
31857 (vw, sdx): New mode attributes.
31858 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
31859 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
31860 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
31861 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
31862 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
31863 also the new vector floating point modes. Renaming to ...
31865 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
31866 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
31867 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
31868 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
31869 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
31870 ("vec_unordered<mode>"): ... these.
31872 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
31873 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
31874 ("*vec_extendv2df"): New insn definitions.
31876 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31878 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
31879 ("mulditi3_2", "*muldi3_sign"): New patterns.
31880 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
31881 rename the pattern definition.
31883 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31885 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
31887 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
31889 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31891 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
31892 instruction if possible.
31893 * config/s390/vector.md (vec_halfnumelts): New mode
31895 ("*vec_vllezlf<mode>"): New pattern.
31897 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31899 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
31900 ("popcountv4si2", "popcountv2di2"): Rename to ...
31901 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
31902 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
31904 ("popcount<mode>2_vxe"): New pattern.
31906 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31908 * common/config/s390/s390-common.c (processor_flags_table): Add
31910 * config.gcc: Add arch12.
31911 * config/s390/driver-native.c (s390_host_detect_local_cpu):
31912 Default to arch12 for unknown CPU model numbers.
31913 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
31914 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
31915 PROCESSOR_max sanity check.
31916 * config/s390/s390-opts.h (enum processor_type): Add
31918 * config/s390/s390.c (processor_table): Add arch12.
31919 (s390_expand_builtin): Add check for B_VXE flag.
31920 (s390_issue_rate): Add PROCESSOR_ARCH12.
31921 (s390_get_sched_attrmask): Likewise.
31922 (s390_get_unit_mask): Likewise.
31923 (s390_sched_score): Enable z13 scheduling for arch12.
31924 (s390_sched_reorder): Likewise.
31925 (s390_sched_variable_issue): Likewise.
31926 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
31928 (s390_tune_attr): Use z13 scheduling also for arch12.
31929 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
31930 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
31931 (TARGET_VXE_P): New macros.
31932 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
31933 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
31934 * config/s390/s390.opt: Add arch12 as processor_type.
31936 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31938 * config/s390/s390.md
31939 ("fixuns_truncdddi2", "fixuns_trunctddi2")
31940 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
31941 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
31943 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
31944 Rename expanders to ...
31946 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
31947 ("fixuns_truncdddi2_emu"): ... these.
31949 ("fixuns_trunc<mode>si2_emu"): New expander.
31951 ("*fixuns_truncdfdi2_z13"): Rename to ...
31952 ("*fixuns_truncdfdi2_vx"): ... this.
31954 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31956 * config/s390/2964.md: Remove the single element vector compare
31957 instructions which are no longer used.
31958 * config/s390/s390.c (s390_select_ccmode): Remove handling of
31960 (s390_canonicalize_comparison): Remove handling of DFmode
31962 (s390_expand_vec_compare_scalar): Remove function.
31963 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
31964 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
31966 ("*cmp<mode>_ccs"): Add wfcdb instruction.
31968 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31970 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
31972 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
31973 will anyway by matched by mov<mode>_64dfp.
31975 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31977 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
31978 vlef/vstef. Add missing operand to vleif.
31980 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31982 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
31983 pair for all vector types with 64 bit elements.
31984 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
31985 * config/s390/vector.md (V_HW_64): ... here.
31986 (V_128_NOSINGLE): New mode iterator.
31987 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
31988 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
31989 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
31990 ("*vec_load_pairv2di"): Change to ...
31991 ("*vec_load_pair<mode>"): ... this one.
31993 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31995 * config/s390/constraints.md: Add comments.
31996 (jKK): Reject element sizes > 8 bytes.
31997 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
31999 * config/s390/s390.md: Add the s_operand checks formerly in
32000 s390_split_ok_p to various splitters where they are still
32002 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32003 for 128 bit vectors. Plus two splitters.
32005 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32007 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32010 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32013 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32014 error if the boundary argument is not constant.
32016 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32018 PR rtl-optimization/80112
32019 * loop-doloop.c (doloop_condition_get): Don't check condition
32020 if cmp isn't SET with IF_THEN_ELSE src.
32022 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32024 PR tree-optimization/80158
32025 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32026 replacing a candidate statement, also replace it for the
32027 candidate's alternate interpretation.
32028 (replace_rhs_if_not_dup): Likewise.
32029 (replace_one_candidate): Likewise.
32031 2017-03-24 Richard Biener <rguenther@suse.de>
32033 PR tree-optimization/80167
32034 * graphite-isl-ast-to-gimple.c
32035 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32037 (translate_isl_ast_to_gimple::get_rename): Likewise.
32039 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32041 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32042 handling of certain combinations of target options, including the
32043 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32044 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32046 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32049 * config/arm/arm.md (*load_multiple): Add reload_completed to
32050 matching condition.
32052 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32053 Richard Biener <rguenth@suse.de>
32055 PR tree-optimization/79908
32056 PR tree-optimization/80136
32057 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32058 been cast away, gimplify_and_add suffices.
32060 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32062 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32064 2017-03-23 Richard Biener <rguenther@suse.de>
32066 PR tree-optimization/80032
32067 * gimplify.c (gimple_push_cleanup): Forced unconditional
32068 cleanups still have to go to the conditional_cleanups
32071 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32073 PR tree-optimization/80072
32074 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32076 (next_operand_entry_id): Change type to unsigned int.
32077 (sort_by_operand_rank): Make sure to return the right return value
32078 even if unsigned fields are bigger than INT_MAX.
32079 (struct oecount): Change cnt and id type to unsigned int.
32080 (oecount_hasher::equal): Formatting fix.
32081 (oecount_cmp): Make sure to return the right return value
32082 even if unsigned fields are bigger than INT_MAX.
32083 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32086 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32087 TREE_READONLY on result if writing it more than once.
32090 * doc/invoke.texi (-fsanitize=thread): Document that with
32091 -fnon-call-exceptions atomics are not able to throw
32095 * tsan.c: Include tree-eh.h.
32096 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32097 maybe_clean_or_replace_eh_stmt where needed.
32098 (instrument_memory_accesses): Add cfg_changed argument.
32099 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32100 if it returned true.
32101 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32103 PR rtl-optimization/63191
32104 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32105 wrapper function, moved the whole old content into ...
32106 (ix86_delegitimize_address_1): ... this. New inline function.
32107 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32108 true as last argument instead of ix86_delegitimize_address.
32110 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32112 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32113 cortexa57_branch_cost.
32115 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32117 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32119 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32122 * doc/md.texi (Constraints): Document wA constraint.
32123 * config/rs6000/constraints.md (wA): New.
32124 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32125 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32126 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32127 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32129 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32132 * gimplify.c (is_oacc_declared): New function.
32133 (oacc_default_clause): Use it to set default flags for acc declared
32134 variables inside parallel regions.
32135 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32136 declared variables.
32137 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32138 declare attribute to any decl as necessary.
32140 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32143 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32144 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32145 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32146 (arm_arch_lpae): This.
32147 * config/arm/arm.c (arm_arch7ve): Rename into ...
32148 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32149 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32152 2017-03-22 Martin Liska <mliska@suse.cz>
32155 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32156 error message instead of an ICE.
32158 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32160 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32162 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32164 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32166 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32169 2017-03-21 Martin Sebor <msebor@redhat.com>
32171 * doc/extend.texi: Use "cannot" instead of "can't."
32172 * doc/hostconfig.texi: Same.
32173 * doc/install.texi: Same.
32174 * doc/invoke.texi: Same.
32175 * doc/loop.texi: Same.
32176 * doc/md.texi: Same.
32177 * doc/objc.texi: Same.
32178 * doc/rtl.texi: Same.
32179 * doc/tm.texi: Same.
32180 * doc/tm.texi.in: Same.
32181 * doc/trouble.texi: Same.
32183 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32186 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32187 (collect_checksum_attributes): Set it.
32188 (die_checksum_ordered): Use it.
32190 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32192 PR tree-optimization/79908
32193 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32194 change: For a VA_ARG whose LHS has been cast away, use
32195 force_gimple_operand to construct the side effects.
32197 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32199 PR translation/80001
32200 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32201 more amenable to translation.
32202 (oacc_loop_auto_partitions): Likewise.
32204 2017-03-21 Marek Polacek <polacek@redhat.com>
32205 Martin Sebor <msebor@redhat.com>
32207 PR tree-optimization/80109
32208 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32209 on INTEGRAL_TYPE_P.
32211 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32212 Segher Boessenkool <segher@kernel.crashing.org>
32215 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32216 check reg_used_between_p between insn and one of succ or succ2
32217 depending on if succ is artificial insn not inserted into insn
32220 2017-03-21 Martin Liska <mliska@suse.cz>
32222 PR gcov-profile/80081
32223 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32224 * doc/gcc.texi: Include gcov-dump stuff.
32225 * doc/gcov-dump.texi: New file.
32227 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32229 PR rtl-optimization/79150
32230 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32231 conditional jump, if the jump is the last insn of the loop.
32233 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32234 Richard Biener <rguenth@suse.de>
32236 PR tree-optimization/79908
32237 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32238 been cast away, use force_gimple_operand to construct the side
32241 2017-03-21 Martin Liska <mliska@suse.cz>
32243 PR libfortran/79956
32244 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32247 2017-03-21 Brad Spengler <spender@grsecurity.net>
32250 * plugin.c (htab_hash_plugin): New function.
32251 (add_new_plugin): Use it and adjust.
32252 (parse_plugin_arg_opt): Adjust.
32253 (init_one_plugin): Likewise.
32255 2017-03-21 Richard Biener <rguenther@suse.de>
32257 PR tree-optimization/80032
32258 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32259 if set force the cleanup to happen unconditionally.
32260 (gimplify_target_expr): Push inserted clobbers with force_uncond
32261 to avoid them being removed by control-dependent DCE.
32263 2017-03-21 Richard Biener <rguenther@suse.de>
32265 PR tree-optimization/80122
32266 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32267 va_arg_pack_len when the inlined call stmt requires pack
32269 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32271 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32274 * tsan.c (instrument_builtin_call): If the memory model argument
32275 is not a constant, assume it is valid.
32278 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32281 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32283 PR rtl-optimization/79910
32284 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32285 if its dest is used by an insn before I2 (other than the combined
32286 insns themselves, which are properly handled already).
32288 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32291 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32293 * combine.c (record_used_regs): New static function.
32294 (try_combine): Handle situations where there is an additional
32295 instruction between I2 and I3 which needs to have a LOG_LINK
32299 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32301 * combine.c (try_combine): Delete redundant i1 test. Call
32302 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32304 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32307 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32308 alternatives 13/14.
32310 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32312 PR tree-optimization/80054
32313 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32314 the optimization if a PHI or any of its arguments is not dominated
32315 by the candidate's basis. Use gphi* rather than gimple* as
32317 (replace_profitable_candidates): Clean up a gimple* variable that
32318 should be a gphi* variable.
32320 2017-03-20 Martin Sebor <msebor@redhat.com>
32323 * doc/extend.texi (attribute constructor): Document present limitation.
32325 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32328 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32329 __POWER9_VECTOR__ #ifdef control, change template definition to
32330 use Power9-specific built-in function.
32331 (vec_any_eq): Likewise.
32332 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32333 to control outcomes from this test.
32334 (vector_ae_<mode>p): For VEC_F modes, likewise.
32336 2017-03-20 Ian Lance Taylor <iant@google.com>
32338 * config/i386/i386.c (ix86_function_regparm): Save an extra
32339 register for -fsplit-stack with DECL_STATIC_CHAIN.
32341 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32344 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32345 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32347 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32349 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32351 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32354 2017-03-20 Marek Polacek <polacek@redhat.com>
32357 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32359 2017-03-20 Richard Biener <rguenther@suse.de>
32361 PR tree-optimization/80113
32362 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32363 allocate extra SSA name for PHI def.
32364 (add_close_phis_to_outer_loops): Likewise.
32365 (add_close_phis_to_merge_points): Likewise.
32366 (copy_loop_close_phi_args): Likewise.
32367 (copy_cond_phi_nodes): Likewise.
32369 2017-03-20 Martin Liska <mliska@suse.cz>
32371 PR middle-end/79753
32372 * tree-chkp.c (chkp_build_returned_bound): Do not build
32373 returned bounds for a LHS that's not a BOUNDED_P type.
32375 2017-03-20 Martin Liska <mliska@suse.cz>
32379 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32380 COMPLEX_CST and VECTOR_CST.
32382 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32385 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32386 target operand. A new splitter adds the clobber statement in case
32387 the target operand is dead anyway.
32389 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32391 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32392 to age-old versions of binutils and glibc.
32394 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32396 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32398 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32400 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32402 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32404 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32405 requirement for binutils 2.13.
32407 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32409 * combine.c (try_combine): Delete redundant i1 test. Call
32410 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32412 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32414 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32415 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32417 <riscv64-*-elf>: Re-arrange section
32418 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32419 <riscv32-*-linux>: Likewise.
32420 <riscv64-*-elf>: Likewise
32421 <riscv64-*-linux>: Likewise.
32423 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32426 * aarch64.opt(verbose-cost-dump): Fix typo.
32428 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32431 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32432 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32434 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32436 * reload.c (find_reloads): When reloading a nonoffsettable address,
32437 use RELOAD_OTHER for it and its address reloads.
32439 PR rtl-optimization/79910
32440 * combine.c (record_used_regs): New static function.
32441 (try_combine): Handle situations where there is an additional
32442 instruction between I2 and I3 which needs to have a LOG_LINK
32445 2017-03-17 Jeff Law <law@redhat.com>
32447 PR tree-optimization/71437
32448 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32449 conditional in the hash table first.
32450 (vrp_dom_walker::before_dom_children): Extract condition from
32451 ASSERT_EXPR. Record condition, its inverion and any implied
32452 conditions as well.
32454 2017-03-17 Marek Polacek <polacek@redhat.com>
32455 Markus Trippelsdorf <markus@trippelsdorf.de>
32457 PR tree-optimization/80079
32458 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32461 2017-03-17 Richard Biener <rguenther@suse.de>
32463 PR middle-end/80075
32464 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32465 Properly verify the LHS before the RHS possibly claims to be
32467 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32470 2017-03-17 Martin Jambor <mjambor@suse.cz>
32472 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32473 (List of -O2 options): Likewise.
32474 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32477 2017-03-17 Tom de Vries <tom@codesourcery.com>
32479 * gcov-dump.c (print_usage): Print bug_report_url.
32481 2017-03-17 Richard Biener <rguenther@suse.de>
32483 PR middle-end/80050
32484 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32485 (parser::peek): Likewise.
32487 2017-03-17 Richard Biener <rguenther@suse.de>
32489 PR tree-optimization/80048
32490 * sese.c (free_sese_info): Properly release rename_map and
32491 copied_bb_map elements.
32493 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32495 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32496 Add linked-list forward and backlinks. Insert on
32497 construction, remove on destruction.
32498 (class pass_store_merging): Add m_stores_head field.
32499 (pass_store_merging::terminate_and_process_all_chains):
32500 Iterate over m_stores_head list.
32501 (pass_store_merging::terminate_all_aliasing_chains):
32503 (pass_store_merging::execute): Check for debug stmts first.
32504 Push new chains onto the m_stores_head stack.
32506 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32509 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32510 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32511 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32513 2017-03-16 Jeff Law <law@redhat.com>
32515 PR tree-optimization/71437
32516 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32517 member function. Implementation moved into after_dom_children
32518 member function and into the threader's thread_outgoing_edges
32520 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32521 some code into new thread_outgoing_edges.
32522 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32523 definition. Simplify marker handling (do it here). Assume we always
32524 have the available expression and the const/copies tables.
32525 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32527 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32528 * tree-vrp.c (equiv_stack): No longer file scoped.
32529 (vrp_dom_walker): New class.
32530 (vrp_dom_walker::before_dom_children): New member function.
32531 (vrp_dom_walker::after_dom_children): Likewise.
32532 (identify_jump_threads): Setup domwalker. Use it rather than
32533 walking edges in a random order by hand. Simplify setup/finalization.
32534 (finalize_jump_threads): Remove.
32535 (vrp_finalize): Do not call identify_jump_threads here.
32536 (execute_vrp): Do it here instead and call thread_through_all_blocks
32539 PR tree-optimization/71437
32540 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32542 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32544 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32545 (dom_opt_dom_walker::thread_across_edge): Remove
32546 handle_dominating_asserts argument. All callers changed.
32547 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32548 changes. Remove calls to lhs_of_dominating_assert. Other
32549 uses of handle_dominating_asserts turn into unconditional code
32550 (simplify_control_stmt_condition_1): Likewise.
32551 (simplify_control_stmt_condition): Likewise.
32552 (thread_through_normal_block, thread_across_edge): Likewise.
32553 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32554 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32555 object if it is not an SSA_NAME.
32556 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32557 before calling into the VRP specific simplifiers.
32558 (identify_jump_threads): Remove handle_dominating_asserts
32561 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32564 * tree-diagnostic.c (default_tree_printer): No longer static.
32565 * tree-diagnostic.h (default_tree_printer): New prototype.
32567 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32569 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32570 Change ins into fmov.
32572 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32574 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32575 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32576 Use h_con constraint for operand 1.
32577 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32578 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32580 2017-03-15 Jeff Law <law@redhat.com>
32582 PR tree-optimization/71437
32583 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32584 (record_temporary_equivalences): Use it.
32586 PR tree-optimization/71437
32587 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32588 tree-ssa-scopedtables.
32589 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32590 (record_conditions, record_cond, vuse_eq): Likewise.
32591 (record_edge_info): Adjust to API tweak of record_conditions.
32592 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32593 (record_temporary_equivalences, optimize_stmt): Likewise.
32594 (eliminate_redundant_computations): Likewise.
32595 (record_equivalences_from_stmt): Likewise.
32596 * tree-ssa-scopedtables.c: Include options.h and params.h.
32597 (vuse_eq): New function, moved from tree-ssa-dom.c
32598 (build_and_record_new_cond): Likewise.
32599 (record_conditions): Likewise. Accept vector of conditions rather
32600 than edge_equivalence structure for first argument.
32601 for the first argument.
32602 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32603 from tree-ssa-dom.c.
32604 (avail_exprs_stack::record_cond): Likewise.
32605 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32606 from tree-ssa-dom.c.
32607 (avail_exprs_stack): Add new member functions lookup_avail_expr
32609 (record_conditions): Declare.
32611 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32614 * lra-constraints.c (process_alt_operands): Increase reject for
32615 reloading an input/output operand.
32617 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32620 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32621 insns to convert from signed/unsigned char/short to IEEE 128-bit
32623 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32625 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32628 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32629 subreg of inner mode for values already in registers.
32631 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32633 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32634 iteration reg is used after the loop.
32636 2017-03-14 Martin Sebor <msebor@redhat.com>
32638 PR tree-optimization/79800
32639 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32640 precision in negative-positive range.
32641 (format_floating): Call non-const overload with adjusted precision.
32643 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32646 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32649 2017-03-14 Martin Sebor <msebor@redhat.com>
32651 PR middle-end/80020
32652 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32653 * builtins.def (aligned_alloc): Use it.
32656 * Makefile.in (GTFILES): Add calls.c.
32657 * calls.c: Include "gt-calls.h".
32659 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32661 PR rtl-optimization/79728
32662 * regs.h (struct target_regs): New field
32663 x_contains_allocatable_regs_of_mode.
32664 (contains_allocatable_regs_of_mode): New macro.
32665 * reginfo.c (init_reg_sets_1): Initialize it, and change
32666 contains_reg_of_mode so it includes global regs as well.
32667 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32668 rather than contains_regs_of_mode.
32670 2017-03-14 Martin Liska <mliska@suse.cz>
32672 * doc/invoke.texi: Document options that can't be combined with
32673 -fcheck-pointer-bounds.
32675 2017-03-14 Martin Liska <mliska@suse.cz>
32677 PR middle-end/79831
32678 * doc/invoke.texi (-Wchkp): Document the option.
32680 2017-03-14 Martin Liska <mliska@suse.cz>
32682 * Makefile.in: Install gcov-dump.
32684 2017-03-14 Martin Liska <mliska@suse.cz>
32686 * multiple_target.c (expand_target_clones): Bail out for
32687 an invalid attribute.
32689 2017-03-14 Richard Biener <rguenther@suse.de>
32691 * alias.c (struct alias_set_entry): Pack properly.
32692 * cfgloop.h (struct loop): Likewise.
32693 * cse.c (struct set): Likewise.
32694 * ipa-utils.c (struct searchc_env): Likewise.
32695 * loop-invariant.c (struct invariant): Likewise.
32696 * lra-remat.c (struct cand): Likewise.
32697 * recog.c (struct change_t): Likewise.
32698 * rtl.h (struct address_info): Likewise.
32699 * symbol-summary.h (function_summary): Likewise.
32700 * tree-loop-distribution.c (struct partition): Likewise.
32701 * tree-object-size.c (struct object_size_info): Likewise.
32702 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32703 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32704 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32705 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32706 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32707 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32708 (struct _stmt_vec_info): Likewise.
32710 2017-03-14 Martin Liska <mliska@suse.cz>
32713 * multiple_target.c (create_dispatcher_calls): Check that
32714 a target can create a function dispatcher.
32716 2017-03-14 Martin Liska <mliska@suse.cz>
32719 * multiple_target.c (expand_target_clones): Drop local.local
32720 flag for default implementation.
32722 2017-03-14 Richard Biener <rguenther@suse.de>
32724 PR tree-optimization/80030
32725 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32727 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32729 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32730 gcc_fallthrough() instead of __attribute__((fallthrough));
32732 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32734 * doc/gcc.texi: Remove "up" link to (DIR).
32735 * doc/gccint.texi: Ditto.
32737 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32739 * doc/install.texi (Specific) <avr>: Remove reference to
32742 2017-03-13 Jeff Law <law@redhat.com>
32744 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32745 attribute rather than comments.
32747 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32748 match_scratch operand is highest.
32750 2017-03-13 Martin Liska <mliska@suse.cz>
32752 PR middle-end/78339
32753 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32754 is a CHKP clone, use original declaration.
32756 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32758 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32759 (arc_conditional_register_usage): Use a different allocation order
32760 when optimizing for size.
32761 * common/config/arc/arc-common.c (arc_option_optimization_table):
32762 Section anchors default on when optimizing for size.
32764 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32766 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32768 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32770 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32771 * config/arc/arc.md (cpu_facility): Add cd variant.
32772 (*movqi_insn): Add code density variant.
32773 (*movhi_insn): Likewise.
32774 (*movqi_insn): Likewise.
32775 (*addsi3_mixed): Likewise.
32776 (subsi3_insn): Likewise.
32778 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32780 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32782 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32784 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32785 expressions with MINUS and UNARY ops.
32787 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32790 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32792 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32793 between vec_select and vector argument.
32794 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32795 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32796 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32797 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32798 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32799 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32801 2017-03-13 Richard Biener <rguenther@suse.de>
32804 * params.def (vect-max-peeling-for-alignment): Fix typo.
32806 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32808 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32809 issue that only occurred with binutils below 2.18.
32811 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32813 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32814 refer to binutils 2.11/2.12 minimum.
32816 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32818 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
32819 ftp.kernel.org and simplify binutils requirement.
32821 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
32823 * doc/invoke.texi (Warning Options): Fix spelling of link-time
32825 (Optimize Options): Ditto. Also remove redundancy.
32827 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32829 PR translation/79848
32830 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
32832 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
32833 to G_ to avoid double translation.
32835 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32837 PR translation/79923
32838 * auto-profile.c (get_combined_location): Convert leading
32839 character of diagnostics to lower case and remove trailing period.
32840 (read_profile): Likewise for various diagnostics.
32841 * config/arm/arm.c (arm_option_override): Remove trailing period
32842 from various diagnostics.
32843 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
32844 (msp430_expand_delay_cycles): Likewise.
32846 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32849 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
32850 full command-line argument, rather than just "str".
32851 (aarch64_validate_march): Likewise.
32852 (aarch64_validate_mtune): Likewise.
32854 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
32856 PR rtl-optimization/78911
32857 * lra-assigns.c (must_not_spill_p): New function.
32858 (spill_for): Use it.
32860 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32862 PR tree-optimization/79981
32863 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
32864 ATOMIC_COMPARE_EXCHANGE ifn result.
32865 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
32866 IFN_ATOMIC_COMPARE_EXCHANGE.
32868 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32871 * opts.c (parse_sanitizer_options): Add missing question mark to
32872 "did you mean" message.
32874 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32876 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
32878 (VMULEUH_UNS): Likewise.
32879 (VMULOUB_UNS): Likewise.
32880 (VMULOUH_UNS): Likewise.
32881 * config/rs6000/rs6000.c (builtin_function_type): Remove
32882 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
32884 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32887 * read-rtl-function.c (function_reader::read_rtx_operand): Update
32888 x with result of extra_parsing_for_operand_code_0.
32889 (function_reader::extra_parsing_for_operand_code_0): Convert
32890 return type from void to rtx, returning x. When reading
32891 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
32892 larger size containing struct block_symbol.
32894 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
32896 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
32897 -mfloat128-hardware without -m64.
32899 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
32902 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
32903 entries to the case statement that marks unsigned arguments to
32904 overloaded functions.
32906 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32908 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
32909 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
32911 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
32914 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
32915 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
32917 2017-03-10 Martin Liska <mliska@suse.cz>
32921 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
32922 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
32925 2017-03-10 Olivier Hainque <hainque@adacore.com>
32927 * tree-switch-conversion (array_value_type): Start by resetting
32928 candidate type to it's main variant.
32930 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32932 PR rtl-optimization/79909
32933 * combine.c (try_combine): Use simplify_replace_rtx on individual
32934 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
32935 of the whole CALL_INSN_FUNCTION_USAGE.
32937 PR tree-optimization/79972
32938 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
32939 get_range_info on SSA_NAMEs. Formatting fixes.
32941 2017-03-10 Richard Biener <rguenther@suse.de>
32942 Jakub Jelinek <jakub@redhat.com>
32944 PR tree-optimization/77975
32945 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
32946 edge to be constant.
32947 (get_val_for): For constant x return it. Formatting fix.
32948 (loop_niter_by_eval): Avoid pointless looping if the next iteration
32949 would use the same bases as the current one.
32951 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32953 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
32954 instead of vec_select for V1TImode.
32955 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
32957 (VSX_LE_128): Add V1TI to this mode iterator.
32958 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
32959 (*vsx_le_perm_store_<mode>): Likewise.
32960 (pre-reload splitter for VSX stores): Likewise.
32961 (post-reload splitter for VSX stores): Likewise.
32962 (*vsx_xxpermdi2_le_<mode>): Likewise.
32963 (*vsx_lxvd2x2_le_<mode>): Likewise.
32964 (*vsx_stxvd2x2_le_<mode>): Likewise.
32966 2017-03-09 Michael Eager <eager@eagercon.com>
32968 Correct failures with --enable-checking=yes,rtl.
32970 * config/microblaze/microblaze.c (microblaze_expand_shift):
32971 Replace GET_CODE test with CONST_INT_P and INTVAL test with
32972 test for const0_rtx.
32973 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
32974 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
32976 2017-03-09 Richard Biener <rguenther@suse.de>
32978 PR tree-optimization/79977
32979 * graphite-scop-detection.c (scop_detection::merge_sese):
32980 Handle the case of extra exits to blocks dominating the entry.
32982 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
32984 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
32987 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
32989 PR rtl-optimization/79949
32990 * lra-constraints.c (process_alt_operands): Check memory when
32991 trying to predict a cycle. Print about the overall increase.
32993 2017-03-09 Richard Biener <rguenther@suse.de>
32995 PR middle-end/79971
32996 * gimple-expr.c (useless_type_conversion_p): Preserve
32997 TYPE_SATURATING for fixed-point types.
32999 2017-03-09 Richard Biener <rguenther@suse.de>
33002 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33003 alignment of BLKmode params.
33005 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33008 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33009 (VALL_NO_V2Q): Likewise.
33011 * config/aarch64/aarch64-simd.md
33012 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33014 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33015 VALL_NO_V2Q mode iterator.
33016 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33018 2017-03-09 Martin Liska <mliska@suse.cz>
33020 PR tree-optimization/79631
33021 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33022 tree_int_cst_sign_bit just for INTEGER constants.
33024 2017-03-09 Martin Liska <mliska@suse.cz>
33028 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33031 2017-03-09 Marek Polacek <polacek@redhat.com>
33034 * tree.c (inchash::add_expr): Handle TREE_VEC.
33036 2017-03-09 Martin Liska <mliska@suse.cz>
33039 (chkp_narrow_size_and_offset): New function.
33040 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33041 (void chkp_parse_bit_field_ref): New function.
33042 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33043 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33045 2017-03-09 Martin Liska <mliska@suse.cz>
33048 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33049 (chkp_find_bounds_1): Remove gcc_unreachable.
33051 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33054 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33055 BUILT_IN_SYNC*, determine the access type from the size suffix and
33056 always build a MEM_REF with that type. Handle forgotten
33057 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33060 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33061 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33062 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33063 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33064 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33065 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33066 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33067 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33068 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33069 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33070 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33071 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33072 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33073 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33074 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33075 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33076 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33077 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33078 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33079 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33080 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33081 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33082 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33083 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33084 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33085 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33086 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33087 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33088 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33089 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33090 definitions outside of __OPTIMIZE__ guarded section.
33093 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33094 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33095 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33096 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33099 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33101 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33102 ("vfenez<mode>"): Add missing constraints.
33104 2017-03-08 Martin Sebor <msebor@redhat.com>
33107 * config/nds32/nds32.c (nds32_option_override):
33108 Fix misspelled diagnostic.
33110 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33113 * gimplify.c (gimplify_omp_for): Replace index var in outer
33114 taskloop statement with an artificial variable and add
33115 OMP_CLAUSE_PRIVATE clause for it.
33117 2017-03-08 Richard Biener <rguenther@suse.de>
33119 PR tree-optimization/79955
33120 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33121 for accesses that are completely outside of the variable.
33123 2017-03-08 Andrew Haley <aph@redhat.com>
33125 PR tree-optimization/79943
33126 * tree-ssa-loop-split.c (compute_new_first_bound): When
33127 calculating the new upper bound, (END-BEG) should be added, not
33130 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33132 * config/avr/avr.md (setmemhi): Make sure match_dup
33133 operand number comes before match_scratch.
33135 2017-03-08 Richard Biener <rguenther@suse.de>
33137 PR tree-optimization/79920
33138 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33139 with ncopies == 1 to ...
33140 (vect_transform_slp_perm_load): ... here. Properly compute
33141 all element loads by iterating VF times over the group. Do
33142 not handle ncopies (computed in a broken way) in
33143 vect_create_mask_and_perm.
33145 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33148 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33149 is a uniform vector, use uniform_vector_p return value instead of
33150 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33152 2017-03-07 Marek Polacek <polacek@redhat.com>
33154 PR middle-end/79809
33155 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33156 (alloca_call_type): Likewise.
33158 2017-03-07 Martin Liska <mliska@suse.cz>
33160 * gcov.c (process_args): Put comment to correct location.
33162 2017-03-07 Martin Liska <mliska@suse.cz>
33164 PR middle-end/68270
33165 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33166 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33167 (chkp_narrow_bounds_for_field): Likewise.
33168 (chkp_parse_array_and_component_ref): Pass one more argument to
33171 2017-03-07 Richard Biener <rguenther@suse.de>
33173 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33176 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33178 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33179 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33181 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33184 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33185 to end of description.
33186 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33188 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33190 PR rtl-optimization/79901
33191 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33193 (*avx512f_<code><mode>3<mask_name>): ... this.
33194 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33195 iterator instead of VI8_AVX2_AVX512BW.
33197 PR rtl-optimization/79901
33198 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33199 min/max expander, expand it using expand_vec_cond_expr.
33202 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33205 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33208 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33209 to void * for PCH reasons.
33210 * dwarf2out.c (output_loc_operands, output_die): Cast
33211 v.val_vec.array to unsigned char *.
33213 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33216 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33219 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33221 PR rtl-optimization/79571
33222 * lra-constraints.c (process_alt_operands): Calculate static
33223 reject and subtract it from overall when only addresses will be
33226 2017-03-06 Julia Koval <julia.koval@intel.com>
33229 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33230 incoming stack boundary to 128 for 64-bit targets.
33232 2017-03-06 Richard Biener <rguenther@suse.de>
33234 PR tree-optimization/79894
33235 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33236 to NULL after folding it.
33238 2017-03-06 Richard Biener <rguenther@suse.de>
33240 PR tree-optimization/79824
33241 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33242 check disabling peeling for gaps.
33244 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33246 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33247 attributes): Document gettimeofday.
33249 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33251 * config/s390/s390.c (s390_option_override_internal): Set
33252 PARAM_MIN_VECT_LOOP_BOUND
33254 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33256 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33257 * config/s390/s390.md: Likewise.
33259 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33262 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33263 (<avx2_avx512>_perm<mode>): Rename to ...
33264 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33266 (<avx512>_perm<mode>_mask): Rename to ...
33267 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33269 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33270 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33271 instead of VI8F_256_512.
33272 (avx512f_perm<mode>): New define_expand.
33273 (avx512f_perm<mode>_mask): Likewise.
33274 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33275 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33277 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33279 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33280 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33282 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33284 2017-03-06 Martin Liska <mliska@suse.cz>
33287 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33288 when having a SSA NAME w/o VAR_DECL assigned to it.
33290 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33292 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33293 msa_dpsub_<su>_d): Fix MODE for vec_select.
33295 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33297 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33299 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33301 2017-03-06 Richard Biener <rguenther@suse.de>
33303 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33304 * plugin.c (register_plugin_info): Likewise.
33305 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33307 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33309 * config/i386/sse.md (sse_storehps, sse_storelps,
33310 avx_<castmode><avxsizesuffix>_<castmode>,
33311 avx512f_<castmode><avxsizesuffix>_<castmode>,
33312 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33313 in condition that at least one operand is not a MEM.
33315 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33317 PR middle-end/79805
33318 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33319 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33321 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33322 gimple_call_nothrow_p flag based on whether original builtin can throw.
33323 If it can, emit following stmts on the fallthrough edge.
33324 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33325 don't create new bb if inserting just debug stmts on the edge, try to
33326 insert them on the fallthru bb or just reset debug stmts.
33328 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33331 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33332 restore recog_data (including the operand rtxes inside it) around
33333 the call to get_insn_template.
33335 2017-03-03 Martin Sebor <msebor@redhat.com>
33337 PR tree-optimization/79699
33338 * context.c (context::~context): Free MPFR caches to avoid
33339 a memory leak on program exit.
33341 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33343 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33344 Use wide_int::ulow () instead of .elt (0).
33346 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33348 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33349 (*pushxf): Limit oF constraint to 32bit targets and add oC
33350 constraint for 64bit targets.
33351 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33352 (*pushdf): Change rmF constraint to rmC.
33354 2017-03-03 Martin Liska <mliska@suse.cz>
33356 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33357 Remove unused variable.
33359 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33362 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33363 is a memory operand, increase num_memory.
33364 (ix86_expand_args_builtin): Likewise.
33366 2017-03-03 Jan Hubicka <jh@suse.cz>
33369 * ipa-devirt.c (maybe_record_node): Properly handle
33370 __cxa_pure_virtual visibility.
33372 2017-03-03 Martin Liska <mliska@suse.cz>
33374 PR tree-optimization/79803
33375 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33377 (pass_loop_prefetch::execute): Disabled optimization if an
33378 assumption about L1 cache size is not met.
33380 2017-03-03 Martin Liska <mliska@suse.cz>
33382 PR rtl-optimization/79574
33383 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33384 (hash_scan_set): Likewise.
33385 (dump_hash_table): Likewise.
33386 (hoist_code): Likewise.
33388 2017-03-03 Richard Biener <rguenther@suse.de>
33390 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33391 in place of uhigh ().
33392 (fixed_convert_from_real): Likewise.
33394 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33397 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33399 2017-03-03 Richard Biener <rguenther@suse.de>
33401 PR middle-end/79818
33402 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33403 TYPE_OVERFLOW_UNDEFINED check.
33405 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33407 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33409 (vector_ae_<mode>_p): Likewise.
33410 (vector_nez_<mode>_p): Likewise.
33411 (vector_ne_v2di_p): Likewise.
33412 (vector_ae_v2di_p): Likewise.
33413 (vector_ne_<mode>_p): Likewise.
33414 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33416 (vsx_tsqrt<mode>2_fe): Likewise.
33418 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33421 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33423 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33425 PR rtl-optimization/79780
33426 * cprop.c (one_cprop_pass): When second and further conditional trap
33427 in a single basic block is turned into an unconditional trap, turn it
33428 into a deleted note to avoid RTL verification failures.
33430 2017-03-02 Richard Biener <rguenther@suse.de>
33432 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33434 2017-03-02 Richard Biener <rguenther@suse.de>
33436 PR tree-optimization/79345
33438 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33439 param and abort the walk, returning -1 if it is hit.
33440 (walk_aliased_vdefs): Take a limit param and pass it on.
33441 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33442 defaulting to 0 and return a signed int.
33443 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33444 (check_defs): New helper.
33445 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33446 about uninitialized memory.
33447 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33448 bogus uninitialized warning.
33449 (fixed_convert_from_real): Likewise.
33451 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33453 PR tree-optimization/66768
33454 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33455 iv_use if base object can't be determined.
33457 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33459 PR tree-optimization/79345
33460 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33461 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33462 (get_pattern_stats): Initialize it.
33463 * genemit.c (gen_expand): Verify match_scratch numbers come after
33464 match_operand/match_dup numbers.
33465 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33466 match_scratch numbers.
33467 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33469 * config/s390/s390.md (trunctdsd2): Likewise.
33471 2017-03-02 Richard Biener <rguenther@suse.de>
33473 * wide-int.h (wide_int_storage::operator=): Implement in terms
33476 2017-03-02 Richard Biener <rguenther@suse.de>
33478 PR tree-optimization/79777
33479 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33480 the to insert expression to sth existing.
33482 2017-03-01 Martin Sebor <msebor@redhat.com>
33484 PR middle-end/79692
33485 * gimple-ssa-sprintf.c
33486 (directive::known_width_and_precision): New function.
33487 (format_integer): Use it.
33488 (get_mpfr_format_length): Consider the full range of precision
33489 when computing %g output with the # flag. Set the likely byte
33490 count to 3 rather than 1 when precision is indeterminate.
33491 (format_floating): Correct the lower bound of precision.
33493 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33495 * doc/invoke.texi: Document default code model for 64-bit Linux.
33497 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33500 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33501 udiv rather than div since input pattern is unsigned.
33503 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33505 * config/i386/i386.c (print_reg): Warn for values of
33506 unsupported size in integer register.
33508 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33511 * config/rs6000/predicates.md (current_file_function_operand): Do
33512 not allow self calls to be local if the function is replaceable.
33514 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33517 * config/rs6000/altivec.h (vec_ctz and others): Change the
33518 preprocessor macro that controls conditional compilation from
33519 _ARCH_PWR9 to __POWER9_VECTOR__.
33520 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33521 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33522 control (instead of _ARCH_PWR9 control) so that template
33523 definition uses power9-specific function.
33524 (vec_any_eq): Likewise.
33525 (vec_all_ne): Change macro definition to use a power9-specific
33526 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33527 _ARCH_PWR9 control).
33528 (vec_any_eq) Likewise.
33529 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33530 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33531 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33532 support for xvcmpnedp instruction.
33533 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33534 macro expansion so that Power9 implementation of vec_all_ne does
33535 not use the AltiVec predicate framework.
33536 (VCMPNEH_P): Likewise.
33537 (VCMPNEW_P): Likewise.
33538 (VCMPNED_P): Likewise.
33539 (VCMPNEFP_P): Likewise.
33540 (VCMPNEDP_P): Likewise.
33541 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33542 implementation of vec_any_eq to not use AltiVec predicate
33544 (VCMPAEH_P): Likewise.
33545 (VCMPAEW_P): Likewise.
33546 (VCMPAED_P): Likewise.
33547 (VCMPAEFP_P): Likewise.
33548 (VCMPAEDP_P): Likewise.
33549 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33550 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33551 not use the AltiVec predicate framework.
33552 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33553 of vec_any_eq to not use AltiVec predicate framework.
33554 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33555 support for predefined __POWER9_VECTOR__ macro to indicate that
33556 Power9 instruction selection is enabled.
33557 (altivec_overloaded_builtins): Remove extraneous
33558 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33559 function argument types RS6000_BTI_bool_V16QI and
33560 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33561 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33562 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33563 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33564 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33565 Power9 for implementations of vec_cmpne. Change the signature for
33566 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33567 (representing vec_all_ne) to remove the previously described first
33568 argument of type RS6000_BTI_INTSI, as this was an artifact of
33569 reliance on the AltiVec predicate framework, which is no longer
33570 used in the implementation of these functions. Add
33571 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33572 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33573 since, unlike the AltiVec predicate framework implementation, we
33574 do not share function descriptors between vec_alle and vec_anyeq.
33575 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33576 set of modes that receive special treatment even when
33577 TARGET_P9_VECTOR is true. The special treatment emits code that
33578 does not depend on Power9 instructions.
33579 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33580 define_expand to not rely on AltiVec predicate framework.
33581 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33583 (vector_ne_v2di_p): Change this define_expand to not rely on
33584 AltiVec predicate framework.
33585 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33587 (vector_ne_<mode>_p): Change this define_expand to not rely on
33588 AltiVec predicate framework.
33589 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33591 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33592 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33593 define_insn pattern.
33594 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33595 define_insn pattern because the xvcmpne<VSs>. instruction is not
33597 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33598 instruction is not supported.
33600 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33602 * config/nvptx/nvptx.c: Include intl.h.
33604 2017-03-01 Martin Jambor <mjambor@suse.cz>
33607 * ipa-prop.h (ipa_bits): Removed field known.
33608 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33609 to pointers. Adjusted their comments to warn about their sharing.
33610 (ipcp_transformation_summary): Change bits to a vector of pointers.
33611 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33612 (ipa_get_ipa_bits_for_value): Declare.
33613 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33614 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33615 (ipa_bits_hash_table): Likewise.
33616 (ipa_vr_ggc_hash_traits): Likewise.
33617 (ipa_vr_hash_table): Likewise.
33618 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33619 being pointers and vr_known being removed.
33620 (ipa_set_jf_unknown): Likewise.
33621 (ipa_get_ipa_bits_for_value): New function.
33622 (ipa_set_jfunc_bits): Likewise.
33623 (ipa_get_value_range): New overloaded functions.
33624 (ipa_set_jfunc_vr): Likewise.
33625 (ipa_compute_jump_functions_for_edge): Use the above functions to
33626 construct bits and vr parts of jump functions.
33627 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33628 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33630 (ipcp_grow_transformations_if_necessary): Also allocate
33631 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33633 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33634 them. Fix too long lines.
33635 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33636 vr_known being removed.
33637 (ipa_read_jump_function): Use new setter functions to construct bits
33638 and vr parts of jump functions or set them to NULL.
33639 (write_ipcp_transformation_info): Adjust for bits being pointers.
33640 (read_ipcp_transformation_info): Likewise.
33641 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33643 Include gt-ipa-prop.h.
33644 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33646 (ipcp_store_bits_results): Likewise.
33647 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33648 Do not write to existing jump functions but use a temporary instead.
33650 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33653 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33654 attempt to use its first operand as BIT_FIELD_REF base.
33656 2017-03-01 Richard Biener <rguenther@suse.de>
33658 PR middle-end/79721
33659 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33660 interpolating formula in wrapping arithmetic.
33661 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33663 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33665 PR tree-optimization/79734
33666 * tree-vect-generic.c (expand_vector_condition): Optimize
33667 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33668 Handle VEC_COND_EXPR where comparison has different inner width from
33669 type's inner width.
33671 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33673 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33674 markup, and similar issues. Remove @opindex entries for things
33675 that aren't options. Add missing -mmpy-option entries.
33677 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33679 PR tree-optimization/79737
33680 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33681 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33682 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33683 instead of byte_size. Formatting fix.
33684 (shift_bytes_in_array_right): Formatting fix.
33686 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33689 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33690 condition on optimize for the leaf function test.
33692 2017-02-28 Martin Liska <mliska@suse.cz>
33695 * read-rtl-function.c (function_reader::handle_unknown_directive):
33696 Bail out when one uses -flto.
33698 2017-02-28 Martin Liska <mliska@suse.cz>
33700 * common.opt: Replace space with tabular for options of <number>
33702 * config/i386/i386.opt: Show <number> value for
33703 -mlarge-data-threshold.
33704 * opts.c (print_filtered_help): Do not display number in hexadecimal
33707 2017-02-28 Martin Liska <mliska@suse.cz>
33709 * common.opt: Fix --help=option -Q for options which are of
33712 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33714 * config/i386/i386.c (print_reg): Error out for values
33715 of 8-bit size in invalid integer register.
33717 2017-02-28 Martin Sebor <msebor@redhat.com>
33719 PR tree-optimization/79691
33720 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33722 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33725 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33726 gcc_unreachable with output_operand_lossage.
33728 2017-02-28 Richard Biener <rguenther@suse.de>
33730 PR tree-optimization/79740
33731 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33733 (visit_nary_op): Insert the nary into the hashtable if we
33734 pattern-matched sth.
33735 * tree-ssa-pre.c (eliminate_insert): Robustify.
33737 2017-02-28 Richard Biener <rguenther@suse.de>
33739 PR middle-end/79731
33740 * fold-const.c (decode_field_reference): Reject out-of-bound
33743 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33745 * config/i386/i386.c: Include intl.h.
33746 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33747 instead of just cond ? "..." : "...".
33748 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33749 * coverage.c (read_counts_file): Likewise.
33750 * omp-offload.c: Include intl.h.
33751 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33752 of just cond ? "..." : "...".
33753 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33754 of just cond ? "..." : "...".
33756 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33759 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33761 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33762 'tune for' CPU name.
33763 * config/arm/arm-cpu-data.h: Regenerated.
33765 2017-02-28 Richard Biener <rguenther@suse.de>
33767 PR tree-optimization/79732
33768 * tree-inline.c (expand_call_inline): Do not shadow var.
33770 2017-02-28 Richard Biener <rguenther@suse.de>
33772 PR tree-optimization/79723
33773 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33774 address-space properly.
33776 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33778 * doc/optinfo.texi (Optimization groups): Fix option used for
33780 * doc/invoke.texi (-fopt-info): Document "omp".
33781 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33782 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33783 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33784 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33785 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33787 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33789 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33790 OPTGROUP_OMP with "omp".
33792 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33795 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33796 for arithmetic shift of unsigned V2DI.
33798 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33800 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33801 arc/linux.h headers.
33802 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33803 (LINK_SPEC): Likewise.
33804 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33805 (EXTRA_SPECS): Likewise.
33806 (STARTFILE_SPEC): Likewise.
33807 (ENDFILE_SPEC): Likewise.
33808 (LIB_SPEC): Likewise.
33809 (TARGET_SDATA_DEFAULT): Likewise.
33810 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33811 (MULTILIB_DEFAULTS): Likewise.
33812 (DWARF2_UNWIND_INFO): Likewise.
33813 * config/arc/big.h: New file.
33814 * config/arc/elf.h: Likewise.
33815 * config/arc/linux.h: Likewise.
33816 * config/arc/t-uClibc: Remove.
33818 2017-02-27 Bin Cheng <bin.cheng@arm.com>
33820 PR tree-optimization/77536
33821 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
33822 (tree_transform_and_unroll_loop): Use above function to compute the
33823 estimated niter of unrolled loop and use it when scaling profile.
33824 Also use count info rather than frequency if it's non-zero.
33825 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
33826 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
33827 (vect_transform_loop): Call above function.
33829 2017-02-27 Richard Biener <rguenther@suse.de>
33831 PR tree-optimization/45397
33832 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
33833 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
33834 (visit_nary_op): Add pattern matching for CSEing sign-changed
33835 or truncated operations with wider ones.
33837 2017-02-27 Richard Biener <rguenther@suse.de>
33839 PR tree-optimization/79690
33840 * tree-vect-stmts.c (vectorizable_store): Use vector type
33841 built from the DR with address-space.
33843 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
33845 * doc/invoke.texi (Optimize Options): Refine the description
33846 of asan-use-after-return.
33848 2017-02-25 Alan Modra <amodra@gmail.com>
33850 PR rtl-optimization/79584
33851 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
33852 base, not ad->base_term, the reg within base. Remove assertion
33853 that ad->base == ad->base_term. Replace gen_int_mode using
33854 bogus mode with const0_rtx.
33856 2017-02-25 Jakub Jelinek <jakub@redhat.com>
33858 PR middle-end/79396
33859 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
33860 FMA_EXPR like tcc_binary or tcc_unary.
33862 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
33865 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
33867 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
33868 (output_loc_operands): Handle DW_OP_call_ref and
33869 DW_OP_GNU_variable_value.
33870 (struct variable_value_struct): New type.
33871 (struct variable_value_hasher): Likewise.
33872 (variable_value_hash): New variable.
33873 (string_types): Remove.
33874 (copy_loc_descr): New function.
33875 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
33876 (prepend_loc_descr_to_each): New function.
33877 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
33878 instead of add_loc_descr_to_each if the first argument is single
33879 location list and the second has multiple.
33880 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
33881 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
33882 when looking for variable value which doesn't have other location info.
33883 (loc_list_from_tree): Formatting fix.
33884 (gen_array_type_die): Simplify DW_AT_string_length handling.
33885 (adjust_string_types): Remove.
33886 (gen_subprogram_die): Don't call adjust_string_types nor test/set
33887 string_types. Call resolve_variable_values.
33888 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
33889 (resolve_addr_in_expr): Likewise. Add A argument.
33890 (copy_deref_exprloc): Remove deref argument. Adjust for the
33891 original expression being DW_OP_GNU_variable_value with optionally
33892 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
33893 optionally after it.
33894 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
33895 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
33896 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
33897 (variable_value_hasher::hash, variable_value_hasher::equal): New
33899 (resolve_variable_value_in_expr, resolve_variable_value,
33900 resolve_variable_values, note_variable_value_in_expr,
33901 note_variable_value): New functions.
33902 (dwarf2out_early_finish): Call note_variable_value on all toplevel
33905 2017-02-24 Jakub Jelinek <jakub@redhat.com>
33908 * opts.h (handle_generated_option): Add GENERATED_P argument.
33909 * opts-common.c (handle_option): Adjust function comment.
33910 (handle_generated_option): Add GENERATED_P argument, pass it to
33912 (control_warning_option): Pass false to handle_generated_option
33914 * opts.c (maybe_default_option): Pass true to handle_generated_option
33916 * optc-gen.awk: Likewise.
33918 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33920 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
33921 a REG, look at the REG it is a SUBREG of.
33922 (splitter for cmpeqsi_t): Ditto.
33924 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33926 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
33927 the special USEs with the pattern of the insn, not the insn itself.
33929 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
33932 * doc/invoke.texi: Document -mload-store-pairs.
33934 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33935 Sandra Loosemore <sandra@codesourcery.com>
33937 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
33938 argument isn't a CONST_INT.
33939 (nios2_alternate_compare_const): Assert op is a CONST_INT.
33940 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
33941 (nios2_validate_compare): Bypass alternate compare logic if *op2
33942 is not a CONST_INT.
33943 (ldstwm_operation_p): Return false if first_base is not a REG or
33944 if first_offset is not a CONST_INT.
33946 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33948 * config/cris/cris.md: Use correct operand in a define_peephole2.
33950 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33952 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
33954 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33956 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
33957 this_insn if it is an INSN or JUMP_INSN.
33958 (force_offsettable): Look at base, not at addr.
33959 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
33960 on things that aren't necessarily CONST_INTs.
33962 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
33964 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
33965 -mfpmath=sse is the default also for x86-32 targets with SSE2
33966 instruction set when @option{-ffast-math} is enabled
33968 2017-02-24 Jeff Law <law@redhat.com>
33970 PR rtl-optimizatoin/79286
33971 * ira.c (update_equiv_regs): Drop may_trap_p exception to
33974 2017-02-24 Richard Biener <rguenther@suse.de>
33976 PR tree-optimization/79389
33977 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
33980 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
33982 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
33983 function comment to reflect reality.
33984 (loop_exits_before_overflow): Fix typo in function description.
33986 2017-02-24 Richard Biener <rguenther@suse.de>
33988 PR tree-optimization/79389
33989 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
33990 properly that a threading opportunity exists. Detect conditional
33991 copy/constant propagation opportunities.
33993 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
33995 * config/visium/visium.md (type): Add trap.
33996 (b): New mode attribute.
33997 (*btst): Rename into...
33998 (*btst<mode>): ...this and adjust.
33999 (*cbranchsi4_btst_insn): Rename into...
34000 (*cbranch<mode>4_btst_insn): ...this and adjust.
34001 (trap): New define_insn.
34003 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34005 PR tree-optimization/79389
34006 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34007 (noce_reversed_cond_code): New function.
34008 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34009 reversed_comparison_code. Formatting fix.
34010 (noce_try_store_flag): Test rev_cond != NULL in addition to
34011 reversed_comparison_code.
34012 (noce_try_store_flag_constants): Likewise.
34013 (noce_try_store_flag_mask): Likewise.
34014 (noce_try_addcc): Use rev_cond if non-NULL instead of
34015 reversed_comparison_code.
34016 (noce_try_cmove_arith): Likewise. Formatting fixes.
34017 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34018 (noce_find_if_block): Initialize rev_cond.
34019 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34020 instead of false as last argument never attempt to reverse it
34023 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34025 PR tree-optimization/79663
34026 * tree-predcom.c (combine_chains): Process refs in reverse order
34027 only for ZERO length chains, and add explaining comment.
34029 2017-02-23 Jeff Law <law@redhat.com>
34031 PR tree-optimization/79578
34032 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34033 in call to operand_equal_p.
34035 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34038 * config/i386/cpuid.h: Fix another undefined behavior.
34040 2017-02-23 Richard Biener <rguenther@suse.de>
34042 PR tree-optimization/79683
34043 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34044 vector types for data-refs.
34046 2017-02-23 Martin Liska <mliska@suse.cz>
34048 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34050 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34052 PR middle-end/79665
34053 * internal-fn.c (get_range_pos_neg): Moved to ...
34054 * tree.c (get_range_pos_neg): ... here. No longer static.
34055 * tree.h (get_range_pos_neg): New prototype.
34056 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34057 are known to be in between 0 and signed maximum inclusive, try to
34058 expand both unsigned and signed divmod and use the cheaper one from
34061 2017-02-22 Jeff Law <law@redhat.com>
34063 PR tree-optimization/79578
34064 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34065 to compare base operands.
34067 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34070 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34071 gpc_reg_operand instead of fpr_reg_operand.
34073 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34075 * config/mips/mips.c (mips_return_in_memory): Force FP
34076 vector types to be returned in memory for o32 ABI.
34078 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34080 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34081 instead of DW_TAG_member for static data member declarations and don't
34082 set no_linkage_name for static inline data members.
34083 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34086 2017-02-22 Martin Liska <mliska@suse.cz>
34088 * doc/invoke.texi: Replace inequality signs with square brackets
34091 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34094 * lra-constraints.c (simplify_operand_subreg): Handle
34095 WORD_REGISTER_OPERATIONS targets.
34097 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34100 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34101 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34102 elimination by swapping fld*.
34104 2017-02-22 Richard Biener <rguenther@suse.de>
34106 PR tree-optimization/79673
34107 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34108 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34109 irrelevant address-space qualifiers and avoiding a
34110 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34112 2017-02-22 Richard Biener <rguenther@suse.de>
34114 PR tree-optimization/79666
34115 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34116 to not symbolically negate if that may introduce undefined
34119 2017-02-22 Martin Liska <mliska@suse.cz>
34122 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34123 * data-streamer-out.c (streamer_write_gcov_count_stream):
34125 * value-prof.c (stream_out_histogram_value): Make assert more
34126 precise based on type of counter.
34128 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34131 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34132 Use nonimmediate_operand instead of memory_operand for operand 1.
34133 (float-extend standard_x87sse_constant_load splitter): Ditto.
34135 2017-02-21 Jeff Law <law@redhat.com>
34137 PR tree-optimization/79621
34138 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34139 blocks with edges to themselves.
34141 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34144 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34145 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34146 Use gimple_call_builtin_p.
34149 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34150 on temporarily removed DEBUG_INSNs.
34152 PR tree-optimization/79649
34153 * tree-loop-distribution.c (classify_partition): Give up on
34154 non-generic address space loads/stores.
34156 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34158 * doc/loop.texi (Loop manipulation): Remove nonexistent
34159 tree_ssa_loop_version from the documentation.
34160 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34162 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34165 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34166 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34167 * config/rs6000/rs6000.c: Include except.h.
34168 (rs6000_expand_split_stack_prologue): Call
34169 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34171 2017-02-21 Martin Jambor <mjambor@suse.cz>
34174 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34175 have been analyzed.
34177 2017-02-21 Martin Jambor <mjambor@suse.cz>
34179 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34180 for backward compatibility only.
34181 * doc/invoke.texi (Option Summary): Remove all references to
34182 -fipa-cp-alignment.
34184 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34188 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34190 * lra-constraints.c (curr_insn_transform): Handle
34191 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34193 2017-02-21 Martin Liska <mliska@suse.cz>
34195 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34197 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34199 PR translation/79638
34200 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34202 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34205 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34206 (arm_function_ok_for_sibcall): Return false for an indirect call by
34207 descriptor if all the argument registers are used.
34208 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34209 alignment of the function.
34211 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34213 PR tree-optimization/61441
34214 * simplify-rtx.c (simplify_const_unary_operation): For
34215 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34216 the sNaN unmodified.
34218 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34220 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34221 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34222 instead of SYSTEM_HEADER_DIR.
34224 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34225 Martin Liška <mliska@suse.cz>
34227 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34228 Fix typos and grammar, use active voice, and clarify.
34230 2017-02-20 Marek Polacek <polacek@redhat.com>
34232 PR middle-end/79537
34233 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34236 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34238 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34241 * config/i386/i386.c (ix86_expand_builtin): Handle
34242 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34243 ix86_builtins_isa[fcode].isa as a requirement of those
34244 flags and any other flag in the bitmask.
34245 (ix86_init_mmx_sse_builtins): Use 0 instead of
34246 ~OPTION_MASK_ISA_64BIT as mask.
34247 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34248 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34249 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34250 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34252 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34255 * lra-constraints.c (split_reg): Check requested split mode
34256 is supported by the register.
34258 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34260 * lra-constraints.c (simplify_operand_subreg): Remove early
34263 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34266 * lra-constraints.c (curr_insn_transform): Tighten condition
34267 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34269 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34272 * lra-constraints.c (curr_insn_transform): Handle
34273 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34275 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34278 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34280 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34282 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34285 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34288 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34290 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34293 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34295 * final.c (last_columnnum, override_columnnum): New variables.
34296 (final_start_function): Set last_columnnum, pass it to begin_prologue
34297 hook and pass 0 to dwarf2out_begin_prologue.
34298 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34299 to source_line debug hook.
34300 (notice_source_line): Compute last_columnnum and for debug_column_info
34301 return true on column changes.
34302 * debug.h (struct gcc_debug_hooks): Add column argument to
34303 source_line and begin_prologue hooks.
34304 (debug_nothing_int_charstar_int_bool): Remove prototype.
34305 (debug_nothing_int_int_charstar,
34306 debug_nothing_int_int_charstar_int_bool): New prototypes.
34307 (dwarf2out_begin_prologue): Add column argument.
34308 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34309 begin_prologue hooks.
34310 (debug_nothing_int_charstar_int_bool): Remove.
34311 (debug_nothing_int_int_charstar,
34312 debug_nothing_int_int_charstar_int_bool): New functions.
34313 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34314 through to dwarf2out_source_line.
34315 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34316 (dwarf2out_source_line): Add column argument, emit it if requested.
34317 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34319 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34320 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34321 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34322 through to dwarf2out_begin_prologue.
34323 (vmsdbgout_source_line): Add column argument, pass it through to
34324 dwarf2out_source_line.
34325 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34326 dbxout_source_line caller.
34327 (dbxout_source_line): Add column argument.
34329 * common.opt (gno-column-info, gcolumn-info): New options.
34330 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34331 (check_die): Also test for multiple DW_AT_decl_column attributes.
34332 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34333 DW_AT_decl_column if requested.
34334 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34336 (gen_variable_die): Likewise.
34337 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34338 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34341 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34342 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34343 (ix86_handle_option): Handle OPT_m3dnowa.
34344 * doc/invoke.texi (-m3dnowa): Document.
34345 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34346 -m3dnowa instead of -m3dnow -march=athlon.
34349 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34350 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34352 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34355 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34356 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34357 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34358 generator for vsx_xxpermdi_<mode>_be.
34359 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34360 force big-endian semantics.
34361 (vsx_xxpermdi_<mode>_be): New define_expand with same
34362 implementation as previous version of vsx_xxpermdi_<mode>.
34364 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34366 PR tree-optimization/79327
34367 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34368 variable, its initialization and use.
34370 2017-02-17 Julia Koval <julia.koval@intel.com>
34372 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34373 (OPTION_MASK_ISA_PKU_UNSET): New.
34374 (ix86_handle_option): Handle -mrdpid.
34375 * config/i386/cpuid.h (bit_RDPID): New.
34376 * config/i386/driver-i386.c (host_detect_local_cpu):
34377 Detect RDPID feature.
34378 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34379 * config/i386/i386-c.c (ix86_target_macros_internal):
34381 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34382 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34383 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34384 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34385 * config/i386/i386.md (define_insn "rdpid"): New.
34386 * config/i386/i386.opt Add -mrdpid.
34387 * config/i386/immintrin.h (_rdpid_u32): New.
34389 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34391 PR rtl-optimization/79541
34392 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34393 instead of transforming it into USE.
34395 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34397 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34398 If HONOR_SNANS (SFmode) force the input to a register.
34399 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34400 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34401 an frsp or similar insn.
34403 2017-02-17 Martin Liska <mliska@suse.cz>
34405 PR rtl-optimization/79577
34406 * params.def (selsched-max-sched-times): Increase minimum to 1.
34408 2017-02-17 Martin Liska <mliska@suse.cz>
34410 PR rtl-optimization/79574
34411 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34413 2017-02-17 Martin Liska <mliska@suse.cz>
34415 PR tree-optimization/79529
34416 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34417 ssa_defined_default_def_p to handle cases which are implicitly
34419 * tree-ssa.c (ssa_defined_default_def_p): New function.
34420 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34421 which are implicitly defined.
34422 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34424 2017-02-17 Richard Biener <rguenther@suse.de>
34426 PR middle-end/79576
34427 * params.def (max-ssa-name-query-depth): Limit to 10.
34429 2017-02-17 Richard Biener <rguenther@suse.de>
34431 PR tree-optimization/79552
34432 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34435 2017-02-17 Richard Biener <rguenther@suse.de>
34438 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34440 2017-02-17 Marek Polacek <polacek@redhat.com>
34442 PR middle-end/79536
34443 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34444 (fold_negate_expr): New wrapper.
34446 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34448 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34449 Correct terminology and de-emphasize pre-standard behavior.
34451 2017-02-16 Alan Modra <amodra@gmail.com>
34453 PR rtl-optimization/79286
34454 * ira.c (def_dominates_uses): New function.
34455 (update_equiv_regs): Don't create an equivalence for insns that
34456 may trap where the register def does not dominate the use.
34458 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34460 PR rtl-optimization/78127
34461 * lra.c (lra): Call lra_eliminate before finish the loop after
34464 2017-02-16 Richard Biener <rguenther@suse.de>
34466 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34468 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34469 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34470 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34471 (isl_val_int_from_wi): New function.
34472 (extract_affine_gmp): Rename to ...
34473 (extract_affine_wi): ... this, take a widest_int.
34474 (extract_affine_int): Just wrap extract_affine_wi.
34475 (add_param_constraints): Use isl_val_int_from_wi.
34476 (add_loop_constraints): Likewise, and extract_affine_wi.
34478 2017-02-15 Jeff Law <law@redhat.com>
34480 PR middle-end/79521
34481 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34482 ira_init_register_move_cost_if_necessary.
34484 2017-02-15 Martin Sebor <msebor@redhat.com>
34486 PR middle-end/32003
34487 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34488 removed in a prior commit.
34490 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34492 PR tree-optimization/79347
34493 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34494 counters during peeling.
34496 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34498 * Makefile.in (site.exp): Remove "set ISLVER".
34500 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34503 * real.c (real_from_integer): Call real_convert even for decimal.
34505 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34508 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34510 2017-02-14 Andrew Pinski <apinski@cavium.com>
34512 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34513 cores and change the partno/implementer to be correct.
34514 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34515 the 'B" as the implementer.
34516 * config/aarch64/aarch64-tune.md: Regenerate.
34518 2017-02-14 Carl Love <cel@us.ibm.com>
34520 * config/rs6000/rs6000.c: Add case statement entry to make the
34521 xvcvuxdsp built-in argument unsigned.
34522 * config/rs6000/vsx.md: Fix the source and return operand types so they
34523 match the instruction definitions from the ISA document. Fix typo
34524 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34527 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34530 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34531 member early_clobber_alts.
34532 * lra-lives.c (reg_early_clobber_p): New.
34533 (process_bb_lives): Use it.
34534 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34535 (debug_operand_data): Initialize early_clobber_alts.
34536 (setup_operand_alternative): Set up early_clobber_alts.
34537 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34538 alternatives to new_insn_reg.
34539 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34541 (lra_update_insn_regno_info): Pass the new arg.
34543 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34545 PR middle-end/79505
34546 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34547 (new_oacc_loop_raw): Don't clear already cleared fields.
34550 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34551 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34552 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34554 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34557 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34559 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34562 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34563 the extra instruction to the right place to store 128-bit constant
34566 2017-02-14 Martin Sebor <msebor@redhat.com>
34568 PR middle-end/79448
34569 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34570 warning for strings of unknown length.
34572 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34574 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34576 2017-02-14 Jeff Law <law@redhat.com>
34579 * ira-costs.c (scan_one_insn): Initialize register move costs
34580 for pseudos seen in USE/CLOBBER insns.
34582 PR tree-optimization/79095
34583 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34584 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34585 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34586 if the operands are known to be not equal, then the resulting range
34588 (intersect_ranges): If the new range is ~[0,0] and the old range is
34589 wide, then prefer ~[0,0].
34590 * tree-vrp.c (overflow_comparison_p_1): New function.
34591 (overflow_comparison_p): New function.
34592 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34593 if NAME is used in an overflow test.
34594 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34595 overflow check that can be expressed as an equality test, then adjust
34596 ops to be that equality test.
34598 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34600 * config/s390/s390-builtin-types.def: Remove flags argument.
34601 * config/s390/s390.c (s390_init_builtins): Likewise.
34603 2017-02-14 Martin Liska <mliska@suse.cz>
34605 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34606 vector. Fix trailing white spaces.
34608 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34610 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34613 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34615 PR rtl-optimization/68664
34616 * config/arm/arm.c (arm_sched_can_speculate_insn):
34617 New function. Declare prototype.
34618 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34620 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34622 PR rtl-optimization/68664
34623 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34625 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34627 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34629 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34630 max skip bytes for function, loop and jump.
34632 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34634 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34635 ABS_EXPR for gimple dump.
34637 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34640 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34642 PR tree-optimization/79408
34643 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34644 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34645 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34646 also if rhs1 is INTEGER_CST.
34648 2017-02-14 Richard Biener <rguenther@suse.de>
34650 PR middle-end/79432
34651 * tree-into-ssa.c (insert_phi_nodes): When the function can
34652 have abnormal edges rewrite SSA names with broken use-def
34653 dominance out of SSA and register them for PHI insertion.
34655 2017-02-13 Martin Sebor <msebor@redhat.com>
34657 PR middle-end/79496
34658 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34659 clearing info.nowrite flag when snprintf size argument is a range.
34661 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34663 * cprop.c (cprop_jump): Add missing space in string literal.
34664 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34665 (get_constraint_for_component_ref): Likewise.
34666 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34667 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34668 * lra-constraints.c (process_alt_operands): Likewise.
34669 * ipa-inline.c (inline_small_functions): Likewise.
34670 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34671 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34672 * trans-mem.c (diagnose_tm_1_op): Likewise.
34673 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34674 (grid_parallel_clauses_gridifiable): Likewise.
34676 * config/nvptx/mkoffload.c (process): Add space in between
34679 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34680 "MOD4_SSE_REGS" and "ALL_REGS".
34682 * spellcheck.c (test_data): Add , in between "foo" and "food".
34684 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34687 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34688 boundary crossing check and subsequent code generation agree.
34690 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34692 * config/aarch64/aarch64.c (has_memory_op): Delete.
34693 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34696 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34698 PR rtl-optimization/79388
34699 PR rtl-optimization/79450
34700 * combine.c (distribute_notes): When removing TEM_INSN for which
34701 corresponding dest has last value recorded, invalidate that last
34704 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34706 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34707 of explicit '@'. Add missing assembly comment marker on branch costs
34710 2017-02-13 Nathan Sidwell <nathan@acm.org>
34712 * gengtype-lex.l (<in_struct>): Add '/'.
34714 2017-02-13 Martin Liska <mliska@suse.cz>
34717 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34719 2017-02-13 Richard Biener <rguenther@suse.de>
34721 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34723 * configure: Re-generate.
34724 * config.in: Likewise.
34725 * graphite-dependences.c: Simplify as if
34726 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34727 * graphite-isl-ast-to-gimple.c: Likewise.
34728 * graphite-optimize-isl.c: Likewise.
34729 * graphite-poly.c: Likewise.
34730 * graphite-sese-to-poly.c: Likewise.
34731 * graphite.h: Likewise.
34732 * toplev.c: Include isl/version.h and use isl_version () for
34733 printing the ISL version.
34734 * doc/install.texi: Update ISL requirement.
34736 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34738 * doc/standards.texi (Standards): Update reference to
34741 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34743 * doc/extend.texi (Named Address Spaces): sourceware.org now
34745 * doc/install.texi (Binaries): Ditto.
34748 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34750 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34751 standard terminology "stringize"/"stringizing" throughout.
34752 * doc/cppinternals.texi: Likewise.
34754 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34756 * doc/extend.texi: Fix some spelling mistakes and typos.
34757 * doc/invoke.texi: Likewise.
34759 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34762 * params.def (inline-min-speedup) Change from 10 to 8.
34764 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34766 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34769 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34772 * ipa-inline-analysis.c (get_minimal_bb): New function.
34773 (record_modified): Use it.
34774 (remap_edge_change_prob): Handle also ancestor functions.
34776 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34778 * doc/contrib.texi (Contributors): Remove broken link into
34779 the Mauve CVS repository.
34781 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34783 PR middle-end/79454
34784 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34785 result computation whenever lhs doesn't have vector mode, not
34786 just when it has BLKmode.
34788 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34790 * doc/makefile.texi (profiledbootstrap): Refer to the
34791 installation instructions only in textual form.
34793 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34796 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34798 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34800 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34801 (Specific): Update mingw-w64 reference.
34803 (Specific): Remove broken link to Renesas RX processor.
34805 2017-02-10 Richard Biener <rguenther@suse.de>
34807 * toplev.c (process_options): Do not mention obsolete graphite
34808 options when printing sorry message about missing graphite support.
34809 Mention -floop-nest-optimize.
34811 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34813 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34814 (vtst_p16): Likewise.
34815 (vtstq_p8): Likewise.
34816 (vtstq_p16): Likewise.
34818 (vtstq_p64): Likewise.
34819 * config/arm/arm_neon.h (vgetq_lane_p64): New.
34820 (vset_lane_p64): New.
34821 (vsetq_lane_p64): New.
34823 2017-02-10 Jakub Jelinek <jakub@redhat.com>
34825 PR tree-optimization/79411
34826 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
34827 stmt operands are SSA_NAMEs used in abnormal phis.
34828 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
34831 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34834 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
34837 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34839 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
34841 2017-02-09 Jakub Jelinek <jakub@redhat.com>
34843 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
34847 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
34848 not arbitrary TREE_CONSTANT.
34851 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
34852 "omp declare target link" attribute unless is_global_var.
34853 * omp-offload.c (find_link_var_op): Likewise.
34855 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34856 Chung-Lin Tang <cltang@codesourcery.com>
34858 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
34860 (gimplify_adjust_omp_clauses): Don't delete TILE.
34861 (gimplify_omp_for): Deal with TILE.
34862 * internal-fn.c (expand_GOACC_TILE): New function.
34863 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
34865 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
34866 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
34868 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
34869 avoid DIV for outermost collapse var.
34870 (expand_oacc_for): Insert tile element loop as needed. Adjust.
34871 Remove out of date comments, fix whitespace.
34872 * omp-general.c (omp_extract_for_data): Deal with tiling.
34873 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
34874 adjust OLF_DIM_BASE value.
34875 (struct omp_for_data): Add tiling field.
34876 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
34877 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
34878 for auto loops. Remove default auto determining, moved to
34879 oacc_loop_fixed_partitions.
34880 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
34881 stmts, add e_mask field.
34882 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
34883 (oacc_thread_numbers): Use oacc_dim_call.
34884 (oacc_xform_tile): New.
34885 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
34886 (finish_oacc_loop): Adjust for ifns vector.
34887 (oacc_loop_discover_walk): Append loop abstraction sites to list,
34888 add case for GOACC_TILE fns.
34889 (oacc_loop_xform_loop): Delete.
34890 (oacc_loop_process): Iterate over call list directly, and add
34891 handling for GOACC_TILE fns.
34892 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
34894 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
34895 vector partitioning to outer loops. Assign 2 partitions to loops
34896 when available. Add TILE handling.
34897 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
34898 (execite_oacc_device_lower): Process GOACC_TILE fns,
34899 ignore unknown specs.
34900 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
34901 * tree.c (omp_clause_num_ops): Adjust TILE ops.
34902 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
34904 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
34906 * configure.ac (ACX_BUGURL): Update.
34907 * configure: Regenerate.
34909 2017-02-09 Richard Biener <rguenther@suse.de>
34911 PR tree-optimization/69823
34912 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
34913 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
34915 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34917 * config/arc/arc-c.def: Add __NPS400__ definition.
34918 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
34919 (TARGET_NPS400): Define.
34921 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34923 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
34925 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
34926 pointer, arch_info.
34927 (arc_cpu_types): Fill the arch_info field with a pointer into the
34928 arc_arch_types table.
34929 (arc_selected_cpu): Declare.
34930 * config/arc/arc.c (arc_selected_cpu): Make global.
34931 (arc_selected_arch): Delete.
34932 (arc_base_cpu): Delete.
34933 (arc_override_options): Remove references to deleted variables,
34934 update access to arch information.
34935 (ARC_OPT): Update access to arch information.
34936 (ARC_OPTX): Likewise.
34937 * config/arc/arc.h (arc_base_cpu): Remove declaration.
34938 (TARGET_ARC600): Update access to arch information.
34939 (TARGET_ARC601): Likewise.
34940 (TARGET_ARC700): Likewise.
34941 (TARGET_EM): Likewise.
34942 (TARGET_HS): Likewise.
34943 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
34946 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
34949 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
34950 condition/operands for integer GE/LE/GEU/LEU operations.
34952 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
34954 PR translation/79397
34955 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
34958 2017-02-08 Martin Jambor <mjambor@suse.cz>
34961 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
34962 whether allocation happened.
34963 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
34964 nothing was allocated.
34966 2017-02-08 Jakub Jelinek <jakub@redhat.com>
34968 PR tree-optimization/79408
34969 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
34970 constant, but SSA_NAME with a known integer range, use the minimum
34971 of that range instead of op1 to determine if modulo can be replaced
34972 with its first operand.
34974 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34976 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
34978 2017-02-08 Richard Biener <rguenther@suse.de>
34980 PR tree-optimization/71824
34981 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34982 Check all loops contained in the merged region.
34984 2017-02-07 Andrew Pinski <apinski@cavium.com>
34986 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
34988 2017-02-07 Andrew Pinski <apinski@cavium.com>
34990 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
34991 (thunderxt88): Likewise.
34992 (thunderxt81): Disable LSE and change v8.1 to v8.
34993 (thunderxt83): Likewise.
34995 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34996 Richard Biener <rguenther@suse.de>
34998 PR middle-end/79399
34999 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35000 type from int to size_t.
35001 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35003 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35005 PR rtl-optimization/79386
35006 * cprop.c (bypass_conditional_jumps): Initialize
35007 bypass_last_basic_block already before splitting bbs after
35008 unconditional traps...
35009 (bypass_conditional_jumps): ... rather than here.
35012 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35013 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35014 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35015 fix -masm=intel patterns.
35017 2017-02-07 Richard Biener <rguenther@suse.de>
35019 PR tree-optimization/79256
35020 PR middle-end/79278
35021 * builtins.c (get_object_alignment_2): Use min_align_of_type
35022 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35023 and ADJUST_FIELD_ALIGN.
35025 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35027 * doc/tm.texi: Regenerate.
35028 * stor-layout.c (layout_decl): Adjust.
35029 (update_alignment_for_field): Likewise.
35030 (place_field): Likewise.
35031 (min_align_of_type): Likewise.
35032 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35033 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35034 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35035 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35036 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35037 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35038 * config/i386/i386.c (x86_field_alignment): Likewise.
35039 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35040 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35041 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35042 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35043 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35044 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35048 2017-01-30 Richard Biener <rguenther@suse.de>
35050 PR tree-optimization/79256
35051 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35052 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35055 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35057 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35058 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35059 builtins to SImode and emit a zero-extend, if necessary.
35061 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35063 * docs/invoke.texi (RISC-V Options): Alphabetize.
35065 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35067 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35070 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35072 * config/riscv/riscv.c: New file.
35073 * common/config/riscv/riscv-common.c: Likewise.
35074 * config.gcc: Likewise.
35075 * config/riscv/constraints.md: Likewise.
35076 * config/riscv/elf.h: Likewise.
35077 * config/riscv/generic.md: Likewise.
35078 * config/riscv/linux.h: Likewise.
35079 * config/riscv/multilib-generator: Likewise.
35080 * config/riscv/peephole.md: Likewise.
35081 * config/riscv/pic.md: Likewise.
35082 * config/riscv/predicates.md: Likewise.
35083 * config/riscv/riscv-builtins.c: Likewise.
35084 * config/riscv/riscv-c.c: Likewise.
35085 * config/riscv/riscv-ftypes.def: Likewise.
35086 * config/riscv/riscv-modes.def: Likewise.
35087 * config/riscv/riscv-opts.h: Likewise.
35088 * config/riscv/riscv-protos.h: Likewise.
35089 * config/riscv/riscv.h: Likewise.
35090 * config/riscv/riscv.md: Likewise.
35091 * config/riscv/riscv.opt: Likewise.
35092 * config/riscv/sync.md: Likewise.
35093 * config/riscv/t-elf-multilib: Likewise.
35094 * config/riscv/t-linux: Likewise.
35095 * config/riscv/t-linux-multilib: Likewise.
35096 * config/riscv/t-riscv: Likewise.
35097 * configure.ac: Likewise.
35098 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35099 Waterman as RISC-V maintainers.
35100 * doc/install.texi: Add RISC-V entries.
35101 * doc/invoke.texi: Add RISC-V options section.
35102 * doc/md.texi: Add RISC-V constraints section.
35103 * configure: Regenerated.
35105 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35108 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35109 false values to be constant vectors with all 0 or all 1 bits set.
35110 (vcondu<mode><mode>): Likewise.
35111 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35113 (fpmask_comparison_operator): Update comment.
35114 (vecint_comparison_operator): New predicate.
35115 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35116 vector conditionals when the true and false values are constant
35117 vectors with all 0 bits or all 1 bits set.
35119 2017-02-06 Martin Sebor <msebor@redhat.com>
35121 PR tree-optimization/79376
35122 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35124 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35126 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35127 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35128 to simplify split condition.
35130 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35132 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35133 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35136 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35138 PR rtl-optimization/68664
35139 * target.def (can_speculate_insn): New hook.
35140 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35141 * doc/tm.texi: Regenerate.
35142 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35143 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35144 (rs6000_sched_can_speculate_insn): New function.
35146 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35148 PR tree-optimization/79284
35149 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35150 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35151 vectorizable_mask_load_store, vectorizable_operation,
35152 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35153 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35154 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35155 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35156 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35157 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35158 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35159 is_gimple_assign (stmt). Replace another such test with
35160 is_gimple_assign (stmt).
35162 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35165 * config/avr/avr.c (rtl-iter.h): Include it.
35166 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35167 (avr_legitimate_combined_insn): ...and implementation.
35169 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35171 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35172 * config/s390/s390.c (s390_const_operand_ok)
35173 (s390_canonicalize_comparison, s390_extract_part)
35174 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35175 (s390_contiguous_bitmask_p, s390_rtx_costs)
35176 (legitimize_pic_address): Likewise.
35177 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35178 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35179 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35180 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35181 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35183 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35185 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35186 REGNO($0) == REGNO($1).
35188 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35190 * config/s390/linux.h(SIZE_TYPE): Add comment.
35192 2017-02-06 Julian Brown <julian@codesourcery.com>
35193 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35194 Virendra Pathak <virendra.pathak@broadcom.com>
35196 * config/aarch64/aarch64-cores.def: Change the scheduler
35198 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35199 * config/aarch64/thunderx2t99.md: New file.
35201 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35203 * doc/standards.texi (Go Language): Update link to language
35206 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35208 * tree-eh.c (lower_resx): Sanitize profile.
35209 (cleanup_empty_eh_move_lp): Likewise.
35211 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35214 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35216 * cfgloopmanip.h (loop_version): Update prototype.
35217 * modulo-sched.c (sms_schedule): Update call of loop_version.
35218 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35219 * tree-parloops.c (gen_parallel_loop): Likewise.
35220 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35221 * tree-ssa-loop-split.c (split_loop): Likewise.
35222 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35223 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35225 2017-02-05 Martin Liska <mliska@suse.cz>
35228 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35230 (print_operand_address): Initialize a struct to zero.
35232 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35234 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35235 garbage collector only in textual form.
35237 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35239 * doc/extend.texi (x86 specific memory model extensions for
35240 transactional memory): Simplify a phrase.
35242 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35245 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35246 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35247 (atomic_storedi_1): Likewise.
35249 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35251 PR tree-optimization/79338
35252 * tree-parloops.c (gather_scalar_reductions): Don't call
35253 vect_analyze_loop_form for loop->inner before destroying loop's
35256 2017-02-03 Martin Sebor <msebor@redhat.com>
35258 PR tree-optimization/79327
35259 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35260 when precision has resulted in leading zeros.
35261 (format_integer): Adjust the likely counter to assume an unknown
35262 argument that may be zero is non-zero.
35264 2017-02-03 Jason Merrill <jason@redhat.com>
35267 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35268 avoid copying non-taken branch.
35270 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35272 PR tree-optimization/79340
35273 * tree-vect-loop.c (vectorizable_reduction): Release
35274 vec_defs elements after safe_splicing them into other vectors.
35277 PR tree-optimization/79327
35278 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35279 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35281 (format_integer): Use wide_int_to_tree instead of build_int_cst
35282 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35283 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35284 of shortest and longest sequence.
35286 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35288 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35289 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35291 2017-02-03 Walter Lee <walt@tilera.com>
35294 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35295 after initial stackframe link reg save.
35296 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35298 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35301 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35302 wu for stxssp alternative.
35304 2017-02-03 Martin Sebor <msebor@redhat.com>
35306 PR tree-optimization/79352
35307 * gimple-fold.c (get_range_strlen): Add argument.
35308 (get_range_strlen): Change return type to bool.
35309 (get_maxval_strlen): Pass in a dummy argument.
35310 * gimple-fold.h (get_range_strlen): Change return type to bool.
35311 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35312 * tree.h (array_at_struct_end_p): Add argument.
35313 * tree.c (array_at_struct_end_p): Handle it.
35315 2017-02-03 Martin Liska <mliska@suse.cz>
35318 * multiple_target.c (create_dispatcher_calls): Redirect edge
35319 from a caller of a dispatcher.
35320 (expand_target_clones): Make the clones local.
35321 (ipa_target_clone): Do both target clones and resolvers.
35322 (ipa_dispatcher_calls): Remove the pass.
35323 (pass_dispatcher_calls::gate): Likewise.
35324 (make_pass_dispatcher_calls): Likewise.
35325 * passes.def (pass_target_clone): Put as very first IPA early
35328 2017-02-03 Martin Liska <mliska@suse.cz>
35330 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35331 in case of a function with ifunc attribute.
35333 2017-02-03 Martin Liska <mliska@suse.cz>
35335 * cgraph.c (cgraph_node::dump): Dump function version info.
35336 * symtab.c (symtab_node::dump_base): Add missing new line.
35338 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35340 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35341 (ifcombine_ifandif): Use it.
35343 2017-02-03 Martin Liska <mliska@suse.cz>
35345 * doc/invoke.texi: Document default value for
35346 use-after-scope-direct-emission-threshold.
35348 2017-02-03 Martin Liska <mliska@suse.cz>
35350 PR tree-optimization/79339
35351 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35352 (format_floating): Likewise.
35354 2017-02-03 Martin Liska <mliska@suse.cz>
35357 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35359 (ipa_node_params_t::remove): Likewise.
35360 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35361 initialization from removed ipa_node_params_t::insert.
35362 (ipa_node_params::~ipa_node_params): Move from removed
35363 ipa_node_params_t::release.
35364 * symbol-summary.h (symbol_summary::m_released): New member.
35365 Do not release a summary twice. Do not allow to call finalizer
35366 for types of a summary that live in GGC memory.
35368 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35370 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35373 2017-02-02 Martin Sebor <msebor@redhat.com>
35375 PR middle-end/79275
35376 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35377 (format_string): Tighten up the range of output for non-constant
35378 strings and correct the expected range for wide non-constant strings.
35380 2017-02-02 Martin Sebor <msebor@redhat.com>
35382 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35384 PR middle-end/32003
35385 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35387 (-fdump-tree-@var): Add to index and document how to come up
35388 with pass-specific option and dump file names.
35389 (-fdump-passes): Clarify where to look for output.
35391 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35393 PR middle-end/77445
35394 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35395 statistics of the analyzed path; allow threading for speed when
35396 any of BBs along the path are optimized for speed.
35398 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35400 PR middle-end/78468
35401 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35402 settings of the virtual registers.
35405 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35407 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35408 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35411 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35413 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35414 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35416 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35418 * config/s390/s390.md: Add missing comments with the expanded
35420 * config/s390/vector.md: Likewise.
35421 * config/s390/vx-builtins.md: Likewise.
35423 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35426 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35427 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35428 conditions on a single line.
35430 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35432 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35433 __S390_VX__ to __VX__.
35435 2017-02-01 Andrew Pinski <apinski@cavium.com>
35437 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35438 stmt_info to record_stmt_cost.
35439 (vect_get_known_peeling_cost): Pass stmt_info if known to
35441 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35442 cpu_vector_cost field into
35443 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35444 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35445 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35446 splitting of scalar_stmt_cost and vec_stmt_cost.
35447 (thunderx_vector_cost): Likewise.
35448 (cortexa57_vector_cost): LIkewise.
35449 (exynosm1_vector_cost): Likewise.
35450 (xgene1_vector_cost): Likewise.
35451 (thunderx2t99_vector_cost): Improve after the splitting of the two
35453 (aarch64_builtin_vectorization_cost): Update for the splitting of
35454 scalar_stmt_cost and vec_stmt_cost.
35456 2017-02-01 Torvald Riegel <triegel@redhat.com>
35457 Richard Henderson <rth@redhat.com>
35459 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35460 conditional on existance of a fast atomic load.
35461 * optabs-query.c (can_atomic_load_p): New function.
35462 * optabs-query.h (can_atomic_load_p): Declare it.
35463 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35464 no fast atomic load is available for the particular size of access.
35465 (expand_atomic_compare_and_swap): Likewise.
35466 (expand_atomic_load): Likewise.
35467 (expand_atomic_store): Likewise.
35468 (expand_atomic_fetch_op): Likewise.
35469 * testsuite/lib/target-supports.exp
35470 (check_effective_target_sync_int_128): Remove x86 because it provides
35471 no fast atomic load.
35472 (check_effective_target_sync_int_128_runtime): Likewise.
35474 2017-02-01 Richard Biener <rguenther@suse.de>
35476 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35477 (graphite_transform_loops): Provide opt-info for optimized nests.
35478 * tree-parloop.c (parallelize_loops): Provide opt-info for
35479 parallelized loops.
35481 2017-02-01 Richard Biener <rguenther@suse.de>
35483 PR middle-end/79315
35484 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35485 was not set before.
35487 2017-02-01 Richard Biener <rguenther@suse.de>
35489 PR tree-optimization/71824
35490 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35491 Verify the loops are valid in the merged SESE region.
35492 (scop_detection::can_represent_loop_1): Check analyzing the
35493 evolution of the number of iterations in the region succeeds.
35495 2017-01-31 Ian Lance Taylor <iant@golang.org>
35497 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35498 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35500 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35502 PR preprocessor/79210
35503 * input.c (get_substring_ranges_for_loc): Replace line_width
35504 assertion with error-handling.
35506 2017-01-31 Richard Biener <rguenther@suse.de>
35508 PR tree-optimization/77318
35509 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35510 (create_pw_aff_from_tree): Take loop parameter.
35511 (add_condition_to_pbb): Pass loop of the condition to
35512 create_pw_aff_from_tree.
35514 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35516 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35517 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35519 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35523 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35525 (convert_int_to_float128): Likewise.
35526 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35527 (convert_int_to_float128): Likewise.
35528 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35529 (UNSPEC_IEEE128_CONVERT): Likewise.
35530 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35531 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35532 Use local variables for IBM extended format.
35533 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35534 (fix_trunc<mode>si2_fprs): Likewise.
35535 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35536 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35537 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35538 to know that we can now have integers of all sizes in vector
35540 (fix<uns>_<mode>di2_hw): Likewise.
35541 (float<uns>_<mode>si2_hw): Likewise.
35542 (fix_<mode>si2_hw): Likewise.
35543 (fixuns_<mode>si2_hw): Likewise.
35544 (float<uns>_<mode>di2_hw): Likewise.
35545 (float_<mode>di2_hw): Likewise.
35546 (float_<mode>si2_hw): Likewise.
35547 (floatuns_<mode>di2_hw): Likewise.
35548 (floatuns_<mode>si2_hw): Likewise.
35549 (xscvqp<su>wz_<mode>): Delete, no longer used.
35550 (xscvqp<su>dz_<mode>): Likewise.
35551 (xscv<su>dqp_<mode>): Likewise.
35552 (ieee128_mfvsrd_64bit): Likewise.
35553 (ieee128_mfvsrd_32bit): Likewise.
35554 (ieee128_mfvsrwz): Likewise.
35555 (ieee128_mtvsrw): Likewise.
35556 (ieee128_mtvsrd_64bit): Likewise.
35557 (ieee128_mtvsrd_32bit): Likewise.
35559 2017-01-31 Martin Liska <mliska@suse.cz>
35562 * ipa-prop.c (ipa_free_all_node_params): Call release method
35563 instead of ~sumbol_summary to not to trigger double times
35566 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35568 PR tree-optimization/71691
35569 * bitmap.h (class auto_bitmap): New.
35570 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35571 is_maybe_undefined instead of ssa_undefined_value_p.
35573 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35575 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35576 __S390_ARCH_LEVEL__ to __ARCH__.
35578 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35580 PR tree-optimization/79267
35581 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35582 if should_remove_lhs_p is true.
35584 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35587 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35588 (add_alignment_attribute): New.
35589 (base_type_die): Add alignment attribute.
35590 (subrange_type_die): Likewise.
35591 (modified_type_die): Likewise.
35592 (gen_array_type_die): Likewise.
35593 (gen_descr_array_type_die: Likewise.
35594 (gen_enumeration_type_die): Likewise.
35595 (gen_subprogram_die): Likewise.
35596 (gen_variable_die): Likewise.
35597 (gen_field_die): Likewise.
35598 (gen_ptr_to_mbr_type_die): Likewise.
35599 (gen_struct_or_union_type_die): Likewise.
35600 (gen_subroutine_type_die): Likewise.
35601 (gen_typedef_die): Likewise.
35602 (base_type_cmp): Compare alignment attribute.
35604 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35607 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35608 (setb_unsigned) New pattern for setb with CCUNS.
35609 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35610 subfc./subfe sequence to avoid overflow problems. Generate a
35611 shorter sequence with cmpld/setb for power9.
35612 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35613 for generating subfc. instruction.
35614 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35615 now uses this instruction.
35617 2017-01-30 Ian Lance Taylor <iant@google.com>
35620 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35621 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35623 2017-01-30 Martin Sebor <msebor@redhat.com>
35625 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35626 Move constant to the right of a relational operator.
35627 (get_mpfr_format_length, format_character, format_string): Ditto.
35628 (should_warn_p, maybe_warn): Same.
35630 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35632 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35635 * asan.c (get_translation_unit_decl): Remove function.
35636 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35638 2017-01-30 Martin Liska <mliska@suse.cz>
35640 PR gcov-profile/79259
35641 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35642 -fprofile-generate.
35644 2017-01-30 Martin Liska <mliska@suse.cz>
35647 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35648 Initialize variables with NULL value.
35650 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35653 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35655 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35657 2017-01-30 Richard Biener <rguenther@suse.de>
35659 PR tree-optimization/79276
35660 * tree-vrp.c (process_assert_insertions): Properly adjust common
35661 when removing a duplicate.
35663 2017-01-30 Richard Biener <rguenther@suse.de>
35665 PR tree-optimization/79256
35666 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35667 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35669 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35671 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35674 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35675 ("*r<noxa>sbg_<mode>_sll_bitmask")
35676 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35677 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35678 Use contiguous_bitmask_nowrap_operand.
35680 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35683 * config/rs6000/altivec.h (vec_xl): Revise #define.
35684 (vec_xst): Likewise.
35686 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35688 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35690 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35692 PR rtl-optimization/79194
35693 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35694 traps before call to bypass_conditional_jumps.
35696 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35698 PR tree-optimization/71374
35699 * lra-constraints.c (check_conflict_input_operands): New.
35700 (match_reload): Use it.
35702 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35705 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35706 account to calculate conflict_set.
35708 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35710 PR rtl-optimization/78559
35711 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35712 other_insn in combine.
35714 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35716 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35717 uint16_type_node for BT_UINT16.
35719 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35721 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35722 "RTL Tests" to menu.
35723 (GIMPLE Tests): New node.
35724 (RTL Tests): New node.
35726 2017-01-27 Richard Biener <rguenther@suse.de>
35728 PR tree-optimization/79245
35729 * tree-loop-distribution.c (distribute_loop): Apply cost
35730 modeling also to detected patterns.
35732 2017-01-27 Richard Biener <rguenther@suse.de>
35734 PR tree-optimization/71433
35735 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35736 (compare_assert_loc): New function.
35737 (process_assert_insertions): Sort and optimize assert locations
35738 to remove duplicates and push down identical assertions on
35739 edges to their destination block.
35741 2017-01-27 Richard Biener <rguenther@suse.de>
35743 PR tree-optimization/79244
35744 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35745 out SSA names even if abnormal.
35747 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35749 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35750 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35751 instead of MPFR_RNDN.
35753 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35756 * arm.c (arm_option_override): Don't call build_target_option_node
35757 until after doing all option overrides.
35758 (arm_valid_target_attribute_tree): Likewise.
35760 2017-01-27 Martin Liska <mliska@suse.cz>
35762 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35763 for {cd}tors and C++ {cd}tors.
35765 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35767 * config/s390/s390.md ("*setmem_long_and")
35768 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35770 2017-01-26 Martin Sebor <msebor@redhat.com>
35772 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35775 2017-01-26 Martin Sebor <msebor@redhat.com>
35777 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35778 HAVE_DFmode before using XFmode or DFmode.
35779 (parse_directive): Avoid using the z length modifier to avoid
35780 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35782 PR middle-end/78703
35783 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35784 to accept adjustment as an array.
35785 (get_int_range): New function.
35786 (struct directive): Make width and prec arrays.
35787 (directive::set_width, directive::set_precision): Call get_int_range.
35788 (format_integer, format_floating): Handle width and precision ranges.
35789 (format_string, parse_directive): Same.
35791 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35794 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35795 comdat_type_p set, just clone them, but keep the children in the
35799 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35800 which have direct callers with -fvar-tracking-assignments enabled
35802 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35803 inside of type units.
35805 2017-01-26 Martin Sebor <msebor@redhat.com>
35807 PR middle-end/78703
35808 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35810 (struct format_result): Replace number_chars, number_chars_min,
35811 and number_chars_max with a single member of struct result_range.
35813 (format_result::operator+=): Adjust.
35814 (struct fmtresult): Remove bounded. Handle likely and unlikely
35816 (fmtresult::adjust_for_width_or_precision): New function.
35817 (fmtresult:type_max_digits): New function.
35818 (bytes_remaining): Handle likely and unlikely counters.
35819 (min_bytes_remaining): Remove.
35820 (format_percent): Simplify.
35821 (format_integer, format_floating): Set likely and unlikely counters.
35822 (get_string_length, format_character, format_string): Same.
35823 (format_plain, should_warn_p): New function.
35824 (maybe_warn): Call should_warn_p. Update diagnostic messages
35825 and handle those for all directives, including plain strings.
35826 (format_directive): Handle likely and unlikely counters.
35827 Remove unnecessary quoting from diagnostics. Add an informational
35829 (add_bytes): Remove.
35830 (pass_sprintf_length::compute_format_length): Simplify.
35831 (try_substitute_return_value): Handle likely and unlikely counters.
35833 2017-01-26 Carl Love <cel@us.ibm.com>
35835 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
35836 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
35838 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
35841 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
35842 endianess for subregs into account.
35843 * lra-constraints.c (lra_constraints): Do risky transformations
35844 always on the first iteration.
35845 * lra-lives.c (check_pseudos_live_through_calls): Add arg
35846 last_call_used_reg_set.
35847 (process_bb_lives): Define and use last_call_used_reg_set.
35848 * lra.c (lra): Always continue after lra_constraints on the first
35851 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
35853 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
35855 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
35857 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35859 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
35860 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
35861 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
35862 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
35863 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
35864 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
35865 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
35866 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
35867 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
35869 2017-01-26 Marek Polacek <polacek@redhat.com>
35872 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
35873 for the third operand.
35875 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35877 PR middle-end/79236
35878 * omp-low.c (struct omp_context): Add simt_stmt field.
35879 (scan_omp_for): Return omp_context *.
35880 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
35881 context to the _simt_ SIMD stmt.
35882 (lower_omp_for): For combined SIMD with sibling _simt_
35883 SIMD, make sure to use the same decls in _looptemp_
35884 clauses as in the sibling.
35886 2017-01-26 David Sherwood <david.sherwood@arm.com>
35888 PR middle-end/79212
35889 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
35892 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35895 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
35896 emit fld b; fld a; if possible.
35898 * brig-builtins.def: Update copyright years.
35899 * config/arm/arm_acle_builtins.def: Update copyright years.
35901 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
35904 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
35905 constraint instead of o for the stxsd instruction.
35907 2017-01-25 Carl Love <cel@us.ibm.com>
35909 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
35910 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
35912 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
35914 * doc/invoke.texi (C++ Dialect Options): Fix typo.
35916 2017-01-25 Richard Biener <rguenther@suse.de>
35918 PR tree-optimization/69264
35919 * target.def (vector_alignment_reachable): Improve documentation.
35920 * doc/tm.texi: Regenerate.
35921 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
35923 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
35924 earlier changes with respect to TYPE_USER_ALIGN.
35925 (vector_alignment_reachable_p): Likewise. Improve dumping.
35927 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35930 * config/arm/arm.md (xordi3): Force constant operand into a register
35933 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35935 * doc/invoke.texi (-fstore-merging): Correct default optimization
35936 levels at which it is enabled.
35937 (-O): Move -fstore-merging from list to...
35940 2017-01-25 Richard Biener <rguenther@suse.de>
35943 * omp-expand.c: Include debug.h.
35944 (expand_omp_taskreg): Make sure to generate early debug before
35945 outlining anything from a function.
35946 (expand_omp_target): Likewise.
35947 (grid_expand_target_grid_body): Likewise.
35949 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
35952 * asan.c (get_translation_unit_decl): New function.
35953 (asan_add_global): Extract modules file name from globals
35954 TRANSLATION_UNIT_DECL name.
35956 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
35959 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
35960 for long calls with APCS frame and VFP.
35962 2017-01-24 David Malcolm <dmalcolm@redhat.com>
35964 * cfg.c (original_copy_tables_initialized_p): New function.
35965 * cfg.h (original_copy_tables_initialized_p): New decl.
35966 * cfgrtl.c (relink_block_chain): Guard the call to
35967 free_original_copy_tables with a call to
35968 original_copy_tables_initialized_p.
35969 * cgraph.h (symtab_node::native_rtl_p): New decl.
35970 * cgraphunit.c (symtab_node::native_rtl_p): New function.
35971 (symtab_node::needed_p): Don't assert for early assembly output
35972 for __RTL functions.
35973 (cgraph_node::finalize_function): Set "force_output" for __RTL
35975 (cgraph_node::analyze): Bail out early for __RTL functions.
35976 (analyze_functions): Update assertion to support __RTL functions.
35977 (cgraph_node::expand): Bail out early for __RTL functions.
35978 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
35980 * function.h (struct function): Update comment for field
35982 * gimple-expr.c: Include "tree-pass.h".
35983 (gimple_has_body_p): Return false for __RTL functions.
35984 * Makefile.in (OBJS): Add run-rtl-passes.o.
35985 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
35987 (gcc::pass_manager::get_clean_slate): New accessor.
35988 * passes.c: Include "insn-addr.h".
35989 (should_skip_pass_p): Add logging. Update logic for running
35990 "expand" to be compatible with both __GIMPLE and __RTL. Guard
35991 property-provider override so it is only done for gimple passes.
35993 (skip_pass): New function.
35994 (execute_one_pass): Call skip_pass when skipping passes.
35995 * read-md.c (md_reader::read_char): Support filtering
35996 the input to a subset of line numbers.
35997 (md_reader::md_reader): Initialize fields
35998 m_first_line and m_last_line.
35999 (md_reader::read_file_fragment): New function.
36000 * read-md.h (md_reader::read_file_fragment): New decl.
36001 (md_reader::m_first_line): New field.
36002 (md_reader::m_last_line): New field.
36003 * read-rtl-function.c (function_reader::create_function): Only
36004 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36005 curr_properties. Set DECL_INITIAL to a dummy block.
36006 (read_rtl_function_body_from_file_range): New function.
36007 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36009 * run-rtl-passes.c: New file.
36010 * run-rtl-passes.h: New file.
36012 2017-01-24 Jeff Law <law@redhat.com>
36014 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36017 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36019 PR tree-optimization/79159
36020 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36021 (record_nonwrapping_iv): Improve boundary using above function if no
36022 value range information.
36024 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36025 Martin Jambor <mjambor@suse.cz>
36027 * brig-builtins.def: New file.
36028 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36029 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36030 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36031 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36032 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36033 * builtin-types.def (BT_INT8): New.
36034 (BT_INT16): Likewise.
36035 (BT_UINT8): Likewise.
36036 (BT_UINT16): Likewise.
36037 (BT_FN_ULONG): Likewise.
36038 (BT_FN_UINT_INT): Likewise.
36039 (BT_FN_UINT_ULONG): Likewise.
36040 (BT_FN_UINT_LONG): Likewise.
36041 (BT_FN_UINT_PTR): Likewise.
36042 (BT_FN_ULONG_PTR): Likewise.
36043 (BT_FN_INT8_FLOAT): Likewise.
36044 (BT_FN_INT16_FLOAT): Likewise.
36045 (BT_FN_UINT32_FLOAT): Likewise.
36046 (BT_FN_UINT16_FLOAT): Likewise.
36047 (BT_FN_UINT8_FLOAT): Likewise.
36048 (BT_FN_UINT64_FLOAT): Likewise.
36049 (BT_FN_UINT16_UINT32): Likewise.
36050 (BT_FN_UINT32_UINT16): Likewise.
36051 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36052 (BT_FN_INT_PTR_INT): Likewise.
36053 (BT_FN_UINT_PTR_UINT): Likewise.
36054 (BT_FN_LONG_PTR_LONG): Likewise.
36055 (BT_FN_ULONG_PTR_ULONG): Likewise.
36056 (BT_FN_VOID_UINT64_UINT64): Likewise.
36057 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36058 (BT_FN_INT8_INT8_INT8): Likewise.
36059 (BT_FN_INT16_INT16_INT16): Likewise.
36060 (BT_FN_INT_INT_INT): Likewise.
36061 (BT_FN_UINT_FLOAT_UINT): Likewise.
36062 (BT_FN_FLOAT_UINT_UINT): Likewise.
36063 (BT_FN_ULONG_UINT_UINT): Likewise.
36064 (BT_FN_ULONG_UINT_PTR): Likewise.
36065 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36066 (BT_FN_UINT_UINT_UINT): Likewise.
36067 (BT_FN_VOID_UINT_PTR): Likewise.
36068 (BT_FN_UINT_UINT_PTR: Likewise.
36069 (BT_FN_UINT32_UINT64_PTR): Likewise.
36070 (BT_FN_INT_INT_UINT_UINT): Likewise.
36071 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36072 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36073 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36074 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36075 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36076 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36077 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36078 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36079 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36080 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36081 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36082 * doc/frontends.texi: List BRIG FE.
36083 * doc/install.texi (Testing): Add BRIG tesring requirements.
36084 * doc/invoke.texi (Overall Options): Mention BRIG.
36085 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36087 2017-01-24 Richard Biener <rguenther@suse.de>
36089 PR translation/79208
36090 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36092 2017-01-24 Martin Jambor <mjambor@suse.cz>
36095 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36096 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36097 and known_contexts.
36099 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36101 PR middle-end/79123
36102 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36103 casts from signed to unsigned really don't have a range.
36105 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36107 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36108 GMP_RNDx for compatiblity.
36110 2017-01-24 Martin Liska <mliska@suse.cz>
36113 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36114 that would prevent us to call alloca with -1 as argument.
36116 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36118 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36119 Avoid -Wformat-security warning.
36121 2017-01-23 Andrew Pinski <apinski@cavium.com>
36123 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36126 2017-01-23 Martin Sebor <msebor@redhat.com>
36128 PR middle-end/78703
36129 * gimple-ssa-sprintf.c (warn_level): New global.
36130 (format_integer): Use it here and throughout the rest of the file.
36131 Use the same switch to compute sign as base.
36132 (maybe_warn): New function.
36133 (format_directive): Factor out warnings into maybe_warn.
36134 Add debugging output. Use warn_level.
36135 (add_bytes): Use warn_level.
36136 (pass_sprintf_length::compute_format_length): Add debugging output.
36137 (try_substitute_return_value): Same.
36138 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36140 PR middle-end/78703
36141 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36142 (struct fmtresult, format_integer, format_floating): Adjust.
36143 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36144 (get_string_length, format_string,format_directive): Same.
36145 (pass_sprintf_length::compute_format_length): Same.
36146 (try_substitute_return_value): Simplify slightly.
36148 PR middle-end/78703
36149 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36150 (fmtresult::operator+=): Outlined.
36151 (struct fmtresult): Add ctors.
36152 (struct conversion_spec): Rename...
36153 (struct directive): ...to this. Add and remove data members.
36154 (directive::set_width, directive::set_precision): New functions.
36155 (format_percent): Use fmtresult ctor.
36156 (get_width_and_precision): Remove.
36157 (format_integer): Make naming changes. Avoid computing width and
36159 (format_floating): Same. Adjust indentation.
36160 (format_character, format_none): New functions.
36161 (format_string): Moved character handling to format_character.
36162 (format_directive): Remove arguments, change return type.
36163 (parse_directive): New function.
36164 (pass_sprintf_length::compute_format_length): Move directive
36165 parsing to parse_directive.
36167 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36169 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36170 (assign_assembler_name_if_needed): ... this.
36171 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36172 (assign_assembler_name_if_needed): ... this.
36173 (free_lang_data_in_cgraph): Adjust callers.
36174 * cgraphunit.c (cgraph_node::analyze): Likewise.
36175 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36178 2017-01-23 Richard Biener <rguenther@suse.de>
36180 PR tree-optimization/79088
36181 PR tree-optimization/79188
36182 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36183 resetting loop bounds after last path deletion. Reset loop
36184 bounds of the target loop, make code match the comments.
36185 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36186 Make sure loops need no fixups.
36188 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36190 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36191 exponent support with double type for first argument.
36192 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36193 type returned by __builtin_vec_extract_sig,
36194 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36195 functions from "vector int" to "vector unsigned int" or from
36196 "vector long long int" to "vector unsigned long long int".
36197 Changed type returned by __builtin_vec_extract_exp,
36198 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36199 functions from "vector int" to "vector unsigned int" or from
36200 "vector long long int" to "vector unsigned long long int".
36201 Changed return type of __builtin_vec_test_data_class,
36202 __builtin_vec_test_data_class_sp, and
36203 __builtin_vec_test_data_class_dp from "vector int" to
36204 "vector bool int" or from "vector long long int" to "vector bool
36205 long long int" and changed second argument type from "unsigned
36206 int" to "int". Added new overloaded function forms "vector float
36207 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36208 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36209 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36210 double, vector unsigned long long int)" and "vector double
36211 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36212 long int)". Changed return type of
36213 __builtin_scalar_test_data_class and
36214 __builtin_scalar_test_data_class_sp and
36215 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36216 int" and changed second argument from "unsigned int" to "int".
36217 Changed type returned by __builtin_scalar_test_neg,
36218 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36219 from "int" to "bool int". Added new overloaded function form
36220 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36221 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36222 exponent double-precision with floating point first argument.
36223 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36224 documentation of scalar_test_data_class, scalar_test_neg,
36225 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36226 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36227 vec_test_data_class built-in functions to reflect refinements in
36228 their type signatures.
36230 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36232 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36234 (aarch64_elf_asm_destructor): Likewise.
36236 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36238 PR rtl-optimization/78634
36239 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36240 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36241 * ifcvt.c (noce_try_cmove): Add missing cost check.
36243 PR rtl-optimization/71724
36244 * combine.c (if_then_else_cond): Look for situations where it is
36245 beneficial to undo the work of one of the recursive calls.
36247 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36249 PR tree-optimization/70754
36250 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36251 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36252 combined stmt before it if not NULL.
36253 (combine_chains): Process refs reversely and compute dominance point
36256 2017-01-23 Martin Liska <mliska@suse.cz>
36258 PR tree-optimization/79196
36259 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36260 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36262 (strlen_optimize_stmt): Call the renamed function.
36264 2017-01-23 Michael Matz <matz@suse.de>
36266 PR tree-optimization/78384
36267 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36269 2017-01-23 Richard Biener <rguenther@suse.de>
36271 PR tree-optimization/79186
36272 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36273 both incoming edges before moving an assert.
36275 2017-01-23 Martin Jambor <mjambor@suse.cz>
36277 * ipa-prop.c (load_from_param_1): Removed.
36278 (load_from_unmodified_param): Bits from load_from_param_1 put back
36280 (load_from_param): Removed.
36281 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36282 with stmt. Reverted back to use of load_from_unmodified_param.
36284 2017-01-23 Martin Jambor <mjambor@suse.cz>
36287 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36288 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36289 field a pointer to garbage collected vector, mark lattices and
36290 ipcp_orig_node with GTY((skip)).
36291 (ipa_get_param_count): Adjust to descriptors being a pointer.
36292 (ipa_get_param): Likewise.
36293 (ipa_get_type): Likewise.
36294 (ipa_get_param_move_cost): Likewise.
36295 (ipa_set_param_used): Likewise.
36296 (ipa_get_controlled_uses): Likewise.
36297 (ipa_set_controlled_uses): Likewise.
36298 (ipa_is_param_used): Likewise.
36299 (ipa_node_params_t): Move into garbage collector. New methods insert
36301 (ipa_node_params_sum): Annotate wth GTY(()).
36302 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36304 (ipa_load_from_parm_agg): Adjust declaration.
36305 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36306 * ipa-profile.c (ipa_profile): Likewise.
36307 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36308 (ipa_populate_param_decls): Make descriptors parameter garbage
36310 (ipa_dump_param): Adjust to descriptors being a pointer.
36311 (ipa_alloc_node_params): Likewise.
36312 (ipa_initialize_node_params): Likewise.
36313 (load_from_param_1): Make descriptors parameter garbage collected.
36314 (load_from_unmodified_param): Likewise.
36315 (load_from_param): Likewise.
36316 (ipa_load_from_parm_agg): Likewise.
36317 (ipa_node_params::~ipa_node_params): Removed.
36318 (ipa_free_all_node_params): Remove call to delete operator.
36319 (ipa_node_params_t::insert): New.
36320 (ipa_node_params_t::remove): Likewise.
36321 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36322 copy known_csts and known_contexts vectors.
36323 (ipa_read_node_info): Adjust to descriptors being a pointer.
36324 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36326 (ipcp_transform_function): Make descriptors variable garbage
36329 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36331 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36332 * config/i386/avx512dqintrin.h: Ditto.
36333 * config/i386/avx512fintrin.h: Ditto.
36334 * config/i386/i386.c: Handle new builtins.
36335 * config/i386/i386-builtin.def: Add new builtins.
36336 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36337 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36339 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36340 Martin Liska <mliska@suse.cz>
36342 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36343 * asan.c (asan_expand_poison_ifn): Support stores and use
36344 appropriate ASAN report function.
36345 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36346 * internal-fn.def (ASAN_POISON_USE): Declare.
36347 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36348 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36349 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36350 ASAN_POISON calls w/o LHS.
36351 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36352 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36353 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36354 * gimplify.c (asan_poison_variables): Add attribute
36355 use_after_scope_memory to variables that really needs to live
36357 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36358 having the attribute.
36360 2017-01-23 Martin Liska <mliska@suse.cz>
36362 * asan.c (create_asan_shadow_var): New function.
36363 (asan_expand_poison_ifn): Likewise.
36364 * asan.h (asan_expand_poison_ifn): New declaration.
36365 * internal-fn.c (expand_ASAN_POISON): Likewise.
36366 * internal-fn.def (ASAN_POISON): New builtin.
36367 * sanopt.c (pass_sanopt::execute): Expand
36368 asan_expand_poison_ifn.
36369 * tree-inline.c (copy_decl_for_dup_finish): Make function
36371 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36372 * tree-ssa.c (is_asan_mark_p): New function.
36373 (execute_update_addresses_taken): Rewrite local variables
36374 (identified just by use-after-scope as addressable) into SSA.
36376 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36378 * doc/install.texi (Specific): opensource.apple.com uses https
36379 now. Remove trailing slash.
36381 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36383 * README.Portability: Remove note on an Irix compatibility issue.
36385 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36387 * gcov.c (INCLUDE_ALGORITHM): Define.
36388 (INCLUDE_VECTOR): Define.
36389 No longer include <vector> and <algorithm> directly.
36391 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36393 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36395 * doc/invoke.texi (Code Gen Options): Ditto.
36397 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36400 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36402 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36404 rtl-optimization/79125
36405 * cprop.c (local_cprop_pass): Handle cases where we make an
36406 unconditional trap.
36408 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36412 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36413 read from, for big endian.
36415 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36417 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36418 register pauth builtins for LP64 only.
36420 2017-01-20 Marek Polacek <polacek@redhat.com>
36423 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36426 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36428 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36430 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36431 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36432 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36434 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36437 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36438 in big-endian mode when they are not a single duplicated value.
36440 2017-01-20 Richard Biener <rguenther@suse.de>
36442 * BASE-VER: Bump to 7.0.1.
36444 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36446 * omp-low.c (omplow_simd_context): New struct. Use it...
36447 (lower_rec_simd_input_clauses): ...here and...
36448 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36449 references to idx, lane, max_vf, is_simt.
36451 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36453 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36456 2017-01-20 Martin Jambor <mjambor@suse.cz>
36458 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36459 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36461 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36462 (GTFILES): Rename hsa.c to hsa-common.c.
36463 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36464 * hsa-dump.c: Likewise.
36465 * hsa-gen.c: Likewise.
36466 * hsa-regalloc.c: Likewise.
36467 * ipa-hsa.c: Likewise.
36468 * omp-expand.c: Likewise.
36469 * omp-low.c: Likewise.
36470 * toplev.c: Likewise.
36472 2017-01-20 Marek Polacek <polacek@redhat.com>
36475 * doc/invoke.texi: Document -Wduplicated-branches.
36476 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36477 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36478 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36479 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36480 return 0 only when not OEP_LEXICOGRAPHIC.
36481 (fold_build_cleanup_point_expr): Use the expression
36482 location when building CLEANUP_POINT_EXPR.
36483 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36484 * tree.c (add_expr): Handle error_mark_node.
36486 2017-01-20 Martin Liska <mliska@suse.cz>
36489 * tree-profile.c (init_ic_make_global_vars): Do not call
36491 (gimple_init_gcov_profiler): Likewise.
36493 2017-01-20 Martin Liska <mliska@suse.cz>
36496 * cgraph.h (maybe_create_reference): Remove argument and
36498 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36500 * ipa-cp.c (create_specialized_node): Likewise.
36501 * symtab.c (symtab_node::maybe_create_reference): Handle
36502 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36504 2017-01-20 Martin Liska <mliska@suse.cz>
36506 * read-rtl-function.c (function_reader::create_function): Use
36507 build_decl instread of build_decl_stat.
36509 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36511 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36512 * config/i386/avx512dqintrin.h: Ditto.
36513 * config/i386/avx512fintrin.h: Ditto.
36514 * config/i386/i386-builtin-types.def: Add new types.
36515 * config/i386/i386.c: Handle new types.
36516 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36517 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36518 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36519 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36520 (__builtin_ia32_kshiftridi): New.
36521 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36523 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36527 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36528 define to rs6000_init_stack_protect_guard.
36529 (rs6000_init_stack_protect_guard): New function.
36531 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36532 Yunqiang Su <yunqiang.su@imgtec.com>
36534 * config.gcc (supported_defaults): Add madd4.
36535 (with_madd4): Add validation.
36536 (all_defaults): Add madd4.
36537 * config/mips/mips.opt (mmadd4): New option.
36538 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36540 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36542 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36543 (ISA_HAS_FUSED_MADD4): Likewise.
36544 * doc/invoke.texi (-mmadd4): Document the new option.
36545 * doc/install.texi (--with-madd4): Document the new option.
36547 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36549 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36550 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36551 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36552 (aarch64_init_pauth_hint_builtins): New.
36553 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36554 (aarch64_expand_builtin): Expand new builtins.
36556 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36558 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36559 * combine-stack-adj.c (no_unhandled_cfa): Handle
36560 REG_CFA_TOGGLE_RA_MANGLE.
36561 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36562 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36563 info for return address signing.
36564 (aarch64_expand_epilogue): Likewise.
36566 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36568 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36569 * config/aarch64/aarch64-protos.h
36570 (aarch64_return_address_signing_enabled): New declaration.
36571 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36573 (aarch64_expand_prologue): Sign return address before it's pushed onto
36575 (aarch64_expand_epilogue): Authenticate return address fetched from
36577 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36578 (aarch64_attributes): New function attributes for "sign-return-address".
36579 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36580 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36581 ("*do_return"): Generate combined instructions according to key index.
36582 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36583 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36585 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36586 * config/aarch64/aarch64.opt (msign-return-address=): New.
36587 * doc/extend.texi (AArch64 Function Attributes): Documents
36588 "sign-return-address=".
36589 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36591 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36593 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36594 overall option summary.
36596 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36598 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36599 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36600 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36601 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36603 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36605 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36606 -mpower9-minmax by default for -mcpu=power9.
36607 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36608 128-bit floating point.
36610 2017-01-20 Alan Modra <amodra@gmail.com>
36612 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36613 optimizing for size.
36615 2017-01-20 Alan Modra <amodra@gmail.com>
36618 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36619 for strcmp and strncmp from corresponding builtin decl.
36621 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36623 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36624 instead of i386/rtems-64.h.
36625 * config/i386/rtems-64.h: Remove.
36627 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36631 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36633 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36635 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36637 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36638 Change int to HOST_WIDE_INT.
36639 * config/aarch64/aarch64-protos.h
36640 (aarch64_simd_gen_const_vector_dup): Likewise.
36641 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36643 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36645 * langhooks-def.h (lhd_type_for_size): New decl.
36646 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36647 * langhooks.c (lhd_type_for_size): New function, taken from
36650 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36652 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36653 define_bypass for CR latency.
36654 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36655 (power9-alu2): Add define_bypass for CR latency.
36657 (power9-mul): Update insn latency.
36658 (power9-mul-compare): Update insn latency, bypass latency and remove
36661 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36663 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36665 * config/aarch64/aarch64.md
36666 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36667 aarch64_nopcrelative_literal_loads.
36668 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36670 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36672 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36673 TARGET_LOONGSON_3A.
36674 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36676 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36679 * config.gcc (supported_defaults): Add lxc1-sxc1.
36680 (with_lxc1_sxc1): Add validation.
36681 (all_defaults): Add lxc1-sxc1.
36682 * config/mips/mips.opt (mlxc1-sxc1): New option.
36683 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36685 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36686 __mips_no_lxc1_sxc1.
36687 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36688 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36689 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36691 2017-01-19 Richard Biener <rguenther@suse.de>
36693 PR tree-optimization/72488
36694 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36695 sure to restore SSA info.
36696 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36698 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36700 PR rtl-optimization/79121
36701 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36702 of the inner type when shifting an extended value.
36704 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36707 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36708 interposable aliases.
36710 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36713 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36714 Use the evmergelohi instruction.
36715 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36716 (mov_sitf_e500_subreg8_2_be): Likewise.
36717 (mov_sitf_e500_subreg12_2_le): Likewise.
36718 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36719 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36720 (mov_sitf_e500_subreg8_2_le): Likewise.
36721 (mov_sitf_e500_subreg12_2_be): Likewise.
36723 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36725 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36726 attribute from vecsimple to vecperm.
36727 (altivec_vbpermq2): Likewise.
36729 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36732 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36734 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36735 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36736 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36737 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36738 case where N arg is SIZE_MAX.
36739 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36740 (cmpstrsi): Add pattern.
36742 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36744 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36745 __builtin_vec_revb builtins.
36746 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36747 built-in functions to support generation of the ISA 3.0 XXBR<x>
36748 vector byte reverse instructions.
36749 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36750 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36751 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36752 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36753 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36754 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36755 (P9V_BUILTIN_VEC_REVB): Likewise.
36756 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36757 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36758 (p9_xxbrq_v16qi): Likewise.
36759 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36760 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36761 (p9_xxbrh_v8hi): Likewise.
36762 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36763 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36764 vec_revb built-in functions.
36766 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36768 PR rtl-optimization/78952
36769 * config/i386/i386.md (any_extract): New code iterator.
36770 (*insvqi_2): Use any_extract for source operand.
36771 (*insvqi_3): Use any_shiftrt for source operand.
36773 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36775 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36777 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36779 2017-01-18 Matthias Klose <doko@ubuntu.com>
36781 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36783 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36785 * config/rs6000/altivec.h (vec_bperm): Change #define.
36786 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36787 (altivec_vbpermq2): New define_insn.
36788 (altivec_vbpermd): Likewise.
36789 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36790 function interface.
36791 (VBPERMD): Likewise.
36792 (VBPERM): New polymorphic function interface.
36793 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36794 Add entries for P9V_BUILTIN_VEC_VBPERM.
36795 * doc/extend.texi: Add interfaces for vec_bperm.
36797 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36799 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36800 first letter of error messages.
36801 (s390_resolve_overloaded_builtin): Likewise.
36802 * config/s390/s390.c (s390_expand_builtin): Likewise.
36803 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36804 (s390_valid_target_attribute_inner_p): Likewise.
36805 * config/s390/s390.md ("tabort"): Likewise.
36807 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36809 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36810 (ISA_AVOID_DIV_HILO): New macro.
36811 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36812 (ISA_HAS_DDIV): Likewise.
36814 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36816 * doc/invoke.texi (fabi-version): Correct number of occurrences.
36818 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36820 * doc/invoke.texi (fabi-version): Spelling fix.
36822 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36825 * doc/invoke.texi (fabi-version): Mention mangling fix for
36828 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36831 * doc/invoke.texi (fabi-version): Document discriminator mangling.
36833 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
36836 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
36837 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
36839 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
36841 (stack_protect_test): Ditto.
36842 * config/rs6000/rs6000.opt (mstack-protector-guard=,
36843 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
36845 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
36846 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
36847 -mstack-protector-guard-offset=.
36848 (RS/6000 and PowerPC Options): Ditto.
36850 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36852 * config/i386/i386.h (MASK_CLASS_P): New define.
36853 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
36854 there are no registers from different register sets also when
36855 mask registers are used. Update function comment.
36856 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
36857 to (*k/*r) and (*k/*km) alternatives.
36859 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
36861 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
36862 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
36863 (EH_RETURN_HANDLER_RTX): New define.
36864 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
36865 Force frame pointer in EH return functions.
36866 (aarch64_expand_epilogue): Add barrier for eh_return.
36867 (aarch64_final_eh_return_addr): Remove.
36868 (aarch64_eh_return_handler_rtx): New function.
36869 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
36871 (aarch64_eh_return_handler_rtx): New prototype.
36873 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36875 * config/rs6000/altivec.h (vec_rlmi): New #define.
36876 (vec_vrlnm): Likewise.
36877 (vec_rlnm): Likewise.
36878 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
36879 (UNSPEC_VRLNM): Likewise.
36880 (VIlong): New mode iterator.
36881 (altivec_vrl<VI_char>mi): New define_insn.
36882 (altivec_vrl<VI_char>nm): Likewise.
36883 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
36885 (VRLDNM): Likewise.
36886 (RLNM): New polymorphic function entry.
36887 (VRLWMI): New monomorphic function entry.
36888 (VRLDMI): Likewise.
36889 (RLMI): New polymorphic function entry.
36890 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
36891 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
36892 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
36895 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36898 * dwarf2out.c (field_byte_offset): Restore the
36899 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
36900 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
36901 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
36904 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36907 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
36909 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36912 * dwarf2out.c (add_data_member_location_attribute): For constant
36913 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
36914 instead of DW_AT_data_member_location, DW_AT_bit_offset and
36915 DW_AT_byte_size attributes.
36917 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36919 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
36920 after forcing to constant memory when the code model is medium.
36922 2017-01-17 Julia Koval <julia.koval@intel.com>
36925 * config/i386/avx512fintrin.h
36926 (_mm512_i32gather_ps): Change __addr type to void const*.
36927 (_mm512_mask_i32gather_ps): Ditto.
36928 (_mm512_i32gather_pd): Ditto.
36929 (_mm512_mask_i32gather_pd): Ditto.
36930 (_mm512_i64gather_ps): Ditto.
36931 (_mm512_mask_i64gather_ps): Ditto.
36932 (_mm512_i64gather_pd): Ditto.
36933 (_mm512_mask_i64gather_pd): Ditto.
36934 (_mm512_i32gather_epi32): Ditto.
36935 (_mm512_mask_i32gather_epi32): Ditto.
36936 (_mm512_i32gather_epi64): Ditto.
36937 (_mm512_mask_i32gather_epi64): Ditto.
36938 (_mm512_i64gather_epi32): Ditto.
36939 (_mm512_mask_i64gather_epi32): Ditto.
36940 (_mm512_i64gather_epi64): Ditto.
36941 (_mm512_mask_i64gather_epi64): Ditto.
36942 (_mm512_i32scatter_ps): Change __addr type to void*.
36943 (_mm512_mask_i32scatter_ps): Ditto.
36944 (_mm512_i32scatter_pd): Ditto.
36945 (_mm512_mask_i32scatter_pd): Ditto.
36946 (_mm512_i64scatter_ps): Ditto.
36947 (_mm512_mask_i64scatter_ps): Ditto.
36948 (_mm512_i64scatter_pd): Ditto.
36949 (_mm512_mask_i64scatter_pd): Ditto.
36950 (_mm512_i32scatter_epi32): Ditto.
36951 (_mm512_mask_i32scatter_epi32): Ditto.
36952 (_mm512_i32scatter_epi64): Ditto.
36953 (_mm512_mask_i32scatter_epi64): Ditto.
36954 (_mm512_i64scatter_epi32): Ditto.
36955 (_mm512_mask_i64scatter_epi32): Ditto.
36956 (_mm512_i64scatter_epi64): Ditto.
36957 (_mm512_mask_i64scatter_epi64): Ditto.
36958 * config/i386/avx512pfintrin.h
36959 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
36960 (_mm512_mask_prefetch_i32gather_ps): Ditto.
36961 (_mm512_mask_prefetch_i64gather_pd): Ditto.
36962 (_mm512_mask_prefetch_i64gather_ps): Ditto.
36963 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
36964 (_mm512_prefetch_i32scatter_ps): Ditto.
36965 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
36966 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
36967 (_mm512_prefetch_i64scatter_pd): Ditto.
36968 (_mm512_prefetch_i64scatter_ps): Ditto.
36969 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
36970 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
36971 * config/i386/avx512vlintrin.h
36972 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
36973 (_mm_mmask_i32gather_ps): Ditto.
36974 (_mm256_mmask_i32gather_pd): Ditto.
36975 (_mm_mmask_i32gather_pd): Ditto.
36976 (_mm256_mmask_i64gather_ps): Ditto.
36977 (_mm_mmask_i64gather_ps): Ditto.
36978 (_mm256_mmask_i64gather_pd): Ditto.
36979 (_mm_mmask_i64gather_pd): Ditto.
36980 (_mm256_mmask_i32gather_epi32): Ditto.
36981 (_mm_mmask_i32gather_epi32): Ditto.
36982 (_mm256_mmask_i32gather_epi64): Ditto.
36983 (_mm_mmask_i32gather_epi64): Ditto.
36984 (_mm256_mmask_i64gather_epi32): Ditto.
36985 (_mm_mmask_i64gather_epi32): Ditto.
36986 (_mm256_mmask_i64gather_epi64): Ditto.
36987 (_mm_mmask_i64gather_epi64): Ditto.
36988 (_mm256_i32scatter_ps): Change __addr type to void*.
36989 (_mm256_mask_i32scatter_ps): Ditto.
36990 (_mm_i32scatter_ps): Ditto.
36991 (_mm_mask_i32scatter_ps): Ditto.
36992 (_mm256_i32scatter_pd): Ditto.
36993 (_mm256_mask_i32scatter_pd): Ditto.
36994 (_mm_i32scatter_pd): Ditto.
36995 (_mm_mask_i32scatter_pd): Ditto.
36996 (_mm256_i64scatter_ps): Ditto.
36997 (_mm256_mask_i64scatter_ps): Ditto.
36998 (_mm_i64scatter_ps): Ditto.
36999 (_mm_mask_i64scatter_ps): Ditto.
37000 (_mm256_i64scatter_pd): Ditto.
37001 (_mm256_mask_i64scatter_pd): Ditto.
37002 (_mm_i64scatter_pd): Ditto.
37003 (_mm_mask_i64scatter_pd): Ditto.
37004 (_mm256_i32scatter_epi32): Ditto.
37005 (_mm256_mask_i32scatter_epi32): Ditto.
37006 (_mm_i32scatter_epi32): Ditto.
37007 (_mm_mask_i32scatter_epi32): Ditto.
37008 (_mm256_i32scatter_epi64): Ditto.
37009 (_mm256_mask_i32scatter_epi64): Ditto.
37010 (_mm_i32scatter_epi64): Ditto.
37011 (_mm_mask_i32scatter_epi64): Ditto.
37012 (_mm256_i64scatter_epi32): Ditto.
37013 (_mm256_mask_i64scatter_epi32): Ditto.
37014 (_mm_i64scatter_epi32): Ditto.
37015 (_mm_mask_i64scatter_epi32): Ditto.
37016 (_mm256_i64scatter_epi64): Ditto.
37017 (_mm256_mask_i64scatter_epi64): Ditto.
37018 (_mm_i64scatter_epi64): Ditto.
37019 (_mm_mask_i64scatter_epi64): Ditto.
37020 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37021 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37022 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37023 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37024 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37025 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37026 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37027 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37028 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37029 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37030 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37031 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37032 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37033 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37034 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37035 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37036 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37037 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37038 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37039 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37040 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37041 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37042 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37043 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37044 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37045 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37046 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37047 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37048 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37049 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37050 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37051 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37052 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37053 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37054 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37055 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37056 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37057 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37058 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37059 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37060 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37061 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37062 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37063 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37064 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37065 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37066 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37067 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37068 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37069 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37070 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37071 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37072 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37073 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37074 definitions accordingly.
37076 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37077 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37080 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37083 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37086 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37087 hard regs for inner regno.
37089 2017-01-17 Martin Liska <mliska@suse.cz>
37092 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37093 assumption and add comment.
37095 2017-01-17 Nathan Sidwell <nathan@acm.org>
37097 * ipa-visibility.c (localize_node): New function, broken out of ...
37098 (function_and_variable_visibility): ... here. Call it.
37100 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37102 PR middle-end/77445
37103 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37104 correctly set frequency of oudgoing edge.
37105 (duplicate_thread_path): Fix profile updating.
37107 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37110 * configure.ac: Add GCC_BASE_VER.
37111 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37112 version from BASE-VER file.
37113 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37114 (gcc.o): Depend on $(BASEVER).
37115 * common.opt (dumpfullversion): New option.
37116 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37117 * doc/invoke.texi: Document -dumpfullversion.
37118 * doc/install.texi: Document --with-gcc-major-version-only.
37119 * configure: Regenerated.
37121 2017-01-17 Richard Biener <rguenther@suse.de>
37123 PR tree-optimization/71433
37124 * tree-vrp.c (register_new_assert_for): Merge same asserts
37125 on all incoming edges.
37126 (process_assert_insertions_for): Handle insertions at the
37129 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37131 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37132 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37134 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37137 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37140 2017-01-17 Alan Modra <amodra@gmail.com>
37143 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37144 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37145 symbolic stack limit when pic.
37147 2017-01-16 Martin Sebor <msebor@redhat.com>
37149 PR tree-optimization/78608
37150 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37152 2017-01-16 Jeff Law <law@redhat.com>
37155 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37156 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37157 for several include directories that may be relative to sysroot.
37158 * config/i386/x-mingw32 (gplus_includedir): Define.
37159 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37160 (native_system_includedir): Likewise.
37161 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37162 override if TARGET_SYSTEM_ROOT is defined.
37163 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37165 PR tree-optimization/79090
37166 PR tree-optimization/33562
37167 PR tree-optimization/61912
37168 PR tree-optimization/77485
37169 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37170 and computed trims into the dump file.
37172 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37174 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37176 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37179 * gimplify.c (gimplify_init_constructor): If want_value and
37180 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37184 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37185 sequence. Formatting fixes.
37186 (doloop_optimize): Formatting fixes.
37189 * gcc.c (debug_level_greater_than_spec_func): New function.
37190 (static_spec_functions): Add debug-level-gt spec function.
37191 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37193 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37194 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37195 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37196 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37197 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37198 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37200 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37202 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37203 QImode fixups to general and mask registers only.
37205 2017-01-16 Carl Love <cel@us.ibm.com>
37207 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37208 for built-in functions
37209 vector signed char vec_nabs (vector signed char)
37210 vector signed short vec_nabs (vector signed short)
37211 vector signed int vec_nabs (vector signed int)
37212 vector signed long long vec_nabs (vector signed long long)
37213 vector float vec_nabs (vector float)
37214 vector double vec_nabs (vector double)
37215 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37217 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37218 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37219 * doc/extend.texi: Update the documentation file for the new built-in
37222 2017-01-16 Martin Sebor <msebor@redhat.com>
37224 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37227 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37229 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37230 UNSPEC_VSX__XXSPLTD to require special splat handling.
37232 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37235 * system.h: Poison strndup.
37237 2017-01-16 Alan Modra <amodra@gmail.com>
37240 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37243 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37245 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37247 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37249 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37250 call recog here. Assert that INSN_CODE (insn) is non-negative.
37252 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37255 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37257 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37258 in the currently scheduled RTL fragment.
37260 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37262 PR rtl-optimization/78751
37263 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37266 2017-01-14 Jeff Law <law@redhat.com>
37268 PR tree-optimization/79090
37269 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37270 variable length stores.
37271 (compute_trims): Delete dead assignment to *trim_tail.
37272 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37275 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37277 PR rtl-optimization/78626
37278 PR rtl-optimization/78727
37279 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37280 of a block, and split such blocks after everything else is finished.
37282 2017-01-14 Alan Modra <amodra@gmail.com>
37285 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37286 target legitimate_combined_insn.
37287 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37288 (rs6000_legitimate_combined_insn): New function.
37289 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37291 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37292 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37293 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37295 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37297 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37299 2017-01-13 Jeff Law <law@redhat.com>
37301 PR tree-optimization/33562
37302 PR tree-optimization/61912
37303 PR tree-optimization/77485
37304 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37306 (delete_dead_assignment): Likewise.
37307 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37308 statement to delete_dead_call and delete_dead_assignment.
37310 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37313 * substring-locations.c (format_warning_va): Strengthen case 1 so
37314 that both endpoints of the substring must be within the format
37315 range for just the substring to be printed.
37317 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37319 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37320 * config/i386/i386.c (ix86_target_string): Add missing options
37321 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37322 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37323 flags_other and ix86_target_other to flags2_other. Display unknown
37325 (ix86_valid_target_attribute_inner_p): Add missing options and
37326 reorder options by implied ISAs, as in ix86_target_string.
37328 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37330 * hash-table.h (hash_table::too_empty_p): New function.
37331 (hash_table::expand): Use it.
37332 (hash_table::traverse): Likewise.
37333 (hash_table::empty_slot): Use sizeof (value_type) instead of
37334 sizeof (PTR) to convert bytes to elements. Shrink the table
37335 if the current size is excessive for the current number of
37338 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37340 * ira-costs.c (record_reg_classes): Break from the inner loop
37341 early once alt_fail is known to be true. Update outer loop
37342 handling accordingly.
37344 2017-01-13 Jeff Law <law@redhat.com>
37346 * tree-ssa-dse.c (decrement_count): New function.
37347 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37348 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37349 when we know the partially dead statement is a mem* function.
37351 PR tree-optimization/61912
37352 PR tree-optimization/77485
37353 * tree-ssa-dse.c: Include expr.h.
37354 (maybe_trim_constructor_store): New function.
37355 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37357 PR tree-optimization/33562
37358 PR tree-optimization/61912
37359 PR tree-optimization/77485
37360 * doc/invoke.texi: Document new dse-max-object-size param.
37361 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37362 * tree-ssa-dse.c: Include params.h.
37363 (dse_store_status): New enum.
37364 (initialize_ao_ref_for_dse): New, partially extracted from
37366 (valid_ao_ref_for_dse, normalize_ref): New.
37367 (setup_live_bytes_from_ref, compute_trims): Likewise.
37368 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37369 (maybe_trim_partially_dead_store): Likewise.
37370 (maybe_trim_complex_store): Likewise.
37371 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37372 Track what bytes live from the original store. Return tri-state
37373 for dead, partially dead or live.
37374 (dse_dom_walker): Add constructor, destructor and new private members.
37375 (delete_dead_call, delete_dead_assignment): New extracted from
37377 (dse_optimize_stmt): Make a member of dse_dom_walker.
37378 Use initialize_ao_ref_for_dse.
37380 PR tree-optimization/33562
37381 PR tree-optimization/61912
37382 PR tree-optimization/77485
37383 * sbitmap.h (bitmap_count_bits): Prototype.
37384 (bitmap_clear_range, bitmap_set_range): Likewise.
37385 * sbitmap.c (bitmap_clear_range): New function.
37386 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37388 2017-01-13 Martin Liska <mliska@suse.cz>
37391 * function.c (set_cfun): Add new argument force.
37392 * function.h (set_cfun): Likewise.
37393 * ipa-inline-transform.c (inline_call): Use the function when
37394 strict alising from is dropped for function we inline to.
37396 2017-01-13 Richard Biener <rguenther@suse.de>
37398 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37399 for dumping GIMPLE INTEGER_CSTs.
37401 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37403 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37404 to 201112L since C++17.
37406 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37409 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37410 if -fsanitize=kernel-address is present.
37412 2017-01-13 Richard Biener <rguenther@suse.de>
37414 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37415 as _Literal ( type ) number in case usual suffixes do not
37416 preserve all information.
37418 2017-01-13 Richard Biener <rguenther@suse.de>
37420 PR tree-optimization/77283
37421 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37422 and ssa-iterators.h.
37423 (is_feasible_trace): Implement a cost model based on joiner
37426 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37429 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37430 char or short to __float128/_Float128 directly.
37432 2017-01-12 Martin Sebor <msebor@redhat.com>
37434 to -Wformat-overflow.
37435 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37436 (min_bytes_remaining): Same.
37437 (get_string_length): Same.
37438 (format_string): Same.
37439 (format_directive): Same.
37441 (pass_sprintf_length::handle_gimple_call): Same.
37443 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37445 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37446 info.nowrite calls with no lhs that can't throw. Return bool
37447 whether gsi_remove has been called or not.
37448 (pass_sprintf_length::handle_gimple_call): Return bool whether
37449 try_substitute_return_value called gsi_remove. Formatting fix.
37450 (pass_sprintf_length::execute): Don't use gsi_remove if
37451 handle_gimple_call returned true.
37454 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37455 be removed due to side-effects, don't remove following barrier nor
37456 turn the successor edge into fallthru edge.
37458 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37461 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37462 element-reversing loads and stores as not swappable.
37464 2017-01-12 Nathan Sidwell <nathan@acm.org>
37465 Nicolai Stange <nicstange@gmail.com>
37467 * combine.c (try_combine): Don't ignore result of overlap checking
37468 loop. Combine overlap & asm check into single loop.
37470 2017-01-12 Richard Biener <rguenther@suse.de>
37472 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37473 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37475 2017-01-12 Richard Biener <rguenther@suse.de>
37477 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37478 and TS_TARGET_OPTION directly derive from TS_BASE.
37479 * tree-core.h (tree_optimization_option): Derive from tree_base.
37480 (tree_target_option): Likewise.
37482 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37484 * config/i386/i386.c (memory_address_length): Increase len
37485 only when rip_relative_addr_p returns false.
37487 2017-01-11 Julia Koval <julia.koval@intel.com>
37489 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37490 (OPTION_MASK_ISA_SGX_SET): New.
37491 (ix86_handle_option): Handle OPT_msgx.
37492 * config.gcc: Added sgxintrin.h.
37493 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37494 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37495 * config/i386/i386.c (ix86_target_string): Add -msgx.
37497 (ix86_option_override_internal): Handle new options.
37498 (ix86_valid_target_attribute_inner_p): Add sgx.
37499 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37500 * config/i386/i386.opt: Add msgx.
37501 * config/i386/sgxintrin.h: New file.
37502 * config/i386/x86intrin.h: Add sgxintrin.h.
37504 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37507 * fold-const.c (maybe_nonzero_address): Return 1 for function
37509 (tree_single_nonzero_warnv_p): Don't handle function local objects
37513 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37516 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37519 * opts.c: Include "spellcheck.h"
37520 (struct string_fragment): New struct.
37521 (struct edit_distance_traits<const string_fragment &>): New
37523 (get_closest_sanitizer_option): New function.
37524 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37526 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37528 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37530 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37531 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37532 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37533 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37534 for initial die_offset if dwarf_split_debug_info.
37535 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37536 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37538 (output_skeleton_debug_sections): Formatting fix. Use
37539 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37540 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37542 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37544 * config/arm/cortex-a53.md: Add bypasses for
37545 cortex_a53_r2f_cvt.
37546 (cortex_a53_r2f): Only use for transfers.
37547 (cortex_a53_f2r): Likewise.
37548 (cortex_a53_r2f_cvt): Add reservation for conversions.
37549 (cortex_a53_f2r_cvt): Likewise.
37551 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37553 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37554 to all inlined functions, change static to extern.
37556 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37559 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37561 (arm_assemble_integer): Likewise.
37563 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37565 * config.gcc: Use new awk script to check CPU, FPU and architecture
37566 parameters for --with-... options.
37567 * config/arm/parsecpu.awk: New file
37568 * config/arm/arm-cpus.in: New file.
37569 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37571 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37573 * config/arm/t-arm: Update dependency rules.
37574 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37575 of processing .def files.
37576 * config/arm/genopt.sh: Deleted.
37577 * config/arm/gentune.sh: Deleted.
37578 * config/arm/arm-cores.def: Deleted.
37579 * config/arm/arm-arches.def: Deleted.
37580 * config/arm/arm-fpus.def: Deleted.
37581 * config/arm/arm-tune.md: Regenerated.
37582 * config/arm/arm-tables.opt: Regenerated.
37583 * config/arm/arm-cpu.h: New generated file.
37584 * config/arm/arm-cpu-data.h: New generated file.
37585 * config/arm/arm-cpu-cdata.h: New generated file.
37587 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37590 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37592 (input_varpool_node): Unpack dynamically_initialized bit.
37594 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37596 PR rtl-optimization/79032
37597 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37598 the alignment of the adjusted memory reference against that of MODE,
37599 instead of the alignment of the original memory reference.
37601 2017-01-11 Martin Jambor <mjambor@suse.cz>
37603 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37605 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37606 decorated functions.
37608 2017-01-11 Richard Biener <rguenther@suse.de>
37610 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37611 set range/nonnull info for PHI results. Do not set it on
37612 stmts marked for removal.
37614 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37616 * expr.c (store_field): In the bitfield case, fetch the return value
37617 from the registers before applying a single big-endian adjustment.
37618 Always do a final load for a BLKmode value not larger than a word.
37620 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37623 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37624 that we correctly handle column numbers greater than
37625 LINE_MAP_MAX_COLUMN_NUMBER.
37627 2017-01-10 Martin Sebor <msebor@redhat.com>
37629 PR middle-end/78245
37630 * gimple-ssa-sprintf.c (get_destination_size): Call
37631 {init,fini}object_sizes.
37632 * tree-object-size.c (addr_object_size): Adjust.
37633 (pass_through_call): Adjust.
37634 (pass_object_sizes::execute): Adjust.
37635 * tree-object-size.h (fini_object_sizes): Declare.
37637 2017-01-10 Martin Sebor <msebor@redhat.com>
37639 PR tree-optimization/78775
37640 * builtins.c (get_size_range): Move...
37641 * calls.c: ...to here.
37642 (alloc_max_size): Accept zero argument.
37643 (operand_signed_p): Remove.
37644 (maybe_warn_alloc_args_overflow): Call get_size_range.
37645 * calls.h (get_size_range): Declare.
37647 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37649 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37650 from TI's devices.csv file as of September 2016.
37651 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37653 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37655 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37656 * doc/invoke.texi: Likewise.
37657 * doc/md.texi: Likewise.
37658 * doc/objc.texi: Likewise.
37660 2017-01-10 Joshua Conner <joshconner@google.com>
37662 * config/arm/fuchsia-elf.h: New file.
37663 * config/fuchsia.h: New file.
37664 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37665 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37667 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37669 2016-01-10 Richard Biener <rguenther@suse.de>
37671 PR tree-optimization/79034
37672 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37673 Propagate out degenerate PHIs in the joiner.
37675 2017-01-10 Martin Liska <mliska@suse.cz>
37677 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37678 (sort_congruence_classes_by_decl_uid): Likewise.
37679 (sort_congruence_class_groups_by_decl_uid): Likewise.
37680 (sem_item_optimizer::merge_classes): Sort class, groups in these
37681 classes and members in the groups by DECL_UID of declarations.
37682 This would make merge operations stable.
37684 2017-01-10 Martin Liska <mliska@suse.cz>
37686 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37687 usage of m_classes_vec.
37688 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37689 (sem_item_optimizer::get_group_by_hash): Likewise.
37690 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37691 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37692 (sem_item_optimizer::verify_classes): Likewise.
37693 (sem_item_optimizer::process_cong_reduction): Likewise.
37694 (sem_item_optimizer::dump_cong_classes): Likewise.
37695 (sem_item_optimizer::merge_classes): Likewise.
37696 * ipa-icf.h (congruence_class_hash): Rename from
37697 congruence_class_group_hash. Remove declaration of m_classes_vec.
37699 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37701 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37702 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37703 * config.gcc: Add avx512vpopcntdqintrin.h.
37704 * config/i386/avx512vpopcntdqintrin.h: New.
37705 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37706 * config/i386/i386-builtin-types.def: Add new types.
37707 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37708 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37709 __builtin_ia32_vpopcountq_v8di_mask): New.
37710 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37711 __AVX512VPOPCNTDQ__.
37712 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37713 (PTA_AVX512VPOPCNTDQ): Define.
37714 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37715 TARGET_AVX512VPOPCNTDQ_P): Define.
37716 * config/i386/i386.opt: Add mavx512vpopcntdq.
37717 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37718 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37720 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37722 PR middle-end/77484
37723 * predict.def (PRED_CALL): Set to 67.
37725 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37727 * expr.c (store_field): In the bitfield case, if the value comes from
37728 a function call and is of an aggregate type returned in registers, do
37729 not modify the field mode; extract the value in all cases if the mode
37730 is BLKmode and the size is not larger than a word.
37732 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37735 * config/i386/cpuid.h: Fix undefined behavior.
37737 2017-01-04 Jeff Law <law@redhat.com>
37739 PR tree-optimization/79007
37740 PR tree-optimization/67955
37741 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37742 conservative for pt.null when flag_non_call_exceptions is on.
37744 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37746 PR translation/79019
37747 PR translation/79020
37748 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37749 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37750 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37752 * config/avr/avr.opt (maccumulate-args): Likewise.
37753 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37754 * common.opt (freport-bug): Likewise.
37755 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37756 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37757 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37758 translatable string.
37759 * config/i386/i386.c (function_value_32): Likewise.
37760 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37761 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37763 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37764 * common/config/msp430/msp430-common.c (msp430_handle_option):
37766 * symtab.c (symtab_node::verify_base): Likewise.
37767 * opts.c (set_debug_level): Likewise.
37768 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37769 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37770 missing whitespace to translatable strings.
37771 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37772 * config/sh/superh.h: Likewise.
37773 * config/i386/xopintrin.h: Likewise.
37774 * config/i386/znver1.md: Likewise.
37775 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37776 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37777 * double-int.h (struct double_int): Likewise.
37778 * double-int.c (div_and_round_double): Likewise.
37779 * wide-int.cc: Likewise.
37780 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37781 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37782 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37783 (crossjumps_occurred): ... this.
37784 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37787 PR tree-optimization/78899
37788 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37789 returning bool return struct loop *, NULL for failure and the new
37791 (versionable_outer_loop_p): Don't version outer loop if it has
37792 dont_vectorized bit set.
37793 (tree_if_conversion): When versioning outer loop, ensure
37794 tree_if_conversion is performed also on the inner loop of the
37795 non-vectorizable outer loop copy.
37796 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37797 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37798 prevent vectorization of it.
37799 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37800 the outer loop vectorization of the non-scalar version is attempted
37801 before vectorization of the inner loop in scalar version. If
37802 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37803 vectorization of its inner loop.
37804 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37805 has 2 inner loops, rename also on edges from bb whose single pred
37806 is outer_loop->header. Fix typo in function comment.
37808 2017-01-09 Martin Sebor <msebor@redhat.com>
37811 * asan.c (asan_emit_stack_protection): Increase local buffer size
37812 to avoid snprintf truncation warning.
37814 2017-01-09 Andrew Pinski <apinski@cavium.com>
37816 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
37817 to reference thunderx2t99 for the tuning structure
37818 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
37820 (thunderx2t99_extra_costs): This.
37821 * config/aarch64/aarch64-tune.md: Regenerate.
37822 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
37823 (vulcan_addrcost_table): This.
37824 (vulcan_regmove_cost): Rename to ...
37825 (thunderx2t99_regmove_cost): This.
37826 (vulcan_vector_cost): Rename to ...
37827 (thunderx2t99_vector_cost): this.
37828 (vulcan_branch_cost): Rename to ...
37829 (thunderx2t99_branch_cost): This.
37830 (vulcan_tunings): Rename to ...
37831 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
37832 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
37834 2017-01-09 Martin Jambor <mjambor@suse.cz>
37838 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
37839 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
37840 (propagate_vr_accross_jump_function): Use the above function for all
37841 value range computations for pass-through jump functions and type
37842 converasion from explicit value range values.
37843 (ipcp_propagate_stage): Do not attempt to deduce types of formal
37844 parameters from TYPE_ARG_TYPES.
37845 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
37846 (ipa_write_node_info): Stream type of the actual argument.
37847 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
37849 2017-01-09 Martin Liska <mliska@suse.cz>
37852 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
37853 (lookup_compiler): Do not show error message with have_E.
37855 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37857 PR tree-optimization/78938
37858 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
37859 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
37860 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
37861 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
37864 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37866 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
37869 2017-01-09 Richard Biener <rguenther@suse.de>
37871 PR tree-optimization/78997
37872 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
37873 name condition properly.
37875 2017-01-09 Richard Biener <rguenther@suse.de>
37878 * dwarf2out.c (is_cxx): New overload with context.
37879 (is_naming_typedef_decl): Use it.
37881 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37883 * invoke.texi (Option Summary): Correct spacing in option lists
37884 and add line breaks to fix over-long lines.
37886 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37888 PR middle-end/17660
37890 * extend.texi (Common Variable Attributes): Add xref to GCC
37891 Internals manual to explain mode attribute keywords.
37893 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37896 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
37897 and Preprocessor Options.
37898 (Options for Linking): Document -pthread here....
37899 (RS/6000 and PowerPC Options): ...not here.
37900 (Solaris 2 Options): ...or here.
37901 * doc/cppopts.texi: Document -pthread.
37903 2017-01-08 Martin Sebor <msebor@redhat.com>
37905 PR middle-end/77708
37906 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
37907 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
37908 New member functions.
37909 (format_directive): Used them.
37911 (pass_sprintf_length::handle_gimple_call): Same.
37912 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
37913 to avoid truncation for any argument.
37914 (extract_affine_mul): Same.
37915 * tree.c (get_file_function_name): Same.
37917 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37919 PR middle-end/77484
37920 * predict.def (PRED_INDIR_CALL): Set to 86.
37922 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37924 PR preprocessor/54124
37925 * doc/cppopts.texi: Reformat -d subtable to list the full name
37926 of the options. Add cross-reference to the docs for the general
37927 compiler -d options.
37928 * doc/invoke.texi (Developer Options): Add cross-reference to the
37929 preprocessor-specific -d option documentation.
37931 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37933 PR preprocessor/13498
37934 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
37935 redudant material, and reflect new command-line options.
37936 (System Headers): Likewise.
37938 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37940 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
37941 -isystem, and -idirafter. Copy-edit.
37942 * doc/cppopts.texi: Copy-edit. Remove contradiction about
37943 default for -ftrack-macro-expansion. Delete obsolete and
37944 badly-formatted implementation details about -fdebug-cpp output.
37945 * doc/cppwarnopts.texi: Copy-edit.
37947 2017-01-07 David Malcolm <dmalcolm@redhat.com>
37950 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37951 that the transition from a max line width >= 1<<10 to narrower
37952 lines works correctly.
37954 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
37956 * doc/options.texi (PerFunction): New.
37957 * opt-functions.awk (switch_flags): Map both Optimization and
37958 PerFunction to CL_OPTIMIZATION.
37959 * opth-gen.awk: Test for PerFunction flag along with
37961 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
37962 it only when the latter is present. Skip those that don't in
37963 the hash function generator.
37964 * common.opt (fvar-tracking): Mark as PerFunction instead of
37966 (fvar-tracking-assignments): Likewise.
37967 (fvar-tracking-assignments-toggle): Likewise.
37968 (fvar-tracking-uninit): Likewise.
37970 2017-01-07 Jakub Jelinek <jakub@redhat.com>
37972 PR translation/79018
37973 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
37976 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
37979 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
37980 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
37981 TARGET_LONG_JUMP_TABLE_OFFSETS.
37982 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
37983 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
37984 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
37985 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
37986 * config/m68k/m68k.md (tablejump expander): Likewise.
37987 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
37988 TARGET_LONG_JUMP_TABLE_OFFSETS.
37989 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
37990 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
37992 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
37993 David Holsgrove <david.holsgrove@xilinx.com>
37995 * common/config/microblaze/microblaze-common.c
37996 (TARGET_EXCEPT_UNWIND_INFO): Remove.
37997 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
37999 * config/microblaze/microblaze.c (microblaze_must_save_register)
38000 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38002 (microblaze_eh_return): New function.
38003 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38004 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38005 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38006 * config/microblaze/microblaze.md (eh_return): New pattern.
38008 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38010 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38011 GCC_DIAGNOSTIC_STRINGIFY): Define.
38013 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38015 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38017 * config/arm/arm.md (<mcrr>): New.
38019 * config/arm/arm.c (arm_arch5te): New.
38020 (arm_option_override): Set arm_arch5te.
38021 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38023 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38024 (arm_mcrr_qualifiers): ... this. New.
38025 (MRRC_QUALIFIERS): Define to...
38026 (arm_mrrc_qualifiers): ... this. New.
38027 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38029 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38030 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38031 (MRRCI, mrrc, MRRC): New.
38032 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38033 VUNSPEC_MRRC2): New.
38035 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38037 * config/arm/arm.md (<mcr>): New.
38039 * config/arm/arm.c (arm_coproc_builtin_available): Add
38040 support for mcr, mrc, mcr2 and mrc2.
38041 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38042 (arm_mcr_qualifiers): ... this. New.
38043 (MRC_QUALIFIERS): Define to ...
38044 (arm_mrc_qualifiers): ... this. New.
38045 (MCR_QUALIFIERS): Define to ...
38046 (arm_mcr_qualifiers): ... this. New.
38047 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38049 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38050 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38051 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38052 VUNSPEC_MRC2): New.
38054 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38056 * config/arm/arm.md (*ldc): New.
38060 * config/arm/arm.c (arm_coproc_builtin_available): Add
38061 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38062 (arm_coproc_ldc_stc_legitimate_address): New.
38063 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38064 'qualifier_const_pointer'.
38065 (LDC_QUALIFIERS): Define to...
38066 (arm_ldc_qualifiers): ... this. New.
38067 (STC_QUALIFIERS): Define to...
38068 (arm_stc_qualifiers): ... this. New.
38069 * config/arm/arm-protos.h
38070 (arm_coproc_ldc_stc_legitimate_address): New.
38071 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38072 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38073 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38074 stc2, stcl, stc2l): New.
38075 * config/arm/constraints.md (Uz): New.
38076 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38077 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38078 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38079 VUNSPEC_STC2L): New.
38081 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38083 * config/arm/arm.md (<cdp>): New.
38084 * config/arm/arm.c (neon_const_bounds): Rename this ...
38085 (arm_const_bounds): ... this.
38086 (arm_coproc_builtin_available): New.
38087 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38088 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38089 (CDP_QUALIFIERS): Define to...
38090 (arm_cdp_qualifiers): ... this. New.
38092 (arm_expand_builtin_args): Add case for 6 arguments.
38093 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38094 (arm_const_bounds): ... this.
38095 (arm_coproc_builtin_available): New.
38096 * config/arm/arm_acle.h (__arm_cdp): New.
38098 * config/arm/arm_acle_builtins.def (cdp): New.
38100 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38101 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38102 'arm_const_bounds'.
38103 * config/arm/types.md (coproc): New.
38104 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38105 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38106 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38107 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38109 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38111 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38112 (UBINOP_QUALIFIERS): New.
38114 (acle_builtin_data): New. Change comment.
38115 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38116 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38117 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38118 arm_acle_builtins.def.
38119 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38120 (arm_init_acle_builtins): New.
38121 (CRC32_BUILTIN): Remove.
38122 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38123 crc32cb, crc32ch and crc32cw.
38124 (arm_init_crc32_builtins): Remove.
38125 (arm_init_builtins): Use arm_init_acle_builtins rather
38126 than arm_init_crc32_builtins.
38127 (arm_expand_acle_builtin): New.
38128 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38129 * config/arm/arm_acle_builtins.def: New.
38131 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38133 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38134 (arm_builtin_datum): ... this.
38135 (arm_init_neon_builtin): Rename to ...
38136 (arm_init_builtin): ... this. Add a new parameters PREFIX
38137 and USE_SIG_IN_NAME.
38138 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38139 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38140 'arm_builtin_datum'.
38141 (arm_init_vfp_builtins): Likewise.
38142 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38143 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38144 (arm_expand_neon_args): Rename to ...
38145 (arm_expand_builtin_args): ... this. Rename builtin_arg
38146 enum values and differentiate between ARG_BUILTIN_MEMORY
38147 and ARG_BUILTIN_NEON_MEMORY.
38148 (arm_expand_neon_builtin_1): Rename to ...
38149 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38150 values, arm_expand_builtin_args and add bool parameter NEON.
38151 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38152 (arm_expand_vfp_builtin): Likewise.
38153 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38155 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38157 PR middle-end/77484
38158 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38159 * predict.c (tree_estimate_probability_bb): Reverse direction of
38160 polymorphic call predictor.
38162 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38164 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38165 (determine_pass_name_match): ...this new function and...
38166 (should_skip_pass_p): ...this new function.
38168 2017-01-06 Nathan Sidwell <nathan@acm.org>
38170 * ipa-visibility.c (function_and_variable_visibility): Reformat
38171 comments and long lines. Remove extrneous if.
38172 * symtab.c (symtab_node::make_decl_local): Fix code format.
38173 (symtab_node::set_section_for_node): Fix comment typo.
38175 2017-01-06 Martin Liska <mliska@suse.cz>
38178 * lra-constraints.c: Rename invariant to lra_invariant.
38179 * predict.c (set_even_probabilities): Initialize e to NULL.
38181 2017-01-05 Martin Sebor <msebor@redhat.com>
38183 PR tree-optimization/78910
38184 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38185 (format_integer): Correct off-by-one error in the handling
38186 of precision with negative numbers in signed conversions..
38188 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38190 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38192 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38194 PR tree-optimization/71016
38195 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38196 factor_out_conditional_conversion. Formatting fix.
38197 (factor_out_conditional_conversion): Add cond_stmt argument.
38198 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38199 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38202 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38204 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38205 read-rtl-function.o, and selftest-rtl.o.
38206 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38207 (selftest::aarch64_test_loading_full_dump): New function.
38208 (selftest::aarch64_run_selftests): New function.
38209 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38210 selftest::aarch64_run_selftests.
38211 * config/i386/i386.c
38212 (selftest::ix86_test_loading_dump_fragment_1): New function.
38213 (selftest::ix86_test_loading_call_insn): New function.
38214 (selftest::ix86_test_loading_full_dump): New function.
38215 (selftest::ix86_test_loading_unspec): New function.
38216 (selftest::ix86_run_selftests): Call the new functions.
38217 * emit-rtl.c (maybe_set_max_label_num): New function.
38218 * emit-rtl.h (maybe_set_max_label_num): New decl.
38219 * function.c (instantiate_decls): Guard call to
38220 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38221 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38223 * gensupport.c (gen_reader::gen_reader): Pass "false"
38224 for new "compact" param of rtx_reader.
38225 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38226 rather than an empty string for NULL strings.
38227 * read-md.c: Potentially include config.h rather than bconfig.h.
38228 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38229 (have_error): New global, copied from errors.c.
38230 (md_reader::read_name): Rename to...
38231 (md_reader::read_name_1): ...this, adding "out_loc" param,
38232 and converting "missing name or number" to returning false, rather
38234 (md_reader::read_name): Reimplement in terms of read_name_1.
38235 (md_reader::read_name_or_nil): New function.
38236 (md_reader::read_string): Handle "(nil)" by returning NULL.
38237 (md_reader::md_reader): Add new param "compact".
38238 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38239 (md_reader::read_file): New method.
38240 * read-md.h (md_reader::md_reader): Add new param "compact".
38241 (md_reader::read_file): New method.
38242 (md_reader::is_compact): New accessor.
38243 (md_reader::read_name): Convert return type from void to file_location.
38244 (md_reader::read_name_or_nil): New decl.
38245 (md_reader::read_name_1): New decl.
38246 (md_reader::m_compact): New field.
38247 (noop_reader::noop_reader): Pass "false" for new "compact" param
38249 (rtx_reader::rtx_reader): Add new "compact" param.
38250 (rtx_reader::read_rtx_operand): Make virtual and convert return
38251 type from void to rtx.
38252 (rtx_reader::read_until): New decl.
38253 (rtx_reader::handle_any_trailing_information): New virtual function.
38254 (rtx_reader::postprocess): New virtual function.
38255 (rtx_reader::finalize_string): New virtual function.
38256 (rtx_reader::m_in_call_function_usage): New field.
38257 (rtx_reader::m_reuse_rtx_by_id): New field.
38258 * read-rtl-function.c: New file.
38259 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38260 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38261 (selftest::verify_three_block_rtl_cfg): New decl.
38262 * read-rtl-function.h: New file.
38263 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38264 For host, include function.h, memmodel.h, and emit-rtl.h.
38265 (one_time_initialization): New function.
38266 (struct compact_insn_name): New struct.
38267 (compact_insn_names): New array.
38268 (find_code): Handle insn codes in compact dumps.
38269 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38270 (bind_subst_iter_and_attr): Likewise.
38271 (add_condition_to_string): Likewise.
38272 (add_condition_to_rtx): Likewise.
38273 (apply_attribute_uses): Likewise.
38274 (add_current_iterators): Likewise.
38275 (apply_iterators): Likewise.
38276 (initialize_iterators): Guard usage of apply_subst_iterator with
38277 #ifdef GENERATOR_FILE.
38278 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38279 (md_reader::read_mapping): Likewise.
38280 (add_define_attr_for_define_subst): Likewise.
38281 (add_define_subst_attr): Likewise.
38282 (read_subst_mapping): Likewise.
38283 (check_code_iterator): Likewise.
38284 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38286 (one_time_initialization): New function.
38287 (rtx_reader::read_until): New method.
38288 (read_flags): New function.
38289 (parse_reg_note_name): New function.
38290 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38291 Handle reuse_rtx ids.
38292 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38293 Add parsing support for RTL dumps, mirroring the special-cases in
38294 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38295 values, and calling handle_any_trailing_information.
38296 (rtx_reader::read_rtx_operand): Convert return type from void
38297 to rtx, returning return_rtx. Handle case 'e'. Call
38298 finalize_string on XSTR and XTMPL fields.
38299 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38300 "(nil)" values were omitted. Call the postprocess vfunc on the
38302 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38303 class ctor. Initialize m_in_call_function_usage. Call
38304 one_time_initialization.
38305 * rtl-tests.c (selftest::test_uncond_jump): Call
38306 set_new_first_and_last_insn.
38307 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38308 * selftest-rtl.c: New file.
38309 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38310 (selftest::get_insn_by_uid): New decl.
38311 * selftest-run-tests.c (selftest::run_tests): Call
38312 read_rtl_function_c_tests.
38313 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38314 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38317 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38319 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38320 operands in a special way. Assert that pos+len <= mode precision.
38322 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38324 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38325 3 argument Alias with unlimited for the negative form.
38326 (fno-vect-cost-model): Removed.
38328 2017-01-05 Martin Liska <mliska@suse.cz>
38330 * hsa-gen.c (gen_hsa_divmod): New function.
38331 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38333 2017-01-05 Martin Liska <mliska@suse.cz>
38336 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38339 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38341 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38342 small constant length operands.
38344 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38346 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38347 between loop iterations.
38349 2017-01-05 Martin Liska <mliska@suse.cz>
38352 * gimplify.c (gimplify_decl_expr): Compare to
38353 asan_poisoned_variables instread of checking flags.
38354 (gimplify_target_expr): Likewise.
38355 (gimplify_expr): Likewise.
38356 (gimplify_function_tree): Conditionally initialize
38357 asan_poisoned_variables.
38359 2017-01-04 Jeff Law <law@redhat.com>
38361 PR tree-optimizatin/78812
38362 * rtl.h (contains_mem_rtx_p): Prototype.
38363 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38364 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38365 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38366 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38367 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38369 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38371 * input.c (assert_char_at_range): Default-initialize actual_range.
38373 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38375 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38376 to match the caller.
38378 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38380 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38381 insns after final jump in test to emit dummy move.
38383 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38385 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38386 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38388 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38390 * multiple_target.c (create_dispatcher_calls): Init e_next.
38391 * tree-ssa-loop-split.c (split_loop): Init border.
38392 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38395 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38400 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38401 (altivec_register_operand): Do not return true if the operand
38402 contains a SUBREG mixing SImode and SFmode.
38403 (vsx_register_operand): Likewise.
38404 (vsx_reg_sfsubreg_ok): New predicate.
38405 (vfloat_operand): Do not return true if the operand contains a
38406 SUBREG mixing SImode and SFmode.
38407 (vint_operand): Likewise.
38408 (vlogical_operand): Likewise.
38409 (gpc_reg_operand): Likewise.
38410 (int_reg_operand): Likewise.
38411 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38412 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38413 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38415 (rs6000_emit_move_si_sf_subreg): New helper function.
38416 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38417 fixup SUBREGs involving SImode and SFmode.
38418 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38419 numbers for the new peephole2 optimization.
38420 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38421 the GLIBC math library that do AND/IOR/XOR operations on single
38422 precision floating point.
38423 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38424 target macros to say whether we need to avoid SUBREGs mixing
38426 (TARGET_ALLOW_SF_SUBREG): Likewise.
38427 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38428 (UNSPEC_SI_FROM_SF): Likewise.
38429 (iorxor): Change spacing.
38430 (and_ior_xor): New iterator for AND, IOR, and XOR.
38431 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38432 (movdi_from_sf_zero_ext): Likewise.
38433 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38434 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38435 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38436 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38437 (fms<mode>4): Likewise.
38438 (fnma<mode>4): Likewise.
38439 (fnms<mode>4): Likewise.
38440 (nfma<mode>4): Likewise.
38441 (nfms<mode>4): Likewise.
38443 2017-01-04 Marek Polacek <polacek@redhat.com>
38446 * doc/invoke.texi: Document -Wpointer-compare.
38448 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38450 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38453 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38454 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38457 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38460 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38461 documentation of the powerpc_popcntb_ok attribute.
38462 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38463 code to issue warning messages if a requested CPU configuration is
38464 not supported by the binary (assembler and loader) toolchain.
38465 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38466 made to define a built-in function that has been disabled.
38467 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38468 made to define a built-in function that has been disabled.
38469 (altivec_init_builtins): Add comment explaining why definition
38470 of the DST built-in functions is not preceded by an assertion
38471 check. Add assertions to prevent ICE if attempts are made to
38472 define an altivec predicate or an abs* built-in function that has
38474 (htm_init_builtins): Add comment explaining why definition of the
38475 htm built-in functions is not preceded by an assertion check.
38477 2017-01-04 Jeff Law <law@redhat.com>
38479 PR tree-optimizatin/67955
38480 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38481 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38482 the points-to solution does not include pt_null. Use DECL_PT_UID
38485 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38487 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38488 Use gen_int_mode instead of gen_lopwart for const_int operands.
38490 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38492 PR tree-optimization/71563
38493 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38494 out of range value - has low bits known to be zero.
38496 2017-01-04 Alan Modra <amodra@gmail.com>
38498 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38499 * configure: Regenerate.
38500 * config.in: Regenerate.
38502 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38505 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38506 a substring of the message, but strcmp with the whole message. Ifdef
38507 ENABLE_NLS, translate the message first using dgettext.
38509 2017-01-03 Jeff Law <law@redhat.com>
38511 PR tree-optimizatin/78856
38512 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38513 (mark_threaded_blocks): Remove code to truncate thread paths that
38514 cross multiple loop headers. Instead invalidate the cached loop
38515 iteration information and handle case of a thread path walking
38516 into an irreducible region.
38518 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38521 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38522 assertions. Add support for doing the signbit if the IEEE 128-bit
38523 floating point value is in a GPR.
38524 * config/rs6000/rs6000.md (Fsignbit): Delete.
38525 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38526 Update the length attribute if the value is in a GPR.
38527 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38528 the sign or zero extension instruction, since the value is always 0/1.
38529 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38532 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38533 extracting SImode to a GPR register so that we can generate a
38534 store, limit the vector to be in a traditional Altivec register
38535 for the vextuwrx instruction.
38537 2017-01-03 Ian Lance Taylor <iant@google.com>
38539 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38541 2017-01-03 Martin Sebor <msebor@redhat.com>
38543 PR tree-optimization/78696
38544 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38545 precision. Use MPFR for %f for greater fidelity. Correct handling
38547 (pass_sprintf_length::compute_format_length): Set width and precision
38548 specified by asrerisk to void_node for vararg functions.
38549 (try_substitute_return_value): Adjust dump output.
38551 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38553 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38555 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38557 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38558 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38559 -mlra/-mno-lra was passed to the compiler.
38561 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38563 PR rtl-optimization/65618
38564 * emit-rtl.c (try_split): Move initialization of "before" and
38565 "after" to just before the call to emit_insn_after_setloc.
38567 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38569 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38571 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38573 * dwarf2out.c (gen_enumeration_type_die): When
38574 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38576 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38578 PR tree-optimization/78965
38579 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38580 Change first argument from const call_info & to call_info &. For %n
38581 set info.nowrite to false.
38583 PR middle-end/78901
38584 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38585 possibly throwing calls.
38587 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38588 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38589 and fns handling, rather than in a separate case SSA_NAME.
38591 2017-01-02 Jeff Law <law@redhat.com>
38593 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38594 fixes for first_period and second_period variables.
38596 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38599 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38600 (*insvqi_1): New insn pattern.
38601 (*insvqi_1_mem_rex64): Ditto.
38602 (*insvqi_2): Ditto.
38603 (*insvqi_3): Rename from *insvqi.
38605 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38607 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38609 * doc/cfg.texi (Edges): Remove reference to Java.
38610 (Maintaining the CFG): Ditto.
38612 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38614 PR middle-end/77674
38615 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38616 transparent aliases.
38618 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38620 PR middle-end/77484
38621 * predict.def (PRED_CALL): Update hitrate.
38622 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38623 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38624 into direct/indirect/polymorphic variants.
38626 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38628 Update copyright years.
38630 * gcc.c (process_command): Update copyright notice dates.
38631 * gcov-dump.c (print_version): Ditto.
38632 * gcov.c (print_version): Ditto.
38633 * gcov-tool.c (print_version): Ditto.
38634 * gengtype.c (create_file): Ditto.
38635 * doc/cpp.texi: Bump @copying's copyright year.
38636 * doc/cppinternals.texi: Ditto.
38637 * doc/gcc.texi: Ditto.
38638 * doc/gccint.texi: Ditto.
38639 * doc/gcov.texi: Ditto.
38640 * doc/install.texi: Ditto.
38641 * doc/invoke.texi: Ditto.
38643 Copyright (C) 2017 Free Software Foundation, Inc.
38645 Copying and distribution of this file, with or without modification,
38646 are permitted in any medium without royalty provided the copyright
38647 notice and this notice are preserved.