1 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
5 for -fsanitize=thread link of executables.
6 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
12 (<shift_insn><mode>3<mask_name>): Change the last of the 3
13 define_insns for logical vector shifts to use VI248_AVX512BW
14 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
15 condition, useless isa and prefix attributes. Change the first
16 2 of these define_insns to ...
17 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
18 define_insn for avx512vl.
19 (<shift_insn><mode>3): ... and this, new define_insn without
20 masking for non-avx512vl.
23 * config/i386/sse.md (*andnot<mode>3,
24 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
25 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
26 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
27 not applied use empty suffix even for TARGET_AVX512VL.
28 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
29 is applied, supply evex,evex or evex,evex,evex instead of just
32 2017-10-20 Julia Koval <julia.koval@intel.com>
34 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
35 (OPTION_MASK_ISA_GFNI_UNSET): New.
36 (ix86_handle_option): Handle OPT_mgfni.
37 * config/i386/cpuid.h (bit_GFNI): New.
38 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
39 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
40 * config/i386/i386.c (ix86_target_string): Add -mgfni.
41 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
42 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
43 * config/i386/i386.opt: Add mgfni.
45 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
47 * config/msp430/msp430.c (msp430_option_override): Disable
48 -fdelete-null-pointer-checks.
49 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
51 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
53 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
54 of x87 and SSE instructions.
56 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
58 * asan.c (create_cond_insert_point): Do not update edge count.
59 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
60 (afdo_propagate_circuit): Likewise.
61 (afdo_calculate_branch_prob): Likewise.
62 (afdo_annotate_cfg): Likewise.
63 * basic-block.h (struct edge_def): Remove count.
64 (edge_def::count): New accessor.
65 * bb-reorder.c (rotate_loop): Update.
66 (find_traces_1_round): Update.
67 (connect_traces): Update.
68 (sanitize_hot_paths): Update.
69 * cfg.c (unchecked_make_edge): Update.
70 (make_single_succ_edge): Update.
71 (check_bb_profile): Update.
72 (dump_edge_info): Update.
73 (update_bb_profile_for_threading): Update.
74 (scale_bbs_frequencies_int): Update.
75 (scale_bbs_frequencies_gcov_type): Update.
76 (scale_bbs_frequencies_profile_count): Update.
77 (scale_bbs_frequencies): Update.
78 * cfganal.c (connect_infinite_loops_to_exit): Update.
79 * cfgbuild.c (compute_outgoing_frequencies): Update.
80 (find_many_sub_basic_blocks): Update.
81 * cfgcleanup.c (try_forward_edges): Update.
82 (try_crossjump_to_edge): Update
83 * cfgexpand.c (expand_gimple_cond): Update
84 (expand_gimple_tailcall): Update
85 (construct_exit_block): Update
86 * cfghooks.c (verify_flow_info): Update
87 (redirect_edge_succ_nodup): Update
89 (make_forwarder_block): Update
90 (duplicate_block): Update
91 (account_profile_record): Update
92 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
93 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
94 * cfgloopmanip.c (scale_loop_profile): Update.
96 (lv_adjust_loop_entry_edge): Update.
97 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
98 (force_nonfallthru_and_redirect): Update.
99 (purge_dead_edges): Update.
100 (rtl_flow_call_edges_add): Update.
101 * cgraphunit.c (init_lowered_empty_function): Update.
102 (cgraph_node::expand_thunk): Update.
103 * gimple-pretty-print.c (dump_probability): Update.
104 (dump_edge_probability): Update.
105 * gimple-ssa-isolate-paths.c (isolate_path): Update.
106 * haifa-sched.c (sched_create_recovery_edges): Update.
107 * hsa-gen.c (convert_switch_statements): Update.
108 * ifcvt.c (dead_or_predicable): Update.
109 * ipa-inline-transform.c (inline_transform): Update.
110 * ipa-split.c (split_function): Update.
111 * ipa-utils.c (ipa_merge_profiles): Update.
112 * loop-doloop.c (add_test): Update.
113 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
114 * lto-streamer-in.c (input_cfg): Update.
115 (input_function): Update.
116 * lto-streamer-out.c (output_cfg): Update.
117 * modulo-sched.c (sms_schedule): Update.
118 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
119 * predict.c (maybe_hot_edge_p): Update.
120 (unlikely_executed_edge_p): Update.
121 (probably_never_executed_edge_p): Update.
122 (dump_prediction): Update.
123 (drop_profile): Update.
124 (propagate_unlikely_bbs_forward): Update.
125 (determine_unlikely_bbs): Update.
126 (force_edge_cold): Update.
127 * profile.c (compute_branch_probabilities): Update.
128 * reg-stack.c (better_edge): Update.
129 * shrink-wrap.c (handle_simple_exit): Update.
130 * tracer.c (better_p): Update.
131 * trans-mem.c (expand_transaction): Update.
132 (split_bb_make_tm_edge): Update.
133 * tree-call-cdce.c: Update.
134 * tree-cfg.c (gimple_find_sub_bbs): Update.
135 (gimple_split_edge): Update.
136 (gimple_duplicate_sese_region): Update.
137 (gimple_duplicate_sese_tail): Update.
138 (gimple_flow_call_edges_add): Update.
139 (insert_cond_bb): Update.
140 (execute_fixup_cfg): Update.
141 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
142 * tree-complex.c (expand_complex_div_wide): Update.
143 * tree-eh.c (lower_resx): Update.
144 (unsplit_eh): Update.
145 (cleanup_empty_eh_move_lp): Update.
146 * tree-inline.c (copy_edges_for_bb): Update.
147 (freqs_to_counts): Update.
148 (copy_cfg_body): Update.
149 * tree-ssa-dce.c (remove_dead_stmt): Update.
150 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
151 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
152 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
153 (unloop_loops): Update.
154 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
155 * tree-ssa-loop-split.c (connect_loops): Update.
156 (split_loop): Update.
157 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
158 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
159 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
160 * tree-ssa-reassoc.c (branch_fixup): Update.
161 * tree-ssa-tail-merge.c (replace_block_by): Update.
162 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
163 (compute_path_counts): Update.
164 (update_profile): Update.
165 (recompute_probabilities): Update.
166 (update_joiner_offpath_counts): Update.
167 (estimated_freqs_path): Update.
168 (freqs_to_counts_path): Update.
169 (clear_counts_path): Update.
170 (ssa_fix_duplicate_block_edges): Update.
171 (duplicate_thread_path): Update.
172 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
173 (case_bit_test_cmp): Update.
174 (collect_switch_conv_info): Update.
175 (gen_inbound_check): Update.
176 (do_jump_if_equal): Update.
177 (emit_cmp_and_jump_insns): Update.
178 * tree-tailcall.c (decrease_profile): Update.
179 (eliminate_tail_call): Update.
180 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
181 (vect_do_peeling): Update.
182 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
183 * ubsan.c (ubsan_expand_null_ifn): Update.
184 (ubsan_expand_ptr_ifn): Update.
185 * value-prof.c (gimple_divmod_fixed_value): Update.
186 (gimple_mod_pow2): Update.
187 (gimple_mod_subtract): Update.
189 (gimple_stringop_fixed_value): Update.
191 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
194 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
196 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
198 PR rtl-optimization/82395
199 * ira-color.c (allocno_priority_compare_func): Fix comparison step
200 based on non_spilled_static_chain_regno_p.
202 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
204 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
205 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
206 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
208 2017-10-19 Martin Sebor <msebor@redhat.com>
210 PR tree-optimization/82596
211 * tree.c (array_at_struct_end_p): Handle STRING_CST.
213 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
215 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
216 (get_mem_refs_of_builtin_call): Likewise.
217 * builtins.c (expand_builtin_apply): Adjust call to
218 allocate_dynamic_stack_space.
219 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
220 the third argument to allocate_dynamic_stack_space, otherwise -1.
221 (expand_builtin): Deal with all alloca variants.
222 (is_inexpensive_builtin): Likewise.
223 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
224 * calls.c (special_function_p): Deal with all alloca variants.
225 (initialize_argument_information): Adjust call to
226 allocate_dynamic_stack_space.
227 (expand_call): Likewise.
228 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
229 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
230 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
231 use it for the stack usage computation.
232 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
233 * function.c (gimplify_parameters): Call build_alloca_call_expr.
234 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
235 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
236 (in_loop_p): Remove first argument and useless check.
237 (pass_walloca::execute): Remove useless test and adjust call to above.
238 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
239 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
240 (gimplify_call_expr): Deal with all alloca variants.
241 * hsa-gen.c (gen_hsa_alloca): Likewise.
242 (gen_hsa_insns_for_call): Likewise.
243 * ipa-pure-const.c (special_builtin_state): Likewise.
244 * tree-chkp.c (chkp_build_returned_bound): Likewise.
245 * tree-object-size.c (alloc_object_size): Likewise.
246 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
247 (call_may_clobber_ref_p_1): Likewise.
248 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
249 (ccp_fold_stmt): Likewise.
250 (optimize_stack_restore): Likewise.
251 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
252 (mark_all_reaching_defs_necessary_1): Likewise.
253 (propagate_necessity): Likewise.
254 (eliminate_unnecessary_stmts): Likewise.
255 * tree.c (build_common_builtin_nodes): Build
256 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
257 (build_alloca_call_expr): New function.
258 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
259 (CASE_BUILT_IN_ALLOCA): Likewise.
260 (build_alloca_call_expr): Declare.
261 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
263 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
266 * dwarf2out.c (new_die_raw): New static inline function.
267 (new_die): Use it to create the DIE.
268 (add_AT_external_die_ref): Likewise.
269 (clone_die): Likewise.
270 (clone_as_declaration): Likewise.
271 (dwarf2out_vms_debug_main_pointer): Likewise.
272 (base_type_die): Likewise. Remove early return for corner cases.
273 Do not call add_pubtype on the DIE here.
274 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
275 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
276 typedefs for base types with DW_AT_endianity. Make sure a DIE with
277 native order exists for base types, attach the DIE manually and call
278 add_pubtype on it. Do not equate a reverse order DIE to the type.
280 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
282 * config/arm/arm.c (align_ok_ldrd_strd): New function.
283 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
285 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
287 2017-10-19 Jakub Jelinek <jakub@redhat.com>
289 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
290 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
291 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
292 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
293 * opts.c (sanitizer_opts): Add builtin.
294 * ubsan.c (instrument_builtin): New function.
295 (pass_ubsan::execute): Call it.
296 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
297 * doc/invoke.texi: Document -fsanitize=builtin.
299 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
300 builtins, store max (log2 (align), 0) into uchar field instead of
301 align into uptr field.
302 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
303 store uchar 0 field instead of uptr 0 field.
304 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
305 instead of passing one address of struct with 2 locations pass
306 two addresses of structs with 1 location each.
307 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
308 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
309 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
310 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
311 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
312 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
313 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
314 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
316 2017-10-19 Martin Liska <mliska@suse.cz>
319 * file-find.c (remove_prefix): Remove.
320 * file-find.h (remove_prefix): Likewise.
321 * gcc-ar.c: Remove smartness of lookup.
323 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
325 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
326 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
327 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
329 2017-10-19 Jakub Jelinek <jakub@redhat.com>
332 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
333 (setcc + and to xor + setcc): New peephole2.
335 2017-10-19 Tom de Vries <tom@codesourcery.com>
337 * doc/sourcebuild.texi (Test Directives, Variants of
338 dg-require-support): Add dg-require-stack-size.
340 2017-10-19 Martin Liska <mliska@suse.cz>
343 * gimplify.c (gimplify_decl_expr): Do not instrument variables
344 that have a large alignment.
345 (gimplify_target_expr): Likewise.
347 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
349 PR rtl-optimization/82602
350 * ira.c (rtx_moveable_p): Return false for volatile asm.
352 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
355 * config/i386/i386-modes.def (CCGZ): New CC mode.
356 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
357 * config/i386/predicates.md (ix86_comparison_operator):
359 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
360 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
361 with double-word subtraction.
362 (put_condition_code): Handle CCGZmode.
364 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
366 * wide-int.cc (debug (const wide_int &)): New.
367 (debug (const wide_int *)): New.
368 (debug (const widest_int &)): New.
369 (debug (const widest_int *)): New.
371 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
374 * lra-constraints.c (curr_insn_transform): Use non-input operand
375 instead of output one for matched reload.
377 2017-10-18 Bin Cheng <bin.cheng@arm.com>
379 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
380 (tree-ssa-loop-ivopts.h): New header file.
381 (struct builtin_info): New fields.
382 (classify_builtin_1): Compute and record base and offset parts for
383 memset builtin partition by calling strip_offset.
384 (offset_cmp, fuse_memset_builtins): New functions.
385 (finalize_partitions): Fuse adjacent memset partitions by calling
387 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
388 Expose the interface.
389 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
391 2017-10-18 Bin Cheng <bin.cheng@arm.com>
393 PR tree-optimization/82574
394 * tree-loop-distribution.c (find_single_drs): New parameter. Check
395 that data reference must be executed exactly once per iteration
396 against the outermost loop in nest.
397 (classify_partition): Update call to above function.
399 2017-10-18 Richard Biener <rguenther@suse.de>
401 PR tree-optimization/82591
402 * graphite.c (graphite_transform_loops): Move code gen message
404 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
405 Here. Handle scop_to_isl_ast failing.
406 (scop_to_isl_ast): Limit the number of ISL operations.
408 2017-10-18 Richard Biener <rguenther@suse.de>
410 * graphite-isl-ast-to-gimple.c
411 (translate_isl_ast_to_gimple::set_rename): Simplify.
412 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
413 (graphite_copy_stmts_from_block): ... here.
414 (copy_bb_and_scalar_dependences): Simplify.
415 (add_parameters_to_ivs_params): Canonicalize.
416 (generate_entry_out_of_ssa_copies): Simplify.
417 * graphite-sese-to-poly.c (extract_affine_name): Simplify
418 by passing in ISL dimension.
419 (parameter_index_in_region_1): Rename to ...
420 (parameter_index_in_region): ... this.
421 (extract_affine): Adjust assert, pass down parameter index.
422 (add_param_constraints): Use range-info when available.
423 (build_scop_context): Adjust.
424 * sese.c (new_sese_info): Adjust.
425 (free_sese_info): Likewise.
426 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
427 Remove unused typedefs.
428 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
430 2017-10-18 Martin Liska <mliska@suse.cz>
432 * combine.c (simplify_compare_const): Add gcc_fallthrough.
434 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
436 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
437 (s390_sched_init): Do not reset s390_sched_state if we entered the
438 current basic block via a fallthru edge and all others are unlikely.
440 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
442 * config/s390/s390.c (NUM_SIDES): New variable.
443 (LONGRUNNING_THRESHOLD): New variable.
444 (LATENCY_FACTOR): New variable.
445 (s390_sched_score): Decrease score for long-running instructions on
447 (s390_sched_variable_issue): Perform bookkeeping for long-running
450 2017-10-18 Richard Biener <rguenther@suse.de>
452 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
453 Simplify with removal of the parameter rename map.
454 (set_rename): Likewise.
455 (should_copy_to_new_region): Likewise.
456 (graphite_copy_stmts_from_block): Likewise.
457 (copy_bb_and_scalar_dependences): Remove initialization of
458 unused copied_bb_map.
460 (copy_internal_parameters): Likewise.
461 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
462 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
464 (parameter_index_in_region_1): Rename to ...
465 (assign_parameter_index_in_region): ... this. Assert we have
466 a parameter we handle.
467 (scan_tree_for_params): Adjust.
468 * sese.h (parameter_rename_map_t): Remove.
469 (struct sese_info_t): Remove unused parameter_rename_map and
470 copied_bb_map members.
471 * sese.c (new_sese_info): Adjust.
472 (free_sese_info): Likewise.
474 2017-10-18 Martin Liska <mliska@suse.cz>
477 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
480 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
482 * doc/invoke.texi (ffunction-sections and fdata-sections):
485 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
487 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
488 the use statement can throw internally.
490 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
492 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
493 any RTX present on the RHS of a SET.
494 * compare-elim.c (try_eliminate_compare): Restore comment.
496 2017-10-17 Jakub Jelinek <jakub@redhat.com>
498 * langhooks.h (struct lang_hooks): Document that tree_size langhook
499 may be also called on tcc_type nodes.
500 * langhooks.c (lhd_tree_size): Likewise.
502 2017-10-17 David Malcolm <dmalcolm@redhat.com>
504 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
505 format_warning_at_substring.
506 (maybe_warn): Convert source_range * param to a location_t. Pass
507 UNKNOWN_LOCATION rather than NULL to fmtwarn.
508 (format_directive): Remove code to extract source_ranges and
509 source_range * in favor of just a location_t.
510 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
512 * substring-locations.c (format_warning_va): Convert
513 source_range * param to a location_t.
514 (format_warning_at_substring): Likewise.
515 * substring-locations.h (format_warning_va): Likewise.
516 (format_warning_at_substring): Likewise.
518 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
520 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
522 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
523 and scatter/gather ops.
525 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
526 vec_gather_load and vec_scatter_store.
527 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
528 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
529 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
530 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
531 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
532 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
534 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
536 * reg-stack.c (compare_for_stack_reg): Add bool argument.
537 Detect FTST instruction and handle its register pops. Only pop
538 second operand if can_pop_second_op is true.
539 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
540 set can_pop_second_op to false in the compare_for_stack_reg call.
542 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
543 output_fp_compare for stack register operands.
544 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
545 instructions here. Do not emit stack register pops here. Assert
546 that FCOMPP pops next to top stack register. Rewrite function.
548 2017-10-17 Nathan Sidwell <nathan@acm.org>
551 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
552 use DECL_ASSEMBLER_NAME_RAW.
555 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
558 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
559 Wilco Dijkstra <wilco.dijkstra@arm.com>
561 * builtins.c (expand_builtin_update_setjmp_buf): Add a
562 converstion to Pmode from the buf_addr.
564 2017-10-17 Richard Biener <rguenther@suse.de>
566 * graphite-dependences.c (scop_get_reads_and_writes): Change
567 output parameters to references.
569 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
571 PR 71026/tree-optimization
572 * fold-const.c (distribute_real_division): Removed.
573 (fold_binary_loc): Remove calls to distribute_real_divison.
575 2017-10-17 Richard Biener <rguenther@suse.de>
577 * graphite-scop-detection.c
578 (scop_detection::stmt_has_simple_data_refs_p): Always use
579 the full nest as region.
580 (try_generate_gimple_bb): Likewise.
581 * sese.c (scalar_evolution_in_region): Simplify now that
582 SCEV can handle instantiation in regions.
583 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
584 in the non-loop part of a function if requested.
586 2017-10-17 Richard Biener <rguenther@suse.de>
588 PR tree-optimization/82563
589 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
591 (graphite_regenerate_ast_isl): Call it.
592 * graphite-scop-detection.c (build_scops): Remove entry edge split.
594 2017-10-17 Jakub Jelinek <jakub@redhat.com>
596 PR tree-optimization/82549
597 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
598 Formatting fixes. Instead of calling make_bit_field_ref with negative
601 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
603 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
604 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
605 _mm_maskz_reduce_ss): New.
606 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
607 __builtin_ia32_reducess_mask): Ditto..
608 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
609 * config/i386/sse.md (reduces<mode>): Renamed to ...
610 (reduces<mode><mask_scalar_name>): ... this.
611 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
613 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
614 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
616 2017-10-16 David Malcolm <dmalcolm@redhat.com>
618 * Makefile.in (OBJS): Add unique-ptr-tests.o.
619 * selftest-run-tests.c (selftest::run_tests): Call
620 selftest::unique_ptr_tests_cc_tests.
621 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
622 * unique-ptr-tests.cc: New file.
624 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
627 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
629 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
630 (make_hard_regno_born, make_hard_regno_dead): Update
631 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
633 2017-10-16 Jeff Law <law@redhat.com>
635 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
637 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
639 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
641 2017-10-16 Olivier Hainque <hainque@adacore.com>
643 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
644 with_cpu if we were configured for an e500v2 target cpu name.
646 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
648 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
649 * doc/invoke.texi: Document +nodsp as a valid extension for
652 2017-10-16 Martin Liska <mliska@suse.cz>
654 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
655 (test_set_range): Likewise.
656 (test_range_functions): Rename to ...
657 (test_bit_in_range): ... this.
658 (sbitmap_c_tests): Add new test.
660 2017-10-16 Tamar Christina <tamar.christina@arm.com>
662 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
663 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
664 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
666 2017-10-16 Tamar Christina <tamar.christina@arm.com>
668 * config/aarch64/aarch64-builtins.c
669 (aarch64_types_quadopu_lane_qualifiers): New.
670 (TYPES_QUADOPU_LANE): New.
671 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
672 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
673 (aarch64_<sur>dot_laneq<vsi2qi>): New.
674 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
675 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
676 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
677 (Vdottype, DOTPROD): New.
678 (sur): Add SDOT and UDOT.
680 2017-10-16 Tamar Christina <tamar.christina@arm.com>
682 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
683 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
684 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
685 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
686 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
687 (cortex-a75.cortex-a55): Likewise.
688 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
690 2017-10-16 Tamar Christina <tamar.christina@arm.com>
692 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
693 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
694 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
695 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
696 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
697 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
698 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
699 * config/arm/types.md (neon_dot, neon_dot_q): New.
700 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
702 2017-10-16 Tamar Christina <tamar.christina@arm.com>
704 * config/arm/arm.h (TARGET_DOTPROD): New.
705 * config/arm/arm.c (arm_arch_dotprod): New.
706 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
707 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
708 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
709 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
710 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
711 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
712 * doc/invoke.texi (armv8.2-a): Document dotprod
714 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
716 * i386.c (ix86_vec_cost): New function.
717 (ix86_rtx_costs): Handle vector operations better.
718 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
719 * x86-tune-costs.h: Add new costs to all tables.
721 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
723 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
725 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
726 divsd, sqrtss and sqrtsd
727 * x86-tune-costs.h: Add new entries to all costs.
728 (znver1_cost): Fix to match real instruction latencies.
730 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
731 Michael Collison <michael.collison@arm.com>
733 * compare-elim.c: Include emit-rtl.h.
734 (can_merge_compare_into_arith): New function.
735 (try_validate_parallel): Likewise.
736 (try_merge_compare): Likewise.
737 (try_eliminate_compare): Call the above when no previous clobber
739 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
742 2017-10-14 Jakub Jelinek <jakub@redhat.com>
746 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
747 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
751 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
752 to be any operand_equal_p operands. For & (B - 1) require
753 B to be power of 2. Recognize
754 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
756 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
759 * optabs.c (expand_memory_blockage): Fix call of
760 targetm.have_memory_blockage.
762 2017-10-14 Jakub Jelinek <jakub@redhat.com>
765 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
766 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
767 objects to extra_objs instead of overwriting it.
769 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
771 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
772 Use any_fp_register_operand as operand[3] predicate. Simplify
773 equality test for operands[2] and operands[4] memory location.
774 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
775 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
776 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
777 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
778 any_fp_register_operand as operand[1] predicate. Simplify
779 equality test for operands[0] and operands[3] memory location.
780 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
781 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
782 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
784 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
786 * target-insns.def: Add memory_blockage.
787 * optabs.c (expand_memory_blockage): New function.
788 (expand_asm_memory_barrier): Rename ...
789 (expand_asm_memory_blockage): ... to this.
790 (expand_mem_thread_fence): Call expand_memory_blockage
791 instead of expand_asm_memory_barrier.
792 (expand_mem_singnal_fence): Ditto.
793 (expand_atomic_load): Ditto.
794 (expand_atomic_store): Ditto.
795 * doc/md.texi (Standard Pattern Names For Generation):
796 Document memory_blockage instruction pattern.
798 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
800 * config/rl78/rl78.c (rl78_emit_libcall): New function.
801 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
802 * config/rl78/rl78.md: New define_expand "adddi3".
804 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
806 * cfghooks.c (verify_flow_info): Disable check that all probabilities
809 2017-10-13 Jeff Law <law@redhat.com>
811 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
813 2017-10-13 Jakub Jelinek <jakub@redhat.com>
816 * internal-fn.c (expand_mul_overflow): If both operands have
817 the same highpart of -1 or 0 and the topmost bit of lowpart
818 is different, overflow is if res <= 0 rather than res < 0.
820 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
822 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
823 TARGET_P9_VECTOR code for unaligned_load case.
825 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
827 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
829 2017-10-13 Nathan Sidwell <nathan@acm.org>
831 * tree-core.h (tree_contains_struct): Make bool.
832 * tree.c (tree_contains_struct): Likewise.
833 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
834 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
835 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
836 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
837 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
839 2017-10-13 Richard Biener <rguenther@suse.de>
841 * graphite-isl-ast-to-gimple.c
842 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
843 parameters and dominance check.
844 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
845 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
846 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
847 Do not update SSA form here or do intermediate IL verification.
848 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
849 (graphite_initialize): Remove check on the number of loops in
850 the function and inline into graphite_transform_loops.
851 (graphite_finalize): Inline into graphite_transform_loops.
852 (graphite_transform_loops): Perform SSA update and IL verification
854 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
856 2017-10-13 Richard Biener <rguenther@suse.de>
858 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
859 graphite_expression_type_precision): Avoid global constructor
861 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
862 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
863 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
864 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
865 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
866 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
868 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
871 * config/i386/i386.h (ix86_red_zone_size): New.
872 * config/i386/i386.md (push peephole2s): Replace
873 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
875 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
876 Alan Hayward <alan.hayward@arm.com>
877 David Sherwood <david.sherwood@arm.com>
879 * combine.c (can_change_dest_mode): Reject changes in
880 REGMODE_NATURAL_SIZE.
882 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
883 Alan Hayward <alan.hayward@arm.com>
884 David Sherwood <david.sherwood@arm.com>
886 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
887 (expand_debug_source_expr): Likewise.
888 * combine.c (combine_simplify_rtx): Likewise.
889 * cse.c (fold_rtx): Likewise.
890 * fwprop.c (canonicalize_address): Likewise.
891 * targhooks.c (default_shift_truncation_mask): Likewise.
893 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
894 Alan Hayward <alan.hayward@arm.com>
895 David Sherwood <david.sherwood@arm.com>
897 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
898 (widened_mode): Likewise.
899 (expand_unop): Likewise.
900 * ree.c (transform_ifelse): Likewise.
901 (merge_def_and_ext): Likewise.
902 (combine_reaching_defs): Likewise.
903 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
905 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
906 Alan Hayward <alan.hayward@arm.com>
907 David Sherwood <david.sherwood@arm.com>
909 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
910 * combine.c (gen_lowpart_for_combine): Likewise.
911 * dwarf2out.c (rtl_for_decl_location): Likewise.
912 * final.c (alter_subreg): Likewise.
913 * rtlhooks.c (gen_lowpart_general): Likewise.
914 (gen_lowpart_if_possible): Likewise.
916 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
917 Alan Hayward <alan.hayward@arm.com>
918 David Sherwood <david.sherwood@arm.com>
920 * calls.c (expand_call): Use subreg_lowpart_offset.
921 * cse.c (cse_insn): Likewise.
922 * regcprop.c (copy_value): Likewise.
923 (copyprop_hardreg_forward_1): Likewise.
925 2017-10-13 Jakub Jelinek <jakub@redhat.com>
928 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
929 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
930 =Q constraints to +Q and into insn condition add check
931 that operands[0] and operands[1] are equal.
932 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
933 =Q constraints to +Q and into insn condition add check
934 that operands[0] is equal to either operands[1] or operands[2].
937 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
938 instead of handling MINUS_EXPR twice (once for each argument),
939 canonicalize operand order and handle just once, use rtype where
940 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
943 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
944 any values of __C while still being pattern recognizable as a simple
947 2017-10-13 Richard Biener <rguenther@suse.de>
949 PR tree-optimization/82451
951 2017-10-02 Richard Biener <rguenther@suse.de>
953 PR tree-optimization/82355
954 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
955 a mapping for the enclosing loop but avoid generating one for
957 (copy_bb_and_scalar_dependences): Remove premature codegen
958 error on PHIs in blocks duplicated into multiple places.
959 * graphite-scop-detection.c
960 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
961 in the region use it as loop and nest to analyze the DR in.
962 (try_generate_gimple_bb): Likewise.
963 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
964 (add_loop_constraints): For blocks in a loop not in the region
965 create a dimension with a single iteration.
966 * sese.h (gbb_loop_at_index): Remove assert.
968 * cfgloop.c (loop_preheader_edge): For the loop tree root
969 return the single successor of the entry block.
970 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
971 Reset the SCEV hashtable and niters.
972 * graphite-scop-detection.c
973 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
974 assert that we only have POLYNOMIAL_CHREC that vary in loops
975 contained in the region.
976 (scop_detection::graphite_can_represent_expr): Adjust.
977 (scop_detection::stmt_has_simple_data_refs_p): For loops
978 not in the region set loop to NULL. The nest is now the
979 entry edge to the region.
980 (try_generate_gimple_bb): Likewise.
981 * sese.c (scalar_evolution_in_region): Adjust for
982 instantiate_scev change.
983 * tree-data-ref.h (graphite_find_data_references_in_stmt):
984 Make nest parameter the edge into the region.
985 (create_data_ref): Likewise.
986 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
987 entry edge into a region and adjust instantiate_scev calls.
988 (create_data_ref): Likewise.
989 (graphite_find_data_references_in_stmt): Likewise.
990 (find_data_references_in_stmt): Pass the loop preheader edge
991 from the nest argument.
992 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
993 parameter the edge into the region.
994 (instantiate_parameters): Use the loop preheader edge as entry.
995 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
997 (get_instantiated_value_entry): Make instantiate_below parameter
998 the edge into the region.
999 (instantiate_scev_name): Likewise. Adjust dominance checks,
1000 when we cannot use loop-based instantiation instantiate by
1001 walking use-def chains.
1002 (instantiate_scev_poly): Adjust.
1003 (instantiate_scev_binary): Likewise.
1004 (instantiate_scev_convert): Likewise.
1005 (instantiate_scev_not): Likewise.
1006 (instantiate_array_ref): Remove.
1007 (instantiate_scev_3): Likewise.
1008 (instantiate_scev_2): Likewise.
1009 (instantiate_scev_1): Likewise.
1010 (instantiate_scev_r): Do not blindly handle N-operand trees.
1011 Do not instantiate array-refs. Handle all constants and invariants.
1012 (instantiate_scev): Make instantiate_below parameter
1013 the edge into the region.
1014 (resolve_mixers): Use the loop preheader edge for the region
1015 parameter to instantiate_scev_r.
1016 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1018 2017-10-13 Richard Biener <rguenther@suse.de>
1020 PR tree-optimization/82525
1021 * graphite-isl-ast-to-gimple.c
1022 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1024 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1025 Fail code generation when we cannot represent the isl integer.
1026 (binary_op_to_tree): Elide modulo operations that are no-ops
1027 in the type we code generate. Remove now superfluous code
1030 2017-10-13 Richard Biener <rguenther@suse.de>
1032 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1033 (scop_detection::harmful_loop_in_region): Remove premature
1034 IV type restriction.
1035 (scop_detection::graphite_can_represent_scev): We can handle
1036 pointer IVs just fine.
1038 2017-10-13 Alan Modra <amodra@gmail.com>
1040 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1041 "Clobbers and Scratch Registers". Add paragraph on
1042 alternative to clobbers for scratch registers and OpenBLAS
1045 2017-10-13 Alan Modra <amodra@gmail.com>
1047 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1048 example of a memory input for a string of known length. Move
1049 commentary out of table. Add a number of new examples
1050 covering array memory inputs.
1052 2017-10-12 Martin Liska <mliska@suse.cz>
1054 PR tree-optimization/82493
1055 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1056 (test_range_functions): New function.
1057 (sbitmap_c_tests): Likewise.
1058 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1059 * selftest.h (sbitmap_c_tests): New function.
1061 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1063 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1065 * config/rs6000/amo.h: Fix spacing issue.
1067 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1070 * config/i386/i386.md (*ashl<mode>3_mask_1,
1071 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1072 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1075 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1077 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
1078 (profile_probability): Set max_probability
1079 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
1081 * profile-count.c (profile_probability::differs_from_p): Do not
1082 rely on max_probaiblity == 10000
1084 2017-10-12 Jeff Law <law@redhat.com>
1086 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
1089 2017-10-12 Martin Sebor <msebor@redhat.com>
1093 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
1094 (handle_alias_pairs): Call it.
1095 * common.opt (-Wattribute-alias): New option.
1096 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
1097 * doc/invoke.texi (-Wattribute-alias): Document.
1099 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
1102 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1104 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1106 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1107 (make_hard_regno_born, make_hard_regno_dead): Update
1108 bb_killed_pseudos and bb_gen_pseudos.
1110 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1112 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
1114 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
1116 * config/alpha/alpha.c (alpha_split_conditional_move):
1117 Use std::swap instead of manually swapping.
1118 (alpha_stdarg_optimize_hook): Ditto.
1119 (alpha_canonicalize_comparison): Ditto.
1121 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1123 * tree-loop-distribution.c (struct builtin_info): New struct.
1124 (struct partition): Refactor fields into struct builtin_info.
1125 (partition_free): Free struct builtin_info.
1126 (build_size_arg_loc, build_addr_arg_loc): Delete.
1127 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
1128 information from struct builtin_info.
1129 (find_single_drs): New function refactored from classify_partition.
1130 Also moved builtin validity checks to this function.
1131 (compute_access_range, alloc_builtin): New functions.
1132 (classify_builtin_st, classify_builtin_ldst): New functions.
1133 (classify_partition): Refactor code into functions find_single_drs,
1134 classify_builtin_st and classify_builtin_ldst.
1135 (distribute_loop): Don't do runtime alias check when distributing
1137 (find_seed_stmts_for_distribution): New function.
1138 (pass_loop_distribution::execute): Refactor code finding seed
1139 stmts into above function. Support distribution for the innermost
1140 two-level loop nest. Adjust dump information.
1142 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1144 * tree-loop-distribution.c: Adjust the general comment.
1145 (NUM_PARTITION_THRESHOLD): New macro.
1146 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
1147 (classify_partition): Skip builtin pattern of loop nest's inner loop.
1148 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
1149 in call to build_partition_graph.
1150 (finalize_partitions): New parameter. Make loop distribution more
1151 conservative by fusing more partitions.
1152 (distribute_loop): Don't do runtime alias check in case of loop nest
1154 (find_seed_stmts_for_distribution): New function.
1155 (prepare_perfect_loop_nest): New function.
1156 (pass_loop_distribution::execute): Refactor code finding seed stmts
1157 and loop nest into above functions. Support loop nest distribution.
1158 Adjust dump information accordingly.
1160 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1162 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
1163 and set PTYPE_SEQUENTIAL for merged partition.
1165 2017-10-12 Richard Biener <rguenther@suse.de>
1167 PR tree-optimization/69728
1169 2017-09-19 Richard Biener <rguenther@suse.de>
1171 PR tree-optimization/69728
1172 * graphite-sese-to-poly.c (schedule_error): New global.
1173 (add_loop_schedule): Handle empty domain by failing the
1175 (build_original_schedule): Handle schedule_error.
1177 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
1178 domain by returning an unchanged schedule.
1180 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1182 * genrecog.c (validate_pattern): For VEC_SELECT verify that
1183 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
1185 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
1187 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
1190 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1193 * expr.c (store_field): Don't optimize away bitsize == 0 store
1194 from CALL_EXPR with addressable return type.
1196 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1198 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
1199 * config/rs6000/rs6000.md (sel): Delete mode attribute.
1200 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
1201 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
1202 TARGET_ISEL instead of TARGET_ISEL<sel>.
1204 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
1206 * config/rs6000/rs6000.c
1207 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
1209 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1211 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
1213 (reg_or_cint_operand): Fix comment.
1214 (reg_or_zero_operand): New predicate.
1215 * config/rs6000/rs6000-protos.h (output_isel): Delete.
1216 * config/rs6000/rs6000.c (output_isel): Delete.
1217 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
1218 instead of reg_or_cint_operand. Output instruction directly (not via
1220 (isel_unsigned_<mode>): Ditto.
1221 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
1222 gpc_reg_operand. Add an instruction alternative for this. Output
1223 instruction directly.
1224 (*isel_reversed_unsigned_<mode>): Ditto.
1226 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1228 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
1229 (TARGET_CANONICALIZE_COMPARISON): Define.
1231 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
1234 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1235 Check whether the dest is REG before adding REG_EQUIV note.
1237 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1240 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1242 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1243 (make_hard_regno_born, make_hard_regno_dead): Update
1244 bb_killed_pseudos and bb_gen_pseudos.
1246 2017-10-11 Nathan Sidwell <nathan@acm.org>
1248 * incpath.h (enum incpath_kind): Name enum, prefix values.
1249 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
1250 * incpath.c (heads, tails): Use INC_MAX.
1251 (add_env_var_paths, add_standard_paths): Use incpath_kind.
1252 (merge_include_chains, split_quote_chain,
1253 register_include_chains): Update incpath_kind names.
1254 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
1255 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
1257 (add_framework_path, darwin_register_objc_includes): Likewise.
1258 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
1260 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1262 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1263 Do not use float_operator operator predicate.
1264 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1265 * config/i386/predicates.md (float_operator): Remove predicate.
1267 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1269 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
1270 (*jccxf_i387): Ditto.
1271 (*jcc<mode>_i387): Ditto.
1272 (*jccu<mode>_i387): Ditto.
1273 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
1274 (*jcc_*_i387 splitters): Remove.
1275 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
1276 * config/i386/i386.c (ix86_split_fp_branch): Remove.
1277 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
1280 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1282 * profile-count.h (slow_safe_scale_64bit): New function.
1283 (safe_scale_64bit): New inline.
1284 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
1285 * profile-count.c: Include wide-int.h
1286 (slow_safe_scale_64bit): New.
1288 2017-10-11 Nathan Sidwell <nathan@acm.org>
1290 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
1291 HAS_DECL_ASSEMBLER_NAME_P.
1292 * gimple-expr.c (gimple_decl_printable_name: Check
1293 HAS_DECL_ASSEMBLER_NAME_P too.
1294 * ipa-utils.h (type_in_anonymous_namespace_p): Check
1295 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
1296 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
1297 * passes.c (rest_of_decl_compilation): Check
1298 HAS_DECL_ASSEMBLER_NAME_P too.
1299 * recog.c (verify_changes): Likewise.
1300 * tree-pretty-print.c (dump_decl_name): Likewise.
1301 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
1303 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
1304 (SET_DECL_ASSEMBLER_NAME): Use it.
1305 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
1306 (COPY_DECL_ASSEMBLER_NAME): Likewise.
1307 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
1309 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1311 * config.gcc (i386, x86_64): Add extra objects.
1312 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
1313 (ix86_min_insn_size): Declare.
1314 (ix86_issue_rate): Declare.
1315 (ix86_adjust_cost): Declare.
1316 (ia32_multipass_dfa_lookahead): Declare.
1317 (ix86_macro_fusion_p): Declare.
1318 (ix86_macro_fusion_pair_p): Declare.
1319 (ix86_bd_has_dispatch): Declare.
1320 (ix86_bd_do_dispatch): Declare.
1321 (ix86_core2i7_init_hooks): Declare.
1322 (ix86_atom_sched_reorder): Declare.
1323 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
1324 (COSTS_N_BYTES): Move to x86-tune-costs.h.
1325 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
1326 (rip_relative_addr_p): Rename to ...
1327 (ix86_rip_relative_addr_p): ... this one; export.
1328 (memory_address_length): Update.
1329 (ix86_issue_rate): Move to x86-tune-sched.c.
1330 (ix86_flags_dependent): Move to x86-tune-sched.c.
1331 (ix86_agi_dependent): Move to x86-tune-sched.c.
1332 (exact_dependency_1): Move to x86-tune-sched.c.
1333 (exact_store_load_dependency): Move to x86-tune-sched.c.
1334 (ix86_adjust_cost): Move to x86-tune-sched.c.
1335 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
1336 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
1337 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
1338 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
1339 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
1340 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
1341 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
1342 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
1343 (min_insn_size): Rename to ...
1344 (ix86_min_insn_size): ... this one; export.
1345 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
1346 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
1347 (core2i7_first_cycle_multipass_backtrack): Move to
1348 x86-tune-sched-core.c.
1349 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
1350 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
1351 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
1352 (ix86_avoid_jump_mispredicts): Update.
1353 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
1354 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
1355 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
1356 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
1357 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
1358 (MAX_INSN): Move to ix86-tune-sched-bd.c.
1359 (MAX_IMM): Move to ix86-tune-sched-bd.c.
1360 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
1361 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
1362 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
1363 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
1364 (MAX_STORE): Move to ix86-tune-sched-bd.c.
1365 (BIG): Move to ix86-tune-sched-bd.c.
1366 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
1367 (enum insn_path): Move to ix86-tune-sched-bd.c.
1368 (get_mem_group): Move to ix86-tune-sched-bd.c.
1369 (is_cmp): Move to ix86-tune-sched-bd.c.
1370 (dispatch_violation): Move to ix86-tune-sched-bd.c.
1371 (is_branch): Move to ix86-tune-sched-bd.c.
1372 (is_prefetch): Move to ix86-tune-sched-bd.c.
1373 (init_window): Move to ix86-tune-sched-bd.c.
1374 (allocate_window): Move to ix86-tune-sched-bd.c.
1375 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
1376 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
1377 (process_end_window): Move to ix86-tune-sched-bd.c.
1378 (allocate_next_window): Move to ix86-tune-sched-bd.c.
1379 (find_constant): Move to ix86-tune-sched-bd.c.
1380 (get_num_immediates): Move to ix86-tune-sched-bd.c.
1381 (has_immediate): Move to ix86-tune-sched-bd.c.
1382 (get_insn_path): Move to ix86-tune-sched-bd.c.
1383 (get_insn_group): Move to ix86-tune-sched-bd.c.
1384 (count_num_restricted): Move to ix86-tune-sched-bd.c.
1385 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
1386 (add_insn_window): Move to ix86-tune-sched-bd.c.
1387 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
1388 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
1389 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
1390 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
1391 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
1392 (do_dispatch): Move to ix86-tune-sched-bd.c.
1393 (has_dispatch): Move to ix86-tune-sched-bd.c.
1394 * config/i386/t-i386: Add new object files.
1395 * config/i386/x86-tune-costs.h: New file.
1396 * config/i386/x86-tune-sched-atom.c: New file.
1397 * config/i386/x86-tune-sched-bd.c: New file.
1398 * config/i386/x86-tune-sched-core.c: New file.
1399 * config/i386/x86-tune-sched.c: New file.
1401 2017-10-11 Liu Hao <lh_mouse@126.com>
1403 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
1404 the generic version below instead.
1405 (should_colorize): Recognize Windows consoles as terminals
1407 * pretty-print.c [__MINGW32__] (write_all): New function.
1408 [__MINGW32__] (find_esc_head): Likewise.
1409 [__MINGW32__] (find_esc_terminator): Likewise.
1410 [__MINGW32__] (eat_esc_sequence): Likewise.
1411 [__MINGW32__] (mingw_ansi_fputs): New function that handles
1413 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
1416 2017-10-11 Richard Biener <rguenther@suse.de>
1418 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
1419 Properly call analyze_scalar_evolution with the loop of the stmt.
1421 2017-10-11 Richard Biener <rguenther@suse.de>
1423 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
1424 * tree-core.h (tree_base): Add chrec_var union member.
1425 * tree.h (CHREC_VAR): Remove.
1426 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
1427 * tree-chrec.h (build_polynomial_chrec): Adjust.
1428 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
1429 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
1431 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
1433 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
1434 * match.pd: ... here.
1435 ((T) X == (T) Y): Relax condition.
1437 2017-10-11 Bin Cheng <bin.cheng@arm.com>
1439 PR tree-optimization/82472
1440 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
1442 (break_alias_scc_partitions): Update postorder number.
1444 2017-10-11 Martin Liska <mliska@suse.cz>
1447 * opts.c (parse_no_sanitize_attribute): Do not use error_value
1449 * opts.h (parse_no_sanitize_attribute): Remove last argument.
1451 2017-10-11 Martin Liska <mliska@suse.cz>
1453 * print-rtl.c (print_insn): Move declaration of idbuf
1454 to same scope as name.
1456 2017-10-11 Martin Liska <mliska@suse.cz>
1461 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1464 2017-10-11 Martin Liska <mliska@suse.cz>
1467 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1470 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
1472 * config.gcc (default_gnu_indirect_function): Default to yes for
1473 arm*-*-linux* with glibc.
1475 2017-10-11 Richard Biener <rguenther@suse.de>
1477 * tree-scalar-evolution.c (get_scalar_evolution): Handle
1478 default-defs and types we do not want to analyze.
1479 (interpret_loop_phi): Replace unreachable code with an assert.
1480 (compute_scalar_evolution_in_loop): Remove and inline ...
1481 (analyze_scalar_evolution_1): ... here, replacing condition with
1482 what makes the intent clearer. Remove handling of cases
1483 get_scalar_evolution now handles.
1485 2017-10-10 Jim Wilson <wilson@tuliptree.org>
1487 PR rtl-optimization/81434
1488 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
1489 comment for main loop. In sched_group_found if, also add checks for
1490 pass and min_cost_group.
1492 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
1494 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
1495 (rs6000_insn_cost): New function.
1496 * config/rs6000/rs6000.md (cost): New attribute.
1498 2017-10-10 Jakub Jelinek <jakub@redhat.com>
1499 H.J. Lu <hongjiu.lu@intel.com>
1503 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
1504 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
1505 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
1506 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
1507 to OPTION_MASK_ISA_AVX512VL - builtins that have both
1508 OPTION_MASK_ISA_MMX and some other bit set require both
1509 mmx and the ISAs without the mmx bit.
1510 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
1511 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
1512 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
1513 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
1514 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
1515 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
1516 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
1517 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
1518 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
1519 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
1520 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
1521 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
1522 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
1523 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
1524 Add OPTION_MASK_ISA_MMX.
1526 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
1528 * config.gcc (armv7*-*-freebsd*): New target.
1529 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
1531 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
1533 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
1534 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
1537 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
1539 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
1540 a template parameter.
1541 (WIDE_INT_REF_FOR): Update accordingly.
1542 * tree.h (wi::int_traits <const_tree>): Delete.
1543 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
1544 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
1545 (wi::tree_to_wide_ref): New typedef.
1546 (wi::to_wide): New function.
1547 * calls.c (get_size_range): Use wi::to_wide when operating on
1549 * cgraph.c (cgraph_node::create_thunk): Likewise.
1550 * config/i386/i386.c (ix86_data_alignment): Likewise.
1551 (ix86_local_alignment): Likewise.
1552 * dbxout.c (stabstr_O): Likewise.
1553 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
1554 * expr.c (const_vector_from_tree): Likewise.
1555 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
1556 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
1557 (fold_negate_expr_1, int_const_binop_1, const_binop)
1558 (fold_convert_const_int_from_real, optimize_bit_field_compare)
1559 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
1560 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
1561 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
1562 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
1563 (fold_not_const, round_up_loc): Likewise.
1564 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
1565 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
1566 (alloca_call_type): Likewise.
1567 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
1568 * godump.c (go_output_typedef): Likewise.
1569 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1570 * internal-fn.c (get_min_precision): Likewise.
1571 * ipa-cp.c (ipcp_store_vr_results): Likewise.
1572 * ipa-polymorphic-call.c
1573 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1574 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
1575 (ipa_modify_call_arguments): Likewise.
1576 * match.pd: Likewise.
1577 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
1578 * print-tree.c (print_node_brief, print_node): Likewise.
1579 * stmt.c (expand_case): Likewise.
1580 * stor-layout.c (layout_type): Likewise.
1581 * tree-affine.c (tree_to_aff_combination): Likewise.
1582 * tree-cfg.c (group_case_labels_stmt): Likewise.
1583 * tree-data-ref.c (dr_analyze_indices): Likewise.
1584 (prune_runtime_alias_test_list): Likewise.
1585 * tree-dump.c (dequeue_and_dump): Likewise.
1586 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
1587 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
1588 * tree-pretty-print.c (dump_generic_node): Likewise.
1589 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
1590 (simple_iv_with_niters): Likewise.
1591 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1592 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
1593 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
1594 * tree-ssa-loop-niter.c (split_to_var_and_offset)
1595 (refine_value_range_using_guard, number_of_iterations_ne_max)
1596 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
1597 (get_cst_init_from_scev, record_nonwrapping_iv)
1598 (scev_var_range_cant_overflow): Likewise.
1599 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
1600 * tree-ssa-pre.c (compute_avail): Likewise.
1601 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1602 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
1603 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1604 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
1605 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
1606 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
1607 (dump_case_nodes, try_switch_expansion): Likewise.
1608 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
1609 (vect_do_peeling): Likewise.
1610 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
1611 * tree-vect-stmts.c (vectorizable_load): Likewise.
1612 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
1613 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
1614 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
1615 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
1616 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
1617 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
1618 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
1619 (evrp_dom_walker::before_dom_children): Likewise.
1620 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
1621 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
1622 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
1623 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
1624 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
1625 (get_range_pos_neg): Likewise.
1626 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
1627 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
1628 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
1629 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
1630 * config/avr/avr.c (avr_fold_builtin): Likewise.
1631 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
1632 * config/msp430/msp430.c (msp430_attr): Likewise.
1633 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
1634 * config/powerpcspe/powerpcspe-c.c
1635 (altivec_resolve_overloaded_builtin): Likewise.
1636 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
1637 (rs6000_expand_ternop_builtin): Likewise.
1638 * config/rs6000/rs6000-c.c
1639 (altivec_resolve_overloaded_builtin): Likewise.
1640 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
1641 (rs6000_expand_ternop_builtin): Likewise.
1642 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
1644 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1646 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
1647 when copying loop nest with only one inner loop.
1649 2017-10-10 Richard Biener <rguenther@suse.de>
1651 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
1652 blocks if SCEV is active.
1653 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
1655 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
1656 (scev_initialize): Assert we are not yet initialized.
1658 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1660 * tree-loop-distribution.c (generate_loops_for_partition): Remove
1661 inner loop's exit stmt by making it always exit the loop, otherwise
1662 we would generate an infinite empty loop.
1664 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1666 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
1667 renaming variables in new preheader if it's deleted.
1669 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1671 * tree-loop-distribution.c (struct partition): Remove unused field
1672 loops of the structure.
1673 (partition_alloc, partition_free): Ditto.
1674 (build_rdg_partition_for_vertex): Ditto.
1676 2017-10-09 Jeff Law <law@redhat.com>
1678 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
1679 return type to match prototype and documentation.
1681 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1683 * config/rs6000/rs6000.c (processor_costs): Move to ...
1684 * config/rs6000/rs6000.h: ... here.
1685 (rs6000_cost): Declare.
1687 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
1689 * except.c (setjmp_fn): New global variable.
1690 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
1691 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
1692 if DONT_USE_BUILTIN_SETJMP is defined.
1694 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1696 * target.def (insn_cost): New hook.
1697 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
1698 * doc/tm.texi: Regenerate.
1699 * rtlanal.c (insn_cost): Use the new hook.
1701 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1703 * combine.c (combine_validate_cost): Compute the new insn_cost,
1704 not just pattern_cost.
1705 (try_combine): Adjust comment.
1707 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1709 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
1711 * combine.c (uid_insn_cost): Adjust comment.
1712 (combine_validate_cost): Adjust comment. Use pattern_cost instead
1714 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
1715 * dse.c (find_shift_sequence): Ditto.
1716 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
1717 (bb_valid_for_noce_process_p): Use pattern_cost.
1718 * rtl.h (insn_rtx_cost): Delete.
1719 (pattern_cost): New prototype.
1720 (insn_cost): New prototype.
1721 * rtlanal.c (insn_rtx_cost): Rename to...
1722 (pattern_cost): ... this.
1725 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
1727 * config/i386/i386.md (*jcc_2): Remove insn pattern.
1728 (*jcc<mode>_0_r_i387): Ditto.
1729 (*jccxf_r_i387): Ditto.
1730 (*jcc<mode>_r_i387): Ditto.
1731 (*jccu<mode>_r_i387): Ditto.
1732 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
1733 (*jcc): Rename from *jcc_1.
1735 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1737 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
1738 deferred rescans after the lvx/stvx recombination pre-pass.
1740 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1742 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
1743 memory operation instruction support.
1744 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
1745 (rs6000-ibm-aix[789]*): Likewise.
1746 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
1747 Document new functions.
1749 2017-10-09 Richard Biener <rguenther@suse.de>
1751 PR tree-optimization/82397
1752 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
1753 equality only for semantically equal trees.
1755 2017-10-09 Richard Biener <rguenther@suse.de>
1757 PR tree-optimization/82449
1758 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
1759 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
1760 allow constant addresses.
1761 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
1764 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1766 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
1769 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1772 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
1775 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1778 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
1780 2017-10-09 Jakub Jelinek <jakub@redhat.com>
1783 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
1784 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
1786 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
1788 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
1789 (WI_BINARY_PREDICATE_RESULT): Likewise.
1790 (wi::binary_traits::operator_result): New type.
1791 (wi::binary_traits::predicate_result): Likewise.
1792 (generic_wide_int::operator~, unary generic_wide_int::operator-)
1793 (generic_wide_int::operator==, generic_wide_int::operator!=)
1794 (generic_wide_int::operator&, generic_wide_int::and_not)
1795 (generic_wide_int::operator|, generic_wide_int::or_not)
1796 (generic_wide_int::operator^, generic_wide_int::operator+
1797 (binary generic_wide_int::operator-, generic_wide_int::operator*):
1799 (operator~, unary operator-, operator==, operator!=, operator&)
1800 (operator|, operator^, operator+, binary operator-, operator*): New
1802 * expr.c (get_inner_reference): Use wi::bit_and_not.
1803 * fold-const.c (fold_binary_loc): Likewise.
1804 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
1805 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
1806 (bit_value_binop): Likewise.
1807 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
1808 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
1809 (extract_range_from_binary_expr_1): Likewise.
1810 (masked_increment): Likewise.
1811 (simplify_bit_ops_using_ranges): Likewise.
1813 2017-10-09 Martin Jambor <mjambor@suse.cz>
1816 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
1817 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
1818 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
1820 (hsa_fixup_mov_insn_type): New function.
1821 (hsa_op_with_type::get_in_type): Use it.
1822 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
1823 immediates in an assert.
1824 (hsa_op_with_type::extend_int_to_32bit): New method.
1825 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
1826 types. Convert to dest type if necessary.
1827 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
1828 (reg_for_gimple_ssa): Pass false as min32int to
1829 hsa_type_for_scalar_tree_type.
1830 (gen_hsa_addr): Fixup type when creating addresable temporary.
1831 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
1832 (gen_hsa_unary_operation): Extend operands and convert to dest type if
1833 necessary. Call hsa_fixup_mov_insn_type.
1834 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
1835 extend operands and convert to dest type if necessary.
1836 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
1837 to dest type if necessary.
1838 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
1839 if conversion nt necessary and size matches.
1840 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
1841 to dest type if necessary.
1842 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
1843 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
1845 (gen_hsa_clrsb): Likewise.
1846 (gen_hsa_ffs): Likewise.
1847 (gen_hsa_divmod): Extend operands and convert to dest type if
1849 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
1851 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1853 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
1854 Remove empty default arguments. Use a brace block as output
1856 (conditional return): Ditto.
1858 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
1859 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
1861 (group_ending_nop): Ditto.
1862 (doloop_end): Ditto.
1863 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
1864 (splitters for those): Ditto.
1866 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1868 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
1869 a conditional jump (and the compare for it) so that pc_rtx is the
1871 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
1872 for the deleted and renamed ctr<mode>_internal[234] patterns.
1873 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
1874 Delete second conditional return pattern.
1875 (ctr<mode>_internal2): Delete this second bdnz pattern.
1876 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
1877 (ctr<mode>_internal4): Delete this second bdz pattern.
1879 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1881 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
1882 (always_initialized_rtx_for_ssa_name_p): New predicate.
1883 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
1884 (finish_out_of_ssa): Free new field of SA.
1885 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
1886 * tree-ssa-coalesce.c: Include tree-ssa.h.
1887 (get_parm_default_def_partitions): Remove extern keyword.
1888 (get_undefined_value_partitions): New function.
1889 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
1890 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
1892 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
1894 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1896 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
1898 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1900 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
1901 for targets that preffer 128bit.
1903 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1905 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
1907 2017-10-08 Olivier Hainque <hainque@adacore.com>
1909 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
1910 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
1912 (thumb_set_return_address): Likewise.
1914 2017-10-08 Olivier Hainque <hainque@adacore.com>
1916 * common/config/arm/arm-common.c (arm_except_unwind_info):
1917 Handle DWARF2_UNWIND_INFO.
1919 2017-10-07 Michael Collison <michael.collison@arm.com>
1921 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
1924 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
1926 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
1927 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
1928 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
1929 defined, force the creation of a new block for a dispatch label.
1931 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
1933 * invoke.texi (Wsuggest-attribute=cold): Document.
1934 * common.opt (Wsuggest-attribute=cold): New
1935 * ipa-pure-const.c (warn_function_cold): New function.
1936 * predict.c (compute_function_frequency): Use it.
1937 * predict.h (warn_function_cold): Declare.
1939 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
1941 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
1944 2017-10-06 Martin Liska <mliska@suse.cz>
1946 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
1947 keyword for member functions.
1948 (struct sanopt_tree_couple): New struct.
1949 (struct sanopt_tree_couple_hash): New function.
1950 (struct sanopt_ctx): Add new hash_map.
1951 (has_dominating_ubsan_ptr_check): New function.
1952 (record_ubsan_ptr_check_stmt): Likewise.
1953 (maybe_optimize_ubsan_ptr_ifn): Likewise.
1954 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
1955 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
1957 2017-10-06 Sudakshina Das <sudi.das@arm.com>
1960 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
1961 aarch64_simd_valid_immediate on CONST_VECTORs.
1962 (aarch64_reg_or_bic_imm): Likewise.
1964 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
1966 PR rtl-optimization/82396
1967 * haifa-sched.c (ready_sort_real): Disable qsort checking.
1969 2017-10-06 Sebastian Pop <sebpop@gmail.com>
1971 * graphite-dependences.c (scop_get_reads): Move code to...
1972 (scop_get_must_writes): Move code to...
1973 (scop_get_may_writes): Move code to...
1974 (scop_get_reads_and_writes): ... here.
1975 (scop_get_dependences): Call scop_get_reads_and_writes.
1977 2017-10-06 Jakub Jelinek <jakub@redhat.com>
1979 PR tree-optimization/82434
1980 * fold-const.h (can_native_encode_type_p,
1981 can_native_encode_string_p): Remove.
1982 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
1983 don't encode anything, just return what would be otherwise returned.
1984 (native_encode_fixed, native_encode_complex, native_encode_vector):
1986 (native_encode_string): Likewise. Inline by hand
1987 can_native_encode_string_p.
1988 (can_native_encode_type_p): Remove.
1989 (can_native_encode_string_p): Remove.
1990 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
1991 STRING_CSTs using can_native_encode_string_p, test all
1992 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
1993 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
1994 argument from native_encode_expr.
1995 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
1996 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
1999 2017-10-06 Richard Biener <rguenther@suse.de>
2001 PR tree-optimization/82397
2002 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2003 operand_equal_p but rely on data_ref_compare_tree for detecting
2005 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2006 to match up with dr_group_sort_cmp.
2008 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2011 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2013 * config/s390/s390-builtin-types.def: Regenerate.
2015 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2018 * config/s390/s390-builtin-types.def: Regenerate.
2019 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2020 Change flag from B_VXE to B_VX.
2021 (s390_vec_min_dbl): Remove B_VXE flag.
2023 2017-10-06 Richard Biener <rguenther@suse.de>
2025 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2026 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2027 translate_isl_ast_to_gimple::is_valid_rename,
2028 translate_isl_ast_to_gimple::get_rename,
2029 translate_isl_ast_to_gimple::get_def_bb_for_const,
2030 translate_isl_ast_to_gimple::get_new_name,
2031 translate_isl_ast_to_gimple::collect_all_ssa_names,
2032 translate_isl_ast_to_gimple::copy_loop_phi_args,
2033 translate_isl_ast_to_gimple::collect_all_ssa_names,
2034 translate_isl_ast_to_gimple::copy_loop_phi_args,
2035 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2036 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2037 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2038 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2039 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2040 translate_isl_ast_to_gimple::copy_cond_phi_args,
2041 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2042 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2043 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2044 translate_isl_ast_to_gimple::rename_uses,
2045 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2046 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2047 (set_rename_for_each_def): Likewise.
2048 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2049 here. Handle rewriting SCEV analyzable uses here.
2050 (copy_bb_and_scalar_dependences): Generate code for PHI
2052 (graphite_regenerate_ast_isl): Adjust.
2053 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2054 (add_write, add_read): New functions.
2055 (build_cross_bb_scalars_def): Use it and simplify.
2056 (build_cross_bb_scalars_use): Likewise.
2057 (graphite_find_cross_bb_scalar_vars): Inline into...
2058 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2059 simulating out-of-SSA. Compute liveout and add dependencies.
2060 (build_scops): Force an empty entry block.
2061 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2063 (sese_build_liveouts): Declare.
2064 (sese_trivially_empty_bb_p): Likewise.
2065 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2066 compute liveout and debug_liveout.
2067 (sese_bad_liveouts_use): Remove.
2068 (sese_reset_debug_liveouts_bb): Likewise.
2069 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2070 (sese_build_liveouts): Build liveout and debug_liveout and store
2072 (new_sese_info): Adjust.
2073 (free_sese_info): Likewise.
2074 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2075 do not build liveout here.
2076 (move_sese_in_condition): Adjust region entry.
2077 (scev_analyzable_p): Match up with chrec_apply requirements.
2078 (sese_trivially_empty_bb_p): New.
2079 * tree-into-ssa.c (get_reaching_def): Properly support generating
2080 default-defs for incremental rewrite of anonymous names.
2082 2017-10-06 Richard Biener <rguenther@suse.de>
2084 * graphite-sese-to-poly.c (extract_affine): For casts increasing
2085 precision do not perform modulo reduction.
2087 2017-10-06 Richard Biener <rguenther@suse.de>
2089 PR tree-optimization/82436
2090 * tree-vect-slp.c (vect_supported_load_permutation_p): More
2091 conservatively choose the vectorization factor when checking
2092 whether we can perform the required load permutation.
2093 (vect_transform_slp_perm_load): Assert when we may not fail.
2095 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2097 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
2098 message for incompatible -msdata=* and -mcall-* options.
2100 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2102 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
2103 rate for post-reload scheduling.
2105 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2107 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
2109 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2111 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
2112 to improve monte carlo in scimark.
2114 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2116 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
2117 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
2118 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
2119 pentium4_cost, nocona_cost): Set reassociation width to 1.
2120 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
2121 width to 2 for fp operations and 1 otherwise.
2122 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
2124 (atom_cost): Set reassociation width to 2.
2125 (slm_cost, generic_cost): Set fp reassociation width
2126 to 2 and 1 otherwise.
2127 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
2128 (core_cost): Set fp reassociation width to 4 and vector to 2.
2129 (ix86_reassociation_width): Rewrite using cost table; special case
2130 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
2131 and TARGET_AVX128_OPTIMAL.
2132 * config/i386/i386.h (processor_costs): Add
2133 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
2134 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
2135 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
2136 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
2137 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
2138 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
2140 2017-10-05 Nathan Sidwell <nathan@acm.org>
2142 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
2144 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2146 * config/arm/arm.c (arm_test_fpu_data): New.
2147 (arm_run_selftests): Call arm_test_fpu_data.
2149 2017-10-04 Nathan Sidwell <nathan@acm.org>
2151 * toplev.c (toplev::main): Remove excess parens on pretty_printer
2153 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
2155 2017-10-04 Sudakshina Das <sudi.das@arm.com>
2157 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
2158 check type for aarch64_simd_valid_immediate.
2159 (aarch64_output_simd_mov_immediate): Update prototype.
2160 (aarch64_simd_valid_immediate): Update prototype.
2161 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
2162 support for ORR-immediate.
2163 (and<mode>3): modified pattern to add support for BIC-immediate.
2164 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
2165 now checks for valid immediate for BIC and ORR based on new enum
2167 (aarch64_output_simd_mov_immediate): Function now used to output
2168 BIC/ORR imm as well based on new enum argument.
2169 * config/aarch64/constraints.md (Do): New vector immediate constraint.
2171 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
2172 (aarch64_reg_or_bic_imm): Likewise.
2174 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2176 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
2177 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
2179 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2183 PR rtl-optimization/82396
2184 * haifa-sched.c (autopref_multipass_init): Simplify
2186 (autopref_rank_data): Simplify sort order.
2187 * sched-int.h (autopref_multipass_data_): Remove
2188 multi_mem_insn_p, min_offset and max_offset.
2190 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2192 * doc/sourcebuild.texi: Document vect_peeling_profitable.
2194 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2196 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
2199 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2201 * doc/sourcebuild.texi: Document vect_long_mult.
2203 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
2205 PR tree-optimization/82413
2206 * fold-const.c (build_range_check): Use widest_int when comparing
2207 the maximum ETYPE value with HIGH.
2209 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2211 PR rtl-optimization/82396
2212 * haifa-sched.c (autopref_multipass_init): Simplify
2214 (autopref_rank_data): Simplify sort order.
2215 * sched-int.h (autopref_multipass_data_): Remove
2216 multi_mem_insn_p, min_offset and max_offset.
2218 2017-10-04 Jakub Jelinek <jakub@redhat.com>
2220 PR tree-optimization/82381
2221 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
2222 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
2225 PR tree-optimization/82374
2226 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
2227 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
2228 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
2229 current_function_decl to the new decl.
2231 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
2233 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
2234 helper macro for IEEE float128 hardware built-in functions.
2235 (SQRTF128_ODD): Add built-in functions with the round-to-odd
2237 (TRUNCF128_ODD): Likewise.
2238 (ADDF128_ODD): Likewise.
2239 (SUBF128_ODD): Likewise.
2240 (MULF128_ODD): Likewise.
2241 (DIVF128_ODD): Likewise.
2242 (FMAF128_ODD): Likewise.
2243 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
2244 UNSPEC_TRUNC_ROUND_TO_ODD.
2245 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
2246 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
2247 floating point round to odd instructions.
2248 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
2249 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
2250 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
2251 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
2252 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
2253 (trunc<mode>sf2_hw): Change the truncate with round to odd
2254 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
2255 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
2256 to odd hardware instructions.
2257 (sub<mode>3_odd): Likewise.
2258 (mul<mode>3_odd): Likewise.
2259 (div<mode>3_odd): Likewise.
2260 (sqrt<mode>2_odd): Likewise.
2261 (fma<mode>4_odd): Likewise.
2262 (fms<mode>4_odd): Likewise.
2263 (nfma<mode>4_odd): Likewise.
2264 (nfms<mode>4_odd): Likewise.
2265 (trunc<mode>df2_odd): Change the truncate with round to odd
2266 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
2268 * doc/extend.texi (PowerPC built-in functions): Update documentation
2269 for existing IEEE float128-bit built-in functions. Add built-in
2270 functions that generate the IEEE 128-bit floating point round to
2273 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
2275 PR rtl-optimization/77729
2276 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
2277 to (X&(C1&~C2))|C2 transformations.
2279 2017-10-03 Martin Jambor <mjambor@suse.cz>
2281 PR tree-optimization/82363
2282 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
2283 mismatch, mark lacc written regardless of racc.
2285 2017-10-03 Jakub Jelinek <jakub@redhat.com>
2287 PR tree-optimization/82381
2288 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
2289 stmt_to_insert nor wheather SSA_NAMEs are default defs.
2290 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
2291 fallthrough into reassoc_stmt_dominates_stmt_p.
2294 * combine.c (combine_instructions): Don't combine in unreachable
2297 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2300 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
2301 function to not use the have_cpu variable. Do not set cpu_index,
2302 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
2304 (rs6000_valid_attribute_p): Remove duplicate initializations of
2305 old_optimize and func_optimize.
2306 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
2307 (rs6000_activate_target_options): Make global.
2308 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
2311 2017-10-02 Jakub Jelinek <jakub@redhat.com>
2313 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
2314 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
2315 Set *poffset to 0 and *psize and *pmax_size to -1 if
2316 *poffset + *psize overflows in HOST_WIDE_INT.
2318 PR tree-optimization/82387
2319 PR tree-optimization/82388
2320 PR tree-optimization/82389
2321 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
2322 instead of live_bytes non-NULL.
2324 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
2327 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
2330 2017-10-02 Richard Biener <rguenther@suse.de>
2332 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
2333 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
2334 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
2336 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2338 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
2339 requested precision matches the type's.
2340 * calls.c (alloc_max_size): Calculate the new candidate size as
2341 a widest_int and use wi::to_widest when comparing it with the
2342 current candidate size.
2343 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
2344 zero rather than integer_zero_node.
2345 * match.pd: Check for a no-op conversion before using wi::add
2346 rather than after. Use tree_to_uhwi when summing small shift
2347 counts into an unsigned int.
2349 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2350 Alan Hayward <alan.hayward@arm.com>
2351 David Sherwood <david.sherwood@arm.com>
2354 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
2355 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
2356 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
2357 POINTER_AND_FP_REGS.
2359 2017-10-02 Richard Biener <rguenther@suse.de>
2361 PR tree-optimization/82355
2362 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2363 a mapping for the enclosing loop but avoid generating one for
2365 (copy_bb_and_scalar_dependences): Remove premature codegen
2366 error on PHIs in blocks duplicated into multiple places.
2367 * graphite-scop-detection.c
2368 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2369 in the region use it as loop and nest to analyze the DR in.
2370 (try_generate_gimple_bb): Likewise.
2371 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2372 (add_loop_constraints): For blocks in a loop not in the region
2373 create a dimension with a single iteration.
2374 * sese.h (gbb_loop_at_index): Remove assert.
2376 2017-10-01 Kevin Buettner <kevinb@redhat.com>
2378 * omp-expand.c (adjust_context_scope): New function.
2379 (expand_parallel_call): Call adjust_context_scope.
2381 2017-10-01 Jeff Law <law@redhat.com>
2383 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
2384 dom_opt_dom_walker class with direct access to private members.
2385 Add comments. Call test_for_singularity.
2386 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
2387 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
2388 m_dummy_cond anymore.
2389 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
2391 (pass_dominator:execute): Build the dummy_cond here and pass it
2392 to the dom_opt_dom_walker ctor.
2393 (test_for_singularity): New function.
2395 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
2396 Maya Rashish <coypu@sdf.org>
2398 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
2399 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
2400 (alpha*-*-netbsd*) Use nbsd_tm_file.
2401 (arm*-*-netbsdelf*) Likewise.
2402 (i[34567]86-*-netbsdelf*) Likewise.
2403 (x86_64-*-netbsd*) Likewise.
2404 (mips*-*-netbsd*) Likewise.
2405 (powerpc-*-netbsd*) Likewise.
2406 (sh*-*-netbsd*) Likewise.
2407 (sparc-*-netbsdelf*) Likewise.
2408 (sparc64-*-netbsd*) Likewise.
2409 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
2411 (vax-*-netbsdelf*) Likewise.
2412 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
2413 (UINT_FAST8_TYPE) Likewise.
2414 (INT_FAST16_TYPE) Check CHAR_FAST16.
2415 (UINT_FAST16_TYPE) Likewise.
2417 2017-09-30 Jakub Jelinek <jakub@redhat.com>
2420 * config/i386/i386.md
2421 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
2422 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
2423 *divmodsi4_zext_2): New define_insn_and_split.
2424 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
2425 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
2426 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
2427 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
2428 New define_insn_and_split.
2429 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
2430 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
2431 operands[1] having DImode when mode is SImode.
2433 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
2434 always SImode for DIV and MOD in REG_EQUAL notes.
2436 2017-09-29 Yury Gribov <tetra2005@gmail.com>
2439 * match.pd: Fix handling of NaNs in pattern.
2441 2017-09-29 Jeff Law <law@redhat.com>
2443 * sbitmap.c (bitmap_bit_in_range_p): New function.
2444 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
2445 * tree-ssa-dse.c (live_bytes_read): New function.
2446 (dse_classify_store): Ignore reads of dead bytes.
2448 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
2449 typos and whitespace errors.
2450 * config/i386/predicates.md (address_no_seg_operand): Likewise.
2451 * config/s390/s390.c (s390_emit_prologue): Likewise.
2453 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2456 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
2457 with a symbol for LRA.
2459 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2461 PR rtl-optimization/82338
2462 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
2464 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
2466 * genmodes.c (calc_wider_mode): Suppress qsort macro.
2467 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
2468 (qsort_chk): Declare.
2469 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
2470 (qsort_chk): New function.
2472 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2474 PR tree-optimization/82337
2475 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
2476 phi definition if the PHI result appears in an abnormal PHI.
2477 (find_basis_for_base_expr): Don't record a basis if the LHS of the
2478 basis appears in an abnormal PHI.
2480 2017-09-29 Richard Biener <rguenther@suse.de>
2482 * graphite-isl-ast-to-gimple.c
2483 (translate_isl_ast_to_gimple::set_codegen_error): New function.
2484 (binary_op_to_tree): Use it.
2485 (get_rename_from_scev): Likewise.
2486 (copy_loop_phi_nodes): Likewise.
2487 (copy_bb_and_scalar_dependences): Likewise.
2488 (translate_pending_phi_nodes): Likewise.
2490 2017-09-29 Jakub Jelinek <jakub@redhat.com>
2493 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
2494 for movabsq $(i32 << shift), r64.
2496 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
2498 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
2499 index when encoding %esp as %rsp to avoid 0x67 prefix.
2501 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
2503 * config/i386/i386.md (*movsf_internal, *movdf_internal):
2504 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
2506 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
2508 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
2509 Extensions with more than 16 double VFP registers.
2510 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
2511 to_clear_mask and all code related to it. Replace the remaining
2512 entry by a sbitmap and adapt code accordingly.
2514 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
2516 * brig-builtins.def: Change pure attributes to const.
2518 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2520 * config.gcc (default_gnu_indirect_function): Default to yes for
2521 sparc*-*-linux* with glibc.
2523 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2525 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
2526 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
2527 when creating .init_array and .fini_array sections with priority
2530 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
2533 * config/aarch64/aarch64.c
2534 (aarch64_builtin_support_vector_misalignment): Always return false
2535 when misalignment is unknown.
2537 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
2539 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
2540 this function to return false if the definition used by the swap
2541 instruction is artificial, or if the memory address from which the
2542 constant value is loaded is not represented by a base address held
2543 in a register or if the base address register is a frame or stack
2544 pointer. Additionally, return false if the base address of the
2545 loaded constant is a SYMBOL_REF but is not considered to be a
2547 (replace_swapped_load_constant): New function.
2548 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
2549 loaded constant vector with a load of a swapped constant vector.
2551 2017-09-27 Carl Love <cel@us.ibm.com>
2553 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
2554 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
2555 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
2558 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
2560 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
2561 first, always call autopref_rank_data otherwise.
2563 2017-09-27 Richard Biener <rguenther@suse.de>
2565 * graphite-scop-detection.c (find_scop_parameters): Move
2566 loop bound handling ...
2567 (gather_bbs::before_dom_children): ... here, avoiding the need
2568 to build scop_info->loop_nest.
2569 (record_loop_in_sese): Remove.
2570 * sese.h (sese_info_t::loop_nest): Remove.
2571 * sese.c (new_sese_info): Do not allocate loop_nest.
2572 (free_sese_info): Do not free loop_nest.
2574 2017-09-27 Jakub Jelinek <jakub@redhat.com>
2577 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
2578 lhs from calls if the lhs has addressable type.
2580 2017-09-27 Richard Biener <rguenther@suse.de>
2582 * graphite.h (scop::max_alias_set): New member.
2583 * graphite-scop-detection.c: Remove references to non-existing
2584 --param in comments.
2585 (build_alias_sets): Record the maximum alias set used for drs.
2586 (build_scops): Support zero as unlimited for
2587 --param graphite-max-arrays-per-scop.
2588 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
2590 (build_poly_sr_1): ... here. Compute alias set based on the
2591 maximum alias set used for drs rather than
2592 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
2594 2017-09-27 Richard Biener <rguenther@suse.de>
2596 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
2597 --param loop-block-tile-size=0 to disable tiling.
2599 2017-09-27 Richard Biener <rguenther@suse.de>
2601 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
2602 (graphite-max-nb-scop-params): Document special value zero.
2603 * domwalk.h (dom_walker::STOP): New symbolical constant.
2604 (dom_walker::dom_walker): Add optional parameter for bb to
2606 (dom_walker::~dom_walker): Declare.
2607 (dom_walker::before_dom_children): Document STOP return value.
2608 (dom_walker::m_user_bb_to_rpo): New member.
2609 (dom_walker::m_bb_to_rpo): Likewise.
2610 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
2611 mapping here if not provided by the user.
2612 (dom_walker::~dom_walker): Free bb to RPO mapping if not
2613 provided by the user.
2614 (dom_walker::STOP): Define.
2615 (dom_walker::walk): Do not compute bb to RPO mapping here.
2616 Support STOP return value from before_dom_children to stop
2618 * graphite-optimize-isl.c (optimize_isl): If the schedule
2619 is the same still generate code if -fgraphite-identity
2620 or -floop-parallelize-all are given.
2621 * graphite-scop-detection.c: Include cfganal.h.
2622 (gather_bbs::gather_bbs): Get and pass through bb to RPO
2624 (gather_bbs::before_dom_children): Return STOP for BBs
2626 (build_scops): Compute bb to RPO mapping and pass it to
2627 the domwalk. Treat --param graphite-max-nb-scop-params=0
2628 as not limiting the number of params.
2629 * graphite.c (graphite_initialize): Remove limit on the
2630 number of basic-blocks in a function.
2631 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
2632 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
2633 default value of 10.
2635 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2637 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
2638 Adjust code to eliminate needing to do the shift right 32-bits
2639 operation after XSCVDPSPN.
2641 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2643 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
2644 ((X / Y) != 0 -> X >= Y): Likewise.
2646 2017-09-26 Carl Love <cel@us.ibm.com>
2648 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
2649 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
2650 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
2651 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
2652 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
2653 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
2654 definitions and overloading.
2655 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
2656 statement for P9V_BUILTIN_XST_LEN_R.
2657 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
2658 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
2659 define_expand and define_insn for the instructions and builtins.
2660 * doc/extend.texi: Update the built-in documentation file for the new
2662 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
2663 define_insn for the instructions
2665 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
2668 * gcc/config/netbsd-protos.h: New file.
2669 * gcc/config/netbsd.c: New file.
2670 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
2671 * gcc/config/t-netbsd: New file.
2672 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
2673 (tmake_file) Add t-netbsd.
2674 (extra_objs) Add netbsd.o.
2676 2017-09-26 Janus Weil <janus@gcc.gnu.org>
2680 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
2682 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2684 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
2685 sign extension from a vector register to a GPR by doing a 32-bit
2686 direct move and then an EXTSW.
2687 (extendsi<mode>2 splitter): Likewise.
2688 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
2689 right or vector extract after doing XSCVDPSPN. Use
2690 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
2692 (movdi_from_sf_zero_ext): Likewise.
2693 (reload_gpr_from_vsxsf): Likewise.
2694 (p8_mfvsrd_4_disf): Delete, no longer used.
2695 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
2696 then needing to move the SFmode to a GPR to use the XSCVDPSP
2697 instruction instead of FRSP and XSCVDPSPN.
2698 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
2699 it is adjacent to the other XSCVSPDP insns.
2700 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
2701 SFmode to be in traditional Altivec registers.
2702 (vsx_xscvdpspn): Eliminate useless alternative constraint.
2703 (vsx_xscvspdpn): Likewise.
2704 (vsx_xscvspdpn_scalar): Likewise.
2706 2017-09-26 Martin Jambor <mjambor@suse.cz>
2708 * tree-sra.c (compare_access_positions): Put integral types first,
2709 stabilize sorting of integral types, remove conditions putting
2710 non-full-precision integers last.
2711 (sort_and_splice_var_accesses): Disable scalarization if a
2712 non-integert would be represented by a non-full-precision integer.
2714 2017-09-26 Joseph Myers <joseph@codesourcery.com>
2716 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
2717 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
2718 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
2719 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
2720 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
2721 conditionals inside the function instead of around it. Call
2722 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
2723 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
2725 2017-09-26 Richard Biener <rguenther@suse.de>
2727 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
2729 (scop_detection::build_scop_breadth): ... this. Removed.
2730 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
2731 (scop_detection::harmful_stmt_in_bb): Likewise.
2732 (scop_detection::graphite_can_represent_stmt): Likewise.
2733 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
2734 (scop_detection::can_represent_loop): Remove recursion, fold in ...
2735 (scop_detection::can_represent_loop_1): ... this. Removed.
2736 (scop_detection::harmful_loop_in_region): Simplify after inlining
2737 the above and remove more quadraticness.
2738 (build_scops): Adjust.
2739 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
2742 2017-09-26 Jakub Jelinek <jakub@redhat.com>
2745 * config/i386/i386.c (ix86_print_operand_address_as): Only test
2746 REGNO (base) == SP_REG if base is a REG.
2749 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
2750 if it is different SSA_NAME.
2751 (optimize_range_tests_cmp_bitwise): New function.
2752 (optimize_range_tests): Call it.
2754 2017-09-26 Richard Biener <rguenther@suse.de>
2756 PR tree-optimization/82321
2757 * graphite.c (canonicalize_loop_closed_ssa): Properly check
2758 for the def being inside the loop.
2760 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2762 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
2764 * config/s390/s390-builtins.def: Fix constraint on op4.
2766 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2768 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
2769 independent expanders.
2770 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
2771 ("vec_ordered", "vec_unordered"): New expanders.
2773 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2775 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
2778 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2780 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
2781 vec_unpacks_lo_v16qi.
2782 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
2784 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2786 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
2787 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
2788 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
2790 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2792 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
2794 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
2796 ("*vec_slb<mode>"): New insn pattern.
2797 ("vec_shr_<mode>"): New expander.
2798 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
2799 and force the shift count operand to V16QImode.
2800 ("vec_srb<mode>"): Set shift count mode to V16QI.
2802 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2804 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
2805 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
2806 ("vec_widen_smult_hi_<mode>"): New expander definitions.
2808 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
2811 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
2813 2017-09-26 Richard Biener <rguenther@suse.de>
2815 PR tree-optimization/82320
2816 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
2819 2017-09-25 Jeff Law <law@redhat.com>
2821 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
2822 prototype for new argument.
2823 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
2824 mostly extracted from rs6000_emit_allocate_stack.
2825 (rs6000_emit_probe_stack_range_stack_clash): New function.
2826 (rs6000_emit_allocate_stack): Call
2827 rs6000_emit_probe_stack_range_stack_clash as needed.
2828 (rs6000_emit_probe_stack_range): Add additional argument
2829 to call to gen_probe_stack_range{si,di}.
2830 (output_probe_stack_range): New.
2831 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
2832 (output_probe_stack_range_stack_clash): New.
2833 (rs6000_emit_prologue): Emit notes into dump file as requested.
2834 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
2835 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
2836 Add additional operand and pass it to output_probe_stack_range.
2838 2017-09-25 Bin Cheng <bin.cheng@arm.com>
2840 PR tree-optimization/82163
2841 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
2842 (checking_verify_loop_closed_ssa): New parameter.
2843 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
2844 (check_loop_closed_ssa_stmt): Delete.
2845 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
2846 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
2847 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
2850 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
2852 * brig-builtins.def: Treat HSAIL barrier builtins as
2853 setjmp/longjump style functions.
2855 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2857 * target.def (constant_alignment): New hook.
2858 * defaults.h (CONSTANT_ALIGNMENT): Delete.
2859 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
2860 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
2861 * doc/tm.texi: Regenerate.
2862 * targhooks.h (default_constant_alignment): Declare.
2863 (constant_alignment_word_strings): Likewise.
2864 * targhooks.c (default_constant_alignment): New function.
2865 (constant_alignment_word_strings): Likewise.
2866 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
2867 instead of CONSTANT_ALIGNMENT.
2868 * varasm.c (align_variable, get_variable_align, build_constant_desc)
2869 (force_const_mem): Likewise.
2870 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
2871 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
2872 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
2873 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2874 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
2876 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
2877 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2878 constant_alignment_word_strings.
2879 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
2880 (CONSTANT_ALIGNMENT): Likewise.
2881 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2882 (arm_constant_alignment): New function.
2883 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
2884 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2885 constant_alignment_word_strings.
2886 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
2887 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2888 constant_alignment_word_strings.
2889 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
2890 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2891 (cris_constant_alignment): New function.
2892 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
2893 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2894 (epiphany_constant_alignment): New function.
2895 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
2896 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2897 constant_alignment_word_strings.
2898 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
2899 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2900 constant_alignment_word_strings.
2901 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
2902 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2903 constant_alignment_word_strings.
2904 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
2905 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
2906 * config/i386/i386.c (ix86_constant_alignment): Make static.
2907 Use the same interface as the target hook.
2908 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2909 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
2910 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2911 constant_alignment_word_strings.
2912 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
2913 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
2914 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2915 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
2916 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2917 constant_alignment_word_strings.
2918 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
2919 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2920 constant_alignment_word_strings.
2921 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
2922 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2923 constant_alignment_word_strings.
2924 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
2925 * config/microblaze/microblaze.c (microblaze_constant_alignment):
2927 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2928 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
2929 * config/mips/mips.c (mips_constant_alignment): New function.
2930 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2931 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
2932 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
2933 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2934 (mmix_constant_alignment): Make static. Use the same interface
2936 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
2937 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2938 constant_alignment_word_strings.
2939 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
2940 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2941 constant_alignment_word_strings.
2942 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
2943 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2944 constant_alignment_word_strings.
2945 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
2946 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2947 (rs6000_constant_alignment): New function.
2948 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
2949 * config/riscv/riscv.c (riscv_constant_alignment): New function.
2950 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2951 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
2952 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2953 (rs6000_constant_alignment): New function.
2954 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
2955 * config/s390/s390.c (s390_constant_alignment): New function.
2956 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2957 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
2958 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2959 constant_alignment_word_strings.
2960 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
2961 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2962 (sparc_constant_alignment): New function.
2963 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
2964 * config/spu/spu.c (spu_constant_alignment): New function.
2965 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2966 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
2967 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2968 constant_alignment_word_strings.
2969 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
2970 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2971 constant_alignment_word_strings.
2972 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
2973 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2974 constant_alignment_word_strings.
2975 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
2976 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2977 (visium_constant_alignment): New function.
2978 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
2979 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2980 (xtensa_constant_alignment): New function.
2981 * system.h (CONSTANT_ALIGNMENT): Poison.
2983 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
2985 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2986 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
2987 (rs6000_builtin_valid_without_lhs): New helper function.
2988 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2989 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
2991 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2993 * target.h (vec_perm_indices): Use unsigned short rather than
2995 (auto_vec_perm_indices): Likewise.
2996 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
2997 Use unsigned int rather than unsigned char.
2998 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3000 2017-09-25 Richard Biener <rguenther@suse.de>
3002 * cfgloop.h (sort_sibling_loops): Declare.
3003 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3004 (sort_sibling_loops): New function sorting the sibling loop list
3006 * graphite.c (graphite_transform_loops): Sort sibling loops.
3008 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3010 * target.def (vec_perm_const_ok): Change sel parameter to
3012 * optabs-query.c (can_vec_perm_p): Update accordingly.
3013 * doc/tm.texi: Regenerate.
3014 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3015 auto_vec_perm_indices and remove separate nelt field.
3016 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3017 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3018 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3019 (aarch64_expand_vec_perm_const): Update accordingly.
3020 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3021 to vec_perm_indices.
3022 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3023 auto_vec_perm_indices and remove separate nelt field.
3024 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3025 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3026 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3028 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3029 to vec_perm_indices.
3030 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3031 sel to vec_perm_indices.
3032 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3033 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3034 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3036 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3039 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3042 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3043 on the FUNCTION_DECL function context if it has a DIE that is a
3046 2017-09-25 Richard Biener <rguenther@suse.de>
3048 PR tree-optimization/82285
3049 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3052 2017-09-25 Tom de Vries <tom@codesourcery.com>
3056 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3059 2017-09-25 Richard Biener <rguenther@suse.de>
3061 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3062 ISL errors other than isl_error_quota happen. Dump if the
3063 schedule is the same.
3064 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3065 errors instead of aborting inside ISL.
3067 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3070 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3071 of libgcc_eh for m64.
3072 * config/i386/darwin64.h: Likewise.
3074 2017-09-25 Richard Biener <rguenther@suse.de>
3077 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
3078 attribute for incomplete types nor twice for complete ones.
3080 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
3083 * config/i386/i386.c (ix86_print_operand_address_as): Encode
3084 %esp as %rsp to avoid 0x67 prefix if there is no index or base
3087 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
3090 * config/i386/i386.opt (mprefer-avx256): Use
3091 ix86_target_flags variable.
3092 * config/i386/i386.c (ix86_target_string): Move
3093 -mprefer-avx256 to flag2_opts.
3095 2017-09-22 Jakub Jelinek <jakub@redhat.com>
3098 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
3099 and x != -1 | y != -1 into (x & y) != -1.
3101 2017-09-22 Steve Ellcey <sellcey@cavium.com>
3103 * config.gcc: Add new case statement to set
3104 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
3105 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
3106 s390x-*-linux* case statements. Added aarch64 to the list of
3107 supported architectures.
3109 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3111 PR tree-optimization/82289
3112 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
3113 STMT_VINFO_RELEVANT_P.
3115 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3116 Alan Hayward <alan.hayward@arm.com>
3117 David Sherwood <david.sherwood@arm.com>
3119 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
3120 for VR_RANGE only; don't allow VR_ANTI_RANGE.
3121 (extract_range_from_binary_expr_1): Don't call
3122 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
3124 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3125 Alan Hayward <alan.hayward@arm.com>
3126 David Sherwood <david.sherwood@arm.com>
3128 * target.def (preferred_vector_alignment): New hook.
3129 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
3131 * doc/tm.texi: Regenerate.
3132 * targhooks.h (default_preferred_vector_alignment): Declare.
3133 * targhooks.c (default_preferred_vector_alignment): New function.
3134 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
3136 (DR_TARGET_ALIGNMENT): New macro.
3137 (aligned_access_p): Update commentary.
3138 (vect_known_alignment_in_bytes): New function.
3139 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
3141 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
3142 Calculate the misalignment based on the target alignment rather than
3144 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
3145 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
3146 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
3147 the target alignment, rather than masking the element misalignment
3148 with the number of elements in a vector. Also use the target
3149 alignment when calculating the maximum number of peels.
3150 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
3151 instead of TYPE_ALIGN_UNIT.
3152 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
3153 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
3154 (vect_create_addr_base_for_vector_ref): Update call accordingly.
3155 (vect_create_data_ref_ptr): Likewise.
3156 (vect_setup_realignment): Realign by ANDing with
3157 -DR_TARGET_MISALIGNMENT.
3158 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
3159 the number of peels based on DR_TARGET_ALIGNMENT.
3160 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
3161 with the guaranteed alignment boundary when deciding whether
3163 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
3164 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
3165 (ensure_base_align): Remove stmt_info parameter. Get the
3166 target base alignment from DR_TARGET_ALIGNMENT.
3167 (vectorizable_store): Update call accordingly. Interpret
3168 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
3170 (vectorizable_load): Likewise.
3172 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3173 Alan Hayward <alan.hayward@arm.com>
3174 David Sherwood <david.sherwood@arm.com>
3176 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
3177 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
3178 (vect_enhance_data_refs_alignment): Likewise.
3180 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3182 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3183 error. Only quit immediately if parsing is complete.
3184 (BEGIN): Initialize fatal_err and parse_done.
3185 (begin fpu, end fpu): Check number of arguments.
3186 (begin arch, end arch): Likewise.
3187 (begin cpu, end cpu): Likewise.
3188 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3189 (optalias): Likewise.
3191 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3193 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3194 * config/arm/arm-isa.h: Delete. Move definitions to ...
3195 * arm-cpus.in: ... here. Use new feature and fgroup values.
3196 * config/arm/arm.c (arm_option_override): Use lower case for feature
3198 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3199 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3200 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3202 (print_isa_bits_for): New function.
3203 (gen_isa): New function.
3204 (gen_comm_data): Use print_isa_bits_for.
3205 (define feature): New keyword.
3206 (define fgroup): New keyword.
3207 * config/arm/t-arm (TM_H): Remove.
3208 (GTM_H): Add arm-isa.h.
3209 (arm-isa.h): Add rule to generate file.
3210 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3211 case for feature bit names.
3213 2017-09-22 Richard Biener <rguenther@suse.de>
3215 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
3217 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
3218 print of no dependency loops ...
3219 * graphite.c (graphite_transform_loops): ... here.
3220 (canonicalize_loop_closed_ssa_form): Work from inner to outer
3222 (same_close_phi_node, remove_duplicate_close_phi,
3223 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
3224 (canonicalize_loop_closed_ssa): ... here and simplify.
3225 * graphite-optimize-isl.c: Include tree-vectorizer.h.
3226 (optimize_isl): Use dump_printf_loc to tell when we stopped
3227 optimizing because of an ISL timeout.
3229 2017-09-22 Richard Biener <rguenther@suse.de>
3231 PR tree-optimization/82291
3232 * tree-if-conv.c (predicate_mem_writes): Make sure to
3233 remove writes in blocks predicated with false.
3235 2017-09-22 Richard Biener <rguenther@suse.de>
3237 * sese.c: Include cfganal.h.
3238 (if_region_set_false_region): Remove.
3239 (create_if_region_on_edge): Likewise.
3240 (move_sese_in_condition): Re-implement without destroying
3243 2017-09-22 Richard Biener <rguenther@suse.de>
3245 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
3246 Verify both BBs contain loop PHI nodes before dispatching to
3248 (graphite_regenerate_ast_isl): Do not recompute dominators,
3249 do not verify three times. Restructure for clarity.
3250 * graphite-scop-detection.c (same_close_phi_node,
3251 remove_duplicate_close_phi, make_close_phi_nodes_unique,
3252 defined_in_loop_p, canonicalize_loop_closed_ssa,
3253 canonicalize_loop_closed_ssa_form): Simplify, remove excess
3254 checking and SSA rewrite, move to ...
3255 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
3256 (graphite_initialize): Do not pass in ctx, do not reset the
3257 SCEV cache, compute only dominators.
3258 (graphite_transform_loops): Allocate ISL ctx after
3259 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
3260 Maintain post-dominators only around build_scops.
3261 * sese.c (if_region_set_false_region): Make static. Free
3262 and recompute dominators.
3263 (move_sese_in_condition): Assert we don't get called with
3264 post-dominators computed.
3265 * sese.h (if_region_set_false_region): Remove.
3267 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
3269 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
3270 mode attribute for TARGET_AVX512VL.
3272 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
3274 * config/i386/i386.opt (mprefer-avx256): New option.
3275 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
3277 (ix86_preferred_simd_mode): Return 256-bit AVX modes
3278 for TARGET_PREFER_AVX256.
3279 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
3281 2017-09-21 Jeff Law <law@redhat.com>
3283 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
3284 Fix dump output if the only stack space is for pushed registers.
3286 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3288 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
3291 2017-09-21 Martin Sebor <msebor@redhat.com>
3294 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
3295 code (in C++) or code that triggers warnings.
3297 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
3299 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
3301 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
3303 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
3304 * sched-rgn.c: Ditto.
3305 * sel-sched-ir.c: Ditto.
3307 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
3309 * toplev.h (set_random_seed): Adjust return type.
3310 * toplev.c (init_local_tick): Move eager initialization of random_seed
3311 to get_random_seed. Adjust comment.
3312 (init_random_seed): Inline to get_random_seed, delete.
3313 (get_random_seed): Initialize random_seed lazily.
3314 (set_random_seed): Do not return previous value.
3315 (print_switch_value): Do not call get_random_seed.
3317 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
3319 * cgraph.c (delete_function_version): New, broken out from...
3320 (cgraph_node::delete_function_version): ...here. Rename to
3321 cgraph_node::delete_function_version_by_decl. Update all uses.
3322 (cgraph_node::remove): Call delete_function_version.
3324 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3327 * tree-inline.c (expand_call_inline): Emit clobber stmts for
3328 VAR_DECLs to which addressable non-volatile parameters are mapped
3329 and for id->retvar after the return value assignment. Clear
3330 id->retval and id->retbnd after inlining.
3332 2017-09-21 Richard Biener <rguenther@suse.de>
3334 PR tree-optimization/82276
3335 PR tree-optimization/82244
3336 * tree-vrp.c (build_assert_expr_for): Set
3337 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
3339 (remove_range_assertions): Revert earlier change.
3341 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
3344 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
3346 2017-09-21 Richard Biener <rguenther@suse.de>
3348 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3349 Restore valid IL after code generation errors.
3350 * graphite.c (graphite_transform_loops): Diagnose code
3351 generation issues as MSG_MISSED_OPTIMIZATION and continue
3352 with processing SCOPs.
3354 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3355 Alan Hayward <alan.hayward@arm.com>
3356 David Sherwood <david.sherwood@arm.com>
3358 * calls.c (compute_argument_addresses): Use simplify_gen_binary
3359 rather than choosing between plus_constant and gen_rtx_<CODE>.
3360 * expr.c (emit_push_insn): Likewise.
3361 (expand_expr_real_2): Likewise.
3363 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3364 Alan Hayward <alan.hayward@arm.com>
3365 David Sherwood <david.sherwood@arm.com>
3367 * loop-unroll.c (split_iv): Call copy_rtx on the step.
3369 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3370 Alan Hayward <alan.hayward@arm.com>
3371 David Sherwood <david.sherwood@arm.com>
3373 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
3374 calling tree_to_uhwi.
3376 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3377 Alan Hayward <alan.hayward@arm.com>
3378 David Sherwood <david.sherwood@arm.com>
3380 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
3381 INTEGER_CST rather than a negative test for ADDR_EXPR.
3383 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3384 Alan Hayward <alan.hayward@arm.com>
3385 David Sherwood <david.sherwood@arm.com>
3387 * tree-vrp.c (extract_range_from_binary_expr_1): Check
3388 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
3390 2017-09-21 Richard Biener <rguenther@suse.de>
3392 PR tree-optimization/71351
3393 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
3394 graphite_create_new_loop_guard): Remove, fold remaining parts
3396 (translate_isl_ast_node_for): ... here and simplify.
3398 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3401 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
3402 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
3403 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
3404 alternative always use QI mode, for -Os imov (=R,R) alternative
3405 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
3408 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3409 Jeff Law <law@redhat.com>
3411 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
3412 (allocate_stack_space): New function, partially extracted from
3414 (s390_emit_prologue): Track offset to most recent stack probe.
3415 Code to allocate space moved into allocate_stack_space.
3416 Dump actions when no stack is allocated.
3417 (s390_prologue_plus_offset): New function.
3418 (s390_emit_stack_probe): Likewise.
3420 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
3422 * common.opt (Wa, Wl, Wp, g, gz=): Add
3424 (gno-column-info): Remove.
3425 (gcolumn-info): Drop RejectNegative.
3427 (gno-record-gcc-switches): Remove.
3428 (grecord-gcc-switches): Drop RejectNegative.
3429 (gno-split-dwarf): Remove.
3430 (gsplit-dwarf): Drop RejectNegative.
3431 (gno-strict-dwarf): Remove.
3432 (gstrict-dwarf): Drop RejectNegative.
3433 * config/darwin.opt (gfull, gused): Add RejectNegative.
3434 * dwarf2out.c (gen_producer_string): Drop
3435 gno-record-gcc-switches handler.
3436 * optc-gen.awk: Add g to prefixes with negative forms.
3437 * opts-common.c (remapping_prefix_p): New.
3438 (find_opt): Check it.
3439 (generate_canonical_option): Test g prefix.
3440 (option_map): Add -gno- mapping.
3441 (add_misspelling_candidates): Check remapping_prefix_p.
3443 2017-09-20 Jeff Law <law@redhat.com>
3445 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
3446 thinko in stack clash protection support.
3448 * explow.c (compute_stack_clash_protection_loop_data): Use
3449 CONST_INT_P instead of explicit test. Verify object is a
3450 CONST_INT_P before looking at INTVAL.
3451 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
3452 instead of explicit test.
3454 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3457 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
3458 address instead of to r1 and r11.
3460 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
3462 * config.gcc: Support "knm".
3463 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
3464 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3466 * config/i386/i386.c (m_KNM): Define.
3467 (processor_target_table): Add "knm".
3469 (ix86_option_override_internal): Add "knm".
3470 (ix86_issue_rate): Add PROCESSOR_KNM.
3471 (ix86_adjust_cost): Ditto.
3472 (ia32_multipass_dfa_lookahead): Ditto.
3473 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
3474 (fold_builtin_cpu): Add M_INTEL_KNM.
3475 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
3476 (processor_type): Add PROCESSOR_KNM.
3477 * config/i386/x86-tune.def: Add m_KNM.
3478 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
3480 2017-09-20 Richard Biener <rguenther@suse.de>
3482 PR tree-optimization/80213
3483 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
3484 are allowed in empty BBs as well.
3485 (canonicalize_loop_closed_ssa): Also look for other complex
3487 (scop_detection::get_sese): Include the loop-closed PHI block
3489 (scop_detection::merge_sese): Remove code adding extra blocks.
3490 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
3491 (build_scops): Assert the final returned scop is invalid.
3493 2017-09-20 Richard Biener <rguenther@suse.de>
3495 PR tree-optimization/82264
3496 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
3498 (vn_phi_lookup): Likewise.
3499 (vn_phi_insert): Likewise.
3501 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3503 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
3504 that fits into uhwi or shwi, add DW_AT_const_value regardless
3505 of early_dwarf without going through RTL, using add_AT_unsigned
3508 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
3509 (DEBUG_LTO_ABBREV_SECTION): Likewise.
3510 (DEBUG_LTO_MACINFO_SECTION): Likewise.
3511 (DEBUG_MACRO_SECTION): Likewise.
3512 (DEBUG_LTO_MACRO_SECTION): Likewise.
3513 (DEBUG_STR_DWO_SECTION): Likewise.
3514 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
3515 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
3516 (DEBUG_LTO_DWO_LINE_SECTION): Define.
3517 (DEBUG_LTO_LINE_STR_SECTION): Define.
3518 (init_sections_and_labels): Initialize debug_line_str_section
3519 variable. Initialize debug_loc_section for -gdwarf-5 to
3520 DEBUG_LOCLISTS_SECTION. Formatting fixes.
3522 2017-09-20 Richard Biener <rguenther@suse.de>
3524 * graphite-sese-to-poly.c (extract_affine): Properly handle
3525 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
3527 2017-09-20 Richard Biener <rguenther@suse.de>
3529 PR tree-optimization/81373
3530 * graphite-scop-detection.c (build_cross_bb_scalars_def):
3531 Force SESE live-out defs to be handled even if they are
3534 2017-09-19 Jeff Law <law@redhat.com>
3536 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
3537 nothing for stack adjustments with REG_STACK_CHECK.
3538 * sched-deps.c (parse_add_or_inc): Reject insns with
3539 REG_STACK_CHECK from dependency breaking.
3540 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
3541 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
3542 * reg-notes.def (STACK_CHECK): New note.
3544 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
3545 (ix86_expand_prologue): Dump stack clash info as needed.
3546 Call ix86_adjust_stack_and_probe_stack_clash as needed.
3548 * function.c (dump_stack_clash_frame_info): New function.
3549 * function.h (dump_stack_clash_frame_info): Prototype.
3550 (enum stack_clash_probes): New enum.
3552 * config/alpha/alpha.c (alpha_expand_prologue): Also check
3553 flag_stack_clash_protection.
3554 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
3555 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
3556 (arm_frame_pointer_required): Likewise.
3557 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
3558 (ia64_expand_prologue): Likewise.
3559 * config/mips/mips.c (mips_expand_prologue): Likewise.
3560 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
3561 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3562 (sparc_flat_expand_prologue): Likewise.
3563 * config/spu/spu.c (spu_expand_prologue): Likewise.
3565 * explow.c: Include "params.h".
3566 (anti_adjust_stack_and_probe_stack_clash): New function.
3567 (get_stack_check_protect): Likewise.
3568 (compute_stack_clash_protection_loop_data): Likewise.
3569 (emit_stack_clash_protection_loop_start): Likewise.
3570 (emit_stack_clash_protection_loop_end): Likewise.
3571 (allocate_dynamic_stack_space): Use get_stack_check_protect.
3572 Use anti_adjust_stack_and_probe_stack_clash.
3573 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
3574 (emit_stack_clash_protection_loop_start): Likewise.
3575 (emit_stack_clash_protection_loop_end): Likewise.
3576 * rtl.h (get_stack_check_protect): Prototype.
3577 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
3578 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
3579 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
3581 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
3583 * doc/tm.texi: Rebuilt.
3584 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3585 get_stack_check_protect.
3586 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
3587 * config/arm/arm.c (arm_expand_prologue): Likewise.
3588 (arm_frame_pointer_required): Likewise.
3589 * config/i386/i386.c (ix86_expand_prologue): Likewise.
3590 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
3591 * config/mips/mips.c (mips_expand_prologue): Likewise.
3592 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
3593 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
3594 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3595 (sparc_flat_expand_prologue): Likewise.
3597 * common.opt (-fstack-clash-protection): New option.
3598 * flag-types.h (enum stack_check_type): Note difference between
3599 -fstack-check= and -fstack-clash-protection.
3600 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
3601 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
3602 * toplev.c (process_options): Issue warnings/errors for cases
3603 not handled with -fstack-clash-protection.
3604 * doc/invoke.texi (-fstack-clash-protection): Document new option.
3605 (-fstack-check): Note additional problem with -fstack-check=generic.
3606 Note that -fstack-check is primarily for Ada and refer users
3607 to -fstack-clash-protection for stack-clash-protection.
3608 Document new params for stack clash protection.
3610 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
3612 * config/i386/i386.c (ix86_split_long_move): Do not handle
3613 address used for LEA in a special way.
3615 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3617 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
3619 2017-09-19 Martin Sebor <msebor@redhat.com>
3622 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
3623 of incompatible types.
3625 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
3627 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3628 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
3629 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3630 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
3632 2017-09-19 Richard Biener <rguenther@suse.de>
3634 PR tree-optimization/82244
3635 * tree-vrp.c (remove_range_assertions): Do not propagate
3636 a constant to abnormals but replace the assert with a copy.
3638 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
3640 PR rtl-optimization/57878
3641 PR rtl-optimization/68988
3642 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
3643 avoidance test involving non_reload_pseudos. Move frequency test
3644 below the general fragmentation avoidance test.
3646 2017-09-19 Richard Biener <rguenther@suse.de>
3648 PR tree-optimization/69728
3649 * graphite-sese-to-poly.c (schedule_error): New global.
3650 (add_loop_schedule): Handle empty domain by failing the
3652 (build_original_schedule): Handle schedule_error.
3654 2017-09-19 Richard Biener <rguenther@suse.de>
3656 * graphite-scop-detection.c (scop_detection::can_represent_loop):
3657 Do not iterate to sibling loops but only to siblings of inner
3660 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
3663 * config/m68k/m68k.md (moveq feeding equality comparison): Check
3664 that the registers are different.
3666 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
3668 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
3669 to processor_model and "amdfam17h" to arch_names_table.
3670 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
3672 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3675 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
3677 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3678 Alan Hayward <alan.hayward@arm.com>
3679 David Sherwood <david.sherwood@arm.com>
3681 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
3683 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
3685 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
3686 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
3687 vect_schedule_slp_instance.
3688 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
3689 Update call to vect_slp_analyze_node_operations. Simplify return
3691 (vect_slp_analyze_bb_1): Update call accordingly.
3692 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
3693 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
3694 (vect_schedule_slp): Update call accordingly.
3696 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3697 Alan Hayward <alan.hayward@arm.com>
3698 David Sherwood <david.sherwood@arm.com>
3700 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
3701 with types that aren't in fact scalar.
3703 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3705 * tree-vect-slp.c (vect_record_max_nunits): New function,
3707 (vect_build_slp_tree_1): ...here.
3708 (vect_build_slp_tree_2): Call it for phis too.
3710 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3712 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
3713 to vect_get_vec_def_for_operand when getting the mask operand.
3715 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3716 Alan Hayward <alan.hayward@arm.com>
3717 David Sherwood <david.sherwood@arm.com>
3719 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
3722 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3723 Alan Hayward <alan.hayward@arm.com>
3724 David Sherwood <david.sherwood@arm.com>
3726 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
3727 calculation for vector booleans.
3729 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3730 Alan Hayward <alan.hayward@arm.com>
3731 David Sherwood <david.sherwood@arm.com>
3733 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
3735 (vect_transform_stmt): ...here.
3736 (vect_analyze_stmt): Use it instead of calling
3737 vectorizable_live_operation directly.
3739 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
3741 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
3742 non-SIMT targets in acc vector loops.
3744 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
3746 * configure.ac: Add arc and check if assembler supports gdwarf2.
3747 * configure: Regenerate.
3749 2017-09-18 Richard Biener <rguenther@suse.de>
3751 PR tree-optimization/82220
3752 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
3753 epilogue niters from the min_profitable_iters compute.
3755 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3758 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
3759 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
3760 (ix86_init_pic_reg): Revert 2017-09-01 changes.
3762 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
3765 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
3766 switching to a new text section.
3768 2017-09-18 Richard Biener <rguenther@suse.de>
3770 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
3772 (build_alias_set): Reject aliases with no access function.
3774 2017-09-18 Richard Biener <rguenther@suse.de>
3776 PR tree-optimization/79622
3777 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
3779 (build_cross_bb_scalars_use): Likewise.
3781 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
3783 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
3785 2017-09-18 Alan Modra <amodra@gmail.com>
3788 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
3789 stack_pointer_rtx for count 0. Update comments. Break up
3790 large rtl expression.
3792 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
3794 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
3795 Increase to 20 bytes.
3796 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
3797 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
3798 or avx version of the stub.
3800 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
3803 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
3804 compute the minimum stack alignment. Also update preferred stack
3805 boundary for leaf functions.
3807 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
3809 PR tree-optimization/82228
3810 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
3813 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3815 * common/config/nds32/nds32-common.c
3816 (nds32_option_optimization_table): Refine formatting.
3817 (nds32_option_optimization_table): Use -fsched-pressure and
3818 -fomit-frame-pointer for specific optimization level.
3820 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3822 * config/nds32/nds32.c: Refine formatting and comments.
3823 * config/nds32/nds32.h: Likewise.
3824 * config/nds32/nds32.md: Likewise.
3825 * config/nds32/nds32-cost.c: Likewise.
3826 * config/nds32/nds32-isr.c: Likewise.
3827 * config/nds32/nds32-md-auxiliary.c: Likewise.
3828 * config/nds32/nds32-multiple.md: Likewise.
3829 * config/nds32/nds32-predicates.c: Likewise.
3831 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
3832 Jakub Jelinek <jakub@redhat.com>
3834 Add support for -std=c++2a.
3835 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
3837 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
3839 2017-09-15 Steve Ellcey <sellcey@cavium.com>
3842 * doc/extend.texi (Common Function Attributes): Add
3843 references to ARM, AArch64, and S/390 specific attributes.
3844 (Function Specific Option Pragmas): Add AArch64 and S/390
3845 to list of back ends that support the target pragma.
3847 2017-09-15 Nathan Sidwell <nathan@acm.org>
3849 * doc/standards.texi: Fix C++17 description. Update URLs for
3852 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3854 * common.opt (Wcast-align=strict): New warning option.
3855 * doc/invoke.texi: Document -Wcast-align=strict.
3857 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
3859 * cgraph.h (cgraph_thunk_info): Add comments.
3860 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
3861 assert for VIRTUAL_* arguments stricter.
3863 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
3865 PR tree-optimization/71026
3866 * match.pd: Move RDIV patterns from fold-const.c
3867 * fold-const.c (distribute_real_division): Removed.
3868 (fold_binary_loc): Remove calls to distribute_real_divison.
3870 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3872 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
3873 c++1z and gnu++1z as deprecated. Change other references to
3874 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
3875 Change -Wc++1z-compat to -Wc++17-compat.
3876 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
3877 * dwarf2out.c (highest_c_language): Handle C++17.
3878 (gen_compile_unit_die): Likewise.
3880 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3882 PR rtl-optimization/82192
3883 * combine.c (make_extraction): Don't look through non-paradoxical
3884 SUBREGs or TRUNCATE if pos + len is or might be bigger than
3887 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3888 Alan Hayward <alan.hayward@arm.com>
3889 David Sherwood <david.sherwood@arm.com>
3891 * target.def (function_arg_offset): New hook.
3892 * targhooks.h (default_function_arg_offset): Declare.
3893 * targhooks.c (default_function_arg_offset): New function.
3894 * function.c (locate_and_pad_parm): Use
3895 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
3896 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
3897 (TARGET_FUNCTION_ARG_OFFSET): ...this.
3898 * doc/tm.texi: Regenerate.
3899 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
3900 * config/spu/spu.c (spu_function_arg_offset): New function.
3901 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
3902 * system.h (FUNCTION_ARG_OFFSET): Poison.
3904 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3905 Alan Hayard <alan.hayward@arm.com>
3906 David Sherwood <david.sherwood@arm.com>
3908 * target.def (truly_noop_truncation): New hook.
3909 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
3910 than TRULY_NOOP_TRUNCATION.
3911 * hooks.h (hook_bool_uint_uint_true): Declare.
3912 * hooks.c (hook_bool_uint_uint_true): New function.
3913 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
3914 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
3915 * doc/tm.texi: Regenerate.
3916 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
3917 rather than TRULY_NOOP_TRUNCATION in comments.
3918 (simplify_comparison): Likewise.
3919 (record_truncated_value): Likewise.
3920 * expmed.c (extract_bit_field_1): Likewise.
3921 (extract_split_bit_field): Likewise.
3922 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
3923 instead of TRULY_NOOP_TRUNCATION.
3924 * function.c (assign_parm_setup_block): Likewise.
3925 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
3926 * rtlhooks.c: Include target.h.
3927 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
3928 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
3929 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
3930 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
3931 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
3932 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
3933 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
3934 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
3935 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
3936 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
3937 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
3938 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
3939 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
3940 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
3941 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
3942 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
3943 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
3944 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
3945 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
3946 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
3947 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
3948 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
3949 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
3950 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
3951 * config/mips/mips.c (mips_truly_noop_truncation): New function.
3952 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3953 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3954 rather than TRULY_NOOP_TRUNCATION in comments.
3955 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
3956 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
3957 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
3958 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
3959 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
3960 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
3961 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
3962 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
3963 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
3964 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
3965 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
3966 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3967 rather than TRULY_NOOP_TRUNCATION in comments.
3968 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
3969 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
3970 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
3971 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
3972 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
3973 TRULY_NOOP_TRUNCATION condition.
3974 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
3975 (TRULY_NOOP_TRUNCATION): Delete.
3976 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
3977 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
3978 * config/spu/spu.c (spu_truly_noop_truncation): New function.
3979 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3980 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
3981 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
3982 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
3983 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3984 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3985 rather than TRULY_NOOP_TRUNCATION in comments.
3986 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
3987 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
3988 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
3989 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
3990 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
3991 * system.h (TRULY_NOOP_TRUNCATION): Poison.
3993 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
3996 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
3997 (*cmp_ior): Likewise.
3998 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
3999 (*ior_scc_scc_cmp): Likewise.
4000 (*and_scc_scc): Likewise.
4001 (*and_scc_scc_cmp): Likewise.
4003 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4004 Alan Hayard <alan.hayward@arm.com>
4005 David Sherwood <david.sherwood@arm.com>
4007 * target.def (can_change_mode_class): New hook.
4008 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4009 (hard_regno_nregs): Likewise.
4010 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4011 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4012 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4013 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4014 (LOAD_EXTEND_OP): Update accordingly.
4015 * doc/tm.texi: Regenerate.
4016 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4017 CANNOT_CHANGE_MODE_CLASS.
4018 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4019 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4020 * combine.c (simplify_set): Update accordingly.
4021 * emit-rtl.c (validate_subreg): Likewise.
4022 * recog.c (general_operand): Likewise.
4023 * regcprop.c (mode_change_ok): Likewise.
4024 * reload1.c (choose_reload_regs): Likewise.
4025 (inherit_piecemeal_p): Likewise.
4026 * rtlanal.c (simplify_subreg_regno): Likewise.
4027 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4028 instead of CANNOT_CHANGE_MODE_CLASS.
4029 (reload_cse_simplify_operands): Likewise.
4030 * reload.c (push_reload): Use targetm.can_change_mode_class
4031 instead of CANNOT_CHANGE_MODE_CLASS.
4032 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4033 REG_CANNOT_CHANGE_MODE_P.
4034 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4035 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4036 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4037 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4038 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4039 (arm_can_change_mode_class): New function.
4040 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4041 than CANNOT_CHANGE_MODE_CLASS in comments.
4042 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4043 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4044 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4045 (ix86_can_change_mode_class): ...this new function, inverting the
4046 sense of the return value.
4047 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4048 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4049 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4050 (ia64_can_change_mode_class): New function.
4051 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4052 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4053 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4054 (m32c_can_change_mode_class): ...this new function, inverting the
4055 sense of the return value.
4056 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4057 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4058 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4059 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4060 (mips_can_change_mode_class): ...this new function, inverting the
4061 sense of the return value.
4062 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4063 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4064 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4065 (msp430_can_change_mode_class): New function.
4066 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4067 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4068 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4069 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4070 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4071 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4072 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4073 (pa_cannot_change_mode_class): Replace with...
4074 (pa_can_change_mode_class): ...this new function, inverting the
4075 sense of the return value.
4076 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4077 than CANNOT_CHANGE_MODE_CLASS in comments.
4078 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4079 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
4080 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4081 (pdp11_cannot_change_mode_class): Replace with...
4082 (pdp11_can_change_mode_class): ...this new function, inverting the
4083 sense of the return value.
4084 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4085 * config/powerpcspe/powerpcspe-protos.h
4086 (rs6000_cannot_change_mode_class_ptr): Delete.
4087 * config/powerpcspe/powerpcspe.c
4088 (rs6000_cannot_change_mode_class_ptr): Delete.
4089 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4090 (rs6000_option_override_internal): Assign to
4091 targetm.can_change_mode_class instead of
4092 rs6000_cannot_change_mode_class_ptr.
4093 (rs6000_cannot_change_mode_class): Replace with...
4094 (rs6000_can_change_mode_class): ...this new function, inverting the
4095 sense of the return value.
4096 (rs6000_debug_cannot_change_mode_class): Replace with...
4097 (rs6000_debug_can_change_mode_class): ...this new function.
4098 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4099 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
4100 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4101 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4102 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
4104 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
4105 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4106 (rs6000_option_override_internal): Assign to
4107 targetm.can_change_mode_class instead of
4108 rs6000_cannot_change_mode_class_ptr.
4109 (rs6000_cannot_change_mode_class): Replace with...
4110 (rs6000_can_change_mode_class): ...this new function, inverting the
4111 sense of the return value.
4112 (rs6000_debug_cannot_change_mode_class): Replace with...
4113 (rs6000_debug_can_change_mode_class): ...this new function.
4114 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4115 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
4116 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
4117 (s390_can_change_mode_class): ...this new function, inverting the
4118 sense of the return value.
4119 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4120 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4121 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
4122 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4123 (sh_cannot_change_mode_class): Replace with...
4124 (sh_can_change_mode_class): ...this new function, inverting the
4125 sense of the return value.
4126 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4127 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4128 (sparc_can_change_mode_class): New function.
4129 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4130 * config/spu/spu.c (spu_can_change_mode_class): New function.
4131 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4132 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4133 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4134 (visium_can_change_mode_class): New function.
4135 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
4137 2017-09-15 Richard Biener <rguenther@suse.de>
4139 PR tree-optimization/82217
4140 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
4141 but not undefined case.
4143 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4146 * postreload.c (reload_cse_simplify_operands): Skip
4147 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
4149 2017-09-15 Richard Biener <rguenther@suse.de>
4151 PR tree-optimization/68823
4152 * graphite-scop-detection.c (build_alias_set): If we have a
4153 possible dependence check whether we can handle them by just
4154 looking at the DRs DR_ACCESS_FNs.
4155 (build_scops): If build_alias_set fails, fail the SCOP.
4157 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4159 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4160 to support float128 built-in functions that require the ISA 3.0
4162 (BU_FLOAT128_3_HW): Likewise.
4163 (SQRTF128): Add support for the IEEE 128-bit square root and fma
4165 (FMAF128): Likewise.
4167 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4168 support for built-in functions that need the ISA 3.0 IEEE 128-bit
4169 floating point instructions.
4170 (rs6000_invalid_builtin): Likewise.
4171 (rs6000_builtin_mask_names): Likewise.
4172 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
4173 (RS6000_BTM_FLOAT128_HW): Likewise.
4174 (RS6000_BTM_COMMON): Likewise.
4175 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
4177 * doc/extend.texi (RS/6000 built-in functions): Document the
4178 IEEE 128-bit floating point square root and fused multiply-add
4181 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
4183 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
4184 reg (r2) isn't in the set of registers defined in the prologue.
4186 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4187 Alan Hayward <alan.hayward@arm.com>
4188 David Sherwood <david.sherwood@arm.com>
4190 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
4191 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
4192 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
4193 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
4194 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
4196 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
4197 max_vectorization_factor.
4198 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
4200 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4201 Alan Hayward <alan.hayward@arm.com>
4202 David Sherwood <david.sherwood@arm.com>
4204 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
4205 (vect_worthwhile_without_simd_p): Declare.
4206 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
4207 (vectorizable_reduction): Use it.
4208 * tree-vect-stmts.c (vectorizable_shift): Likewise.
4209 (vectorizable_operation): Likewise.
4211 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4212 Alan Hayward <alan.hayward@arm.com>
4213 David Sherwood <david.sherwood@arm.com>
4215 * tree-vectorizer.h (vect_get_num_copies): New function.
4216 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
4217 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4218 (vectorizable_induction): Likewise.
4219 (vectorizable_live_operation): Likewise.
4220 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4221 (vectorizable_bswap): Likewise.
4222 (vectorizable_call): Likewise.
4223 (vectorizable_conversion): Likewise.
4224 (vectorizable_assignment): Likewise.
4225 (vectorizable_shift): Likewise.
4226 (vectorizable_operation): Likewise.
4227 (vectorizable_store): Likewise.
4228 (vectorizable_load): Likewise.
4229 (vectorizable_condition): Likewise.
4230 (vectorizable_comparison): Likewise.
4231 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
4233 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4234 Alan Hayward <alan.hayward@arm.com>
4235 David Sherwood <david.sherwood@arm.com>
4237 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
4238 of vect_init_vector.
4240 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4241 Alan Hayward <alan.hayward@arm.com>
4242 David Sherwood <david.sherwood@arm.com>
4244 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
4245 an inline wrapper that provides a location.
4246 (gimple_build_vector): Likewise.
4247 * gimple-fold.c (gimple_build_vector_from_val): New function.
4248 (gimple_build_vector): Likewise.
4249 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
4250 functions to build the initial value. Always return a gimple value.
4251 (get_initial_defs_for_reduction): Likewise. Only compute
4253 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
4254 vect_init_vector on the results from get_initial_def(s)_for_reduction.
4255 (vectorizable_induction): Use gimple_build_vector rather than
4258 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4259 Alan Hayward <alan.hayward@arm.com>
4260 David Sherwood <david.sherwood@arm.com>
4262 * target.h (vec_perm_indices): New typedef.
4263 (auto_vec_perm_indices): Likewise.
4264 * optabs-query.h: Include target.h
4265 (can_vec_perm_p): Take a vec_perm_indices *.
4266 * optabs-query.c (can_vec_perm_p): Likewise.
4267 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
4268 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4269 * tree-vect-generic.c (lower_vec_perm): Likewise.
4270 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
4271 (vect_grouped_load_supported): Likewise.
4272 (vect_shift_permute_load_chain): Likewise.
4273 (vect_permute_store_chain): Use auto_vec_perm_indices.
4274 (vect_permute_load_chain): Likewise.
4275 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
4276 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
4277 Update uses of can_vec_perm_p.
4278 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
4279 mode with a number of elements. Take a vec_perm_indices *.
4280 (vect_create_epilog_for_reduction): Update accordingly.
4281 Use auto_vec_perm_indices.
4282 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
4283 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
4284 (vect_transform_slp_perm_load): Likewise.
4285 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
4286 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
4287 (vect_gen_perm_mask_checked): Likewise.
4288 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
4289 (vect_gen_perm_mask_checked): Likewise.
4290 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
4291 (vectorizable_store): Likewise.
4292 (vectorizable_load): Likewise.
4293 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
4294 (vectorizable_bswap): Likewise.
4296 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4297 Alan Hayward <alan.hayward@arm.com>
4298 David Sherwood <david.sherwood@arm.com>
4300 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
4301 * tree.c (build_vector): Likewise.
4302 (build_vector_from_ctor): Update accordingly.
4303 (build_vector_from_val): Likewise.
4304 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4305 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4306 * tree-vect-generic.c (add_rshift): Likewise.
4307 (expand_vector_divmod): Likewise.
4308 (optimize_vector_constructor): Likewise.
4309 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4310 (vect_transform_slp_perm_load): Likewise.
4311 (vect_schedule_slp_instance): Likewise.
4312 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
4313 (vectorizable_call): Likewise.
4314 (vect_gen_perm_mask_any): Likewise. Add elements in order.
4315 * expmed.c (make_tree): Likewise.
4316 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
4317 a vector passed to build_vector.
4318 (fold_convert_const): Likewise.
4319 (exact_inverse): Likewise.
4320 (fold_ternary_loc): Likewise.
4321 (fold_relational_const): Likewise.
4322 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
4323 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
4324 (const_unop): Likewise. Store the reduction accumulator in a
4325 variable rather than an array.
4326 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
4327 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
4328 the new vector, rather than constructing it after the input arrays.
4329 (native_interpret_vector): Use auto_vec<tree> when building
4330 a vector passed to build_vector. Add elements in order.
4331 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
4332 auto_vec<tree> when building a vector passed to build_vector.
4333 (vect_create_epilog_for_reduction): Likewise.
4334 (vectorizable_induction): Likewise.
4335 (get_initial_def_for_reduction): Likewise. Fix indentation of
4337 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
4339 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
4340 passed to build_vector.
4342 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4343 Alan Hayward <alan.hayward@arm.com>
4344 David Sherwood <david.sherwood@arm.com>
4346 * tree-core.h (tree_base::u): Add an "nelts" field.
4347 (tree_vector): Use VECTOR_CST_NELTS as the length.
4348 * tree.c (tree_size): Likewise.
4349 (make_vector): Initialize VECTOR_CST_NELTS.
4350 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
4351 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
4352 TYPE_VECTOR_SUBPARTS.
4353 * expr.c (const_vector_mask_from_tree): Consistently use "units"
4354 as the number of units, setting it from VECTOR_CST_NELTS.
4355 (const_vector_from_tree): Likewise.
4356 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
4357 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
4358 (fold_negate_expr_1): Likewise.
4359 (fold_convert_const): Likewise.
4360 (const_binop): Likewise. Differentiate the number of output and
4362 (const_unop): Likewise.
4363 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
4364 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
4365 in cases that did the opposite.
4367 2017-09-14 Richard Biener <rguenther@suse.de>
4369 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
4372 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
4374 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
4376 2017-09-14 Jakub Jelinek <jakub@redhat.com>
4379 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
4380 if and where to split a bb, except for splitting before debug insn
4381 sequences followed by non-label real insn. Delete debug insns
4382 in between basic blocks.
4384 * combine.c (make_compound_operation_int): Formatting fixes.
4386 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
4387 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4388 * config/netbsd.h (LINK_EH_SPEC): Likewise.
4389 * config/sol2.h (LINK_EH_SPEC): Likewise.
4390 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4391 * config/s390/linux.h (LINK_SPEC): Likewise.
4392 * config/freebsd.h (LINK_EH_SPEC): Likewise.
4393 * config/openbsd.h (LINK_EH_SPEC): Likewise.
4394 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4395 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
4396 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
4397 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4398 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
4399 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
4401 2017-09-13 Jakub Jelinek <jakub@redhat.com>
4403 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
4405 (ENDFILE_LINUX_SPEC): Likewise.
4406 (LINK_EH_SPEC): Likewise.
4407 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
4408 (LINK_OS_LINUX_SPEC32): Likewise.
4409 (LINK_OS_LINUX_SPEC64): Likewise.
4410 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
4411 (LINK_OS_LINUX_SPEC): Likewise.
4413 2017-09-13 Martin Liska <mliska@suse.cz>
4416 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
4417 CASE_HIGH is NULL_TREE.
4419 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4420 Alan Hayward <alan.hayward@arm.com>
4421 David Sherwood <david.sherwood@arm.com>
4423 * target.def (secondary_memory_needed): New hook.
4424 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
4425 instead of SECONDARY_MEMORY_NEEDED.
4426 (secondary_memory_needed_mode): Likewise.
4427 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
4428 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
4429 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
4430 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
4431 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
4432 * doc/tm.texi: Regenerate.
4433 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
4434 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
4435 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4436 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
4437 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
4438 * config/i386/i386.c (inline_secondary_memory_needed): Put the
4439 mode argument first and change the reg_class arguments to reg_class_t.
4440 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
4441 Make static. Update the call to inline_secondary_memory_needed.
4442 (ix86_register_move_cost): Update the call to
4443 inline_secondary_memory_needed.
4444 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4445 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
4447 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
4448 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4450 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
4451 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
4452 * config/mips/mips.c (mips_secondary_memory_needed): Make static
4453 and match hook interface. Add comment from mips.h.
4454 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4455 * config/mmix/mmix.md (truncdfsf2): Refer to
4456 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4458 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
4459 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
4460 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4461 (pa_secondary_memory_needed): New function.
4462 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
4463 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
4464 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4465 (pdp11_secondary_memory_needed): Make static and match hook interface.
4466 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
4467 * config/powerpcspe/powerpcspe-protos.h
4468 (rs6000_secondary_memory_needed_ptr): Delete.
4469 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
4471 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4472 (rs6000_option_override_internal): Assign to
4473 targetm.secondary_memory_needed rather than
4474 rs6000_secondary_memory_needed_ptr.
4475 (rs6000_secondary_memory_needed): Match hook interface.
4476 (rs6000_debug_secondary_memory_needed): Likewise.
4477 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
4478 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
4479 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
4480 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4481 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
4482 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
4484 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
4485 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4486 (rs6000_option_override_internal): Assign to
4487 targetm.secondary_memory_needed rather than
4488 rs6000_secondary_memory_needed_ptr.
4489 (rs6000_secondary_memory_needed): Match hook interface.
4490 (rs6000_debug_secondary_memory_needed): Likewise.
4491 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
4492 * config/s390/s390.c (s390_secondary_memory_needed): New function.
4493 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4494 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
4495 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4496 (sparc_secondary_memory_needed): New function.
4497 * lra-constraints.c (check_and_process_move): Refer to
4498 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4500 (curr_insn_transform): Likewise.
4501 (process_alt_operands): Use targetm.secondary_memory_needed
4502 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4503 (check_secondary_memory_needed_p): Likewise.
4504 (choose_split_class): Likewise.
4505 * reload.c: Unconditionally include code that was previously
4506 conditional on SECONDARY_MEMORY_NEEDED.
4507 (push_secondary_reload): Use targetm.secondary_memory_needed
4508 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4509 (push_reload): Likewise.
4510 * reload1.c: Unconditionally include code that was previously
4511 conditional on SECONDARY_MEMORY_NEEDED.
4512 (choose_reload_regs): Use targetm.secondary_memory_needed
4513 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4514 (gen_reload): Likewise.
4515 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
4517 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4518 Alan Hayward <alan.hayward@arm.com>
4519 David Sherwood <david.sherwood@arm.com>
4521 * target.def (secondary_memory_needed_mode): New hook:
4522 * targhooks.c (default_secondary_memory_needed_mode): Declare.
4523 * targhooks.h (default_secondary_memory_needed_mode): New function.
4524 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
4525 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
4526 * doc/tm.texi: Regenerate.
4527 * lra-constraints.c (check_and_process_move): Use
4528 targetm.secondary_memory_needed_mode instead of
4529 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
4530 (curr_insn_transform): Likewise.
4531 * reload.c (get_secondary_mem): Likewise.
4532 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4533 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
4535 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4536 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4537 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
4538 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4539 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
4541 * config/powerpcspe/powerpcspe-protos.h
4542 (rs6000_secondary_memory_needed_mode): Delete.
4543 * config/powerpcspe/powerpcspe.c
4544 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4545 (rs6000_secondary_memory_needed_mode): Make static.
4546 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4547 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
4549 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4551 (rs6000_secondary_memory_needed_mode): Make static.
4552 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4553 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
4554 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4555 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4556 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4558 (sparc_secondary_memory_needed_mode): New function.
4559 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
4561 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
4563 * config/aarch64/constraints.md (Umq): New constraint.
4564 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
4566 (mov<mode>): Update condition.
4568 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4570 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
4571 when bitposition is the same.
4573 2017-09-13 Richard Biener <rguenther@suse.de>
4575 * dwarf2out.c (output_die_symbol): Remove.
4576 (output_die): Do not output a DIEs symbol.
4578 2017-09-13 Richard Biener <rguenther@suse.de>
4581 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
4582 default-def to avoid breaking iterator update with the weird
4583 interaction with cgraph_update_edges_for_call_stmt_node.
4585 2017-09-13 Richard Biener <rguenther@suse.de>
4587 * tree-cfg.c (verify_gimple_assign_binary): Add verification
4588 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
4589 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
4590 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
4592 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
4594 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
4595 Disable pc relative literal load irrespective of
4596 TARGET_FIX_ERR_A53_84341 for default.
4598 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
4600 * config/sparc/sparc.c (output_return): Output the source location of
4601 the insn in the delay slot, if any.
4602 (output_sibcall): Likewise.
4604 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
4607 * common.opt (-static-pie): New alias.
4608 (shared): Negate static-pie.
4609 (-no-pie): Update help text.
4611 (static-pie): New option.
4612 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
4613 -static-pie support.
4614 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
4615 (LINK_EH_SPEC): Likewise.
4616 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4617 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4618 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4619 * gcc.c (LINK_COMMAND_SPEC): Likewise.
4620 (init_gcc_specs): Likewise.
4621 (init_spec): Likewise.
4622 (display_help): Update help message for -pie.
4623 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
4626 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
4628 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
4629 (movdi_aarch64): Likewise.
4630 (movti_aarch64): Likewise.
4632 2017-09-12 Simon Wright <simon@pushface.org>
4635 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
4636 calculation of the minor version, always output as 0.
4638 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4641 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4642 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
4643 on it early, rather than manual conversion late. For
4644 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
4645 instead of performing manual conversion.
4647 2017-09-12 Carl Love <cel@us.ibm.com>
4649 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
4650 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
4652 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4653 VMULOSW): Add definitions.
4654 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4655 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4656 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4657 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4658 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4660 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4662 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
4664 (movti_aarch64): Likewise.
4665 (movdf_aarch64): Likewise.
4666 (movtf_aarch64): Likewise.
4667 (load_pairdi): Likewise.
4668 (store_pairdi): Likewise.
4669 (load_pairdf): Likewise.
4670 (store_pairdf): Likewise.
4671 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
4672 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
4673 (ldr_got_small_<mode>): Likewise.
4674 (ldr_got_small_28k_<mode>): Likewise.
4675 (ldr_got_tiny): Likewise.
4676 * config/aarch64/iterators.md (ldst_sz): New.
4677 (ldpstp_sz): Likewise.
4678 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
4680 (thunderx_load): Split load_8 to load_16.
4681 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
4683 (thunderx2t99_storepair_basic): Split store_8 to store_16.
4684 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
4685 (xgene1_store_pair): Split store_8 to store_16.
4686 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
4687 (falkor_st_0_st_sd): Split store_8 to store_16.
4689 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4691 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
4692 and store1/2/3/4 to store_4/8/12/16.
4693 * config/aarch64/aarch64.md: Update for rename.
4694 * config/arm/arm.md: Likewise.: Likewise.
4695 * config/arm/arm.c: Likewise.
4696 * config/arm/thumb1.md: Likewise.
4697 * config/arm/thumb2.md: Likewise.
4698 * config/arm/vfp.md: Likewise.
4699 * config/arm/arm-generic.md: Likewise.
4700 * config/arm/arm1020e.md: Likewise.
4701 * config/arm/arm1026ejs.md: Likewise.
4702 * config/arm/arm1136jfs.md: Likewise.
4703 * config/arm/arm926ejs.md: Likewise.
4704 * config/arm/cortex-a15.md: Likewise.
4705 * config/arm/cortex-a17.md: Likewise.
4706 * config/arm/cortex-a5.md: Likewise.
4707 * config/arm/cortex-a53.md: Likewise.
4708 * config/arm/cortex-a57.md: Likewise.
4709 * config/arm/cortex-a7.md: Likewise.
4710 * config/arm/cortex-a8.md: Likewise.
4711 * config/arm/cortex-a9.md: Likewise.
4712 * config/arm/cortex-m4.md: Likewise.
4713 * config/arm/cortex-m7.md: Likewise.
4714 * config/arm/cortex-r4.md: Likewise.
4715 * config/arm/exynos-m1.md: Likewise.
4716 * config/arm/fa526.md: Likewise.
4717 * config/arm/fa606te.md: Likewise.
4718 * config/arm/fa626te.md: Likewise.
4719 * config/arm/fa726te.md: Likewise.
4720 * config/arm/fmp626.md: Likewise.
4721 * config/arm/iwmmxt.md: Likewise.
4722 * config/arm/ldmstm.md: Likewise.
4723 * config/arm/marvell-pj4.md: Likewise.
4724 * config/arm/xgene1.md: Likewise.
4725 * config/aarch64/thunderx.md: Likewise.
4726 * config/aarch64/thunderx2t99.md: Likewise.
4727 * config/aarch64/falkor.md: Likewise.
4729 2017-09-12 Martin Liska <mliska@suse.cz>
4731 * attribs.c (private_lookup_attribute): New function.
4732 * attribs.h (private_lookup_attribute): Declared here.
4733 (lookup_attribute): Called from this place.
4735 2017-09-12 Richard Biener <rguenther@suse.de>
4737 PR tree-optimization/82157
4738 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
4739 stmts with side-effects.
4741 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4742 Alan Hayward <alan.hayward@arm.com>
4743 David Sherwood <david.sherwood@arm.com>
4745 * target.def (hard_regno_nregs): New hook.
4746 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
4747 * targhooks.h (default_hard_regno_nregs): Declare.
4748 * targhooks.c (default_hard_regno_nregs): New function.
4749 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
4750 (TARGET_HARD_REGNO_NREGS): ...this hook.
4751 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
4752 (CLASS_MAX_NREGS): Likewise.
4753 * doc/tm.texi: Regenerate.
4754 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
4755 instead of HARD_REGNO_NREGS.
4756 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
4757 HARD_REGNO_NREGS in the comment.
4758 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
4759 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
4760 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
4761 Return an unsigned int.
4762 (TARGET_HARD_REGNO_NREGS): Redefine.
4763 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
4764 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
4765 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4766 (arc_hard_regno_nregs): New function.
4767 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
4768 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
4769 (arm_hard_regno_nregs): New function.
4770 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
4771 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
4772 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
4773 (TARGET_HARD_REGNO_NREGS): Redefine.
4774 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
4775 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
4776 (HARD_REGNO_NREGS): Delete.
4777 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
4778 (cr16_hard_regno_nregs): New function.
4779 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
4780 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
4781 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
4782 (cris_hard_regno_nregs): New function.
4783 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
4784 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
4785 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4786 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
4787 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
4788 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
4789 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
4790 (frv_hard_regno_nregs): Make static. Take and return an
4792 (frv_class_max_nregs): Remove outdated copy of documentation.
4793 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
4794 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
4795 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
4796 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
4797 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
4798 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
4799 (TARGET_HARD_REGNO_NREGS): Redefine.
4800 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
4801 (CLASS_MAX_NREGS): Update comment.
4802 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
4803 (ia64_hard_regno_nregs): New function.
4804 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
4805 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
4806 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
4807 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
4808 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
4810 (m32c_hard_regno_nregs): Likewise. Make static.
4811 (TARGET_HARD_REGNO_NREGS): Redefine.
4812 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
4813 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
4814 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
4815 (m68k_hard_regno_nregs): New function.
4816 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
4817 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
4818 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
4819 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
4820 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
4821 Take and return an unsigned int.
4822 (TARGET_HARD_REGNO_NREGS): Redefine.
4823 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
4824 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4825 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
4826 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
4827 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
4828 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
4829 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
4830 (msp430_hard_regno_nregs): Make static. Take and return an
4832 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
4833 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
4834 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
4835 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
4836 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
4837 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
4838 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
4839 (TARGET_HARD_REGNO_NREGS): Redefine.
4840 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
4841 (PA_HARD_REGNO_NREGS): ...this.
4842 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
4843 (PA_HARD_REGNO_NREGS): ...this.
4844 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4845 (pa_hard_regno_nregs): New function.
4846 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
4847 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
4848 (pdp11_hard_regno_nregs): New function.
4849 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
4850 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
4851 (rs6000_hard_regno_nregs_hook): New function.
4852 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
4853 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4854 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
4855 Take and return an unsigned int. Move earlier in file.
4856 (TARGET_HARD_REGNO_NREGS): Redefine.
4857 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
4858 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
4859 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
4860 (rl78_hard_regno_nregs): Make static. Take and return an
4862 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
4863 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
4864 (rs6000_hard_regno_nregs_hook): New function.
4865 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
4866 * config/rx/rx.c (rx_hard_regno_nregs): New function.
4867 (TARGET_HARD_REGNO_NREGS): Redefine.
4868 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
4869 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
4870 instead of HARD_REGNO_NREGS.
4871 (s390_hard_regno_nregs): New function.
4872 (s390_hard_regno_mode_ok): Add comment from s390.h.
4873 (TARGET_HARD_REGNO_NREGS): Redefine.
4874 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
4875 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
4876 (sh_hard_regno_nregs): New function.
4877 (sh_pass_in_reg_p): Use it.
4878 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
4879 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4880 (sparc_hard_regno_nregs): New function.
4881 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
4882 * config/spu/spu.c (spu_hard_regno_nregs): New function.
4883 (spu_function_arg_advance): Use it, supplying a valid register number.
4884 (TARGET_HARD_REGNO_NREGS): Redefine.
4885 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
4886 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
4887 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
4888 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
4889 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
4890 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
4891 (CLASS_MAX_NREGS): Remove copy of old documentation.
4892 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
4893 (visium_hard_regno_nregs): New function.
4894 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
4895 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
4896 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4897 xtensa_hard_regno_nregs): New function.
4898 * system.h (HARD_REGNO_NREGS): Poison.
4900 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4902 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
4903 hard_regno_nregs instead of HARD_REGNO_NREGS.
4904 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
4905 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
4906 (c6x_expand_epilogue): Likewise.
4907 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
4908 (frv_read_iacc_argument): Likewise.
4909 * config/sh/sh.c: Include regs.h.
4910 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4911 (regs_used): Likewise.
4912 (output_stack_adjust): Likewise.
4913 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
4914 * expmed.c: Include regs.h.
4915 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4916 * ree.c: Include regs.h.
4917 (combine_reaching_defs): Use hard_regno_nregs instead of
4919 (add_removable_extension): Likewise.
4921 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4923 * regs.h (hard_regno_nregs): Turn into a function.
4924 (end_hard_regno): Update accordingly.
4925 * caller-save.c (setup_save_areas): Likewise.
4926 (save_call_clobbered_regs): Likewise.
4927 (replace_reg_with_saved_mem): Likewise.
4928 (insert_restore): Likewise.
4929 (insert_save): Likewise.
4930 * combine.c (can_change_dest_mode): Likewise.
4931 (move_deaths): Likewise.
4932 (distribute_notes): Likewise.
4933 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
4934 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
4935 (rs6000_split_multireg_move): Likewise.
4936 (rs6000_register_move_cost): Likewise.
4937 (rs6000_memory_move_cost): Likewise.
4938 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
4939 (rs6000_split_multireg_move): Likewise.
4940 (rs6000_register_move_cost): Likewise.
4941 (rs6000_memory_move_cost): Likewise.
4942 * cselib.c (cselib_reset_table): Likewise.
4943 (cselib_lookup_1): Likewise.
4944 * emit-rtl.c (set_mode_and_regno): Likewise.
4945 * function.c (aggregate_value_p): Likewise.
4946 * ira-color.c (setup_profitable_hard_regs): Likewise.
4947 (check_hard_reg_p): Likewise.
4948 (calculate_saved_nregs): Likewise.
4949 (assign_hard_reg): Likewise.
4950 (improve_allocation): Likewise.
4951 (calculate_spill_cost): Likewise.
4952 * ira-emit.c (modify_move_list): Likewise.
4953 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
4954 (ira_hard_reg_in_set_p): Likewise.
4955 * ira.c (setup_reg_mode_hard_regset): Likewise.
4956 (clarify_prohibited_class_mode_regs): Likewise.
4957 (check_allocation): Likewise.
4958 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4959 (lra_setup_reg_renumber): Likewise.
4960 (setup_try_hard_regno_pseudos): Likewise.
4961 (spill_for): Likewise.
4962 (assign_hard_regno): Likewise.
4963 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
4964 * lra-constraints.c (in_class_p): Likewise.
4965 (lra_constraint_offset): Likewise.
4966 (simplify_operand_subreg): Likewise.
4967 (lra_constraints): Likewise.
4968 (split_reg): Likewise.
4969 (split_if_necessary): Likewise.
4970 (invariant_p): Likewise.
4971 (inherit_in_ebb): Likewise.
4972 * lra-lives.c (process_bb_lives): Likewise.
4973 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
4974 (get_hard_regs): Likewise.
4975 (do_remat): Likewise.
4976 * lra-spills.c (assign_spill_hard_regs): Likewise.
4977 * mode-switching.c (create_pre_exit): Likewise.
4978 * postreload.c (reload_combine_recognize_pattern): Likewise.
4979 * recog.c (peep2_find_free_register): Likewise.
4980 * regcprop.c (kill_value_regno): Likewise.
4981 (set_value_regno): Likewise.
4982 (copy_value): Likewise.
4983 (maybe_mode_change): Likewise.
4984 (find_oldest_value_reg): Likewise.
4985 (copyprop_hardreg_forward_1): Likewise.
4986 * regrename.c (check_new_reg_p): Likewise.
4987 (regrename_do_replace): Likewise.
4988 * reload.c (push_reload): Likewise.
4989 (combine_reloads): Likewise.
4990 (find_dummy_reload): Likewise.
4991 (operands_match_p): Likewise.
4992 (find_reloads): Likewise.
4993 (find_equiv_reg): Likewise.
4994 (reload_adjust_reg_for_mode): Likewise.
4995 * reload1.c (count_pseudo): Likewise.
4996 (count_spilled_pseudo): Likewise.
4997 (find_reg): Likewise.
4998 (clear_reload_reg_in_use): Likewise.
4999 (free_for_value_p): Likewise.
5000 (allocate_reload_reg): Likewise.
5001 (choose_reload_regs): Likewise.
5002 (reload_adjust_reg_for_temp): Likewise.
5003 (emit_reload_insns): Likewise.
5004 (delete_output_reload): Likewise.
5005 * rtlanal.c (subreg_get_info): Likewise.
5006 * sched-deps.c (sched_analyze_reg): Likewise.
5007 * sel-sched.c (init_regs_for_mode): Likewise.
5008 (mark_unavailable_hard_regs): Likewise.
5009 (choose_best_reg_1): Likewise.
5010 (verify_target_availability): Likewise.
5011 * valtrack.c (dead_debug_insert_temp): Likewise.
5012 * var-tracking.c (track_loc_p): Likewise.
5013 (emit_note_insn_var_location): Likewise.
5014 * varasm.c (make_decl_rtl): Likewise.
5015 * reginfo.c (choose_hard_reg_mode): Likewise.
5016 (init_reg_modes_target): Refer directly to
5017 this_target_regs->x_hard_regno_nregs.
5019 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5021 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5022 instead of hard_regno_nregs.
5024 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5026 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5027 end_hard_regno instead of hard_regno_nregs.
5028 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5029 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5030 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5031 * ira-color.c (improve_allocation): Likewise.
5032 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5033 * lra-lives.c (mark_regno_live): Likewise.
5034 (mark_regno_dead): Likewise.
5035 * lra-remat.c (operand_to_remat): Likewise.
5036 * lra.c (collect_non_operand_hard_regs): Likewise.
5037 * postreload.c (reload_combine_note_store): Likewise.
5038 (move2add_valid_value_p): Likewise.
5039 * reload.c (regno_clobbered_p): Likewise.
5041 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5043 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5045 * config/v850/v850.c (v850_reorg): Likewise.
5046 * reload.c (refers_to_regno_for_reload_p): Likewise.
5047 (find_equiv_reg): Likewise.
5048 * reload1.c (reload_reg_reaches_end_p): Likewise.
5050 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5052 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5054 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5055 * config/arm/arm.c (output_move_neon): Likewise.
5056 (arm_attr_length_move_neon): Likewise.
5057 (neon_split_vcombine): Likewise.
5058 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5059 (c6x_mark_reg_written): Likewise.
5060 (c6x_dwarf_register_span): Likewise.
5061 * config/i386/i386.c (ix86_save_reg): Likewise.
5062 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5063 (rws_access_reg): Likewise.
5064 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5065 * mode-switching.c (create_pre_exit): Likewise.
5066 * ree.c (combine_reaching_defs): Likewise.
5067 (add_removable_extension): Likewise.
5068 * regcprop.c (find_oldest_value_reg): Likewise.
5069 (copyprop_hardreg_forward_1): Likewise.
5070 * reload.c (reload_inner_reg_of_subreg): Likewise.
5071 (push_reload): Likewise.
5072 (combine_reloads): Likewise.
5073 (find_dummy_reload): Likewise.
5074 (reload_adjust_reg_for_mode): Likewise.
5075 * reload1.c (find_reload_regs): Likewise.
5076 (forget_old_reloads_1): Likewise.
5077 (reload_reg_free_for_value_p): Likewise.
5078 (reload_adjust_reg_for_temp): Likewise.
5079 (emit_reload_insns): Likewise.
5080 (delete_output_reload): Likewise.
5081 * sel-sched.c (choose_best_reg_1): Likewise.
5082 (choose_best_pseudo_reg): Likewise.
5084 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5085 Alan Hayward <alan.hayward@arm.com>
5086 David Sherwood <david.sherwood@arm.com>
5088 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
5089 * target.def (slow_unaligned_access): New hook.
5090 * targhooks.h (default_slow_unaligned_access): Declare.
5091 * targhooks.c (default_slow_unaligned_access): New function.
5092 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
5093 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
5094 * doc/tm.texi: Regenerate.
5095 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
5096 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
5097 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
5099 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
5100 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
5102 (rs6000_slow_unaligned_access): New function.
5103 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5104 (expand_block_compare): Likewise.
5105 (expand_strn_compare): Likewise.
5106 (rs6000_rtx_costs): Likewise.
5107 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
5108 (riscv_slow_unaligned_access): Likewise.
5109 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
5110 (riscv_slow_unaligned_access_p): ...this and make static.
5111 (riscv_option_override): Update accordingly.
5112 (riscv_slow_unaligned_access): New function.
5113 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5114 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
5115 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5116 (rs6000_slow_unaligned_access): New function.
5117 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5118 (rs6000_rtx_costs): Likewise.
5119 * config/rs6000/rs6000-string.c (expand_block_compare)
5120 (expand_strn_compare): Use targetm.slow_unaligned_access instead
5121 of SLOW_UNALIGNED_ACCESS.
5122 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
5123 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
5124 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
5125 of SLOW_UNALIGNED_ACCESS.
5126 * expmed.c (simple_mem_bitfield_p): Likewise.
5127 * expr.c (alignment_for_piecewise_move): Likewise.
5128 (emit_group_load_1): Likewise.
5129 (emit_group_store): Likewise.
5130 (copy_blkmode_from_reg): Likewise.
5131 (emit_push_insn): Likewise.
5132 (expand_assignment): Likewise.
5133 (store_field): Likewise.
5134 (expand_expr_real_1): Likewise.
5135 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5136 * lra-constraints.c (simplify_operand_subreg): Likewise.
5137 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5138 * gimple-ssa-store-merging.c: Likewise in block comment at start
5140 * tree-ssa-strlen.c: Include target.h.
5141 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
5142 of SLOW_UNALIGNED_ACCESS.
5143 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
5145 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5147 PR rtl-optimization/82185
5148 * expmed.c (emit_store_flag_int): Only test tem if it has been
5151 2017-09-12 Richard Biener <rguenther@suse.de>
5154 * match.pd ((FTYPE) N CMP CST): Fix typo.
5156 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
5158 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
5160 (mips_near_type_p): Add 'short_call' attribute as a synonym
5162 * doc/extend.texi (short_call): Document new function attribute.
5164 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5167 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
5168 assertion check that in the condition.
5169 (get_atomic_generic_size): Likewise. Before testing if parameter
5170 has pointer type, if it has array type, call for C++
5171 default_conversion to perform array-to-pointer conversion.
5173 2017-09-12 Richard Biener <rguenther@suse.de>
5175 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
5176 for operations we cannot scalarize.
5178 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
5180 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
5181 vectors heap vectors. Clean up comments.
5182 Make visited_bbs a reference.
5183 (profitable_jump_thread_path): Make GC
5184 vectors heap vectors. Clean up comments.
5186 (convert_and_register_jump_thread_path): Make GC vectors heap
5188 (check_subpath_and_update_thread_path): Same. Clean up comments.
5189 Make visited_bbs a reference.
5190 (handle_phi): Abstract common code to to
5191 register_jump_thread_path_if_profitable.
5192 Rename VAR_BB to DEF_BB.
5194 Make GC vectors heap vectors.
5195 Make visited_bbs a reference.
5196 (handle_assignment): Same.
5197 (register_jump_thread_path_if_profitable): New.
5198 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
5200 Make GC vectors heap vectors. Clean up comments.
5201 Make visited_bbs a reference.
5202 (find_jump_threads_backwards): Make visited_bbs live in the stack.
5203 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
5206 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
5209 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
5210 words of E_DImode object are reachable by xtensa_uimm8x4 access.
5212 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
5214 Revert r251800 and r251799.
5216 2017-09-11 Martin Jambor <mjambor@suse.cz>
5219 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
5220 arguments in advance.
5221 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
5222 use it to find predecessor edges.
5223 (naive_outof_ssa): Collect vector of predecessors.
5225 2017-09-08 Jason Merrill <jason@redhat.com>
5227 PR c++/70029 - ICE with ref-qualifier and -flto
5228 * langhooks.h (struct lang_hooks_for_types): Add
5229 copy_lang_qualifiers.
5230 * attribs.c (build_type_attribute_qual_variant): Use it.
5231 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
5233 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
5234 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
5236 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
5239 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
5240 (*mulsi3_sp64): New instruction.
5241 (mulsi3): New expander.
5243 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
5245 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
5247 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5249 * sancov.c: Include memmodel.h.
5251 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
5254 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
5257 2017-09-07 Carl Love <cel@us.ibm.com>
5259 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
5260 the sldi instruction.
5262 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
5264 * sancov.c: Include tm_p.h.
5266 2017-09-07 Jakub Jelinek <jakub@redhat.com>
5269 * output.h (switch_to_other_text_partition): New declaration.
5270 * varasm.c (switch_to_other_text_partition): New function.
5271 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
5272 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
5273 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
5274 to the other text partition before emitting LCL label and switch back
5275 after emitting the word after it.
5277 2017-09-07 Richard Biener <rguenther@suse.de>
5279 * passes.def (pass_split_crit_edges): Remove instance before PRE.
5280 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
5281 critical edges here, after loop init.
5282 (pass_data_pre): Remove PROP_no_crit_edges flags.
5283 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
5284 for valueization of call args to avoid leaking VN_TOP.
5285 (visit_use): Assert we do not visit default defs.
5286 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
5287 Use error_mark_node to more easily detect leaking VN_TOP.
5288 All default-defs are varying, not VN_TOP. Mark them visited.
5289 (run_scc_vn): Make code match comment.
5291 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5293 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
5294 OPTION_MASK_FLOAT128_KEYWORD.
5295 (POWERPC_MASKS): Likewise.
5296 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
5297 support for the -mfloat128-type option, and make -mfloat128
5298 default on PowerPC Linux systems. Define or undefine
5299 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
5300 Define __float128 to be __ieee128 if IEEE 128-bit support is
5301 enabled, or undefine it.
5302 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
5303 Delete defining __FLOAT128_TYPE__.
5304 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
5305 -mfloat128-type option and make -mfloat128 default on PowerPC
5307 (TARGET_FLOAT128_TYPE): Likewise.
5308 (-mfloat128-type): Likewise.
5309 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5310 Delete the -mfloat128-type option and make -mfloat128 default on
5311 PowerPC Linux systems. Always use __ieee128 to be the keyword for
5312 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
5313 128-bit floating point is enabled. Change tests from using
5314 -mfloat128-type to -mfloat128.
5315 (rs6000_mangle_type): Use the correct mangling for the __float128
5316 type even if normal long double is restricted to 64-bits.
5317 (floatn_mode): Enable the _Float128 type by default on VSX Linux
5319 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
5320 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
5321 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
5323 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
5324 documentation for -mfloat128.
5326 2017-09-06 Olivier Hainque <hainque@adacore.com>
5328 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
5330 2017-09-06 Wish Wu <wishwu007@gmail.com>
5331 Jakub Jelinek <jakub@redhat.com>
5333 * asan.c (initialize_sanitizer_builtins): Add
5334 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
5335 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
5336 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
5337 BT_FN_VOID_UINT64_PTR variables.
5338 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
5339 (BT_FN_VOID_UINT16_UINT16): Likewise.
5340 (BT_FN_VOID_UINT32_UINT32): Likewise.
5341 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
5342 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
5343 (BT_FN_VOID_UINT64_PTR): Likewise.
5344 * common.opt (flag_sanitize_coverage): New variable.
5345 (fsanitize-coverage=trace-pc): Remove.
5346 (fsanitize-coverage=): Add.
5347 * flag-types.h (enum sanitize_coverage_code): New enum.
5348 * fold-const.c (fold_range_test): Disable non-short-circuit
5349 optimization if flag_sanitize_coverage.
5350 (fold_truth_andor): Likewise.
5351 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
5352 * opts.c (COVERAGE_SANITIZER_OPT): Define.
5353 (coverage_sanitizer_opts): New array.
5354 (get_closest_sanitizer_option): Add OPTS argument, handle also
5355 OPT_fsanitize_coverage_.
5356 (parse_sanitizer_options): Adjusted to also handle
5357 OPT_fsanitize_coverage_.
5358 (common_handle_option): Add OPT_fsanitize_coverage_.
5359 * sancov.c (instrument_comparison, instrument_switch): New function.
5360 (sancov_pass): Add trace-cmp support.
5361 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
5362 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
5363 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
5364 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
5365 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
5366 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
5367 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
5368 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
5369 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
5370 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
5372 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5374 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5375 error. Only quit immediately if parsing is complete.
5376 (BEGIN): Initialize fatal_err and parse_done.
5377 (begin fpu, end fpu): Check number of arguments.
5378 (begin arch, end arch): Likewise.
5379 (begin cpu, end cpu): Likewise.
5380 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5381 (optalias): Likewise.
5383 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5385 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5386 * config/arm/arm-isa.h: Delete. Move definitions to ...
5387 * arm-cpus.in: ... here. Use new feature and fgroup values.
5388 * config/arm/arm.c (arm_option_override): Use lower case for feature
5390 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5391 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5392 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5394 (print_isa_bits_for): New function.
5395 (gen_isa): New function.
5396 (gen_comm_data): Use print_isa_bits_for.
5397 (define feature): New keyword.
5398 (define fgroup): New keyword.
5399 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
5400 (arm-isa.h): Add rule to generate file.
5401 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5402 case for feature bit names.
5404 2017-09-06 Richard Biener <rguenther@suse.de>
5406 * tree-ssa-pre.c (NECESSARY): Remove.
5407 (create_expression_by_pieces): Do not touch pass-local flags.
5408 (insert_into_preds_of_block): Likewise.
5409 (do_pre_regular_insertion): Likewise.
5410 (eliminate_insert): Likewise.
5411 (eliminate_dom_walker::before_dom_children): Likewise.
5412 (fini_eliminate): Do not look at inserted_exprs.
5413 (mark_operand_necessary): Remove.
5414 (remove_dead_inserted_code): Replace with simple work-list
5415 algorithm based on inserted_exprs and SSA uses.
5416 (pass_pre::execute): Re-order fini_eliminate and
5417 remove_dead_inserted_code.
5419 2017-09-06 Olivier Hainque <hainque@adacore.com>
5421 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
5422 for VxWorks 7. Adjust surrounding comments.
5424 2017-09-06 Richard Biener <rguenther@suse.de>
5426 * gimple-ssa-strength-reduction.c
5427 (find_candidates_dom_walker::before_dom_children): Also allow
5430 2017-09-06 Richard Biener <rguenther@suse.de>
5432 PR tree-optimization/82108
5433 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
5434 for gap in the non-permutation SLP case.
5436 2017-09-06 Martin Jambor <mjambor@suse.cz>
5438 PR tree-optimization/82078
5439 * tree-sra.c (sort_and_splice_var_accesses): Move call to
5440 add_access_to_work_queue...
5441 (build_accesses_from_assign): ...here.
5442 (propagate_all_subaccesses): Make sure racc is the group
5443 representative, if there is one.
5445 2017-09-06 Jakub Jelinek <jakub@redhat.com>
5448 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
5451 2017-09-06 Richard Biener <rguenther@suse.de>
5453 * gimple-ssa-strength-reduction.c
5454 (find_candidates_dom_walker::before_doom_children): Use a
5455 type and not a mode check.
5457 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5460 * config/arm/predicates.md (arm_general_adddi_operand): Create new
5462 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
5464 2017-09-05 Jeff Law <law@redhat.com>
5466 PR tree-optimization/64910
5467 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
5468 cases where we have 3 or more operands.
5470 2017-09-05 Jakub Jelinek <jakub@redhat.com>
5473 * omp-low.c (lower_omp_for): Recompute tree invariant if
5474 gimple_omp_for_initial/final is ADDR_EXPR.
5477 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
5478 into gimple val before gimplification fo the COND_EXPR.
5480 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
5482 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
5483 REGION_COPY argument.
5484 (thread_through_all_blocks): Remove unused argument to
5485 duplicate_thread_path.
5487 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5488 Alan Hayward <alan.hayward@arm.com>
5489 David Sherwood <david.sherwood@arm.com>
5491 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
5492 Take a scalar_mode rather than a machine_mode.
5493 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5494 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
5495 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5496 (aarch64_gen_adjusted_ldpstp): Likewise.
5497 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
5499 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5500 Alan Hayward <alan.hayward@arm.com>
5501 David Sherwood <david.sherwood@arm.com>
5503 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
5504 Take a scalar_int_mode instead of a machine_mode.
5505 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5506 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5507 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5508 (aarch64_simd_attr_length_rglist): Delete.
5509 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
5510 a scalar_int_mode instead of a machine_mode.
5511 (aarch64_add_offset): Likewise.
5512 (aarch64_internal_mov_immediate): Likewise
5513 (aarch64_add_constant_internal): Likewise.
5514 (aarch64_add_constant): Likewise.
5515 (aarch64_movw_imm): Likewise.
5516 (aarch64_rtx_arith_op_extract_p): Likewise.
5517 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5518 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5519 Remove assert that the mode isn't a vector.
5520 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5521 (aarch64_expand_mov_immediate): Update calls after above changes.
5522 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
5523 (aarch64_and_bitmask_imm): Check for scalar integer modes.
5524 (aarch64_move_imm): Likewise.
5525 (aarch64_can_const_movi_rtx_p): Likewise.
5526 (aarch64_strip_extend): Likewise.
5527 (aarch64_extr_rtx_p): Likewise.
5528 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
5529 a CONST_INT when the mode parameter is VOIDmode.
5530 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
5532 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5534 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
5535 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5536 (bitwise_type_for_mode): Update accordingly.
5538 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5540 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
5541 * stor-layout.c (mode_for_size_tree): Likewise.
5542 (mode_for_array): Update accordingly.
5543 (layout_decl): Likewise.
5544 (compute_record_mode): Likewise. Only set the mode once.
5546 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5548 * target.def (get_mask_mode): Change return type to opt_mode.
5550 * doc/tm.texi: Regenerate.
5551 * targhooks.h (default_get_mask_mode): Return an opt_mode.
5552 * targhooks.c (default_get_mask_mode): Likewise.
5553 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5554 * optabs-query.c (can_vec_mask_load_store_p): Update use of
5555 targetm.get_mask_mode.
5556 * tree.c (build_truth_vector_type): Likewise.
5558 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5560 * machmode.h (mode_for_vector): Return an opt_mode.
5561 * stor-layout.c (mode_for_vector): Likewise.
5562 (mode_for_int_vector): Update accordingly.
5563 (layout_type): Likewise.
5564 * config/i386/i386.c (emit_memmov): Likewise.
5565 (ix86_expand_set_or_movmem): Likewise.
5566 (ix86_expand_vector_init): Likewise.
5567 (ix86_get_mask_mode): Likewise.
5568 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
5570 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
5571 * expmed.c (extract_bit_field_1): Likewise.
5572 * expr.c (expand_expr_real_2): Likewise.
5573 * optabs-query.c (can_vec_perm_p): Likewise.
5574 (can_vec_mask_load_store_p): Likewise.
5575 * optabs.c (expand_vec_perm): Likewise.
5576 * targhooks.c (default_get_mask_mode): Likewise.
5577 * tree-vect-stmts.c (vectorizable_store): Likewise.
5578 (vectorizable_load): Likewise.
5579 (get_vectype_for_scalar_type_and_size): Likewise.
5581 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5583 * machmode.h (mode_for_int_vector): New function.
5584 * stor-layout.c (mode_for_int_vector): Likewise.
5585 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
5586 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
5587 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5588 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5589 (s390_expand_vcond): Likewise.
5591 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5593 * machmode.h (opt_machine_mode): New type.
5594 (opt_mode<T>): Allow construction from anything that can be
5596 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
5597 (mode_for_size): Return an opt_machine_mode.
5598 * stor-layout.c (mode_for_size): Likewise.
5599 (mode_for_size_tree): Update call accordingly.
5600 (bitwise_mode_for_mode): Likewise.
5601 (make_fract_type): Likewise.
5602 (make_accum_type): Likewise.
5603 * caller-save.c (replace_reg_with_saved_mem): Update call
5605 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5606 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5607 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5608 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5609 * expmed.c (extract_bit_field_1): Likewise.
5610 * reload.c (get_secondary_mem): Likewise.
5611 * varasm.c (assemble_integer): Likewise.
5612 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
5615 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5617 * machmode.h (decimal_float_mode_for_size): New function.
5618 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
5619 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
5620 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
5621 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
5622 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
5623 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
5625 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5627 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
5628 (get_builtin_sync_mode): Likewise.
5629 (expand_ifn_atomic_compare_exchange): Likewise.
5630 (expand_builtin_atomic_clear): Likewise.
5631 (expand_builtin_atomic_test_and_set): Likewise.
5632 (fold_builtin_atomic_always_lock_free): Likewise.
5633 * calls.c (compute_argument_addresses): Likewise.
5634 (emit_library_call_value_1): Likewise.
5635 (store_one_arg): Likewise.
5636 * combine.c (combine_instructions): Likewise.
5637 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
5638 * config/arm/arm.c (arm_function_value): Likewise.
5639 (aapcs_allocate_return_reg): Likewise.
5640 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
5641 * config/i386/i386.c (construct_container): Likewise.
5642 (ix86_gimplify_va_arg): Likewise.
5643 (ix86_expand_sse_cmp): Likewise.
5644 (emit_memmov): Likewise.
5645 (emit_memset): Likewise.
5646 (expand_small_movmem_or_setmem): Likewise.
5647 (ix86_expand_pextr): Likewise.
5648 (ix86_expand_pinsr): Likewise.
5649 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
5650 * config/microblaze/microblaze.c (microblaze_block_move_straight):
5652 * config/mips/mips.c (mips_function_value_1) Likewise.
5653 (mips_block_move_straight): Likewise.
5654 (mips_expand_ins_as_unaligned_store): Likewise.
5655 * config/powerpcspe/powerpcspe.c
5656 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5657 (rs6000_darwin64_record_arg_flush): Likewise.
5658 * config/rs6000/rs6000.c
5659 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5660 (rs6000_darwin64_record_arg_flush): Likewise.
5661 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
5662 (sparc_function_value_1): Likewise.
5663 * config/spu/spu.c (adjust_operand): Likewise.
5664 (spu_emit_branch_or_set): Likewise.
5665 (arith_immediate_p): Likewise.
5666 * emit-rtl.c (gen_lowpart_common): Likewise.
5667 * expr.c (expand_expr_real_1): Likewise.
5668 * function.c (assign_parm_setup_block): Likewise.
5669 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5670 * reload1.c (alter_reg): Likewise.
5671 * stor-layout.c (mode_for_vector): Likewise.
5672 (layout_type): Likewise.
5674 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5676 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
5677 (spu_convert_move): Likewise.
5678 * lower-subreg.c (resolve_simple_move): Likewise.
5680 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5683 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
5684 define_insn to a define_expand.
5685 (altivec_vsum2sws_direct): New define_insn.
5686 (altivec_vsumsws): Convert from a define_insn to a define_expand.
5688 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5690 * config/arm/arm.c (arm_option_params_internal): Improve setting of
5693 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
5698 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
5699 realign stack if stack alignment needed is less than incoming
5702 2017-09-05 Marek Polacek <polacek@redhat.com>
5705 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
5708 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5710 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
5712 2017-09-05 Richard Biener <rguenther@suse.de>
5714 PR tree-optimization/82084
5715 * fold-const.c (can_native_encode_string_p): Handle wide characters.
5717 2017-09-05 Richard Biener <rguenther@suse.de>
5719 PR tree-optimization/82102
5720 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
5722 2017-09-05 Martin Liska <mliska@suse.cz>
5724 PR tree-optimization/82032
5725 * tree-cfg.c (generate_range_test): New function.
5726 * tree-cfg.h (generate_range_test): Declared here.
5727 * tree-cfgcleanup.c (convert_single_case_switch): New function.
5728 (cleanup_control_expr_graph): Use it.
5729 * tree-switch-conversion.c (try_switch_expansion): Remove
5731 (emit_case_nodes): Use generate_range_test.
5733 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5736 * config/i386/i386.md (*<btsc><mode>_mask): Add
5737 TARGET_USE_BT to insn constraint.
5738 (*btr<mode>_mask): Ditto.
5740 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
5742 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
5743 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
5745 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5748 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
5749 TARGET_NEON and TARGET_IWMMXT.
5750 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
5751 TARGET_NEON and TARGET_IWMMXT.
5752 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
5754 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5756 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
5757 (ix86_rewrite_tls_address): Ditto.
5758 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
5759 (ix86_rewrite_tls_address_1): Ditto.
5760 (ix86_rewrite_tls_address): Ditto.
5761 * config/i386/predicates.md (tls_address_pattern): New predicate.
5762 * config/i386/i386.md (TLS address splitter): New splitter.
5764 2017-09-04 Richard Biener <rguenther@suse.de>
5766 PR tree-optimization/82084
5767 * fold-const.h (can_native_encode_string_p): Declare.
5768 * fold-const.c (can_native_encode_string_p): Factor out from ...
5769 (native_encode_string): ... here.
5770 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
5771 vectorizing stores from constants we later cannot handle.
5773 2017-09-04 Marek Polacek <polacek@redhat.com>
5776 * doc/invoke.texi: Update -Wtautological-compare documentation.
5778 2017-09-04 Jeff Law <law@redhat.com>
5780 PR tree-optimization/64910
5781 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
5782 swap the first and last operand if the last is a constant.
5784 2017-09-04 Marek Polacek <polacek@redhat.com>
5787 * convert.c (do_narrow): When sanitizing signed integer overflows,
5788 bail out for signed types.
5789 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
5791 2017-09-04 Richard Biener <rguenther@suse.de>
5793 PR tree-optimization/82060
5794 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5795 Move devirtualization after stmt folding and before EH/AB/noreturn
5796 cleanup to get the stmt refs canonicalized. Use a bool instead
5797 of gimple_modified_p since that doesn't work for NOPs. Schedule
5798 NOPs generated by folding for removal.
5800 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5801 Alan Hayward <alan.hayward@arm.com>
5802 David Sherwood <david.sherwood@arm.com>
5804 * coretypes.h (pad_direction): New enum.
5805 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
5806 (FUNCTION_ARG_PADDING): Likewise.
5807 * target.def (function_arg_padding): New hook.
5808 * targhooks.h (default_function_arg_padding): Declare.
5809 * targhooks.c (default_function_arg_padding): New function.
5810 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
5811 (TARGET_FUNCTION_ARG_PADDING): ...this.
5812 * doc/tm.texi: Regenerate.
5813 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
5814 instead of direction.
5815 (compute_argument_addresses): Likewise.
5816 (load_register_parameters): Likewise.
5817 (emit_library_call_value_1): Likewise.
5818 (store_one_arg): Use targetm.calls.function_arg_padding instead
5819 of FUNCTION_ARG_PADDING.
5820 (must_pass_in_stack_var_size_or_pad): Likewise.
5821 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
5822 (emit_group_store): Likewise.
5823 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
5824 instead of FUNCTION_ARG_PADDING.
5825 (emit_push_insn): Likewise, and propagate enum change throughout
5827 * function.h (direction): Delete.
5828 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
5830 * function.c (assign_parm_find_stack_rtl): Likewise.
5831 (assign_parm_setup_block_p): Likewise.
5832 (assign_parm_setup_block): Likewise.
5833 (gimplify_parameters): Likewise.
5834 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
5835 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
5837 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
5838 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5839 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
5840 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
5841 (aarch64_function_arg_padding): ...this new function.
5842 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
5843 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5844 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
5845 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5846 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
5847 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5848 (arm_pad_arg_upward): Replace with...
5849 (arm_function_arg_padding): ...this new function.
5850 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
5852 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
5853 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
5854 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5855 (ia64_hpux_function_arg_padding): Replace with...
5856 (ia64_function_arg_padding): ...this new function. Use pad_direction
5857 instead of direction. Check for TARGET_HPUX.
5858 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
5859 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5860 (iq2000_function_arg_padding): New function.
5861 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
5862 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
5863 (mips_function_arg_padding): ...this new function.
5864 (mips_pad_reg_upward): Update accordingly.
5865 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5866 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
5867 targetm.calls.function_arg_padding.
5868 (FUNCTION_ARG_PADDING): Delete.
5869 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5870 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
5871 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5872 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
5873 (nios2_block_reg_padding): Return pad_direction instead of direction.
5874 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
5875 instead of direction.
5876 (nios2_function_arg_padding): Likewise. Make static.
5877 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5878 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
5879 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
5880 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
5881 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5882 (pa_function_arg_padding): Make static. Return pad_direction instead
5884 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
5885 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5886 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
5887 instead of direction. Use targetm.calls.function_arg_padding.
5888 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
5889 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5890 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
5891 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
5892 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
5894 (function_arg_padding): Rename to...
5895 (rs6000_function_arg_padding): ...this. Make static. Return
5896 pad_direction instead of direction.
5897 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5898 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
5899 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5900 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
5901 instead of direction. Use targetm.calls.function_arg_padding.
5902 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
5903 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5904 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
5905 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
5906 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5907 (function_arg_padding): Rename to...
5908 (rs6000_function_arg_padding): ...this. Make static. Return
5909 pad_direction instead of direction.
5910 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5911 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
5912 * config/s390/s390.c (s390_function_arg_padding): New function.
5913 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5914 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
5915 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
5916 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5917 (function_arg_padding): Rename to...
5918 (sparc_function_arg_padding): ...this. Make static. Return
5919 pad_direction instead of direction.
5920 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
5921 * config/spu/spu.c (spu_function_arg_padding): New function.
5922 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5923 * system.h (FUNCTION_ARG_PADDING): Poison.
5925 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5926 Alan Hayward <alan.hayward@arm.com>
5927 David Sherwood <david.sherwood@arm.com>
5929 * target.def (modes_tieable_p): New hook.
5930 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
5931 (TARGET_MODES_TIEABLE_P): ...this.
5932 * doc/tm.texi.in: Regenerate.
5933 * hooks.h (hook_bool_mode_mode_true): Declare.
5934 * hooks.c (hook_bool_mode_mode_true): New function.
5935 * combine.c (subst): Use targetm.modes_tieable_p instead of
5937 * dse.c (find_shift_sequence): Likewise.
5938 * expmed.c (extract_low_bits): Likewise.
5939 * lower-subreg.c: Include target.h.
5940 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
5942 * rtlanal.c (rtx_cost): Likewise.
5943 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
5944 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
5945 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
5946 (TARGET_MODES_TIEABLE_P): Redefine.
5947 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
5948 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
5949 (TARGET_MODES_TIEABLE_P): Redefine.
5950 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
5951 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
5952 (arc_modes_tieable_p): New function.
5953 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
5954 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
5955 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
5956 (arm_modes_tieable_p): Make static.
5957 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
5958 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
5959 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
5960 (TARGET_MODES_TIEABLE_P): Redefine.
5961 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
5962 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
5963 (TARGET_MODES_TIEABLE_P): Redefine.
5964 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
5965 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
5966 (cr16_modes_tieable_p): New function.
5967 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
5968 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
5969 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
5970 (TRULY_NOOP_TRUNCATION): Update comment.
5971 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
5972 (TRULY_NOOP_TRUNCATION): Update comment.
5973 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
5974 (frv_modes_tieable_p): New function.
5975 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
5976 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
5977 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
5978 (TARGET_MODES_TIEABLE_P): Redefine.
5979 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
5980 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
5981 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
5982 (TARGET_MODES_TIEABLE_P): Redefine.
5983 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
5984 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
5985 (ia64_modes_tieable_p): New function.
5986 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
5987 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
5988 (iq2000_modes_tieable_p): New function.
5989 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
5990 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
5991 (lm32_modes_tieable_p): New function.
5992 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
5993 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
5994 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
5995 (TARGET_MODES_TIEABLE_P): Redefine.
5996 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
5997 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
5998 (m32r_modes_tieable_p): New function.
5999 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6000 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6001 (m68k_modes_tieable_p): New function.
6002 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6003 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6004 (mcore_modes_tieable_p): New function.
6005 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6006 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6008 (TARGET_MODES_TIEABLE_P): Redefine.
6009 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6010 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6011 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6012 (TARGET_MODES_TIEABLE_P): Redefine.
6013 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6014 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6015 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6016 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6017 (mn10300_modes_tieable_p): ...this and make static.
6018 (TARGET_MODES_TIEABLE_P): Redefine.
6019 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6020 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6021 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6022 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6023 (msp430_modes_tieable_p): Make static.
6024 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6025 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6026 (TARGET_MODES_TIEABLE_P): Redefine.
6027 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6028 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6029 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6030 (TARGET_MODES_TIEABLE_P): Redefine.
6031 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6032 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6033 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6034 (TARGET_MODES_TIEABLE_P): Redefine.
6035 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6036 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6037 (pdp11_modes_tieable_p): New function.
6038 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6039 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6040 (rs6000_modes_tieable_p): New function.
6041 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6042 * config/powerpcspe/powerpcspe.md: Update comment.
6043 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6044 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6045 (TARGET_MODES_TIEABLE_P): Redefine.
6046 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6047 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6048 (rl78_modes_tieable_p): New function.
6049 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6050 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6051 (rs6000_modes_tieable_p): New function.
6052 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6053 * config/rs6000/rs6000.md: Update comment.
6054 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6055 * config/rx/rx.c (rx_modes_tieable_p): New function.
6056 (TARGET_MODES_TIEABLE_P): Redefine.
6057 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6058 * config/s390/s390.c (s390_modes_tieable_p): New function.
6059 (TARGET_MODES_TIEABLE_P): Redefine.
6060 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6061 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6062 (sh_modes_tieable_p): New function.
6063 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6064 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6065 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6066 (sparc_modes_tieable_p): Make static.
6067 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6068 * config/spu/spu.c (spu_modes_tieable_p): New function.
6069 (TARGET_MODES_TIEABLE_P): Redefine.
6070 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6071 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6072 (TARGET_MODES_TIEABLE_P): Redefine.
6073 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6074 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6075 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6076 * config/v850/v850.c (v850_modes_tieable_p): New function.
6077 (TARGET_MODES_TIEABLE_P): Redefine.
6078 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
6079 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
6080 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
6081 (visium_modes_tieable_p): New function.
6082 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
6083 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
6084 (xtensa_modes_tieable_p): New function.
6085 * system.h (MODES_TIEABLE_P): Poison.
6087 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6088 Alan Hayward <alan.hayward@arm.com>
6089 David Sherwood <david.sherwood@arm.com>
6091 * target.def (hard_regno_mode_ok): New hook.
6092 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
6093 (TARGET_HARD_REGNO_MODE_OK): ...this.
6094 * doc/tm.texi.in: Regenerate.
6095 * hooks.h (hook_bool_uint_mode_true): Declare.
6096 * hooks.c (hook_bool_uint_mode_true): New function.
6097 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
6099 * genpreds.c (write_insn_preds_c): Add an include of target.h.
6100 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
6101 instead of HARD_REGNO_MODE_OK.
6102 * caller-save.c: Include target.h.
6103 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
6105 * combine.c (can_combine_p): Likewise.
6106 (combinable_i3pat): Likewise.
6107 (can_change_dest_mode): Likewise.
6108 * expr.c (init_expr_target): Likewise.
6109 (convert_move): Likewise.
6110 (convert_modes): Likewise.
6111 * ira.c (setup_prohibited_class_mode_regs): Likewise.
6112 (setup_prohibited_mode_move_regs): Likewise.
6113 * ira.h (target_ira): Likewise.
6114 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6115 * lra-constraints.c (process_alt_operands): Likewise.
6116 (split_reg): Likewise.
6117 * recog.c (peep2_find_free_register): Likewise.
6118 * ree.c (combine_reaching_defs): Likewise.
6119 * regcprop.c (maybe_mode_change): Likewise.
6120 * reginfo.c (init_reg_sets_1): Likewise.
6121 (choose_hard_reg_mode): Likewise.
6122 (simplifiable_subregs): Likewise.
6123 * regrename.c (check_new_reg_p): Likewise.
6124 * reload.c (find_valid_class): Likewise.
6125 (find_valid_class_1): Likewise.
6126 (reload_inner_reg_of_subreg): Likewise.
6127 (push_reload): Likewise.
6128 (combine_reloads): Likewise.
6129 (find_dummy_reload): Likewise.
6130 (find_reloads): Likewise.
6131 * reload1.c (find_reg): Likewise.
6132 (set_reload_reg): Likewise.
6133 (allocate_reload_reg): Likewise.
6134 (choose_reload_regs): Likewise.
6135 (reload_adjust_reg_for_temp): Likewise.
6136 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
6137 (simplify_subreg_regno): Likewise.
6138 * sel-sched.c (init_regs_for_mode): Likewise.
6139 * varasm.c (make_decl_rtl): Likewise.
6140 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
6141 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
6143 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
6144 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
6145 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6146 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
6147 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
6148 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6149 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
6150 (arc_mode_class): Delete.
6151 (HARD_REGNO_MODE_OK): Delete.
6152 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6153 (arc_hard_regno_mode_ok): Rename old array to...
6154 (arc_hard_regno_mode_ok_modes): ...this.
6155 (arc_conditional_register_usage): Update accordingly.
6156 (arc_mode_class): Make static.
6157 (arc_hard_regno_mode_ok): New function.
6158 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
6159 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
6160 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6161 (arm_hard_regno_mode_ok): Make static.
6162 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
6164 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
6165 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
6166 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
6168 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6169 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
6170 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
6171 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
6172 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
6173 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6174 * config/bfin/predicates.md (valid_reg_operand): Use
6175 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
6176 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
6177 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
6178 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6179 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
6180 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
6181 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6182 (cr16_hard_regno_mode_ok): Make static and return a bool.
6183 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
6184 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6185 (cris_hard_regno_mode_ok): New function.
6186 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
6187 (epiphany_mode_class): Delete.
6188 (HARD_REGNO_MODE_OK): Delete.
6189 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
6190 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6191 (hard_regno_mode_ok): Rename to...
6192 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
6193 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
6194 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
6196 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
6197 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
6198 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6199 (frv_hard_regno_mode_ok): Make static and return a bool.
6200 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
6202 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
6203 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
6204 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
6205 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
6207 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6208 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
6209 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
6210 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
6212 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6213 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
6214 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6215 (ia64_hard_regno_mode_ok): New function.
6216 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
6217 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6218 (iq2000_hard_regno_mode_ok): New function.
6219 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
6220 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6221 (lm32_hard_regno_mode_ok): New function.
6222 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
6223 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
6224 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
6225 instead of HARD_REGNO_MODE_OK.
6226 (m32c_hard_regno_ok): Rename to...
6227 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
6228 (m32c_cannot_change_mode_class): Update accordingly.
6229 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6230 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
6231 (m32r_mode_class): Delete.
6232 (HARD_REGNO_MODE_OK): Delete.
6233 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6234 (m32r_hard_regno_mode_ok): Rename to...
6235 (m32r_hard_regno_modes): ...this.
6236 (m32r_mode_class): Make static.
6237 (m32r_hard_regno_mode_ok): New function.
6238 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
6239 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
6240 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6241 (m68k_hard_regno_mode_ok): Make static.
6242 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
6243 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6244 (mcore_hard_regno_mode_ok): New function.
6245 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
6246 (HARD_REGNO_MODE_OK): Delete.
6247 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
6249 (microblaze_hard_regno_mode_ok_p): ...this and make static.
6250 (microblaze_hard_regno_mode_ok): New function.
6251 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6252 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
6253 (mips_hard_regno_mode_ok): Delete.
6254 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
6255 (mips_hard_regno_mode_ok_p): ...this and make static.
6256 (mips_hard_regno_mode_ok_p): Rename to...
6257 (mips_hard_regno_mode_ok_uncached): ...this.
6258 (mips_hard_regno_mode_ok): New function.
6259 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
6260 of HARD_REGNO_MODE_OK.
6261 (mips_option_override): Update after above name changes.
6262 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6263 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
6264 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
6265 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
6266 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
6267 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6268 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
6269 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
6270 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
6271 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6272 (msp430_hard_regno_mode_ok): Make static and return a bool.
6273 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
6274 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
6275 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
6277 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6278 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
6279 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
6280 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
6281 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
6282 (PA_HARD_REGNO_MODE_OK): ...this
6283 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
6284 (PA_HARD_REGNO_MODE_OK): ...this.
6285 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6286 (pa_hard_regno_mode_ok): New function.
6287 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
6288 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6289 (pdp11_hard_regno_mode_ok): New function.
6290 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
6291 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
6293 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
6295 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6296 (rs6000_hard_regno_mode_ok): Rename to...
6297 (rs6000_hard_regno_mode_ok_uncached): ...this.
6298 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6299 (rs6000_hard_regno_mode_ok): New function.
6300 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
6301 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
6302 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
6303 (riscv_hard_regno_mode_ok): ...this and make static.
6304 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6305 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
6306 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
6307 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6308 (rl78_hard_regno_mode_ok): Make static and return bool.
6309 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
6310 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
6312 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
6313 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6314 (rs6000_hard_regno_mode_ok): Rename to...
6315 (rs6000_hard_regno_mode_ok_uncached): ...this.
6316 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6317 (rs6000_hard_regno_mode_ok): New function.
6318 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
6319 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
6320 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6321 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
6322 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
6323 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
6324 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6325 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
6326 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
6327 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6328 (sh_hard_regno_mode_ok): Make static.
6329 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
6330 instead of HARD_REGNO_MODE_OK.
6331 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
6332 (sparc_mode_class): Delete.
6333 (HARD_REGNO_MODE_OK): Delete.
6334 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6335 (hard_regno_mode_classes): Make static.
6336 (sparc_mode_class): Likewise.
6337 (sparc_hard_regno_mode_ok): New function.
6338 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
6339 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
6340 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
6342 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6343 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
6344 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
6345 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
6346 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
6347 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6348 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
6349 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
6350 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6351 (visium_hard_regno_mode_ok): New function.
6352 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
6353 instead of HARD_REGNO_MODE_OK.
6354 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
6355 (HARD_REGNO_MODE_OK): Delete.
6356 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
6357 (xtensa_hard_regno_mode_ok_p): ...this and make static.
6358 (xtensa_option_override): Update accordingly.
6359 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6360 (xtensa_hard_regno_mode_ok): New function.
6361 * system.h (HARD_REGNO_MODE_OK): Poison.
6363 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6364 Alan Hayward <alan.hayward@arm.com>
6365 David Sherwood <david.sherwood@arm.com>
6367 * target.def (hard_regno_call_part_clobbered): New hook.
6368 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
6369 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
6370 * doc/tm.texi: Regenerate.
6371 * hooks.h (hook_bool_uint_mode_false): Declare.
6372 * hooks.c (hook_bool_uint_mode_false): New function.
6373 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6374 * cselib.c (cselib_process_insn): Use
6375 targetm.hard_regno_call_part_clobbered instead of
6376 HARD_REGNO_CALL_PART_CLOBBERED.
6377 * ira-conflicts.c (ira_build_conflicts): Likewise.
6378 * ira-costs.c (ira_tune_allocno_costs): Likewise.
6379 * lra-constraints.c (need_for_call_save_p): Likewise.
6380 * lra-lives.c: Include target.h.
6381 (check_pseudos_live_through_calls): Use
6382 targetm.hard_regno_call_part_clobbered instead of
6383 HARD_REGNO_CALL_PART_CLOBBERED.
6384 * regcprop.c: Include target.h.
6385 (copyprop_hardreg_forward_1): Use
6386 targetm.hard_regno_call_part_clobbered instead of
6387 HARD_REGNO_CALL_PART_CLOBBERED.
6388 * reginfo.c (choose_hard_reg_mode): Likewise.
6389 * regrename.c (check_new_reg_p): Likewise.
6390 * reload.c (find_equiv_reg): Likewise.
6391 * reload1.c (emit_reload_insns): Likewise.
6392 * sched-deps.c (deps_analyze_insn): Likewise.
6393 * sel-sched.c (init_regs_for_mode): Likewise.
6394 (mark_unavailable_hard_regs): Likewise.
6395 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
6396 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6397 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
6399 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6400 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6401 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
6403 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
6405 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6406 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6407 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
6409 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6410 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6411 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
6413 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6414 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
6416 * config/powerpcspe/powerpcspe.c
6417 (rs6000_hard_regno_call_part_clobbered): New function.
6418 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6419 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6420 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
6422 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6423 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6424 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
6426 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6427 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6428 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
6430 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6431 Alan Hayward <alan.hayward@arm.com>
6432 David Sherwood <david.sherwood@arm.com>
6434 * rtl.h (subreg_memory_offset): Declare.
6435 * emit-rtl.c (subreg_memory_offset): New function.
6436 * expmed.c (store_bit_field_1): Use it.
6437 * expr.c (undefined_operand_subword_p): Likewise.
6438 * simplify-rtx.c (simplify_subreg): Likewise.
6440 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
6442 PR rtl-optimization/57448
6445 * optabs.c (expand_atomic_load): Place compiler memory barriers if
6446 using atomic_load pattern.
6447 (expand_atomic_store): Likewise.
6449 2017-09-04 Jakub Jelinek <jakub@redhat.com>
6452 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
6453 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
6454 handling. Use replace_call_with_value with NULL instead of
6455 gsi_replace, unlink_stmt_vdef and release_defs.
6457 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
6460 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
6462 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6465 * rtl.h (emit_library_call_value_1): Declare.
6466 (emit_library_call): Replace declaration with a series of overloads.
6467 Remove the parameter count argument.
6468 (emit_library_call_value): Likewise.
6469 * calls.c (emit_library_call_value_1): Make global. Replace varargs
6470 with an "rtx_mode_t *".
6471 (emit_library_call_value): Delete.
6472 (emit_library_call): Likewise.
6473 * asan.c (asan_emit_stack_protection): Update calls accordingly.
6474 (asan_emit_allocas_unpoison): Likewise.
6475 * builtins.c (expand_builtin_powi): Likewise.
6476 (expand_asan_emit_allocas_unpoison): Likewise.
6477 * cfgexpand.c (expand_main_function): Likewise.
6478 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
6479 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
6480 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
6481 * config/arm/arm.c (arm_trampoline_init): Likewise.
6482 (arm_call_tls_get_addr): Likewise.
6483 (arm_expand_divmod_libfunc): Likewise.
6484 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
6485 (smulsi3_highpart): Likewise.
6486 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
6487 (c6x_expand_compare): Likewise.
6488 (c6x_expand_movmem): Likewise.
6489 * config/frv/frv.c (frv_trampoline_init): Likewise.
6490 * config/i386/i386.c (ix86_trampoline_init): Likewise.
6491 (ix86_expand_divmod_libfunc): Likewise.
6492 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
6493 (ia64_expand_compare): Likewise.
6494 (ia64_profile_hook): Likewise.
6495 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
6496 (nonlocal_goto): Likewise.
6497 (restore_stack_nonlocal): Likewise.
6498 * config/m32r/m32r.c (block_move_call): Likewise.
6499 (m32r_trampoline_init): Likewise.
6500 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
6501 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
6502 (m68k_call_m68k_read_tp): Likewise.
6503 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
6504 (microblaze_expand_divide): Likewise.
6505 * config/mips/mips.h (mips_args): Likewise.
6506 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
6507 (MIPS_ICACHE_SYNC): Likewise.
6508 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
6509 (nios2_trampoline_init): Likewise.
6510 * config/pa/pa.c (hppa_tls_call): Likewise.
6511 (pa_trampoline_init): Likewise.
6512 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
6513 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
6514 (expand_strn_compare): Likewise.
6515 (rs6000_generate_compare): Likewise.
6516 (rs6000_expand_float128_convert): Likewise.
6517 (output_profile_hook): Likewise.
6518 (rs6000_trampoline_init): Likewise.
6519 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
6520 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
6521 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
6522 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
6523 (rs6000_generate_compare): Likewise.
6524 (rs6000_expand_float128_convert): Likewise.
6525 (output_profile_hook): Likewise.
6526 (rs6000_trampoline_init): Likewise.
6527 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
6528 * config/sh/sh.c (sh_trampoline_init): Likewise.
6529 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
6530 (sparc_emit_float_lib_cmp): Likewise.
6531 (sparc32_initialize_trampoline): Likewise.
6532 (sparc64_initialize_trampoline): Likewise.
6533 (sparc_profile_hook): Likewise.
6534 * config/spu/spu.c (ea_load_store): Likewise.
6535 * config/spu/spu.md (floatunssidf2): Likewise.
6536 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
6537 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
6538 * config/visium/visium.c (expand_block_move_4): Likewise.
6539 (expand_block_move_2): Likewise.
6540 (expand_block_move_1): Likewise.
6541 (expand_block_set_4): Likewise.
6542 (expand_block_set_2): Likewise.
6543 (expand_block_set_1): Likewise.
6544 (visium_trampoline_init): Likewise.
6545 (visium_profile_hook): Likewise.
6546 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
6547 (xtensa_setup_frame_addresses): Likewise.
6548 (xtensa_trampoline_init): Likewise.
6549 * except.c (sjlj_emit_function_enter): Likewise.
6550 (sjlj_emit_function_exit): Likewise.
6551 * explow.c (allocate_dynamic_stack_space): Likewise.
6552 (probe_stack_range): Likewise.
6553 * expr.c (convert_mode_scalar): Likewise.
6554 * optabs.c (expand_binop): Likewise.
6555 (expand_twoval_binop_libfunc): Likewise.
6556 (expand_unop): Likewise.
6557 (prepare_cmp_insn): Likewise.
6558 (prepare_float_lib_cmp): Likewise.
6559 (expand_float): Likewise.
6560 (expand_fix): Likewise.
6561 (expand_fixed_convert): Likewise.
6562 (maybe_emit_sync_lock_test_and_set): Likewise.
6563 (expand_atomic_compare_and_swap): Likewise.
6564 (expand_mem_thread_fence): Likewise.
6565 (expand_atomic_fetch_op): Likewise.
6567 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
6569 * doc/generic.texi (OpenACC): Adjust URL.
6570 * doc/invoke.texi (C Dialect Options): Ditto.
6572 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
6574 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
6575 predicate for operand 1. Add (m,<S>) constraint.
6576 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
6577 Prevent memory operand 1 with register operand 2.
6579 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
6581 PR rtl-optimization/82024
6582 * combine.c (try_combine): If the combination result is a PARALLEL,
6583 and we only need to retain the SET in there that would be placed
6584 at I2, check that we can place that at I3 instead, before doing so.
6586 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6589 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
6591 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
6592 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
6595 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
6596 Jeff Law <law@redhat.com>
6598 * varasm.c (bss_initializer_p): Do not put constants into .bss
6599 (categorize_decl_for_section): Handle bss_initializer_p returning
6600 false when DECL_INITIAL is NULL.
6602 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6605 * config/s390/s390.c (s390_can_inline_p): New function.
6607 2017-09-01 Jeff Law <law@redhat.com>
6609 PR tree-optimization/82052
6610 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
6611 Always initialize the returned slot after a hash table miss
6612 when INSERT is true.
6614 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
6616 * config/s390/s390.md (mem_signal_fence): Remove.
6617 * doc/md.texi (mem_signal_fence): Remove.
6618 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
6620 * target-insns.def (mem_signal_fence): Remove.
6622 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6625 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
6628 * asan.c (create_odr_indicator): Strip name encoding from assembler
6629 name before appending it after __odr_asan_.
6631 2017-09-01 Martin Liska <mliska@suse.cz>
6633 PR tree-optimization/82059
6634 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
6635 frequency only when an edge is redirected.
6637 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6639 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
6640 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
6641 (arc_conditional_register_usage): Remove ARC600 lp_count
6643 (arc_file_start): Emit Tag_ARC_CPU_variation.
6644 (arc_can_use_doloop_p): New conditions to use ZOLs.
6645 (hwloop_fail): New function.
6646 (hwloop_optimize): Likewise.
6647 (hwloop_pattern_reg): Likewise.
6648 (arc_doloop_hooks): New struct, to be used with reorg_loops.
6649 (arc_reorg_loops): New function, calls reorg_loops.
6650 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
6651 (arc600_corereg_hazard): Remove ZOL checking, case handled by
6653 (arc_loop_hazard): Remove function, functionality moved into
6655 (arc_hazard): Remove arc_loop_hazard call.
6656 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
6657 into hwloop_optimize.
6658 (arc_label_align): Remove ZOL handling.
6659 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
6660 * config/arc/arc.md (doloop_begin): Remove pattern.
6661 (doloop_begin_i): Likewise.
6662 (doloop_end_i): Likewise.
6663 (doloop_fallback): Likewise.
6664 (doloop_fallback_m): Likewise.
6665 (doloop_end): Reimplement expand.
6666 (arc_lp): New pattern for LP instruction.
6667 (loop_end): New pattern.
6668 (loop_fail): Likewise.
6669 (decrement_and_branch_until_zero): Likewise.
6670 * config/arc/arc.opt (mlpc-width): New option.
6671 * doc/invoke.texi (mlpc-width): Document option.
6673 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6675 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
6676 (arc_ccfsm_advance): Fix checking for delay slots.
6677 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
6679 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6681 * config/arc/arc.md (movqi_insn): Add stores to save constant long
6683 (movhi_insn): Update store instruction constraint which are saving
6684 6-bit short immediates.
6685 (movsi_insn): Consider also short scaled load operations.
6686 (zero_extendhisi2_i): Use Usd constraint instead of T.
6687 (extendhisi2_i): Add q constraint.
6688 (arc_clzsi2): Add type and length attributes.
6689 (arc_ctzsi2): Likewise.
6690 * config/arc/constraints.md (Usc): Update constraint, the
6691 assembler can parse two relocations for a single instruction.
6693 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6695 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
6696 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
6698 2017-08-31 Olivier Hainque <hainque@adacore.com>
6700 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
6701 match as powerpc-wrs-vxworks*.
6703 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
6705 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
6706 register constraint for by-element operand.
6707 (aarch64_mls_elt_merge<mode>): Likewise.
6709 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6711 * config/arc/arc.c (arc_can_follow_jump): Check for short
6714 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6716 * config.gcc: Use g.opt for arc.
6717 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
6718 functionality moved to ...
6719 (legitimate_scaled_address_p): New function, ...here.
6720 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
6721 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
6722 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
6724 (arc_override_options): Handle G option.
6725 (arc_output_pic_addr_const): Correct function definition.
6726 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
6727 (arc_decl_anon_ns_mem_p): Delete.
6728 (arc_in_small_data_p): Overhaul this function to take into
6729 consideration the value given via G option.
6730 (arc_rewrite_small_data_1): Renamed and corrected old
6731 arc_rewrite_small_data function.
6732 (arc_rewrite_small_data): New function.
6733 (small_data_pattern): Don't use pic_offset_table_rtx.
6734 * config/arc/arc.h (CC1_SPEC): Recognize G option.
6735 * config/arc/simdext.md (movmisalignv2hi): Use
6736 prepare_move_operands function.
6738 (movmisalign*): Likewise.
6739 * doc/invoke.texi (ARC options): Document -G option.
6741 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6743 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
6745 * config/arc/arc.c (arc_print_operand): Output scalled address for
6746 sdata whenever is possible.
6747 (arc_in_small_data_p): Allow sdata for 64bit datum when double
6748 load/stores are available.
6749 (compact_sda_memory_operand): Check for the alignment required by
6750 code density instructions.
6751 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
6753 * config/arc/constraints.md (Usd): Update constraint.
6754 (Us0): New constraint.
6755 (Usc): Update constraint.
6757 2017-08-31 Richard Biener <rguenther@suse.de>
6760 * dwarf2out.c (dwarf2out_early_global_decl): Process each
6763 2017-08-31 Tamar Christina <tamar.christina@arm.com>
6765 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
6766 Resize type_signature.
6768 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6769 Alan Hayward <alan.hayward@arm.com>
6770 David Sherwood <david.sherwood@arm.com>
6772 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
6773 subregs whose inner modes can be stored in GPRs.
6774 (aarch64_classify_index): Likewise.
6776 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6777 Alan Hayward <alan.hayward@arm.com>
6778 David Sherwood <david.sherwood@arm.com>
6780 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6781 (V_INT_EQUIV): ...this.
6782 (v_cmp_result): Rename to...
6783 (v_int_equiv): ...this.
6784 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
6785 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
6786 (copysign<mode>3): Likewise.
6787 (aarch64_simd_bsl<mode>_internal): Likewise.
6788 (aarch64_simd_bsl<mode>): Likewise.
6789 (vec_cmp<mode><mode>): Likewise.
6790 (vcond<mode><mode>): Likewise.
6791 (vcond<v_cmp_mixed><mode>): Likewise.
6792 (vcondu<mode><v_cmp_mixed>): Likewise.
6793 (aarch64_cm<optab><mode>): Likewise.
6794 (aarch64_cmtst<mode>): Likewise.
6795 (aarch64_fac<optab><mode>): Likewise.
6796 (vec_perm_const<mode>): Likewise.
6797 (vcond_mask_<mode><v_cmp_result>): Rename to...
6798 (vcond_mask_<mode><v_int_equiv>): ...this.
6799 (vec_cmp<mode><v_cmp_result>): Rename to...
6800 (vec_cmp<mode><v_int_equiv>): ...this.
6802 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6803 Alan Hayward <alan.hayward@arm.com>
6804 David Sherwood <david.sherwood@arm.com>
6806 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
6808 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
6809 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
6810 (UNSPEC_LD4_DREG): New unspecs.
6811 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
6812 (aarch64_ld2<mode>_dreg_be): Replace with...
6813 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
6815 (aarch64_ld3<mode>_dreg_le)
6816 (aarch64_ld3<mode>_dreg_be): Replace with...
6817 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
6819 (aarch64_ld4<mode>_dreg_le)
6820 (aarch64_ld4<mode>_dreg_be): Replace with...
6821 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
6824 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6826 PR tree-optimization/81987
6827 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
6828 insert an initializer in a location not dominated by the stride
6831 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6833 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
6834 on the entire header of the finally block in the fallthru case.
6836 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6838 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
6840 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
6842 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
6843 rs6000_emit_move_from_cr and call renamed function.
6844 (rs6000_emit_prologue): Call renamed functions.
6845 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
6846 movesi_from_cr, remove volatile CRs.
6848 2017-08-30 Jon Beniston <jon@beniston.com>
6849 Richard Biener <rguenther@suse.de>
6851 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
6852 of VECTOR_MODE_P check.
6853 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
6854 element vector types.
6856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6858 * df.h (df_read_modify_subreg_p): Remove in favor of...
6859 * rtl.h (read_modify_subreg_p): ...this new function. Take a
6860 const_rtx instead of an rtx.
6861 * cprop.c (local_cprop_find_used_regs): Update accordingly.
6862 * df-problems.c (df_word_lr_mark_ref): Likewise.
6863 * ira-lives.c (mark_pseudo_reg_live): Likewise.
6864 (mark_pseudo_reg_dead): Likewise.
6865 (mark_ref_dead): Likewise.
6866 * reginfo.c (init_subregs_of_mode): Likewise.
6867 * sched-deps.c (sched_analyze_1): Likewise.
6868 * df-scan.c (df_def_record_1): Likewise.
6869 (df_uses_record): Likewise.
6870 (df_read_modify_subreg_p): Remove in favor of...
6871 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
6872 const_rtx instead of an rtx.
6874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6875 Alan Hayward <alan.hayward@arm.com>
6876 David Sherwood <david.sherwood@arm.com>
6878 * rtl.h (partial_subreg_p): New function.
6879 * caller-save.c (save_call_clobbered_regs): Use it.
6880 * calls.c (expand_call): Likewise.
6881 * combine.c (combinable_i3pat): Likewise.
6882 (simplify_set): Likewise.
6883 (make_extraction): Likewise.
6884 (make_compound_operation_int): Likewise.
6885 (gen_lowpart_or_truncate): Likewise.
6886 (force_to_mode): Likewise.
6887 (make_field_assignment): Likewise.
6888 (reg_truncated_to_mode): Likewise.
6889 (record_truncated_value): Likewise.
6890 (move_deaths): Likewise.
6891 * cse.c (record_jump_cond): Likewise.
6892 (cse_insn): Likewise.
6893 * cselib.c (cselib_lookup_1): Likewise.
6894 * expmed.c (extract_bit_field_using_extv): Likewise.
6895 * function.c (assign_parm_setup_reg): Likewise.
6896 * ifcvt.c (noce_convert_multiple_sets): Likewise.
6897 * ira-build.c (create_insn_allocnos): Likewise.
6898 * lra-coalesce.c (merge_pseudos): Likewise.
6899 * lra-constraints.c (match_reload): Likewise.
6900 (simplify_operand_subreg): Likewise.
6901 (curr_insn_transform): Likewise.
6902 * lra-lives.c (process_bb_lives): Likewise.
6903 * lra.c (new_insn_reg): Likewise.
6904 (lra_substitute_pseudo): Likewise.
6905 * regcprop.c (mode_change_ok): Likewise.
6906 (maybe_mode_change): Likewise.
6907 (copyprop_hardreg_forward_1): Likewise.
6908 * reload.c (push_reload): Likewise.
6909 (find_reloads): Likewise.
6910 (find_reloads_subreg_address): Likewise.
6911 * reload1.c (alter_reg): Likewise.
6912 (eliminate_regs_1): Likewise.
6913 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6915 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
6917 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
6918 back to if statements, including unpack.
6920 2017-08-30 Martin Liska <mliska@suse.cz>
6923 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
6925 (func_checker::compare_asm_inputs_outputs): ... this function.
6926 (func_checker::compare_gimple_asm): Use the function to compare
6927 also ASM constrains.
6928 * ipa-icf-gimple.h: Rename the function.
6930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6931 Alan Hayward <alan.hayward@arm.com>
6932 David Sherwood <david.sherwood@arm.com>
6934 * coretypes.h (complex_mode): New type.
6935 * gdbhooks.py (build_pretty_printer): Handle it.
6936 * machmode.h (complex_mode): New class.
6937 (complex_mode::includes_p): New function.
6938 (is_complex_int_mode): Likewise.
6939 (is_complex_float_mode): Likewise.
6940 * genmodes.c (get_mode_class): Handle complex mode classes.
6941 * function.c (expand_function_end): Use is_complex_int_mode.
6943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6944 Alan Hayward <alan.hayward@arm.com>
6945 David Sherwood <david.sherwood@arm.com>
6947 * coretypes.h (scalar_mode_pod): New typedef.
6948 * gdbhooks.py (build_pretty_printer): Handle it.
6949 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
6950 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
6951 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
6952 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
6955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6956 Alan Hayward <alan.hayward@arm.com>
6957 David Sherwood <david.sherwood@arm.com>
6959 * machmode.h (mode_for_vector): Take a scalar_mode instead
6961 * stor-layout.c (mode_for_vector): Likewise.
6962 * explow.c (promote_mode): Use as_a <scalar_mode>.
6963 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
6965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6966 Alan Hayward <alan.hayward@arm.com>
6967 David Sherwood <david.sherwood@arm.com>
6969 * target.def (preferred_simd_mode): Take a scalar_mode
6970 instead of a machine_mode.
6971 * targhooks.h (default_preferred_simd_mode): Likewise.
6972 * targhooks.c (default_preferred_simd_mode): Likewise.
6973 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
6974 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
6975 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
6976 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
6977 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
6978 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
6979 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
6980 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
6982 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
6983 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
6984 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
6985 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
6986 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
6987 * doc/tm.texi: Regenerate.
6988 * optabs-query.c (can_vec_mask_load_store_p): Return false for
6991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6992 Alan Hayward <alan.hayward@arm.com>
6993 David Sherwood <david.sherwood@arm.com>
6995 * target.def (scalar_mode_supported_p): Take a scalar_mode
6996 instead of a machine_mode.
6997 * targhooks.h (default_scalar_mode_supported_p): Likewise.
6998 * targhooks.c (default_scalar_mode_supported_p): Likewise.
6999 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7000 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7001 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7002 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7003 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7004 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7005 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7006 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7007 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7008 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7009 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7010 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7012 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7013 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7014 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7015 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7016 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7018 * doc/tm.texi: Regenerate.
7020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7021 Alan Hayward <alan.hayward@arm.com>
7022 David Sherwood <david.sherwood@arm.com>
7024 * coretypes.h (opt_scalar_mode): New typedef.
7025 * gdbhooks.py (build_pretty_printers): Handle it.
7026 * machmode.h (mode_iterator::get_2xwider): Add overload for
7028 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7030 * expr.c (convert_mode_scalar): Likewise.
7031 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7032 * optabs.c (expand_float): Likewise.
7033 (expand_fix): Likewise.
7034 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7037 Alan Hayward <alan.hayward@arm.com>
7038 David Sherwood <david.sherwood@arm.com>
7040 * optabs.c (expand_float): Explicitly check for scalars before
7041 using a branching expansion.
7042 (expand_fix): Likewise.
7044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7045 Alan Hayward <alan.hayward@arm.com>
7046 David Sherwood <david.sherwood@arm.com>
7048 * expr.c (convert_mode): Split scalar handling out into...
7049 (convert_mode_scalar): ...this new function. Treat the modes
7052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7053 Alan Hayward <alan.hayward@arm.com>
7054 David Sherwood <david.sherwood@arm.com>
7056 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7058 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7061 Alan Hayward <alan.hayward@arm.com>
7062 David Sherwood <david.sherwood@arm.com>
7064 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7065 rather than a machine_mode.
7066 (fixed_from_string): Likewise.
7067 (fixed_convert): Likewise.
7068 (fixed_convert_from_int): Likewise.
7069 (fixed_convert_from_real): Likewise.
7070 (real_convert_from_fixed): Likewise.
7071 * fixed-value.c (fixed_from_double_int): Likewise.
7072 (fixed_from_string): Likewise.
7073 (fixed_convert): Likewise.
7074 (fixed_convert_from_int): Likewise.
7075 (fixed_convert_from_real): Likewise.
7076 (real_convert_from_fixed): Likewise.
7077 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
7079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7080 Alan Hayward <alan.hayward@arm.com>
7081 David Sherwood <david.sherwood@arm.com>
7083 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
7084 of separate mode class checks. Do not allow vector modes here.
7085 (immed_wide_int_const): Use as_a <scalar_mode>.
7086 * explow.c (trunc_int_for_mode): Likewise.
7087 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
7088 (wi::shwi): Likewise.
7089 (wi::min_value): Likewise.
7090 (wi::max_value): Likewise.
7091 * dwarf2out.c (loc_descriptor): Likewise.
7092 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
7095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7096 Alan Hayward <alan.hayward@arm.com>
7097 David Sherwood <david.sherwood@arm.com>
7099 * tree.h (SCALAR_TYPE_MODE): New macro.
7100 * expr.c (expand_expr_addr_expr_1): Use it.
7101 (expand_expr_real_2): Likewise.
7102 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
7103 (fold_convert_const_fixed_from_int): Likewise.
7104 (fold_convert_const_fixed_from_real): Likewise.
7105 (native_encode_fixed): Likewise
7106 (native_encode_complex): Likewise
7107 (native_encode_vector): Likewise.
7108 (native_interpret_fixed): Likewise.
7109 (native_interpret_real): Likewise.
7110 (native_interpret_complex): Likewise.
7111 (native_interpret_vector): Likewise.
7112 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
7113 (simd_clone_adjust_argument_types): Likewise.
7114 (simd_clone_init_simd_arrays): Likewise.
7115 (simd_clone_adjust): Likewise.
7116 * stor-layout.c (layout_type): Likewise.
7117 * tree.c (build_minus_one_cst): Likewise.
7118 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7119 * tree-inline.c (estimate_move_cost): Likewise.
7120 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
7121 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
7122 (vectorizable_reduction): Likewise.
7123 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
7124 (vect_recog_mixed_size_cond_pattern): Likewise.
7125 (check_bool_pattern): Likewise.
7126 (adjust_bool_pattern): Likewise.
7127 (search_type_for_mask_1): Likewise.
7128 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7129 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7130 (vectorizable_load): Likewise.
7131 (vectorizable_store): Likewise.
7132 * ubsan.c (ubsan_encode_value): Likewise.
7133 * varasm.c (output_constant): Likewise.
7135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7136 Alan Hayward <alan.hayward@arm.com>
7137 David Sherwood <david.sherwood@arm.com>
7139 * coretypes.h (scalar_mode): New class.
7140 * machmode.h (scalar_mode): Likewise.
7141 (scalar_mode::includes_p): New function.
7142 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
7143 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
7144 * genmodes.c (get_mode_class): Handle remaining scalar modes.
7145 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
7146 * expmed.c (store_bit_field_1): Likewise.
7147 (extract_bit_field_1): Likewise.
7148 * expr.c (write_complex_part): Likewise.
7149 (read_complex_part): Likewise.
7150 (emit_move_complex_push): Likewise.
7151 (expand_expr_real_2): Likewise.
7152 * function.c (assign_parm_setup_reg): Likewise.
7153 (assign_parms_unsplit_complex): Likewise.
7154 * optabs.c (expand_binop): Likewise.
7155 * rtlanal.c (subreg_get_info): Likewise.
7156 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7157 * varasm.c (output_constant_pool_2): Likewise.
7159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7160 Alan Hayward <alan.hayward@arm.com>
7161 David Sherwood <david.sherwood@arm.com>
7163 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7165 (expmed_mult_highpart_optab): Likewise.
7166 (expmed_mult_highpart): Likewise.
7168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7169 Alan Hayward <alan.hayward@arm.com>
7170 David Sherwood <david.sherwood@arm.com>
7172 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
7173 instead of a machine_mode.
7174 (builtin_memset_read_str): Likewise.
7175 * builtins.c (c_readstr): Likewise.
7176 (builtin_memcpy_read_str): Likewise.
7177 (builtin_strncpy_read_str): Likewise.
7178 (builtin_memset_read_str): Likewise.
7179 (builtin_memset_gen_str): Likewise.
7180 (expand_builtin_signbit): Use scalar_int_mode for local variables.
7181 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
7182 instead of a machine_mode.
7183 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
7185 (make_extraction): Likewise.
7186 (try_widen_shift_mode): Take and return scalar_int_modes instead
7188 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
7189 a scalar_int_mode instead of a machine_mode.
7190 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
7191 (avr_addr_space_pointer_mode): Likewise.
7192 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
7193 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
7194 (msp430_unwind_word_mode): Likewise.
7195 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
7196 (spu_addr_space_pointer_mode): Likewise.
7197 (spu_addr_space_address_mode): Likewise.
7198 (spu_libgcc_cmp_return_mode): Likewise.
7199 (spu_libgcc_shift_count_mode): Likewise.
7200 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
7201 (rl78_addr_space_pointer_mode): Likewise.
7202 (fl78_unwind_word_mode): Likewise.
7203 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
7205 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
7206 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
7207 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
7208 (mips_valid_pointer_mode): Likewise.
7209 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
7210 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
7211 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
7213 (ft32_addr_space_address_mode): Likewise.
7214 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
7215 scalar_int_mode instead of a machine_mode.
7216 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
7218 (m32c_addr_space_address_mode): Likewise.
7219 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
7220 (rs6000_eh_return_filter_mode): Likewise.
7221 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
7222 (rs6000_eh_return_filter_mode): Likewise.
7223 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
7224 (s390_libgcc_shift_count_mode): Likewise.
7225 (s390_unwind_word_mode): Likewise.
7226 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
7228 * target.def (mode_rep_extended): Likewise.
7229 (valid_pointer_mode): Likewise.
7230 (addr_space.valid_pointer_mode): Likewise.
7231 (eh_return_filter_mode): Return a scalar_int_mode rather than
7233 (libgcc_cmp_return_mode): Likewise.
7234 (libgcc_shift_count_mode): Likewise.
7235 (unwind_word_mode): Likewise.
7236 (addr_space.pointer_mode): Likewise.
7237 (addr_space.address_mode): Likewise.
7238 * doc/tm.texi: Regenerate.
7239 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
7241 (do_jump): Use scalar_int_mode for local variables.
7242 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
7243 rather than a machine_mode.
7244 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
7245 (scompare_loc_descriptor_wide): Likewise.
7246 (scompare_loc_descriptor_narrow): Likewise.
7247 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
7249 * except.c (sjlj_emit_dispatch_table): Likewise.
7250 (expand_builtin_eh_copy_values): Likewise.
7251 * explow.c (convert_memory_address_addr_space_1): Likewise.
7252 Take a scalar_int_mode rather than a machine_mode.
7253 (convert_memory_address_addr_space): Take a scalar_int_mode rather
7254 than a machine_mode.
7255 (memory_address_addr_space): Use scalar_int_mode for local variables.
7256 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
7257 rather than a machine_mode.
7258 * expmed.c (mask_rtx): Likewise.
7259 (init_expmed_one_conv): Likewise.
7260 (expand_mult_highpart_adjust): Likewise.
7261 (extract_high_half): Likewise.
7262 (expmed_mult_highpart_optab): Likewise.
7263 (expmed_mult_highpart): Likewise.
7264 (expand_smod_pow2): Likewise.
7265 (expand_sdiv_pow2): Likewise.
7266 (emit_store_flag_int): Likewise.
7267 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
7269 (extract_low_bits): Likewise.
7270 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
7272 * expr.c (pieces_addr::adjust): Likewise.
7273 (can_store_by_pieces): Likewise.
7274 (store_by_pieces): Likewise.
7275 (clear_by_pieces_1): Likewise.
7276 (expand_expr_addr_expr_1): Likewise.
7277 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
7278 (expand_expr_real_1): Likewise.
7279 (try_casesi): Likewise.
7280 * final.c (shorten_branches): Likewise.
7281 * fold-const.c (fold_convert_const_int_from_fixed): Change the
7282 type of "mode" to machine_mode.
7283 * internal-fn.c (expand_arith_overflow_result_store): Take a
7284 scalar_int_mode rather than a machine_mode.
7285 (expand_mul_overflow): Use scalar_int_mode for local variables.
7286 * loop-doloop.c (doloop_modify): Likewise.
7287 (doloop_optimize): Likewise.
7288 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
7289 than a machine_mode.
7290 (expand_doubleword_shift_condmove): Likewise.
7291 (expand_doubleword_shift): Likewise.
7292 (expand_doubleword_clz): Likewise.
7293 (expand_doubleword_popcount): Likewise.
7294 (expand_doubleword_parity): Likewise.
7295 (expand_absneg_bit): Use scalar_int_mode for local variables.
7296 (prepare_float_lib_cmp): Likewise.
7297 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
7298 rather than a machine_mode.
7299 (convert_memory_address_addr_space): Likewise.
7300 (get_mode_bounds): Likewise.
7301 (get_address_mode): Return a scalar_int_mode rather than a
7303 * rtlanal.c (get_address_mode): Likewise.
7304 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
7305 than a machine_mode.
7306 * targhooks.c (default_mode_rep_extended): Likewise.
7307 (default_valid_pointer_mode): Likewise.
7308 (default_addr_space_valid_pointer_mode): Likewise.
7309 (default_eh_return_filter_mode): Return a scalar_int_mode rather
7310 than a machine_mode.
7311 (default_libgcc_cmp_return_mode): Likewise.
7312 (default_libgcc_shift_count_mode): Likewise.
7313 (default_unwind_word_mode): Likewise.
7314 (default_addr_space_pointer_mode): Likewise.
7315 (default_addr_space_address_mode): Likewise.
7316 * targhooks.h (default_eh_return_filter_mode): Likewise.
7317 (default_libgcc_cmp_return_mode): Likewise.
7318 (default_libgcc_shift_count_mode): Likewise.
7319 (default_unwind_word_mode): Likewise.
7320 (default_addr_space_pointer_mode): Likewise.
7321 (default_addr_space_address_mode): Likewise.
7322 (default_mode_rep_extended): Take a scalar_int_mode rather than
7324 (default_valid_pointer_mode): Likewise.
7325 (default_addr_space_valid_pointer_mode): Likewise.
7326 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
7328 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
7329 rather than a machine_mode.
7330 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
7331 for local variables.
7332 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7333 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
7334 than a machine_mode.
7336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7337 Alan Hayward <alan.hayward@arm.com>
7338 David Sherwood <david.sherwood@arm.com>
7340 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
7341 the mode argument to scalar_int_mode.
7342 (do_jump_by_parts_zero_rtx): Likewise.
7343 (do_jump_by_parts_equality_rtx): Likewise.
7344 (do_jump_by_parts_greater): Take a mode argument.
7345 (do_jump_by_parts_equality): Likewise.
7346 (do_jump_1): Update calls accordingly.
7348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7349 Alan Hayward <alan.hayward@arm.com>
7350 David Sherwood <david.sherwood@arm.com>
7352 * is-a.h (safe_dyn_cast): New function.
7353 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
7354 (jump_table_for_label): Likewise.
7355 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
7356 instead of an rtx_insn *.
7357 (shorten_branches): Use dyn_cast instead of LABEL_P and
7358 JUMP_TABLE_DATA_P. Use jump_table_for_label and
7359 rtx_jump_table_data::get_data_mode.
7360 (final_scan_insn): Likewise.
7362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7363 Alan Hayward <alan.hayward@arm.com>
7364 David Sherwood <david.sherwood@arm.com>
7366 * combine.c (try_combine): Use is_a <scalar_int_mode> when
7367 trying to combine a full-register integer set with a subreg
7370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7371 Alan Hayward <alan.hayward@arm.com>
7372 David Sherwood <david.sherwood@arm.com>
7374 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
7375 that is always either address_mode or pointer_mode.
7377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7378 Alan Hayward <alan.hayward@arm.com>
7379 David Sherwood <david.sherwood@arm.com>
7381 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
7382 when the two are known to be equal.
7384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7385 Alan Hayward <alan.hayward@arm.com>
7386 David Sherwood <david.sherwood@arm.com>
7388 * simplify-rtx.c (simplify_const_unary_operation): Use
7389 is_a <scalar_int_mode> instead of checking for a nonzero
7390 precision. Forcibly convert op_mode to a scalar_int_mode
7391 in that case. More clearly differentiate the operand and
7392 result modes and use the former when deciding what the value
7393 of a count-bits operation should be. Use is_int_mode instead
7394 of checking for a MODE_INT. Remove redundant check for whether
7395 this mode has a zero precision.
7397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7398 Alan Hayward <alan.hayward@arm.com>
7399 David Sherwood <david.sherwood@arm.com>
7401 * optabs.c (widen_leading): Change the type of the mode argument
7402 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
7403 (widen_bswap): Likewise.
7404 (expand_parity): Likewise.
7405 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
7406 (expand_ffs): Likewise.
7407 (epand_unop): Check for scalar integer modes before calling the
7410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7411 Alan Hayward <alan.hayward@arm.com>
7412 David Sherwood <david.sherwood@arm.com>
7414 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
7416 (expand_expr_real_1): Update call accordingly.
7418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7419 Alan Hayward <alan.hayward@arm.com>
7420 David Sherwood <david.sherwood@arm.com>
7422 * expmed.c (store_bit_field_using_insv): Add op0_mode and
7423 value_mode arguments. Use scalar_int_mode internally.
7424 (store_bit_field_1): Rename the new integer mode from imode
7425 to op0_mode and use it instead of GET_MODE (op0). Update calls
7426 to store_split_bit_field, store_bit_field_using_insv and
7427 store_fixed_bit_field.
7428 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
7429 Use scalar_int_mode internally. Use a bit count rather than a mode
7430 when calculating the largest bit size for get_best_mode.
7431 Update calls to store_split_bit_field and store_fixed_bit_field_1.
7432 (store_fixed_bit_field_1): Add mode and value_mode arguments.
7433 Remove assertion that OP0 has a scalar integer mode.
7434 (store_split_bit_field): Add op0_mode and value_mode arguments.
7435 Update calls to extract_fixed_bit_field.
7436 (extract_bit_field_using_extv): Add an op0_mode argument.
7437 Use scalar_int_mode internally.
7438 (extract_bit_field_1): Rename the new integer mode from imode to
7439 op0_mode and use it instead of GET_MODE (op0). Update calls to
7440 extract_split_bit_field, extract_bit_field_using_extv and
7441 extract_fixed_bit_field.
7442 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
7443 to extract_split_bit_field and extract_fixed_bit_field_1.
7444 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
7445 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
7447 (extract_split_bit_field): Add an op0_mode argument. Update call
7448 to extract_fixed_bit_field.
7450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7451 Alan Hayward <alan.hayward@arm.com>
7452 David Sherwood <david.sherwood@arm.com>
7454 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
7455 * explow.c (hard_function_value): Likewise.
7456 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
7457 convert_to_mode call outside the loop.
7458 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
7459 for the mode iterator. Require the mode specified by max_pieces
7461 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
7463 (copy_blkmode_to_reg): Likewise.
7464 (set_storage_via_setmem): Likewise.
7465 * optabs.c (prepare_cmp_insn): Likewise.
7466 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7467 * stor-layout.c (finish_bitfield_representative): Likewise.
7469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7470 Alan Hayward <alan.hayward@arm.com>
7471 David Sherwood <david.sherwood@arm.com>
7473 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
7474 * expr.c (convert_move): Use them.
7475 (convert_modes): Likewise.
7476 (store_expr_with_bounds): Likewise.
7478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7479 Alan Hayward <alan.hayward@arm.com>
7480 David Sherwood <david.sherwood@arm.com>
7482 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
7483 parameter for the mode of "x". Remove the "known_x", "known_mode"
7484 and "known_ret" arguments. Change the type of the mode argument
7486 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
7487 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
7488 (reg_num_sign_bit_copies_for_combine): Likewise.
7489 * rtlanal.c (nonzero_bits1): Likewise.
7490 (num_sign_bit_copies1): Likewise.
7491 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
7492 (reg_num_sign_bit_copies_general): Likewise.
7493 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
7494 (reg_nonzero_bits_general): Likewise.
7496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7497 Alan Hayward <alan.hayward@arm.com>
7498 David Sherwood <david.sherwood@arm.com>
7500 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
7501 than in subroutines. Return 1 for non-integer modes.
7502 (cached_num_sign_bit_copies): Change the type of the mode parameter
7504 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
7505 classes. Handle CONST_INT_P first and then check whether X also
7506 has a scalar integer mode. Check the same thing for inner registers
7507 of a SUBREG and for values that are being extended or truncated.
7509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7510 Alan Hayward <alan.hayward@arm.com>
7511 David Sherwood <david.sherwood@arm.com>
7513 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
7514 in subroutines. Return the mode mask for non-integer modes.
7515 (cached_nonzero_bits): Change the type of the mode parameter
7517 (nonzero_bits1): Likewise. Remove early exit for other mode
7518 classes. Handle CONST_INT_P first and then check whether X
7519 also has a scalar integer mode.
7521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7522 Alan Hayward <alan.hayward@arm.com>
7523 David Sherwood <david.sherwood@arm.com>
7525 * expr.c (widest_int_mode_for_size): Make the comment match the code.
7526 Return a scalar_int_mode and assert that the size is greater than
7528 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
7529 (op_by_pieces_d::op_by_pieces_d): Likewise.
7530 (op_by_pieces_d::run): Likewise.
7531 (can_store_by_pieces): Likewise.
7533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7534 Alan Hayward <alan.hayward@arm.com>
7535 David Sherwood <david.sherwood@arm.com>
7537 * combine.c (extract_left_shift): Add a mode argument and update
7539 (make_compound_operation_int): Change the type of the mode parameter
7540 to scalar_int_mode and update the call to extract_left_shift.
7542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7543 Alan Hayward <alan.hayward@arm.com>
7544 David Sherwood <david.sherwood@arm.com>
7546 * combine.c (simplify_and_const_int): Change the type of the mode
7547 parameter to scalar_int_mode.
7548 (simplify_and_const_int_1): Likewise. Update recursive call.
7550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7551 Alan Hayward <alan.hayward@arm.com>
7552 David Sherwood <david.sherwood@arm.com>
7554 * combine.c (simplify_compare_const): Check that the mode is a
7555 scalar_int_mode (rather than VOIDmode) before testing its
7557 (simplify_comparison): Move COMPARISON_P handling out of the
7558 loop and restrict the latter part of the loop to scalar_int_modes.
7559 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
7560 and when considering SUBREG_REGs. Use is_int_mode instead of
7561 checking GET_MODE_CLASS against MODE_INT.
7563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7564 Alan Hayward <alan.hayward@arm.com>
7565 David Sherwood <david.sherwood@arm.com>
7567 * combine.c (try_widen_shift_mode): Move check for equal modes to...
7568 (simplify_shift_const_1): ...here. Use scalar_int_mode for
7569 shift_unit_mode and for modes involved in scalar shifts.
7571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7572 Alan Hayward <alan.hayward@arm.com>
7573 David Sherwood <david.sherwood@arm.com>
7575 * combine.c (force_int_to_mode): New function, split out from...
7576 (force_to_mode): ...here. Keep xmode up-to-date and use it
7577 instead of GET_MODE (x).
7579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7580 Alan Hayward <alan.hayward@arm.com>
7581 David Sherwood <david.sherwood@arm.com>
7583 * optabs-query.h (extraction_insn::struct_mode): Change type to
7584 opt_scalar_int_mode and update comment.
7585 (extraction_insn::field_mode): Change type to scalar_int_mode.
7586 (extraction_insn::pos_mode): Likewise.
7587 * combine.c (make_extraction): Update accordingly.
7588 * optabs-query.c (get_traditional_extraction_insn): Likewise.
7589 (get_optab_extraction_insn): Likewise.
7590 * recog.c (simplify_while_replacing): Likewise.
7591 * expmed.c (narrow_bit_field_mem): Change the type of the mode
7592 parameter to opt_scalar_int_mode.
7594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7595 Alan Hayward <alan.hayward@arm.com>
7596 David Sherwood <david.sherwood@arm.com>
7598 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
7599 to a scalar_int_mode instead of a machine_mode.
7600 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
7601 (get_best_mode): Return a boolean and use a pointer argument to store
7602 the selected mode. Replace the limit mode parameter with a bit limit.
7603 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
7604 for the values returned by bit_field_mode_iterator::next_mode.
7605 (store_bit_field): Update call to get_best_mode.
7606 (store_fixed_bit_field): Likewise.
7607 (extract_fixed_bit_field): Likewise.
7608 * expr.c (optimize_bitfield_assignment_op): Likewise.
7609 * fold-const.c (optimize_bit_field_compare): Likewise.
7610 (fold_truth_andor_1): Likewise.
7611 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
7612 Update for new type of m_mode.
7613 (get_best_mode): As above.
7615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7616 Alan Hayward <alan.hayward@arm.com>
7617 David Sherwood <david.sherwood@arm.com>
7619 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
7620 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
7621 (store_bit_field): Check is_a <scalar_int_mode> before calling
7622 strict_volatile_bitfield_p.
7623 (extract_bit_field): Likewise.
7625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7626 Alan Hayward <alan.hayward@arm.com>
7627 David Sherwood <david.sherwood@arm.com>
7629 * target.def (cstore_mode): Return a scalar_int_mode.
7630 * doc/tm.texi: Regenerate.
7631 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
7632 * targhooks.h (default_cstore_mode): Likewise.
7633 * targhooks.c (default_cstore_mode): Likewise, using a forced
7635 * expmed.c (emit_cstore): Expect the target of the cstore to be
7638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7639 Alan Hayward <alan.hayward@arm.com>
7640 David Sherwood <david.sherwood@arm.com>
7642 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
7644 (niter_desc): Likewise mode.
7645 (iv_analyze): Add a mode parameter.
7647 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
7648 and change its type to scalar_int_mode.
7649 * loop-iv.c: Update commentary at head of file.
7650 (iv_constant): Pass the mode paraeter before the rtx it describes
7651 and change its type to scalar_int_mode. Remove VOIDmode handling.
7652 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
7653 (iv_extend): Likewise.
7654 (shorten_into_mode): Likewise.
7655 (iv_add): Use scalar_int_mode.
7656 (iv_mult): Likewise.
7657 (iv_shift): Likewise.
7658 (canonicalize_iv_subregs): Likewise.
7659 (get_biv_step_1): Pass the outer_mode parameter before the rtx
7660 it describes and change its mode to scalar_int_mode. Also change
7661 the type of the returned inner_mode to scalar_int_mode.
7662 (get_biv_step): Likewise, turning outer_mode from a pointer
7663 into a direct parameter. Update call to get_biv_step_1.
7664 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
7665 iv_constant and get_biv_step.
7666 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
7667 and change its type to scalar_int_mode. Don't initialise iv->mode
7668 to VOIDmode and remove later checks for its still being VOIDmode.
7669 Update calls to iv_analyze_op and iv_analyze_expr. Check
7670 is_a <scalar_int_mode> when changing the mode under consideration.
7671 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
7672 Update call to iv_analyze_expr.
7673 (iv_analyze_op): Add a mode parameter. Reject subregs whose
7674 inner register is not also a scalar_int_mode. Update call to
7676 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
7677 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
7678 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
7679 separate mode class checks. Update calls to iv_analyze. Remove
7680 fix-up of VOIDmodes after iv_analyze_biv.
7681 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
7682 don't have a scalar_int_mode. Update call to biv_p.
7684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7685 Alan Hayward <alan.hayward@arm.com>
7686 David Sherwood <david.sherwood@arm.com>
7688 * cfgexpand.c (convert_debug_memory_address): Use
7689 as_a <scalar_int_mode>.
7690 * combine.c (expand_compound_operation): Likewise.
7691 (make_extraction): Likewise.
7692 (change_zero_ext): Likewise.
7693 (simplify_comparison): Likewise.
7694 * cse.c (cse_insn): Likewise.
7695 * dwarf2out.c (minmax_loc_descriptor): Likewise.
7696 (mem_loc_descriptor): Likewise.
7697 (loc_descriptor): Likewise.
7698 * expmed.c (init_expmed_one_mode): Likewise.
7699 (synth_mult): Likewise.
7700 (emit_store_flag_1): Likewise.
7701 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
7702 of a comparison with size.
7703 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
7704 (reduce_to_bit_field_precision): Likewise.
7705 * function.c (expand_function_end): Likewise.
7706 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
7707 * loop-doloop.c (doloop_modify): Likewise.
7708 * optabs.c (expand_binop): Likewise.
7709 (expand_unop): Likewise.
7710 (expand_copysign_absneg): Likewise.
7711 (prepare_cmp_insn): Likewise.
7712 (maybe_legitimize_operand): Likewise.
7713 * recog.c (const_scalar_int_operand): Likewise.
7714 * rtlanal.c (get_address_mode): Likewise.
7715 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7716 (simplify_cond_clz_ctz): Likewise.
7717 * tree-nested.c (get_nl_goto_field): Likewise.
7718 * tree.c (build_vector_type_for_mode): Likewise.
7719 * var-tracking.c (use_narrower_mode): Likewise.
7721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7722 Alan Hayward <alan.hayward@arm.com>
7723 David Sherwood <david.sherwood@arm.com>
7725 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
7726 * builtins.c (expand_builtin_signbit): Use it.
7727 * cfgexpand.c (expand_debug_expr): Likewise.
7728 * dojump.c (do_jump): Likewise.
7729 (do_compare_and_jump): Likewise.
7730 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
7731 * expmed.c (make_tree): Likewise.
7732 * expr.c (expand_expr_real_2): Likewise.
7733 (expand_expr_real_1): Likewise.
7734 (try_casesi): Likewise.
7735 * fold-const-call.c (fold_const_call_ss): Likewise.
7736 * fold-const.c (unextend): Likewise.
7737 (extract_muldiv_1): Likewise.
7738 (fold_single_bit_test): Likewise.
7739 (native_encode_int): Likewise.
7740 (native_encode_string): Likewise.
7741 (native_interpret_int): Likewise.
7742 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
7743 * internal-fn.c (expand_addsub_overflow): Likewise.
7744 (expand_neg_overflow): Likewise.
7745 (expand_mul_overflow): Likewise.
7746 (expand_arith_overflow): Likewise.
7747 * match.pd: Likewise.
7748 * stor-layout.c (layout_type): Likewise.
7749 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7750 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
7751 * tree-ssanames.c (get_range_info): Likewise.
7752 * tree-switch-conversion.c (array_value_type) Likewise.
7753 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
7754 (vect_recog_divmod_pattern): Likewise.
7755 (vect_recog_mixed_size_cond_pattern): Likewise.
7756 * tree-vrp.c (extract_range_basic): Likewise.
7757 (simplify_float_conversion_using_ranges): Likewise.
7758 * tree.c (int_fits_type_p): Likewise.
7759 * ubsan.c (instrument_bool_enum_load): Likewise.
7760 * varasm.c (mergeable_string_section): Likewise.
7761 (narrowing_initializer_constant_valid_p): Likewise.
7762 (output_constant): Likewise.
7764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7765 Alan Hayward <alan.hayward@arm.com>
7766 David Sherwood <david.sherwood@arm.com>
7768 * machmode.h (NARROWEST_INT_MODE): New macro.
7769 * expr.c (alignment_for_piecewise_move): Use it instead of
7770 GET_CLASS_NARROWEST_MODE (MODE_INT).
7771 (push_block): Likewise.
7772 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
7774 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7777 Alan Hayward <alan.hayward@arm.com>
7778 David Sherwood <david.sherwood@arm.com>
7780 * postreload.c (move2add_valid_value_p): Change the type of the
7781 mode parameter to scalar_int_mode.
7782 (move2add_use_add2_insn): Add a mode parameter and use it instead
7784 (move2add_use_add3_insn): Likewise.
7785 (reload_cse_move2add): Update accordingly.
7787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7788 Alan Hayward <alan.hayward@arm.com>
7789 David Sherwood <david.sherwood@arm.com>
7791 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
7793 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
7794 * optabs.c (expand_unop): Likewise.
7796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7797 Alan Hayward <alan.hayward@arm.com>
7798 David Sherwood <david.sherwood@arm.com>
7800 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
7801 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
7802 (popcount_loc_descriptor): Likewise.
7803 (bswap_loc_descriptor): Likewise.
7804 (rotate_loc_descriptor): Likewise.
7805 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
7806 calling the functions above.
7808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7809 Alan Hayward <alan.hayward@arm.com>
7810 David Sherwood <david.sherwood@arm.com>
7812 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
7814 (try_combine): Likewise.
7815 (simplify_if_then_else): Likewise.
7816 * cse.c (cse_insn): Likewise.
7817 * dwarf2out.c (mem_loc_descriptor): Likewise.
7818 * emit-rtl.c (gen_lowpart_common): Likewise.
7819 * simplify-rtx.c (simplify_truncation): Likewise.
7820 (simplify_binary_operation_1): Likewise.
7821 (simplify_const_relational_operation): Likewise.
7822 (simplify_ternary_operation): Likewise.
7823 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
7825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7826 Alan Hayward <alan.hayward@arm.com>
7827 David Sherwood <david.sherwood@arm.com>
7829 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
7830 * reload.c (push_reload): Likewise.
7831 (find_reloads): Likewise.
7833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7834 Alan Hayward <alan.hayward@arm.com>
7835 David Sherwood <david.sherwood@arm.com>
7837 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
7838 (make_compound_operation_int): Likewise.
7839 (change_zero_ext): Likewise.
7840 * expr.c (convert_move): Likewise.
7841 (convert_modes): Likewise.
7842 * fwprop.c (forward_propagate_subreg): Likewise.
7843 * loop-iv.c (get_biv_step_1): Likewise.
7844 * optabs.c (widen_operand): Likewise.
7845 * postreload.c (move2add_valid_value_p): Likewise.
7846 * recog.c (simplify_while_replacing): Likewise.
7847 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7848 (simplify_binary_operation_1): Likewise. Remove redundant
7849 mode equality check.
7851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7852 Alan Hayward <alan.hayward@arm.com>
7853 David Sherwood <david.sherwood@arm.com>
7855 * combine.c (combine_simplify_rtx): Add checks for
7856 is_a <scalar_int_mode>.
7857 (simplify_if_then_else): Likewise.
7858 (make_field_assignment): Likewise.
7859 (simplify_comparison): Likewise.
7860 * ifcvt.c (noce_try_bitop): Likewise.
7861 * loop-invariant.c (canonicalize_address_mult): Likewise.
7862 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7865 Alan Hayward <alan.hayward@arm.com>
7866 David Sherwood <david.sherwood@arm.com>
7868 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
7869 is_a <scalar_int_mode> instead of != BLKmode.
7871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7872 Alan Hayward <alan.hayward@arm.com>
7873 David Sherwood <david.sherwood@arm.com>
7875 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
7876 instead of != VOIDmode.
7877 * combine.c (if_then_else_cond): Likewise.
7878 (change_zero_ext): Likewise.
7879 * dwarf2out.c (mem_loc_descriptor): Likewise.
7880 (loc_descriptor): Likewise.
7881 * rtlanal.c (canonicalize_condition): Likewise.
7882 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
7884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7885 Alan Hayward <alan.hayward@arm.com>
7886 David Sherwood <david.sherwood@arm.com>
7888 * simplify-rtx.c (simplify_binary_operation_1): Use
7889 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
7891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7892 Alan Hayward <alan.hayward@arm.com>
7893 David Sherwood <david.sherwood@arm.com>
7895 * wide-int.h (int_traits<unsigned char>) New class.
7896 (int_traits<unsigned short>) Likewise.
7897 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
7898 Use GET_MODE_UNIT_PRECISION and remove redundant test for
7900 * combine.c (set_nonzero_bits_and_sign_copies): Use
7901 is_a <scalar_int_mode>.
7902 (find_split_point): Likewise.
7903 (combine_simplify_rtx): Likewise.
7904 (simplify_logical): Likewise.
7905 (expand_compound_operation): Likewise.
7906 (expand_field_assignment): Likewise.
7907 (make_compound_operation): Likewise.
7908 (extended_count): Likewise.
7909 (change_zero_ext): Likewise.
7910 (simplify_comparison): Likewise.
7911 * dwarf2out.c (scompare_loc_descriptor): Likewise.
7912 (ucompare_loc_descriptor): Likewise.
7913 (minmax_loc_descriptor): Likewise.
7914 (mem_loc_descriptor): Likewise.
7915 (loc_descriptor): Likewise.
7916 * expmed.c (init_expmed_one_mode): Likewise.
7917 * lra-constraints.c (lra_constraint_offset): Likewise.
7918 * optabs.c (prepare_libcall_arg): Likewise.
7919 * postreload.c (move2add_note_store): Likewise.
7920 * reload.c (operands_match_p): Likewise.
7921 * rtl.h (load_extend_op): Likewise.
7922 * rtlhooks.c (gen_lowpart_general): Likewise.
7923 * simplify-rtx.c (simplify_truncation): Likewise.
7924 (simplify_unary_operation_1): Likewise.
7925 (simplify_binary_operation_1): Likewise.
7926 (simplify_const_binary_operation): Likewise.
7927 (simplify_const_relational_operation): Likewise.
7928 (simplify_subreg): Likewise.
7929 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7930 * var-tracking.c (adjust_mems): Likewise.
7931 (prepare_call_arguments): Likewise.
7933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7934 Alan Hayward <alan.hayward@arm.com>
7935 David Sherwood <david.sherwood@arm.com>
7937 * machmode.h (is_int_mode): New fuction.
7938 * combine.c (find_split_point): Use it.
7939 (combine_simplify_rtx): Likewise.
7940 (simplify_if_then_else): Likewise.
7941 (simplify_set): Likewise.
7942 (simplify_shift_const_1): Likewise.
7943 (simplify_comparison): Likewise.
7944 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
7945 * cse.c (notreg_cost): Likewise.
7946 (cse_insn): Likewise.
7947 * cselib.c (cselib_lookup_1): Likewise.
7948 * dojump.c (do_jump_1): Likewise.
7949 (do_compare_rtx_and_jump): Likewise.
7950 * dse.c (get_call_args): Likewise.
7951 * dwarf2out.c (rtl_for_decl_init): Likewise.
7952 (native_encode_initializer): Likewise.
7953 * expmed.c (emit_store_flag_1): Likewise.
7954 (emit_store_flag): Likewise.
7955 * expr.c (convert_modes): Likewise.
7956 (store_field): Likewise.
7957 (expand_expr_real_1): Likewise.
7958 * fold-const.c (fold_read_from_constant_string): Likewise.
7959 * gimple-ssa-sprintf.c (get_format_string): Likewise.
7960 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
7961 * optabs.c (expand_binop): Likewise.
7962 (expand_unop): Likewise.
7963 (expand_abs_nojump): Likewise.
7964 (expand_one_cmpl_abs_nojump): Likewise.
7965 * simplify-rtx.c (mode_signbit_p): Likewise.
7966 (val_signbit_p): Likewise.
7967 (val_signbit_known_set_p): Likewise.
7968 (val_signbit_known_clear_p): Likewise.
7969 (simplify_relational_operation_1): Likewise.
7970 * tree.c (vector_type_mode): Likewise.
7972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7973 Alan Hayward <alan.hayward@arm.com>
7974 David Sherwood <david.sherwood@arm.com>
7976 * machmode.h (smallest_mode_for_size): Fix formatting.
7977 (smallest_int_mode_for_size): New function.
7978 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
7979 instead of smallest_mode_for_size.
7980 * combine.c (make_extraction): Likewise.
7981 * config/arc/arc.c (arc_expand_movmem): Likewise.
7982 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
7983 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7984 * config/s390/s390.c (s390_expand_insv): Likewise.
7985 * config/sparc/sparc.c (assign_int_registers): Likewise.
7986 * config/spu/spu.c (spu_function_value): Likewise.
7987 (spu_function_arg): Likewise.
7988 * coverage.c (get_gcov_type): Likewise.
7989 (get_gcov_unsigned_t): Likewise.
7990 * dse.c (find_shift_sequence): Likewise.
7991 * expmed.c (store_bit_field_1): Likewise.
7992 * expr.c (convert_move): Likewise.
7993 (store_field): Likewise.
7994 * internal-fn.c (expand_arith_overflow): Likewise.
7995 * optabs-query.c (get_best_extraction_insn): Likewise.
7996 * optabs.c (expand_twoval_binop_libfunc): Likewise.
7997 * stor-layout.c (layout_type): Likewise.
7998 (initialize_sizetypes): Likewise.
7999 * targhooks.c (default_get_mask_mode): Likewise.
8000 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8003 Alan Hayward <alan.hayward@arm.com>
8004 David Sherwood <david.sherwood@arm.com>
8006 * machmode.h (opt_mode::else_blk): New function.
8007 (int_mode_for_mode): Declare.
8008 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8009 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8011 * cfgexpand.c (expand_debug_expr): Likewise.
8012 * combine.c (gen_lowpart_or_truncate): Likewise.
8013 (gen_lowpart_for_combine): Likewise.
8014 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8015 * config/avr/avr.c (avr_to_int_mode): Likewise.
8016 (avr_out_plus_1): Likewise.
8017 (avr_out_plus): Likewise.
8018 (avr_out_round): Likewise.
8019 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8020 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8021 (s390_expand_vcond): Likewise.
8022 * config/spu/spu.c (spu_split_immediate): Likewise.
8023 (spu_expand_mov): Likewise.
8024 * dse.c (get_stored_val): Likewise.
8025 * expmed.c (store_bit_field_1): Likewise.
8026 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8028 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8029 (extract_low_bits): Likewise.
8030 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8031 handling rather than repeating the check.
8032 (emit_group_store): Likewise.
8033 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8034 * optabs.c (expand_absneg_bit): Likewise.
8035 (expand_copysign_absneg): Likewise.
8036 (expand_copysign_bit): Likewise.
8037 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8038 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8039 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8040 * var-tracking.c (prepare_call_arguments): Likewise.
8041 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8042 int_mode_for_mode instead of mode_for_size.
8043 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8046 Alan Hayward <alan.hayward@arm.com>
8047 David Sherwood <david.sherwood@arm.com>
8049 * machmode.h (int_mode_for_size): New function.
8050 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8051 instead of mode_for_size.
8052 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8054 * combine.c (expand_field_assignment): Use int_mode_for_size
8055 instead of mode_for_size.
8056 (make_extraction): Likewise.
8057 (simplify_shift_const_1): Likewise.
8058 (simplify_comparison): Likewise.
8059 * dojump.c (do_jump): Likewise.
8060 * dwarf2out.c (mem_loc_descriptor): Likewise.
8061 * emit-rtl.c (init_derived_machine_modes): Likewise.
8062 * expmed.c (flip_storage_order): Likewise.
8063 (convert_extracted_bit_field): Likewise.
8064 * expr.c (copy_blkmode_from_reg): Likewise.
8065 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8066 * internal-fn.c (expand_mul_overflow): Likewise.
8067 * lower-subreg.c (simple_move): Likewise.
8068 * optabs-libfuncs.c (init_optabs): Likewise.
8069 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8070 * tree.c (vector_type_mode): Likewise.
8071 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8072 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8073 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8074 * tree-vect-stmts.c (vectorizable_load): Likewise.
8075 (vectorizable_store): Likewise.
8077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8078 Alan Hayward <alan.hayward@arm.com>
8079 David Sherwood <david.sherwood@arm.com>
8081 * coretypes.h (pod_mode): New type.
8082 (scalar_int_mode_pod): New typedef.
8083 * machmode.h (pod_mode): New class.
8084 (int_n_data_t::m): Change type to scalar_int_mode_pod.
8085 * genmodes.c (emit_mode_int_n): Update accordingly.
8086 * lower-subreg.h (target_lower_subreg): Change type to
8087 scalar_int_mode_pod.
8088 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
8089 scalar_int_mode_pod.
8091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8092 Alan Hayward <alan.hayward@arm.com>
8093 David Sherwood <david.sherwood@arm.com>
8095 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
8096 machine_mode to scalar_int_mode.
8097 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
8098 (rs6000_option_override_internal): Remove cast to int.
8099 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
8100 machine_mode to scalar_int_mode.
8101 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
8102 (rs6000_option_override_internal): Remove cast to int.
8103 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
8104 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
8106 * config/s390/s390.c (s390_expand_builtin): Likewise.
8107 * coretypes.h (scalar_int_mode): New type.
8108 (opt_scalar_int_mode): New typedef.
8109 * machmode.h (scalar_int_mode): New class.
8110 (scalar_int_mode::includes_p): New function.
8111 (byte_mode): Change type to scalar_int_mode.
8112 (word_mode): Likewise.
8113 (ptr_mode): Likewise.
8114 * emit-rtl.c (byte_mode): Likewise.
8115 (word_mode): Likewise.
8116 (ptr_mode): Likewise.
8117 (init_derived_machine_modes): Update accordingly.
8118 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
8119 and MODE_PARTIAL_INT.
8120 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
8121 opt_scalar_int_mode.
8123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8124 Alan Hayward <alan.hayward@arm.com>
8125 David Sherwood <david.sherwood@arm.com>
8127 * target.def (libgcc_floating_mode_supported_p): Take a
8129 * doc/tm.texi: Regenerate.
8130 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
8132 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8133 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
8136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8137 Alan Hayward <alan.hayward@arm.com>
8138 David Sherwood <david.sherwood@arm.com>
8140 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
8141 * doc/tm.texi: Regenerate.
8142 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
8143 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
8144 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
8145 * targhooks.h (default_floatn_mode): Likewise.
8146 * targhooks.c (default_floatn_mode): Likewise.
8147 * tree.c (build_common_tree_nodes): Update accordingly.
8149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8150 Alan Hayward <alan.hayward@arm.com>
8151 David Sherwood <david.sherwood@arm.com>
8153 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
8154 (mode_iterator::iterate_p): Likewise.
8155 (mode_iterator::get_wider): Likewise.
8156 * expr.c (init_expr_target): Use opt_scalar_float_mode.
8158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8159 Alan Hayward <alan.hayward@arm.com>
8160 David Sherwood <david.sherwood@arm.com>
8162 * coretypes.h (opt_scalar_float_mode): New typedef.
8163 * machmode.h (float_mode_for_size): New function.
8164 * emit-rtl.c (double_mode): Delete.
8165 (init_emit_once): Use float_mode_for_size.
8166 * stor-layout.c (layout_type): Likewise.
8167 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
8169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8170 Alan Hayward <alan.hayward@arm.com>
8171 David Sherwood <david.sherwood@arm.com>
8173 * output.h (assemble_real): Take a scalar_float_mode.
8174 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
8175 * config/arm/arm.md (consttable_4): Likewise.
8176 (consttable_8): Likewise.
8177 (consttable_16): Likewise.
8178 * config/mips/mips.md (consttable_float): Likewise.
8179 * config/s390/s390.c (s390_output_pool_entry): Likewise.
8180 * varasm.c (assemble_real): Take a scalar_float_mode.
8181 (output_constant_pool_2): Update accordingly.
8182 (output_constant): Likewise.
8184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8185 Alan Hayward <alan.hayward@arm.com>
8186 David Sherwood <david.sherwood@arm.com>
8188 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
8189 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
8190 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
8191 (native_encode_real): Likewise.
8192 (native_interpret_real): Likewise.
8193 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
8194 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8197 Alan Hayward <alan.hayward@arm.com>
8198 David Sherwood <david.sherwood@arm.com>
8200 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
8201 <scalar_float_mode>. Simplify.
8202 (gen_extend_conv_libfunc): Likewise.
8204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8205 Alan Hayward <alan.hayward@arm.com>
8206 David Sherwood <david.sherwood@arm.com>
8208 * coretypes.h (scalar_float_mode): New type.
8209 * machmode.h (mode_traits::from_int): Use machine_mode if
8210 USE_ENUM_MODES is defined.
8211 (is_a): New function.
8213 (dyn_cast): Likewise.
8214 (scalar_float_mode): New class.
8215 (scalar_float_mode::includes_p): New function.
8216 (is_float_mode): Likewise.
8217 * gdbhooks.py (MachineModePrinter): New class.
8218 (build_pretty_printer): Use it for scalar_float_mode.
8219 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
8220 (format_helper::format_helper): Turn into a template.
8221 * genmodes.c (get_mode_class): New function.
8222 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
8223 or machine_mode if none.
8224 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
8225 as_a <scalar_float_mode>.
8226 * dwarf2out.c (mem_loc_descriptor): Likewise.
8227 (insert_float): Likewise.
8228 (add_const_value_attribute): Likewise.
8229 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8230 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
8231 (expand_unop): Update accordingly.
8232 (expand_abs_nojump): Likewise.
8233 (expand_copysign_absneg): Take a scalar_float_mode.
8234 (expand_copysign_bit): Likewise.
8235 (expand_copysign): Update accordingly.
8237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8238 Alan Hayward <alan.hayward@arm.com>
8239 David Sherwood <david.sherwood@arm.com>
8241 * coretypes.h (opt_mode): New class.
8242 * machmode.h (opt_mode): Likewise.
8243 (opt_mode::else_void): New function.
8244 (opt_mode::require): Likewise.
8245 (opt_mode::exists): Likewise.
8246 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
8247 (GET_MODE_2XWIDER_MODE): Likewise.
8248 (mode_iterator::get_wider): Update accordingly.
8249 (mode_iterator::get_2xwider): Likewise.
8250 (mode_iterator::get_known_wider): Likewise, turning into a template.
8251 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
8252 forcing a wider mode to exist.
8253 * config/cr16/cr16.h (LONG_REG_P): Likewise.
8254 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8255 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
8256 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
8257 * lower-subreg.c (init_lower_subreg): Likewise.
8258 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
8259 on the final iteration.
8260 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
8261 a wider mode exists before asking for a move pattern.
8262 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
8263 forcing a wider mode to exist.
8264 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
8265 returning false if no such mode exists.
8266 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
8267 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
8268 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
8269 Avoid checking for a MODE_INT if we already know the mode is not a
8271 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
8272 forcing a wider mode to exist.
8273 (expmed_mult_highpart_optab): Likewise.
8274 (expmed_mult_highpart): Likewise.
8275 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
8277 * lto-streamer-in.c (lto_input_mode_table): Likewise.
8278 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
8279 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8280 * internal-fn.c (expand_mul_overflow): Update use of
8281 GET_MODE_2XWIDER_MODE.
8282 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8283 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
8284 GET_MODE_WIDER_MODE.
8285 (convert_plusminus_to_widen): Likewise.
8286 * tree-switch-conversion.c (array_value_type): Likewise.
8287 * var-tracking.c (emit_note_insn_var_location): Likewise.
8288 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8289 Return false inside rather than outside the loop if no wider mode
8291 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
8292 and GET_MODE_2XWIDER_MODE
8293 (can_compare_p): Use else_void.
8294 * gdbhooks.py (OptMachineModePrinter): New class.
8295 (build_pretty_printer): Use it for opt_mode.
8297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8298 Alan Hayward <alan.hayward@arm.com>
8299 David Sherwood <david.sherwood@arm.com>
8301 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
8302 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
8304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8305 Alan Hayward <alan.hayward@arm.com>
8306 David Sherwood <david.sherwood@arm.com>
8308 * machmode.h (mode_traits): New structure.
8309 (get_narrowest_mode): New function.
8310 (mode_iterator::start): Likewise.
8311 (mode_iterator::iterate_p): Likewise.
8312 (mode_iterator::get_wider): Likewise.
8313 (mode_iterator::get_known_wider): Likewise.
8314 (mode_iterator::get_2xwider): Likewise.
8315 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
8316 (FOR_EACH_MODE): Likewise.
8317 (FOR_EACH_MODE_FROM): Likewise.
8318 (FOR_EACH_MODE_UNTIL): Likewise.
8319 (FOR_EACH_WIDER_MODE): Likewise.
8320 (FOR_EACH_2XWIDER_MODE): Likewise.
8321 * builtins.c (expand_builtin_strlen): Use new mode iterators.
8322 * combine.c (simplify_comparison): Likewise
8323 * config/i386/i386.c (type_natural_mode): Likewise.
8324 * cse.c (cse_insn): Likewise.
8325 * dse.c (find_shift_sequence): Likewise.
8326 * emit-rtl.c (init_derived_machine_modes): Likewise.
8327 (init_emit_once): Likewise.
8328 * explow.c (hard_function_value): Likewise.
8329 * expmed.c (extract_fixed_bit_field_1): Likewise.
8330 (extract_bit_field_1): Likewise.
8331 (expand_divmod): Likewise.
8332 (emit_store_flag_1): Likewise.
8333 * expr.c (init_expr_target): Likewise.
8334 (convert_move): Likewise.
8335 (alignment_for_piecewise_move): Likewise.
8336 (widest_int_mode_for_size): Likewise.
8337 (emit_block_move_via_movmem): Likewise.
8338 (copy_blkmode_to_reg): Likewise.
8339 (set_storage_via_setmem): Likewise.
8340 (compress_float_constant): Likewise.
8341 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8342 * optabs-query.c (get_best_extraction_insn): Likewise.
8343 * optabs.c (expand_binop): Likewise.
8344 (expand_twoval_unop): Likewise.
8345 (expand_twoval_binop): Likewise.
8346 (widen_leading): Likewise.
8347 (widen_bswap): Likewise.
8348 (expand_parity): Likewise.
8349 (expand_unop): Likewise.
8350 (prepare_cmp_insn): Likewise.
8351 (prepare_float_lib_cmp): Likewise.
8352 (expand_float): Likewise.
8353 (expand_fix): Likewise.
8354 (expand_sfix_optab): Likewise.
8355 * postreload.c (move2add_use_add2_insn): Likewise.
8356 * reg-stack.c (reg_to_stack): Likewise.
8357 * reginfo.c (choose_hard_reg_mode): Likewise.
8358 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8359 * stor-layout.c (mode_for_size): Likewise.
8360 (smallest_mode_for_size): Likewise.
8361 (mode_for_vector): Likewise.
8362 (finish_bitfield_representative): Likewise.
8363 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
8364 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
8365 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8366 * var-tracking.c (prepare_call_arguments): Likewise.
8368 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8369 Alan Hayward <alan.hayward@arm.com>
8370 David Sherwood <david.sherwood@arm.com>
8372 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
8373 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
8374 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
8375 * machmode.h (mode_size): Move earlier in file.
8376 (mode_precision): Likewise.
8377 (mode_inner): Likewise.
8378 (mode_nunits): Likewise.
8379 (mode_unit_size): Likewise.
8380 (unit_unit_precision): Likewise.
8381 (mode_wider): Likewise.
8382 (mode_2xwider): Likewise.
8383 (machine_mode): New class.
8384 (mode_to_bytes): New function.
8385 (mode_to_bits): Likewise.
8386 (mode_to_precision): Likewise.
8387 (mode_to_inner): Likewise.
8388 (mode_to_unit_size): Likewise.
8389 (mode_to_unit_precision): Likewise.
8390 (mode_to_nunits): Likewise.
8391 (GET_MODE_SIZE): Use mode_to_bytes.
8392 (GET_MODE_BITSIZE): Use mode_to_bits.
8393 (GET_MODE_PRECISION): Use mode_to_precision.
8394 (GET_MODE_INNER): Use mode_to_inner.
8395 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
8396 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
8397 (GET_MODE_NUNITS): Use mode_to_nunits.
8398 * system.h (ALWAYS_INLINE): New macro.
8399 * config/powerpcspe/powerpcspe-c.c
8400 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
8401 int for arg1_mode and arg2_mode.
8403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8404 Alan Hayward <alan.hayward@arm.com>
8405 David Sherwood <david.sherwood@arm.com>
8407 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8408 Prefix mode names with E_ in case statements.
8409 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8410 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
8411 (aarch64_split_simd_move): Likewise.
8412 (aarch64_gen_storewb_pair): Likewise.
8413 (aarch64_gen_loadwb_pair): Likewise.
8414 (aarch64_gen_store_pair): Likewise.
8415 (aarch64_gen_load_pair): Likewise.
8416 (aarch64_get_condition_code_1): Likewise.
8417 (aarch64_constant_pool_reload_icode): Likewise.
8418 (get_rsqrte_type): Likewise.
8419 (get_rsqrts_type): Likewise.
8420 (get_recpe_type): Likewise.
8421 (get_recps_type): Likewise.
8422 (aarch64_gimplify_va_arg_expr): Likewise.
8423 (aarch64_simd_container_mode): Likewise.
8424 (aarch64_emit_load_exclusive): Likewise.
8425 (aarch64_emit_store_exclusive): Likewise.
8426 (aarch64_expand_compare_and_swap): Likewise.
8427 (aarch64_gen_atomic_cas): Likewise.
8428 (aarch64_emit_bic): Likewise.
8429 (aarch64_emit_atomic_swap): Likewise.
8430 (aarch64_emit_atomic_load_op): Likewise.
8431 (aarch64_evpc_trn): Likewise.
8432 (aarch64_evpc_uzp): Likewise.
8433 (aarch64_evpc_zip): Likewise.
8434 (aarch64_evpc_ext): Likewise.
8435 (aarch64_evpc_rev): Likewise.
8436 (aarch64_evpc_dup): Likewise.
8437 (aarch64_gen_ccmp_first): Likewise.
8438 (aarch64_gen_ccmp_next): Likewise.
8439 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8440 (alpha_emit_xfloating_libcall): Likewise.
8441 (emit_insxl): Likewise.
8442 (alpha_arg_type): Likewise.
8443 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
8444 (arc_preferred_simd_mode): Likewise.
8445 (arc_secondary_reload): Likewise.
8446 (get_arc_condition_code): Likewise.
8447 (arc_print_operand): Likewise.
8448 (arc_legitimate_constant_p): Likewise.
8449 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8450 * config/arc/arc.md (casesi_load): Likewise.
8451 (casesi_compact_jump): Likewise.
8452 * config/arc/predicates.md (proper_comparison_operator): Likewise.
8453 (cc_use_register): Likewise.
8454 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8455 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8456 (arm_init_iwmmxt_builtins): Likewise.
8457 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
8458 (neon_expand_vector_init): Likewise.
8459 (arm_attr_length_move_neon): Likewise.
8460 (maybe_get_arm_condition_code): Likewise.
8461 (arm_emit_vector_const): Likewise.
8462 (arm_preferred_simd_mode): Likewise.
8463 (arm_output_iwmmxt_tinsr): Likewise.
8464 (thumb1_output_casesi): Likewise.
8465 (thumb2_output_casesi): Likewise.
8466 (arm_emit_load_exclusive): Likewise.
8467 (arm_emit_store_exclusive): Likewise.
8468 (arm_expand_compare_and_swap): Likewise.
8469 (arm_evpc_neon_vuzp): Likewise.
8470 (arm_evpc_neon_vzip): Likewise.
8471 (arm_evpc_neon_vrev): Likewise.
8472 (arm_evpc_neon_vtrn): Likewise.
8473 (arm_evpc_neon_vext): Likewise.
8474 (arm_validize_comparison): Likewise.
8475 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
8476 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
8477 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
8478 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
8479 (c6x_preferred_simd_mode): Likewise.
8480 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
8481 (epiphany_rtx_costs): Likewise.
8482 * config/epiphany/predicates.md (proper_comparison_operator):
8484 * config/frv/frv.c (condexec_memory_operand): Likewise.
8485 (frv_emit_move): Likewise.
8486 (output_move_single): Likewise.
8487 (output_condmove_single): Likewise.
8488 (frv_hard_regno_mode_ok): Likewise.
8489 (frv_matching_accg_mode): Likewise.
8490 * config/h8300/h8300.c (split_adds_subs): Likewise.
8491 (h8300_rtx_costs): Likewise.
8492 (h8300_print_operand): Likewise.
8493 (compute_mov_length): Likewise.
8494 (output_logical_op): Likewise.
8495 (compute_logical_op_length): Likewise.
8496 (compute_logical_op_cc): Likewise.
8497 (h8300_shift_needs_scratch_p): Likewise.
8498 (output_a_shift): Likewise.
8499 (compute_a_shift_length): Likewise.
8500 (compute_a_shift_cc): Likewise.
8501 (expand_a_rotate): Likewise.
8502 (output_a_rotate): Likewise.
8503 * config/i386/i386.c (classify_argument): Likewise.
8504 (function_arg_advance_32): Likewise.
8505 (function_arg_32): Likewise.
8506 (function_arg_64): Likewise.
8507 (function_value_64): Likewise.
8508 (ix86_gimplify_va_arg): Likewise.
8509 (ix86_legitimate_constant_p): Likewise.
8510 (put_condition_code): Likewise.
8511 (split_double_mode): Likewise.
8512 (ix86_avx256_split_vector_move_misalign): Likewise.
8513 (ix86_expand_vector_logical_operator): Likewise.
8514 (ix86_split_idivmod): Likewise.
8515 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
8516 (ix86_build_const_vector): Likewise.
8517 (ix86_build_signbit_mask): Likewise.
8518 (ix86_match_ccmode): Likewise.
8519 (ix86_cc_modes_compatible): Likewise.
8520 (ix86_expand_branch): Likewise.
8521 (ix86_expand_sse_cmp): Likewise.
8522 (ix86_expand_sse_movcc): Likewise.
8523 (ix86_expand_int_sse_cmp): Likewise.
8524 (ix86_expand_vec_perm_vpermi2): Likewise.
8525 (ix86_expand_vec_perm): Likewise.
8526 (ix86_expand_sse_unpack): Likewise.
8527 (ix86_expand_int_addcc): Likewise.
8528 (ix86_split_to_parts): Likewise.
8529 (ix86_vectorize_builtin_gather): Likewise.
8530 (ix86_vectorize_builtin_scatter): Likewise.
8531 (avx_vpermilp_parallel): Likewise.
8532 (inline_memory_move_cost): Likewise.
8533 (ix86_tieable_integer_mode_p): Likewise.
8534 (x86_maybe_negate_const_int): Likewise.
8535 (ix86_expand_vector_init_duplicate): Likewise.
8536 (ix86_expand_vector_init_one_nonzero): Likewise.
8537 (ix86_expand_vector_init_one_var): Likewise.
8538 (ix86_expand_vector_init_concat): Likewise.
8539 (ix86_expand_vector_init_interleave): Likewise.
8540 (ix86_expand_vector_init_general): Likewise.
8541 (ix86_expand_vector_set): Likewise.
8542 (ix86_expand_vector_extract): Likewise.
8543 (emit_reduc_half): Likewise.
8544 (ix86_emit_i387_round): Likewise.
8545 (ix86_mangle_type): Likewise.
8546 (ix86_expand_round_sse4): Likewise.
8547 (expand_vec_perm_blend): Likewise.
8548 (canonicalize_vector_int_perm): Likewise.
8549 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
8550 (expand_vec_perm_1): Likewise.
8551 (expand_vec_perm_interleave3): Likewise.
8552 (expand_vec_perm_even_odd_pack): Likewise.
8553 (expand_vec_perm_even_odd_1): Likewise.
8554 (expand_vec_perm_broadcast_1): Likewise.
8555 (ix86_vectorize_vec_perm_const_ok): Likewise.
8556 (ix86_expand_vecop_qihi): Likewise.
8557 (ix86_expand_mul_widen_hilo): Likewise.
8558 (ix86_expand_sse2_abs): Likewise.
8559 (ix86_expand_pextr): Likewise.
8560 (ix86_expand_pinsr): Likewise.
8561 (ix86_preferred_simd_mode): Likewise.
8562 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
8563 * config/i386/sse.md (*andnot<mode>3): Likewise.
8564 (<mask_codefor><code><mode>3<mask_name>): Likewise.
8565 (*<code><mode>3): Likewise.
8566 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
8567 (ia64_expand_atomic_op): Likewise.
8568 (ia64_arg_type): Likewise.
8569 (ia64_mode_to_int): Likewise.
8570 (ia64_scalar_mode_supported_p): Likewise.
8571 (ia64_vector_mode_supported_p): Likewise.
8572 (expand_vec_perm_broadcast): Likewise.
8573 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
8574 (iq2000_function_arg_advance): Likewise.
8575 (iq2000_function_arg): Likewise.
8576 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
8577 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
8578 (m68k_libcall_value): Likewise.
8579 (m68k_function_value): Likewise.
8580 (sched_attr_op_type): Likewise.
8581 * config/mcore/mcore.c (mcore_output_move): Likewise.
8582 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
8584 (microblaze_function_arg): Likewise.
8585 * config/mips/mips.c (mips16_build_call_stub): Likewise.
8586 (mips_print_operand): Likewise.
8587 (mips_mode_ok_for_mov_fmt_p): Likewise.
8588 (mips_vector_mode_supported_p): Likewise.
8589 (mips_preferred_simd_mode): Likewise.
8590 (mips_expand_vpc_loongson_even_odd): Likewise.
8591 (mips_expand_vec_unpack): Likewise.
8592 (mips_expand_vi_broadcast): Likewise.
8593 (mips_expand_vector_init): Likewise.
8594 (mips_expand_vec_reduc): Likewise.
8595 (mips_expand_msa_cmp): Likewise.
8596 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
8597 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
8598 (cc_flags_for_mode): Likewise.
8599 * config/msp430/msp430.c (msp430_print_operand): Likewise.
8600 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
8601 (nds32_output_casesi_pc_relative): Likewise.
8602 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8603 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
8604 (nvptx_gen_unpack): Likewise.
8605 (nvptx_gen_pack): Likewise.
8606 (nvptx_gen_shuffle): Likewise.
8607 (nvptx_gen_wcast): Likewise.
8608 (nvptx_preferred_simd_mode): Likewise.
8609 * config/pa/pa.c (pa_secondary_reload): Likewise.
8610 * config/pa/predicates.md (base14_operand): Likewise.
8611 * config/powerpcspe/powerpcspe-c.c
8612 (altivec_resolve_overloaded_builtin): Likewise.
8613 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
8615 (rs6000_preferred_simd_mode): Likewise.
8616 (output_vec_const_move): Likewise.
8617 (rs6000_expand_vector_extract): Likewise.
8618 (rs6000_split_vec_extract_var): Likewise.
8619 (reg_offset_addressing_ok_p): Likewise.
8620 (rs6000_legitimate_offset_address_p): Likewise.
8621 (rs6000_legitimize_address): Likewise.
8622 (rs6000_emit_set_const): Likewise.
8623 (rs6000_const_vec): Likewise.
8624 (rs6000_emit_move): Likewise.
8625 (spe_build_register_parallel): Likewise.
8626 (rs6000_darwin64_record_arg_recurse): Likewise.
8627 (swap_selector_for_mode): Likewise.
8628 (spe_init_builtins): Likewise.
8629 (paired_init_builtins): Likewise.
8630 (altivec_init_builtins): Likewise.
8631 (do_load_for_compare): Likewise.
8632 (rs6000_generate_compare): Likewise.
8633 (rs6000_expand_float128_convert): Likewise.
8634 (emit_load_locked): Likewise.
8635 (emit_store_conditional): Likewise.
8636 (rs6000_output_function_epilogue): Likewise.
8637 (rs6000_handle_altivec_attribute): Likewise.
8638 (rs6000_function_value): Likewise.
8639 (emit_fusion_gpr_load): Likewise.
8640 (emit_fusion_p9_load): Likewise.
8641 (emit_fusion_p9_store): Likewise.
8642 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
8643 (fusion_gpr_mem_load): Likewise.
8644 (fusion_addis_mem_combo_load): Likewise.
8645 (fusion_addis_mem_combo_store): Likewise.
8646 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
8647 (fusion_gpr_mem_load): Likewise.
8648 (fusion_addis_mem_combo_load): Likewise.
8649 (fusion_addis_mem_combo_store): Likewise.
8650 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8652 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
8653 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8654 (rs6000_preferred_simd_mode): Likewise.
8655 (output_vec_const_move): Likewise.
8656 (rs6000_expand_vector_extract): Likewise.
8657 (rs6000_split_vec_extract_var): Likewise.
8658 (reg_offset_addressing_ok_p): Likewise.
8659 (rs6000_legitimate_offset_address_p): Likewise.
8660 (rs6000_legitimize_address): Likewise.
8661 (rs6000_emit_set_const): Likewise.
8662 (rs6000_const_vec): Likewise.
8663 (rs6000_emit_move): Likewise.
8664 (rs6000_darwin64_record_arg_recurse): Likewise.
8665 (swap_selector_for_mode): Likewise.
8666 (paired_init_builtins): Likewise.
8667 (altivec_init_builtins): Likewise.
8668 (rs6000_expand_float128_convert): Likewise.
8669 (emit_load_locked): Likewise.
8670 (emit_store_conditional): Likewise.
8671 (rs6000_output_function_epilogue): Likewise.
8672 (rs6000_handle_altivec_attribute): Likewise.
8673 (rs6000_function_value): Likewise.
8674 (emit_fusion_gpr_load): Likewise.
8675 (emit_fusion_p9_load): Likewise.
8676 (emit_fusion_p9_store): Likewise.
8677 * config/rx/rx.c (rx_gen_move_template): Likewise.
8678 (flags_from_mode): Likewise.
8679 * config/s390/predicates.md (s390_alc_comparison): Likewise.
8680 (s390_slb_comparison): Likewise.
8681 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
8682 (s390_vector_mode_supported_p): Likewise.
8683 (s390_cc_modes_compatible): Likewise.
8684 (s390_match_ccmode_set): Likewise.
8685 (s390_canonicalize_comparison): Likewise.
8686 (s390_emit_compare_and_swap): Likewise.
8687 (s390_branch_condition_mask): Likewise.
8688 (s390_rtx_costs): Likewise.
8689 (s390_secondary_reload): Likewise.
8690 (__SECONDARY_RELOAD_CASE): Likewise.
8691 (s390_expand_cs): Likewise.
8692 (s390_preferred_simd_mode): Likewise.
8693 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
8694 * config/sh/sh.c (sh_print_operand): Likewise.
8695 (dump_table): Likewise.
8696 (sh_secondary_reload): Likewise.
8697 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8698 * config/sh/sh.md (casesi_worker_1): Likewise.
8699 (casesi_worker_2): Likewise.
8700 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
8701 (fcc_comparison_operator): Likewise.
8702 * config/sparc/sparc.c (sparc_expand_move): Likewise.
8703 (emit_soft_tfmode_cvt): Likewise.
8704 (sparc_preferred_simd_mode): Likewise.
8705 (output_cbranch): Likewise.
8706 (sparc_print_operand): Likewise.
8707 (sparc_expand_vec_perm_bmask): Likewise.
8708 (vector_init_bshuffle): Likewise.
8709 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8710 (spu_vector_mode_supported_p): Likewise.
8711 (spu_expand_insv): Likewise.
8712 (spu_emit_branch_or_set): Likewise.
8713 (spu_handle_vector_attribute): Likewise.
8714 (spu_builtin_splats): Likewise.
8715 (spu_builtin_extract): Likewise.
8716 (spu_builtin_promote): Likewise.
8717 (spu_expand_sign_extend): Likewise.
8718 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8719 (tilegx_simd_int): Likewise.
8720 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
8721 (tilepro_simd_int): Likewise.
8722 * config/v850/v850.c (const_double_split): Likewise.
8723 (v850_print_operand): Likewise.
8724 (ep_memory_offset): Likewise.
8725 * config/vax/vax.c (vax_rtx_costs): Likewise.
8726 (vax_output_int_move): Likewise.
8727 (vax_output_int_add): Likewise.
8728 (vax_output_int_subtract): Likewise.
8729 * config/visium/predicates.md (visium_branch_operator): Likewise.
8730 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
8731 (visium_print_operand_address): Likewise.
8732 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8733 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
8734 (xtensa_expand_conditional_branch): Likewise.
8735 (xtensa_copy_incoming_a7): Likewise.
8736 (xtensa_output_literal): Likewise.
8737 * dfp.c (decimal_real_maxval): Likewise.
8738 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8741 Alan Hayward <alan.hayward@arm.com>
8742 David Sherwood <david.sherwood@arm.com>
8744 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
8745 (mode_nunits_inline): Likewise.
8746 (mode_inner_inline): Likewise.
8747 (mode_unit_size_inline): Likewise.
8748 (mode_unit_precision_inline): Likewise.
8749 (emit_insn_modes_h): Likewise. Also emit a #define of the
8751 (emit_mode_wider): Add an E_ prefix to mode names.
8752 (emit_mode_complex): Likewise.
8753 (emit_mode_inner): Likewise.
8754 (emit_mode_adjustments): Likewise.
8755 (emit_mode_int_n): Likewise.
8756 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
8757 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
8758 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
8759 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
8760 (CRC32_BUILTIN, ENTRY): Likewise.
8761 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
8762 (aarch64_pop_regs): Likewise.
8763 (aarch64_process_components): Likewise.
8764 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
8765 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
8766 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
8767 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
8768 * config/arm/arm.c (arm_init_libfuncs): Likewise.
8769 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
8771 * config/i386/i386-builtin.def (pcmpestr): Likewise.
8772 (pcmpistr): Likewise.
8773 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
8774 * config/mmix/mmix.c (mmix_output_condition): Likewise.
8775 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8777 * config/rl78/rl78.c (mduc_regs): Likewise.
8778 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8779 (htm_expand_builtin): Likewise.
8780 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
8781 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
8782 * config/xtensa/xtensa.c (print_operand): Likewise.
8783 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
8784 (NUM_MODE_VECTOR_INT): Likewise.
8785 * genoutput.c (null_operand): Likewise.
8786 (output_operand_data): Likewise.
8787 * genrecog.c (print_parameter_value): Likewise.
8788 * lra.c (debug_operand_data): Likewise.
8790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8792 * dwarf2out.c (scompare_loc_descriptor_wide)
8793 (scompare_loc_descriptor_narrow): New functions, split out from...
8794 (scompare_loc_descriptor): ...here.
8795 * expmed.c (emit_store_flag_int): New function, split out from...
8796 (emit_store_flag): ...here.
8798 2017-08-30 Richard Biener <rguenther@suse.de>
8800 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
8801 (dwarf2out_early_finish): Move setting of AT_pubnames from
8802 early debug output to early finish.
8804 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8806 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
8807 and -mdata-region to the assembler.
8809 2017-08-30 Richard Biener <rguenther@suse.de>
8811 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
8813 (gen_subprogram_die): Add DW_AT_object_pointer only early.
8814 (dwarf2out_early_global_decl): Only generate a DIE for the
8815 abstract origin if it doesn't already exist or is a declaration DIE.
8816 (resolve_addr): Do not add the linkage name twice when
8817 generating a stub DIE for the DW_TAG_GNU_call_site target.
8819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8821 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8822 Use machine_mode rather than int for arg1_mode.
8824 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8827 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
8828 that the second argument of the built-in functions to unpack
8829 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
8830 switch statement instead a lot of if statements.
8831 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
8832 Allow 64-bit values to be in Altivec registers as well as
8833 traditional floating point registers.
8834 (pack<mode>, FMOVE128_VSX iterator): Likewise.
8836 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
8838 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
8839 MAX_REGS_PER_ADDRESS == 1.
8841 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
8843 * config/i386/i386.opt (flag_fentry): Do not init to -1.
8844 * config/i386/i386.c (ix86_option_override_internal): Simplify
8845 setting of opts->x_flag_entry.
8847 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8848 Jakub Jelinek <jakub@redhat.com>
8849 Richard Biener <rguenther@suse.de>
8851 PR tree-optimization/81503
8852 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
8853 folded constant fits in the target type; reorder tests for clarity.
8855 2017-08-29 Martin Liska <mliska@suse.cz>
8857 * passes.def: Include pass_lower_switch.
8858 * stmt.c (dump_case_nodes): Remove and move to
8859 tree-switch-conversion.
8860 (case_values_threshold): Likewise.
8861 (expand_switch_as_decision_tree_p): Likewise.
8862 (emit_case_decision_tree): Likewise.
8863 (expand_case): Likewise.
8864 (balance_case_nodes): Likewise.
8865 (node_has_low_bound): Likewise.
8866 (node_has_high_bound): Likewise.
8867 (node_is_bounded): Likewise.
8868 (emit_case_nodes): Likewise.
8869 (struct simple_case_node): New struct.
8870 (add_case_node): Remove.
8871 (emit_case_dispatch_table): Use vector instead of case_list.
8872 (reset_out_edges_aux): Remove.
8873 (compute_cases_per_edge): Likewise.
8874 (expand_case): Build list of simple_case_node.
8875 (expand_sjlj_dispatch_table): Use it.
8876 * tree-switch-conversion.c (struct case_node): Moved from
8877 stmt.c and adjusted.
8878 (emit_case_nodes): Likewise.
8879 (node_has_low_bound): Likewise.
8880 (node_has_high_bound): Likewise.
8881 (node_is_bounded): Likewise.
8882 (case_values_threshold): Likewise.
8883 (reset_out_edges_aux): Likewise.
8884 (compute_cases_per_edge): Likewise.
8885 (add_case_node): Likewise.
8886 (dump_case_nodes): Likewise.
8887 (balance_case_nodes): Likewise.
8888 (expand_switch_as_decision_tree_p): Likewise.
8889 (emit_jump): Likewise.
8890 (emit_case_decision_tree): Likewise.
8891 (try_switch_expansion): Likewise.
8892 (do_jump_if_equal): Likewise.
8893 (emit_cmp_and_jump_insns): Likewise.
8894 (fix_phi_operands_for_edge): New function.
8895 (record_phi_operand_mapping): Likewise.
8896 (class pass_lower_switch): New pass.
8897 (pass_lower_switch::execute): New function.
8898 (make_pass_lower_switch): Likewise.
8899 (conditional_probability):
8900 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
8901 * tree-pass.h: Add make_pass_lower_switch.
8903 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8906 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
8909 2017-08-29 Richard Biener <rguenther@suse.de>
8911 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
8912 we do not add a DW_AT_inline attribute twice.
8913 (gen_subprogram_die): Remove code setting DW_AT_inline on
8914 DECL_ABSTRACT_P nodes.
8916 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
8918 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
8919 calls to internal functions.
8920 (gimplify_modify_expr): Likewise.
8921 * tree-call-cdce.c (use_internal_fn): Likewise.
8922 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
8923 (convert_to_divmod): Set the nothrow flag.
8924 * tree-if-conv.c (predicate_mem_writes): Likewise.
8925 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8926 (vectorizable_call): Likewise.
8927 (vectorizable_store): Likewise.
8928 (vectorizable_load): Likewise.
8929 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
8930 (vect_recog_mask_conversion_pattern): Likewise.
8932 2017-08-29 Martin Liska <mliska@suse.cz>
8935 * gcc.c (driver_handle_option): Add new argument.
8936 * opts-common.c (handle_option): Pass
8937 target_option_override_hook.
8938 * opts-global.c (lang_handle_option): Add new option.
8939 (set_default_handlers): Add new argument.
8940 (decode_options): Likewise.
8941 * opts.c (target_handle_option): Likewise.
8942 (common_handle_option): Call target_option_override_hook.
8943 * opts.h (struct cl_option_handler_func): Add hook for
8944 target option override.
8945 (struct cl_option_handlers): Likewise.
8946 (set_default_handlers): Add new argument.
8947 (decode_options): Likewise.
8948 (common_handle_option): Likewise.
8949 (target_handle_option): Likewise.
8950 * toplev.c (toplev::main): Pass targetm.target_option.override
8953 2017-08-29 Richard Biener <rguenther@suse.de>
8954 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8956 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
8957 life to the active subtree.
8959 2017-08-28 Jeff Law <law@redhat.com>
8961 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
8962 derive_equivalences.
8963 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
8965 (edge_info::derive_equivalences): New private member function
8967 * tree-ssa-dom.c (class edge_info): Changed from a struct
8968 to a class. Add ctor/dtor, methods and data members.
8969 (edge_info::edge_info): Renamed from allocate_edge_info.
8970 Initialize additional members.
8971 (edge_info::~edge_info): New.
8972 (free_dom_edge_info): Delete the edge info.
8973 (record_edge_info): Use new class & associated member functions.
8974 Tighten forms for testing for edge equivalences.
8975 (record_temporary_equivalences): Iterate over the simple
8976 equivalences rather than assuming there's only one per edge.
8977 (cprop_into_successor_phis): Iterate over the simple
8978 equivalences rather than assuming there's only one per edge.
8979 (optimize_stmt): Use operand_equal_p rather than pointer
8980 equality for mini-DSE code.
8982 2017-08-28 Nathan Sidwell <nathan@acm.org>
8984 * gcc.c (execute): Fold SIGPIPE handling into switch
8985 statement. Adjust internal error message.
8987 2017-08-28 Richard Biener <rguenther@suse.de>
8990 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
8991 Do nothing for removed DIEs.
8993 2017-08-28 Richard Biener <rguenther@suse.de>
8995 PR tree-optimization/81977
8996 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
8999 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9002 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9003 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9004 using targetm.gen_mem_thread_fence.
9006 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9009 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9010 predicate to register_operand. Reorder operands.
9011 (*btr<mode>): Ditto.
9012 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9013 (*btr<mode>_mask): Ditto.
9015 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9017 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9018 * config/rs6000/xmmintrin.h: New file.
9019 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9021 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9024 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9025 parameter and_insn and return it.
9026 (recombine_lvx_pattern): Insert a copy to ensure availability of
9027 the base register of the copied masking operation at the point of
9028 the instruction replacement.
9029 (recombine_stvx_pattern): Likewise.
9031 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9033 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9034 undocumented switches.
9035 (-mpower9-dform-vector): Likewise.
9036 (-mpower9-dform): Likewise.
9037 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9038 comments to delete references to -mpower9-dform* switches.
9039 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9040 Delete reference to -mpower9-dform* switches, test for
9041 -mpower9-vector instead.
9042 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9043 (OTHER_P9_VECTOR_MASKS): Likewise.
9044 (POWERPC_MASKS): Likewise.
9045 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9046 tests against -mpower9-dform* to -mpower9-vector. Delete code
9047 that checked for -mpower9-dform* consistancy with other options.
9048 Add test for -mpower9-misc to enable other power9 switches.
9049 (rs6000_init_hard_regno_mode_ok): Likewise.
9050 (rs6000_option_override_internal): Likewise.
9051 (rs6000_emit_prologue): Likewise.
9052 (rs6000_emit_epilogue): Likewise.
9053 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9054 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9055 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9057 (emit_fusion_p9_store): Likewise.
9058 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9059 resetting these macros if the assembler does not support ISA 3.0
9061 (TARGET_P9_DFORM_VECTOR): Likewise.
9062 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9063 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9065 2017-08-25 Alan Modra <amodra@gmail.com>
9067 PR rtl-optimization/81747
9068 * cse.c (cse_extended_basic_block): Don't attempt to record
9069 equivalences for degenerate conditional branches that branch
9070 to their fall-through.
9072 2017-08-24 Martin Sebor <msebor@redhat.com>
9075 * gimple-fold.c (size_must_be_zero_p): New function.
9076 (gimple_fold_builtin_memory_op): Call it.
9078 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
9080 * config/rs6000/mm_malloc.h: New file.
9082 2017-08-24 Bin Cheng <bin.cheng@arm.com>
9084 PR tree-optimization/81913
9085 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
9086 analysis when either IVs in condition can wrap.
9088 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
9090 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
9091 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
9093 2017-08-24 Richard Biener <rguenther@suse.de>
9096 * targhooks.c (default_target_can_inline_p): Properly
9097 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
9098 is present and always compare.
9099 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
9100 infer -mfpmath=sse from TARGET_SSE_P.
9101 (ix86_can_inline_p): Properly use target_option_default_node when
9102 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
9104 2017-08-24 Richard Biener <rguenther@suse.de>
9107 * dwarf2out.c (output_die): Handle flag_generate_offload like
9109 (output_comp_unit): Likewise.
9110 (gen_array_type_die): Likewise.
9111 (dwarf2out_early_finish): Likewise.
9112 (note_variable_value_in_expr): Likewise.
9113 (dwarf2out_finish): Likewise. Adjust assert.
9114 * cgraphunit.c (symbol_table::compile): Move setting of
9115 flag_generate_offload earlier ...
9116 (symbol_table::finalize_compilation_unit): ... here, before
9117 early debug finalization.
9119 2017-08-24 Richard Biener <rguenther@suse.de>
9121 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
9122 and ipa-fnsummary.h.
9123 (ix86_can_inline_p): When ix86_fpmath flags do not match
9124 check whether the callee uses FP math at all.
9126 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
9129 * tree-ssanames.c (get_nonzero_bits): Use element_precision
9130 instead of TYPE_PRECISION.
9132 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
9133 Alan Hayward <alan.hayward@arm.com>
9134 David Sherwood <david.sherwood@arm.com>
9136 * combine.c (make_extraction): Use subreg_offset_from_lsb.
9138 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
9140 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
9142 (ix86_frame::stack_realign_allocate): New field.
9143 (struct machine_frame_state): Modify comments.
9144 (machine_frame_state::sp_realigned_fp_end): New field.
9145 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
9147 (sp_valid_at): Add assertion to assure no attempt to access invalid
9148 offset of a realigned stack.
9149 (fp_valid_at): Likewise.
9150 (choose_baseaddr): Modify comments.
9151 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
9152 ix86_expand_prologue.
9153 (ix86_expand_prologue): Modify stack realignment and allocation.
9154 (ix86_expand_epilogue): Modify comments.
9155 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
9156 avx2_runtime, avx512f, and avx512f_runtime.
9158 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
9160 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
9161 (mstackrealign): Do not init to -1.
9162 * config/i386/i386.c (ix86_option_override_internal):
9163 Check opts_set, not opts when setting default value of
9164 opts->x_ix86_force_align_arg_pointer.
9166 2017-08-23 Richard Biener <rguenther@suse.de>
9168 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
9169 lang_hooks.decl_printable_name.
9170 * print-rtl-function.c (print_rtx_function): Likewise.
9171 * tree-pretty-print.c (dump_function_header): Likewise.
9173 2017-08-23 Richard Biener <rguenther@suse.de>
9176 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
9177 -g0 at compile-time.
9179 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9182 * doc/sourcebuild.texi (Other hardware attributes):
9185 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9188 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
9189 Fix single-use check.
9191 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
9193 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
9195 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9197 * doc/install.texi: Modify to add more details on running selected
9200 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9202 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
9203 is combined with -mabi=ms.
9204 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
9207 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9209 PR tree-optimization/81488
9210 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
9211 and cached_basis fields.
9212 (MAX_SPREAD): New constant.
9213 (alloc_cand_and_find_basis): Initialize new fields.
9214 (clear_visited): New function.
9215 (create_phi_basis_1): Rename from create_phi_basis, set visited
9216 and cached_basis fields.
9217 (create_phi_basis): New wrapper function.
9218 (phi_add_costs_1): Rename from phi_add_costs, add spread
9219 parameter, set visited field, short-circuit when limits reached.
9220 (phi_add_costs): New wrapper function.
9221 (record_phi_increments_1): Rename from record_phi_increments, set
9223 (record_phi_increments): New wrapper function.
9224 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
9225 (phi_incr_cost): New wrapper function.
9226 (all_phi_incrs_profitable_1): Rename from
9227 all_phi_incrs_profitable, set visited field.
9228 (all_phi_incrs_profitable): New wrapper function.
9230 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9231 Alan Hayward <alan.hayward@arm.com>
9232 David Sherwood <david.sherwood@arm.com>
9234 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
9235 that takes the outer and inner modes.
9236 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
9237 comparison as the canonical test for a paradoxical subreg.
9238 * combine.c (simplify_set): Use paradoxical_subreg_p.
9239 (make_extraction): Likewise.
9240 (force_to_mode): Likewise.
9241 (rtx_equal_for_field_assignment_p): Likewise.
9242 (gen_lowpart_for_combine): Likewise.
9243 (simplify_comparison): Likewise.
9244 * cse.c (equiv_constant): Likewise.
9245 * expmed.c (store_bit_field_1): Likewise.
9246 * final.c (alter_subreg): Likewise.
9247 * fwprop.c (propagate_rtx): Likewise.
9248 (forward_propagate_subreg): Likewise.
9249 * ira-conflicts.c (ira_build_conflicts): Likewise.
9250 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
9251 * lra-constraints.c (curr_insn_transform): Likewise.
9252 (split_reg): Likewise.
9253 * lra-eliminations.c (move_plus_up): Likewise.
9254 (lra_eliminate_regs_1): Likewise.
9255 * recog.c (general_operand): Likewise.
9256 * ree.c (combine_reaching_defs): Likewise.
9257 * reload.c (push_reload): Likewise.
9258 (find_reloads): Likewise.
9259 * reload1.c (elimination_effects): Likewise.
9260 (compute_reload_subreg_offset): Likewise.
9261 (choose_reload_regs): Likewise.
9262 * rtlanal.c (subreg_lsb_1): Likewise.
9263 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9264 (simplify_subreg): Likewise.
9265 * var-tracking.c (track_loc_p): Likewise.
9266 * emit-rtl.c (byte_lowpart_offset): Likewise.
9267 (paradoxical_subreg_p): Delete out-of-line definition.
9269 2017-08-22 Jeff Law <law@redhat.com>
9271 PR tree-optimization/81741
9272 PR tree-optimization/71947
9273 * tree-ssa-dom.c: Include tree-inline.h.
9274 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
9275 equivalences if one is more expensive to compute than the other.
9276 * tree-ssa-scopedtables.h (class const_or_copies): Make
9277 record_const_or_copy_raw method private.
9278 (class avail_exprs_stack): New method simplify_binary_operation.
9279 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
9280 avail_exprs_stack::simplify_binary_operation as needed.
9281 (avail_exprs_stack::simplify_binary_operation): New function.
9283 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
9285 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
9286 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
9287 (DOT_SYMBOLS): Likewise.
9288 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
9289 (RELOCATABLE_NEEDS_FIXUP): Likewise.
9290 (RS6000_ABI_NAME): Likewise.
9291 (TARGET_CMODEL): Likewise.
9292 (TOC_SECTION_ASM_OP): Likewise.
9293 (SET_CMODEL): New macro.
9294 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
9296 2017-08-22 Richard Biener <rguenther@suse.de>
9298 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
9299 to allow for free-lang-data replacements similar to verify_type_variant.
9301 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9302 Alan Hayward <alan.hayward@arm.com>
9303 David Sherwood <david.sherwood@arm.com>
9305 * config/aarch64/aarch64.md (casesi): Use DImode rather than
9306 VOIDmode for the LABEL_REF.
9308 2017-08-22 Richard Biener <rguenther@suse.de>
9310 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
9312 2017-08-22 Richard Biener <rguenther@suse.de>
9314 * common.opt (feliminate-dwarf2-dups): Ignore.
9315 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
9316 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
9317 same_die_p_wrap, compute_section_prefix,
9318 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
9319 (comdat_symbol_id, comdat_symbol_number): Likewise.
9320 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
9322 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
9323 (output_die): Mark unreachable path unreachable.
9324 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
9325 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
9326 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
9327 (dwarf2out_early_finish): Likewise.
9329 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
9331 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
9333 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
9336 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
9337 not VAR_P. Filter attribute warnings with OPT_Wattributes.
9338 (avr_attribute_table) <io, io_low, address>: Initialize
9339 .decl_required with true.
9341 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
9343 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
9344 undocumented debugging options.
9345 (-mvsx-scalar-double): Likewise.
9346 (-mallow-df-permute): Likewise.
9347 (-mvectorize-builtins): Likewise.
9348 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9349 (rs6000_builtin_vectorized_function): Likewise.
9350 (rs6000_builtin_md_vectorized_function): Likewise.
9351 (rs6000_opt_vars): Likewise.
9353 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
9356 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
9357 (*btrq_imm): Rename from *btrq.
9358 (*btcq_imm): Rename from *btcq.
9359 (btsc): New code attribute.
9360 (*<btsc><mode>): New insn pattern.
9361 (*btr<mode>): Ditto.
9362 (*<btsc><mode>_mask): New insn_and_split pattern.
9363 (*btr<mode>_mask): Ditto.
9365 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9366 Alan Hayward <alan.hayward@arm.com>
9367 David Sherwood <david.sherwood@arm.com>
9369 * function.c (pad_below): Simplify padding calculation.
9371 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9372 Alan Hayward <alan.hayward@arm.com>
9373 David Sherwood <david.sherwood@arm.com>
9375 * target.def (function_prologue): Remove frame size argument.
9376 (function_epilogue): Likewise.
9377 * doc/tm.texi: Regenerate.
9378 * final.c (final_start_function): Update call to function_prologue.
9379 (final_end_function): Update call to function_epilogue.
9380 (default_function_pro_epilogue): Remove frame size argument.
9381 * output.h (default_function_pro_epilogue): Likewise.
9382 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
9383 (arm_output_function_prologue): Likewise.
9384 * config/frv/frv.c (frv_function_prologue): Likewise.
9385 (frv_function_epilogue): Likewise.
9386 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
9387 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
9388 (ia64_output_function_epilogue): Likewise.
9389 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
9390 (m32r_output_function_epilogue): Likewise.
9391 * config/microblaze/microblaze.c (microblaze_function_prologue)
9392 (microblaze_function_epilogue): Likewise.
9393 * config/mips/mips.c (mips_output_function_prologue): Likewise.
9394 (mips_output_function_epilogue): Likewise.
9395 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
9396 (mmix_target_asm_function_epilogue): Likewise.
9397 * config/msp430/msp430.c (msp430_start_function): Likewise.
9398 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
9399 (nds32_asm_function_epilogue): Likewise.
9400 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
9401 * config/pa/pa.c (pa_output_function_prologue): Likewise.
9402 (pa_output_function_epilogue): Likewise.
9403 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
9404 (rs6000_output_function_epilogue): Likewise.
9405 * config/rl78/rl78.c (rl78_start_function): Likewise.
9406 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
9407 (rs6000_output_function_epilogue): Likewise.
9408 * config/rx/rx.c (rx_output_function_prologue): Likewise.
9409 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
9410 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
9411 (sparc_asm_function_epilogue): Likewise.
9413 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9415 * tree.h (type_has_mode_precision_p): New function.
9416 * convert.c (convert_to_integer_1): Use it.
9417 * expr.c (expand_expr_real_2): Likewise.
9418 (expand_expr_real_1): Likewise.
9419 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
9420 * match.pd: Likewise.
9421 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
9422 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
9423 * tree-tailcall.c (process_assignment): Likewise.
9424 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9425 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
9426 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
9427 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9428 (vectorizable_assignment): Likewise.
9429 (vectorizable_shift): Likewise.
9430 (vectorizable_operation): Likewise.
9431 * tree-vrp.c (register_edge_assert_for_2): Likewise.
9433 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
9435 * match.pd: Add pow (C, x) simplification.
9437 2017-08-21 Richard Biener <rguenther@suse.de>
9439 PR tree-optimization/81900
9440 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
9441 for blocks with abnormal predecessors.
9442 (compute_antic): Do not set visited flag prematurely.
9444 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
9447 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
9449 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9451 * stor-layout.h (vector_type_mode): Move to...
9452 * tree.h (vector_type_mode): ...here.
9453 * stor-layout.c (vector_type_mode): Move to...
9454 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
9456 2017-08-21 Richard Biener <rguenther@suse.de>
9458 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
9459 register_external_die hooks.
9460 (debug_false_tree_charstarstar_uhwistar): Declare.
9461 (debug_nothing_tree_charstar_uhwi): Likewise.
9462 * debug.c (do_nothing_debug_hooks): Adjust.
9463 (debug_false_tree_charstarstar_uhwistar): New do nothing.
9464 (debug_nothing_tree_charstar_uhwi): Likewise.
9465 * dbxout.c (dbx_debug_hooks): Adjust.
9466 (xcoff_debug_hooks): Likewise.
9467 * sdbout.c (sdb_debug_hooks): Likewise.
9468 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9469 * dwarf2out.c (macinfo_label_base): New global.
9470 (dwarf2out_register_external_die): New function for the
9471 register_external_die hook.
9472 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
9473 (dwarf2_debug_hooks): Use them.
9474 (dwarf2_lineno_debug_hooks): Adjust.
9475 (struct die_struct): Add with_offset flag.
9476 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
9477 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
9478 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
9479 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
9480 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
9481 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
9482 defining section names for the early LTO debug variants.
9483 (reset_indirect_string): New helper.
9484 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
9485 (print_dw_val): Add support for offsetted symbol references.
9486 (get_ultimate_context): Split out from is_cxx.
9487 (is_cxx): Use get_ultimate_context.
9488 (is_fortran): Add decl overload.
9489 (compute_comp_unit_symbol): Split out worker from
9490 compute_section_prefix.
9491 (compute_section_prefix): Call compute_comp_unit_symbol and
9492 set comdat_type_p here.
9493 (output_die): Skip DIE symbol output for the LTO added one.
9494 Handle DIE symbol references with offset.
9495 (output_comp_unit): Guard section name mangling properly.
9496 For LTO debug sections emit a symbol at the section beginning
9497 which we use to refer to its DIEs.
9498 (add_abstract_origin_attribute): For DIEs registered via
9499 dwarf2out_register_external_die directly refer to the early
9500 DIE rather than indirectly through the shadow one we created.
9501 Remove obsolete call to dwarf2out_abstract_function for
9502 non-function/block origins.
9503 (gen_array_type_die): When generating early LTO debug do
9504 not emit DW_AT_string_length.
9505 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
9506 late when in LTO. As suggested place a gcc_unreachable for
9507 the DECL_ABSTRACT_P case.
9508 (gen_subprogram_die): Avoid another specification DIE
9509 for early built declarations/definitions for the late LTO case.
9510 (gen_variable_die): Add type references for late duplicated VLA dies
9512 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
9513 we have the abstract instance already.
9514 (process_scope_var): Adjust decl DIE contexts in LTO which
9515 first puts them in limbo.
9516 (gen_decl_die): Do not generate type DIEs late apart from
9517 types for VLAs or for decls we do not yet have a DIE. Do not
9518 call dwarf2out_abstract_function late.
9519 (dwarf2out_early_global_decl): Make sure to create DIEs
9520 for abstract instances of a decl first.
9521 (dwarf2out_late_global_decl): Adjust comment.
9522 (output_macinfo_op): With multiple macro sections use
9523 macinfo_label_base to distinguish labels.
9524 (output_macinfo): Likewise. Update macinfo_label_base.
9525 Pass in the line info label.
9526 (note_variable_value_in_expr): When generating LTO resolve
9527 all variable values here by generating DIEs as needed.
9528 (init_sections_and_labels): Add early LTO debug flag parameter
9529 and generate different sections and names if set. Add generation
9530 counter for the labels so we can have multiple of them.
9531 (reset_dies): Helper to allow DIEs to be output multiple times.
9532 (dwarf2out_finish): When outputting DIEs to the fat part of an
9533 LTO object first reset DIEs.
9534 (dwarf2out_early_finish): Output early DIEs when generating LTO.
9535 (modified_type_die): Check for decl_ultimate_origin being self
9537 (gen_type_die_with_usage): Likewise.
9538 (gen_typedef_die): Allow decl_ultimate_origin being self.
9539 (set_decl_abstract_flags): Remove.
9540 (set_block_abstract_flags): Likewise.
9541 (dwarf2out_abstract_function): Treat the early generated DIEs
9542 as the abstract copy and only add DW_AT_inline and
9543 DW_AT_artificial here and call set_decl_origin_self.
9544 If the DIE has an abstract origin don't do anything.
9545 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
9546 if we have none yet (Go fails to build one, PR78628).
9547 (variably_modified_type_p): Prevent endless recursion for Ada
9548 cyclic pointer types.
9549 * lto-streamer-in.c: Include debug.h.
9550 (dref_queue): New global.
9551 (lto_read_tree_1): Stream in DIE references.
9552 (lto_input_tree): Register DIE references.
9553 (input_function): Stream DECL_DEBUG_ARGS.
9554 * lto-streamer-out.c: Include debug.h.
9555 (lto_write_tree_1): Output DIE references.
9556 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
9557 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
9558 (output_function): Stream DECL_DEBUG_ARGS.
9559 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
9560 Stream DECL_ABSTRACT_ORIGIN.
9561 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
9562 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
9563 DECL_CONTEXT for file-scope decls.
9564 * lto-streamer.h (struct dref_entry): Declare.
9565 (dref_queue): Likewise.
9566 * cfgexpand.c (pass_expand::execute): Do not call the
9567 outlining_inline_function hook here.
9568 * lto-wrapper.c (debug_obj): New global.
9569 (tool_cleanup): Unlink it if required.
9570 (debug_objcopy): New function.
9571 (run_gcc): Handle early debug sections in the IL files by
9572 extracting them to separate files, partially linkin them and
9573 feeding the result back as result to the linker.
9574 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
9575 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
9576 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
9577 sections into a separate segment.
9578 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
9580 (darwin_asm_dwarf_section): Likewise.
9581 (darwin_asm_output_dwarf_offset): Likewise.
9582 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
9584 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9585 Alan Hayward <alan.hayward@arm.com>
9586 David Sherwood <david.sherwood@arm.com>
9588 * read-md.h (md_reader::record_potential_iterator_use): Replace
9589 pointer argument with an rtx and an index.
9590 * read-rtl.c (iterator_group::apply_iterator): Likewise.
9591 (apply_mode_iterator): Likewise.
9592 (apply_code_iterator): Likewise.
9593 (apply_int_iterator): Likewise.
9594 (apply_subst_iterator): Likewise.
9595 (record_iterator_use): Likewise.
9596 (record_attribute_use): Likewise.
9597 (md_reader::record_potential_iterator_use): Likewise. Update calls
9598 to record_iterator_use and apply_iterator.
9599 (iterator_use): Replace ptr with x and index.
9600 (attribute_use): Likewise.
9601 (apply_attribute_uses): Update calls to apply_iterator.
9602 (apply_iterators): Likewise. Update initialization of iterator_use.
9603 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
9604 and record_potential_iterator_use.
9605 (rtx_reader::read_rtx_operand): Likewise.
9607 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9608 Alan Hayward <alan.hayward@arm.com>
9609 David Sherwood <david.sherwood@arm.com>
9611 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
9614 2017-08-21 Richard Biener <rguenther@suse.de>
9617 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
9618 at struct end conservatively when comparing common bases.
9620 2017-08-21 Richard Biener <rguenther@suse.de>
9622 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
9623 (mem_ref_in_stmt): Remove.
9624 (determine_max_movement): Use ref index to get at the reference.
9625 (invariantness_dom_walker::before_dom_children): Deal with
9626 lim data already initialized.
9627 (gather_mem_refs_stmt): Initialize lim data and record ref index.
9629 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9631 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
9632 (TARGET_ISA_ROUND): Ditto.
9633 (TARGET_ROUND): Ditto.
9634 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
9635 * config/i386/i386.md: Ditto.
9636 * config/i386/sse.md: Ditto.
9637 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
9638 with OPTION_MASK_ISA_SSE4_1.
9640 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9643 * doc/extend.texi (x86 Built-in Functions): Correct the name of
9644 __builtin_ia32_lzcnt_u16.
9646 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9649 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
9650 (rs6000_set_current_function): Rewrite function to use it.
9652 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
9655 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
9656 and TYPE_WARN_IF_NOT_ALIGN.
9657 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
9658 (handle_warn_if_not_align): New.
9659 (place_union_field): Call handle_warn_if_not_align.
9660 (place_field): Call handle_warn_if_not_align.
9661 Copy TYPE_WARN_IF_NOT_ALIGN.
9662 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
9663 (layout_type): Likewise.
9664 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
9666 (tree_decl_common): Add warn_if_not_align.
9667 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
9668 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
9669 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
9670 (DECL_WARN_IF_NOT_ALIGN): Likewise.
9671 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
9672 * doc/extend.texi: Document warn_if_not_aligned attribute.
9673 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
9675 2017-08-17 Martin Liska <mliska@suse.cz>
9678 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
9679 (get_data_dependence): Use it as pointer type.
9680 (distribute_loop): Likewise.
9682 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9684 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
9685 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
9686 (p8_vmrgow_<mode>_direct): New define_insn.
9687 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
9688 handle endianness for vmrgew and vmrgow permute patterns.
9690 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9692 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
9693 * config/rs6000/rs6000-cpus.def: Remove comment.
9694 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
9695 (POWERPC_MASKS): Likewise.
9696 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
9697 use of TARGET_VSX_TIMODE.
9698 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
9699 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
9700 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
9701 (rs6000_option_override_internal): Remove dead code.
9702 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
9703 (rs6000_legitimize_reload_address): Likewise.
9704 (rs6000_legitimate_address_p): Likewise.
9705 (rs6000_opt_masks): Delete "vsx-timode".
9706 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
9707 from function comment.
9708 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
9709 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
9710 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
9712 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
9713 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
9714 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
9717 (define_peephole2): Likewise.
9719 2017-08-17 Martin Sebor <msebor@redhat.com>
9722 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
9723 past the end of an array.
9724 (test_pp_format): Add test cases.
9726 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
9728 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
9729 missing ECF_NOTHROW flags.
9731 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9734 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
9735 operands residing in integer registers.
9736 (*vsx_le_perm_load_<mode>): Likewise.
9737 (*vsx_le_perm_store_<mode>): Likewise.
9738 (define_peephole2): Add peepholes to optimize the above.
9740 2017-08-17 Marek Polacek <polacek@redhat.com>
9743 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
9744 to mimic what shorten_compare did. Change the return type to bool.
9745 (fold_cond_expr_with_comparison): Update call to
9746 operand_equal_for_comparison_p.
9747 (fold_ternary_loc): Likewise.
9749 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
9751 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
9753 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
9755 2017-08-17 Richard Biener <rguenther@suse.de>
9757 * tree-ssa-structalias.c (solve_graph): When propagating
9758 to successors update the graphs succ edges and avoid duplicate work.
9760 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
9763 * config/i386/i386.c (ix86_option_override_internal): Save target
9764 specific options after ix86_stack_protector_guard_reg was changed.
9766 2017-08-17 Richard Biener <rguenther@suse.de>
9768 PR tree-optimization/81827
9769 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
9770 (new_var_info): Initialize it conservatively.
9771 (get_call_vi): Mark register vars.
9772 (new_scalar_tmp_constraint_exp): Likewise.
9773 (handle_rhs_call): Likewise.
9774 (handle_const_call): Likewise.
9775 (create_function_info_for): Likewise.
9776 (solve_constraints): Sort varinfos to separate register from
9777 non-register vars to pack points-to solution bitmaps during
9780 2017-08-17 Marek Polacek <polacek@redhat.com>
9782 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
9784 2017-08-17 Richard Biener <rguenther@suse.de>
9786 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
9787 to true when overflow is undefined and we saturated the result.
9789 2017-08-17 Alan Modra <amodra@gmail.com>
9792 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
9793 Don't use store multiple if only one reg needs saving.
9794 (interesting_frame_related_regno): New function.
9795 (rs6000_frame_related): Don't emit frame info for regs that
9797 (rs6000_emit_epilogue): Likewise.
9799 2017-08-16 Nathan Sidwell <nathan@acm.org>
9801 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
9802 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
9803 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
9804 * tree.c (free_lang_data_in_type): Use else-if chain. Always
9805 clear TYPE_LANG_1. Remove obsolete member-function stripping.
9806 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
9807 (verify_type): Adjust for TYPE_BINFO move.
9808 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
9809 process TYPE_BINFO directly.
9810 (hash_tree): Likewise.
9811 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9813 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9816 2017-08-16 David Malcolm <dmalcolm@redhat.com>
9818 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
9820 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
9823 * config/i386/i386.md (*anddi_1_btr): Change predicates of
9824 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
9825 Add ix86_binary_operator_ok to insn constraint.
9826 (*iordi_1_bts): Ditto.
9827 (*xordi_1_btc): Ditto.
9828 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
9829 Update corresponding peephole2 pattern.
9833 2017-08-16 Bin Cheng <bin.cheng@arm.com>
9835 PR tree-optimization/81832
9836 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
9837 copy loop header which has IFN_LOOP_DIST_ALIAS call.
9839 2017-08-16 Marek Polacek <polacek@redhat.com>
9842 * fold-const.c (fold_indirect_ref_1): Restore original behavior
9843 regarding size_zero_node.
9845 2017-08-16 Martin Liska <mliska@suse.cz>
9848 * config.gcc: Respect previously set extra_objs in case
9851 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
9853 PR tree-optimization/81835
9854 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
9855 the phi SSA_NAME. Check that the condition in a COND_EXPR does
9856 not depend on the phi.
9858 2017-08-16 Alan Modra <amodra@gmail.com>
9860 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
9863 2017-08-16 Alan Modra <amodra@gmail.com>
9865 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
9866 (save_reg_p): ..into this. Update all callers.
9867 (first_reg_to_save): Simplify.
9869 2017-08-16 Alan Modra <amodra@gmail.com>
9871 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
9874 2017-08-15 Joseph Myers <joseph@codesourcery.com>
9878 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
9879 constant count if that count is less than 32.
9881 2017-08-15 Nathan Sidwell <nathan@acm.org>
9883 * gcc.c (execute): Emit friendlier message if inferior is killed
9884 by an external cause.
9886 2017-08-15 Richard Biener <rguenther@suse.de>
9888 PR tree-optimization/81790
9889 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
9890 CONSTRUCTORs from simplifying and VN.
9892 2017-08-14 Martin Sebor <msebor@redhat.com>
9894 * builtin-attrs.def: Add comments.
9896 2017-08-14 Martin Sebor <msebor@redhat.com>
9899 * doc/extend.texi (attribute nonstring): Document new attribute.
9901 2017-08-14 Martin Sebor <msebor@redhat.com>
9904 * tree-diagnostic.c (default_tree_printer): Handle %G.
9905 * gimple-pretty-print.h (percent_G_format): Declare new function.
9906 * gimple-pretty-print.c (percent_G_format): Define.
9907 * tree-pretty-print.c (percent_K_format): Add argument.
9909 2017-08-14 Martin Sebor <msebor@redhat.com>
9911 PR translation/79998
9912 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
9913 Remove a stray space.
9915 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
9918 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
9919 (*iordi_1_bts): Ditto.
9920 (*xordi_1_btc): Ditto.
9922 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9925 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
9926 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9928 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
9929 quoted strings, and make more translator-friendly.
9930 (darwin_rs6000_override_options): Likewise.
9931 (rs6000_option_override_internal): Likewise.
9932 (rs6000_return_in_memory): Fix overlong line.
9933 (init_cmulative_args): Use quoted strings, and make more
9934 translator-friendly.
9935 (rs6000_pass_by_reference): Fix overlong line.
9936 (def_builtin): Use quoted strings.
9937 (altivec_expand_predicate_builtin): Use quoted strings, and make
9938 more translator-friendly.
9939 (htm_expand_builtin): Use quoted strings.
9940 (cpu_expand_builtin): Use quoted strings, and make more
9941 translator-friendly.
9942 (altivec_expand_builtin): Likewise.
9943 (paired_expand_predicate_builtin): Likewise.
9944 (rs6000_invalid_builtin): Likewise.
9945 (builtin_function_type): Use quoted strings.
9946 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
9947 more translator-friendly.
9948 (rs6000_trampoline_init): Likewise.
9949 (rs6000_handle_altivec_attribute): Likewise.
9950 (rs6000_inner_target_options): Use quoted strings.
9951 (rs6000_disable_incompatible_switches): Likewise.
9952 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
9953 strings, and make more translator-friendly.
9954 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
9956 2017-08-14 Bin Cheng <bin.cheng@arm.com>
9958 PR tree-optimization/81799
9959 * tree-loop-distribution.c (version_loop_by_alias_check): Force
9960 cond_expr to simple gimple operand.
9962 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
9965 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
9967 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
9971 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
9972 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
9973 TARGET_GASISR_PROLOGUES.
9974 * config/avr/avr.c (avr_option_override): Same.
9975 (avr_pass_pre_proep::execute): Same.
9977 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
9980 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
9981 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
9983 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
9985 * config/i386/i386.md (*load_tp_<mode>): Redefine as
9986 define_insn_and_split. Split to a memory load from 0 in
9987 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
9988 using PTR mode iterator.
9989 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
9990 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
9991 (*add_tp_<mode>): Redefine as define_insn_and_split.
9992 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
9993 address space. Merge with *add_tp_x32 using PTR mode iterator.
9994 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
9995 Split to an add with a memory load from 0 in
9996 DEFAULT_TLS_SEG_REG address space.
9998 2017-08-12 Andrew Pinski <apinski@cavium.com>
10000 * config/aarch64/aarch64-option-extensions.def (rdma):
10001 Fix feature string to what Linux prints out in /proc/cpuinfo.
10003 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10006 * dwarf2out.c (modified_type_die): For C typedef types that have
10007 an ultimate origin, process the ultimate origin instead of the
10009 (gen_typedef_die): Assert that input DECLs have no ultimate
10011 (gen_type_die_with_usage): For typedef variants that have an
10012 ultimate origin, just call gen_decl_die on the original DECL.
10013 (process_scope_var): Avoid creating DIEs for local typedefs and
10014 concrete static variables.
10016 2017-08-12 Alan Modra <amodra@gmail.com>
10020 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10021 match gnu-user.h startfile.
10022 (ENDFILE_LINUX_SPEC): Similarly.
10024 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10027 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10029 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10031 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10032 * config/aarch64/aarch64.md (mov<mode>): Change.
10033 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10034 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10035 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10037 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10039 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10040 for storage order barriers.
10042 2017-08-11 Martin Liska <mliska@suse.cz>
10045 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10046 variables of void type.
10048 2017-08-11 Martin Liska <mliska@suse.cz>
10050 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10051 TARGET_SUPPORTS_ALIASES.
10052 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10053 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10054 (optimize_weakref): Likewise.
10055 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10056 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10057 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10059 2017-08-11 Martin Liska <mliska@suse.cz>
10062 * config/i386/i386.c (make_resolver_func): Do complete
10063 refactoring of the function.
10065 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10068 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10069 * config/i386/i386.c (ix86_stack_protect_guard): Use
10070 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10071 * doc/invoke.texi (x86 Options): Document
10072 -mstack-protector-guard-symbol= option.
10074 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10076 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10077 * config/i386/i386.c (ix86_split_stack_guard): New function.
10078 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
10079 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
10080 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
10081 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
10082 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
10083 (split_stack_space_check): Call ix86_split_stack_guard.
10085 2017-08-10 Martin Sebor <msebor@redhat.com>
10087 * print-tree.c (print_node): Print location using the established
10089 Replace spaces with colons.
10090 (debug_raw, debug): Ditto.
10092 2017-08-10 Martin Sebor <msebor@redhat.com>
10095 * pretty-print.c (pp_format): Correct the handling of %s precision.
10097 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
10100 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10102 (ix86_finalize_stack_frame_flags): This. Also clear
10103 frame_pointer_needed if -fno-omit-frame-pointer is used without
10105 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10106 with ix86_finalize_stack_frame_flags.
10107 (ix86_expand_epilogue): Likewise.
10108 (ix86_expand_split_stack_prologue): Likewise.
10109 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
10111 2017-08-10 Martin Liska <mliska@suse.cz>
10114 * c-attribs.c (handle_target_attribute):
10115 Report warning for an empty string argument of target attribute.
10117 2017-08-09 Jakub Jelinek <jakub@redhat.com>
10120 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
10122 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
10123 or DECL_NONLOCAL labels.
10124 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
10125 or DECL_NONLOCAL labels here.
10127 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
10129 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
10130 to indicate when early gimple folding has been disabled.
10131 (rs6000_gimple_fold_builtin): Add debug content.
10132 (rs6000_invalid_builtin): Fix whitespace.
10133 (rs6000_expand_builtin): Fix whitespace.
10134 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
10136 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
10139 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
10140 SAVE_MULTIPLE if not all the registers that saves, should be saved.
10142 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
10144 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
10145 (qdf24xx): Likewise.
10146 * config/aarch64/aarch64.md: Include falkor.md.
10147 * config/aarch64/falkor.md: New.
10149 2017-08-09 Marek Polacek <polacek@redhat.com>
10152 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
10153 * diagnostic.c (emit_diagnostic): Add a comment.
10154 (emit_diagnostic_valist): New function.
10156 2017-08-09 Marek Polacek <polacek@redhat.com>
10159 * input.c (make_location): New overload.
10160 * input.h (make_location): Declare.
10162 2017-08-08 Alan Modra <amodra@gmail.com>
10163 H.J. Lu <hongjiu.lu@intel.com>
10166 * gcc.c (NO_PIE_SPEC): Delete.
10167 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
10169 (LINK_PIE_SPEC): ..to here.
10170 (LINK_COMMAND_SPEC): Support -no-pie.
10171 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
10172 chain of crtbegin*.o selection, update for PIE_SPEC changes and
10174 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
10175 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
10176 (ENDFILE_CRTEND_SPEC): Similarly.
10178 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
10181 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
10182 (mstack-protector-guard-offset=): Ditto.
10183 * config/i386/i386.c (ix86_option_override): Handle
10184 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
10186 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
10187 ix86_stack_protect_guard_offset variables.
10188 (TARGET_STACK_PROTECT_GUARD): Always define.
10189 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
10190 and -mstack-protector-guard-offset= options.
10192 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10194 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
10195 boundary case for the last candidate.
10197 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10199 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
10200 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
10202 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10204 PR middle-end/19706
10205 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
10206 * config/aarch64/aarch64-builtins.c
10207 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
10208 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
10209 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
10211 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10212 Andrew Pinski <pinskia@gmail.com>
10214 PR middle-end/19706
10215 * internal-fn.def (XORSIGN): New.
10216 * optabs.def (xorsign_optab): New.
10217 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
10218 (convert_expand_mult_copysign): New.
10219 (pass_optimize_widening_mul::execute): Call
10220 convert_expand_mult_copysign.
10222 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10224 PR tree-optimization/81354
10225 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
10226 Insert on edges rather than explicitly creating landing pads.
10227 (analyze_candidates_and_replace): Commit edge inserts.
10229 2017-08-08 Richard Biener <rguenther@suse.de>
10231 PR middle-end/81719
10232 * tree-ssa-loop-niter.c: Include tree-dfa.h.
10233 (expand_simple_operations): Also look through ADDR_EXPRs with
10234 MEM_REF bases treating them as POINTER_PLUS_EXPR.
10236 2017-08-08 Richard Biener <rguenther@suse.de>
10238 PR tree-optimization/81723
10239 * tree-vect-slp.c (struct bst_traits): New hash traits.
10240 (bst_fail): New global.
10241 (vect_build_slp_tree_2): New worker, split out from ...
10242 (vect_build_slp_tree): ... this now wrapping it with using
10243 bst_fail set to cache SLP tree build fails. Properly handle
10245 (vect_analyze_slp_instance): Allocate and free bst_fail.
10247 2017-08-08 Martin Liska <mliska@suse.cz>
10250 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
10251 LABEL_DECLs that can be from a different function.
10253 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10255 PR tree-optimization/81744
10256 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
10257 loop's number of iterations.
10259 2017-08-08 Martin Liska <mliska@suse.cz>
10261 * asan.c: Include header files.
10262 * attribs.c (build_decl_attribute_variant): New function moved
10264 (build_type_attribute_qual_variant): Likewise.
10265 (cmp_attrib_identifiers): Likewise.
10266 (simple_cst_list_equal): Likewise.
10267 (omp_declare_simd_clauses_equal): Likewise.
10268 (attribute_value_equal): Likewise.
10269 (comp_type_attributes): Likewise.
10270 (build_type_attribute_variant): Likewise.
10271 (lookup_ident_attribute): Likewise.
10272 (remove_attribute): Likewise.
10273 (merge_attributes): Likewise.
10274 (merge_type_attributes): Likewise.
10275 (merge_decl_attributes): Likewise.
10276 (merge_dllimport_decl_attributes): Likewise.
10277 (handle_dll_attribute): Likewise.
10278 (attribute_list_equal): Likewise.
10279 (attribute_list_contained): Likewise.
10280 * attribs.h (lookup_attribute): New function moved from tree.[ch].
10281 (lookup_attribute_by_prefix): Likewise.
10282 * bb-reorder.c: Include header files.
10283 * builtins.c: Likewise.
10284 * calls.c: Likewise.
10285 * cfgexpand.c: Likewise.
10286 * cgraph.c: Likewise.
10287 * cgraphunit.c: Likewise.
10288 * convert.c: Likewise.
10289 * dwarf2out.c: Likewise.
10290 * final.c: Likewise.
10291 * fold-const.c: Likewise.
10292 * function.c: Likewise.
10293 * gimple-expr.c: Likewise.
10294 * gimple-fold.c: Likewise.
10295 * gimple-pretty-print.c: Likewise.
10296 * gimple.c: Likewise.
10297 * gimplify.c: Likewise.
10298 * hsa-common.c: Likewise.
10299 * hsa-gen.c: Likewise.
10300 * internal-fn.c: Likewise.
10301 * ipa-chkp.c: Likewise.
10302 * ipa-cp.c: Likewise.
10303 * ipa-devirt.c: Likewise.
10304 * ipa-fnsummary.c: Likewise.
10305 * ipa-inline.c: Likewise.
10306 * ipa-visibility.c: Likewise.
10308 * lto-cgraph.c: Likewise.
10309 * omp-expand.c: Likewise.
10310 * omp-general.c: Likewise.
10311 * omp-low.c: Likewise.
10312 * omp-offload.c: Likewise.
10313 * omp-simd-clone.c: Likewise.
10314 * opts-global.c: Likewise.
10315 * passes.c: Likewise.
10316 * predict.c: Likewise.
10317 * sancov.c: Likewise.
10318 * sanopt.c: Likewise.
10319 * symtab.c: Likewise.
10320 * toplev.c: Likewise.
10321 * trans-mem.c: Likewise.
10322 * tree-chkp.c: Likewise.
10323 * tree-eh.c: Likewise.
10324 * tree-into-ssa.c: Likewise.
10325 * tree-object-size.c: Likewise.
10326 * tree-parloops.c: Likewise.
10327 * tree-profile.c: Likewise.
10328 * tree-ssa-ccp.c: Likewise.
10329 * tree-ssa-live.c: Likewise.
10330 * tree-ssa-loop.c: Likewise.
10331 * tree-ssa-sccvn.c: Likewise.
10332 * tree-ssa-structalias.c: Likewise.
10333 * tree-ssa.c: Likewise.
10334 * tree-streamer-in.c: Likewise.
10335 * tree-vectorizer.c: Likewise.
10336 * tree-vrp.c: Likewise.
10337 * tsan.c: Likewise.
10338 * ubsan.c: Likewise.
10339 * varasm.c: Likewise.
10340 * varpool.c: Likewise.
10341 * tree.c: Remove functions moved to attribs.[ch].
10342 * tree.h: Likewise.
10343 * config/aarch64/aarch64.c: Add attrs.h header file.
10344 * config/alpha/alpha.c: Likewise.
10345 * config/arc/arc.c: Likewise.
10346 * config/arm/arm.c: Likewise.
10347 * config/avr/avr.c: Likewise.
10348 * config/bfin/bfin.c: Likewise.
10349 * config/c6x/c6x.c: Likewise.
10350 * config/cr16/cr16.c: Likewise.
10351 * config/cris/cris.c: Likewise.
10352 * config/darwin.c: Likewise.
10353 * config/epiphany/epiphany.c: Likewise.
10354 * config/fr30/fr30.c: Likewise.
10355 * config/frv/frv.c: Likewise.
10356 * config/ft32/ft32.c: Likewise.
10357 * config/h8300/h8300.c: Likewise.
10358 * config/i386/winnt.c: Likewise.
10359 * config/ia64/ia64.c: Likewise.
10360 * config/iq2000/iq2000.c: Likewise.
10361 * config/lm32/lm32.c: Likewise.
10362 * config/m32c/m32c.c: Likewise.
10363 * config/m32r/m32r.c: Likewise.
10364 * config/m68k/m68k.c: Likewise.
10365 * config/mcore/mcore.c: Likewise.
10366 * config/microblaze/microblaze.c: Likewise.
10367 * config/mips/mips.c: Likewise.
10368 * config/mmix/mmix.c: Likewise.
10369 * config/mn10300/mn10300.c: Likewise.
10370 * config/moxie/moxie.c: Likewise.
10371 * config/msp430/msp430.c: Likewise.
10372 * config/nds32/nds32-isr.c: Likewise.
10373 * config/nds32/nds32.c: Likewise.
10374 * config/nios2/nios2.c: Likewise.
10375 * config/nvptx/nvptx.c: Likewise.
10376 * config/pa/pa.c: Likewise.
10377 * config/pdp11/pdp11.c: Likewise.
10378 * config/powerpcspe/powerpcspe.c: Likewise.
10379 * config/riscv/riscv.c: Likewise.
10380 * config/rl78/rl78.c: Likewise.
10381 * config/rx/rx.c: Likewise.
10382 * config/s390/s390.c: Likewise.
10383 * config/sh/sh.c: Likewise.
10384 * config/sol2.c: Likewise.
10385 * config/sparc/sparc.c: Likewise.
10386 * config/spu/spu.c: Likewise.
10387 * config/stormy16/stormy16.c: Likewise.
10388 * config/tilegx/tilegx.c: Likewise.
10389 * config/tilepro/tilepro.c: Likewise.
10390 * config/v850/v850.c: Likewise.
10391 * config/vax/vax.c: Likewise.
10392 * config/visium/visium.c: Likewise.
10393 * config/xtensa/xtensa.c: Likewise.
10395 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10398 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
10399 constraints since the -mupper-regs-* switches have been
10401 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
10402 into a vector from a double word element that was extracted from
10403 another vector, and eliminate extra XXPERMDI instructions.
10404 (vsx_concat_<mode>_2): Likewise.
10405 (vsx_concat_<mode>_3): Likewise.
10406 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
10407 concat to allow optimizing inserts from previous extracts.
10409 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
10411 * config/i386/i386.c (ix86_stack_protect_guard): Generate
10412 memory reference to a SSP offset in TLS address space.
10413 (ix86_print_operand) <case '@'>: Remove.
10414 (ix86_print_operand_punct_valid_p): Remove '@' code.
10415 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
10416 UNSPEC_SP_TLS_TEST.
10417 (stack_tls_protect_set_<mode>): Remove.
10418 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
10419 (stack_tls_protect_test_<mode>): Remove.
10420 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
10422 2017-08-07 Olivier Hainque <hainque@adacore.com>
10425 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
10427 2017-08-07 Douglas Rupp <rupp@adacore.com>
10429 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
10430 variable was referenced as multidir in command.
10432 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10435 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
10438 2017-08-07 Martin Liska <mliska@suse.cz>
10440 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
10441 * config/rl78/rl78.c: Add include of attribs.h.
10442 * config/sh/sh.c: Likewise.
10443 * config/v850/v850.c: Likewise.
10445 2017-08-07 Tom de Vries <tom@codesourcery.com>
10447 PR middle-end/78266
10448 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
10450 2017-08-07 Martin Liska <mliska@suse.cz>
10452 * config/mips/mips.c: Include attribs.h.
10454 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
10457 * doc/invoke.texi: Document change in behvaior for -Ofast for
10460 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
10462 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
10464 (aarch64_pop_regs): Likewise.
10465 (aarch64_gen_load_pair): Likewise.
10466 (aarch64_save_callee_saves): Likewise.
10467 (aarch64_restore_callee_saves): Likewise.
10469 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10471 * config/i386/i386.c: Revert the last change.
10473 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10476 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10478 (ix86_finalize_stack_frame_flags): This. Also clear
10479 frame_pointer_needed if -fno-omit-frame-pointer is used without
10481 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10482 with ix86_finalize_stack_frame_flags.
10483 (ix86_expand_epilogue): Likewise.
10484 (ix86_expand_split_stack_prologue): Likewise.
10486 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10489 * config/i386/i386.c (get_builtin_code_for_version): Set priority
10490 to P_AES for Westmere.
10492 2017-08-07 Jonathan Yong <10walls@gmail.com>
10494 * config/i386/mingw.opt (fset-stack-executable): Removed.
10495 * config/i386/cygming.opt (fset-stack-executable): Moved
10497 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
10499 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
10501 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
10503 2017-08-07 Marek Polacek <polacek@redhat.com>
10505 PR middle-end/81737
10506 * fold-const.c (fold_indirect_ref_1): Check type_domain.
10508 2017-08-07 Martin Liska <mliska@suse.cz>
10510 * attribs.h (canonicalize_attr_name): New function.
10511 (cmp_attribs): Move from c-format.c and adjusted.
10512 (is_attribute_p): Moved from tree.h.
10513 * tree-inline.c: Add new includes.
10514 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
10515 (private_is_attribute_p): Remove.
10516 (private_lookup_attribute): Likewise.
10517 (private_lookup_attribute_by_prefix): Simplify.
10518 (remove_attribute): Use is_attribute_p.
10519 * tree.h: Remove removed declarations.
10521 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10523 PR middle-end/81698
10524 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
10525 instead of computing it in the function. Formatting fix.
10526 (expand_case): Don't rely on default_edge being the first edge,
10527 clear it if removing it, pass default_edge to
10528 emit_case_dispatch_table.
10529 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
10532 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
10534 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
10535 insn in the function, emit NOP after the insn.
10537 2017-08-06 Tom de Vries <tom@codesourcery.com>
10539 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
10542 2017-08-06 Tom de Vries <tom@codesourcery.com>
10544 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
10547 2017-08-04 Yury Gribov <tetra2005@gmail.com>
10549 PR tree-optimization/57371
10550 * match.pd: New pattern.
10552 2017-08-04 Marek Polacek <polacek@redhat.com>
10554 PR middle-end/81695
10555 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
10556 perform the computation in offset_int.
10558 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10560 PR tree-optimization/81136
10561 * tree-vectorizer.h: Include tree-hash-traits.h.
10562 (vec_base_alignments): New typedef.
10563 (vec_info): Add a base_alignments field.
10564 (vect_record_base_alignments): Declare.
10565 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
10567 (DR_IS_CONDITIONAL_IN_STMT): New macro.
10568 (create_data_ref): Add an is_conditional_in_stmt argument.
10569 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
10570 the is_conditional_in_stmt field.
10571 (data_ref_loc): Add an is_conditional_in_stmt field.
10572 (get_references_in_stmt): Set the is_conditional_in_stmt field.
10573 (find_data_references_in_stmt): Update call to create_data_ref.
10574 (graphite_find_data_references_in_stmt): Likewise.
10575 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
10576 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
10577 (vect_record_base_alignment): New function.
10578 (vect_record_base_alignments): Likewise.
10579 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
10580 for nested statements even if we fail to compute a misalignment.
10581 Use pooled base alignments for unconditional references.
10582 (vect_find_same_alignment_drs): Compare base addresses instead
10584 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
10585 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
10587 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10589 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
10590 Add an explicit name for the enum. Use auto_vec for slp_instances
10591 and grouped_stores.
10592 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
10594 (_bb_vec_info): Add a constructor and destructor.
10595 (vinfo_for_stmt): Return NULL for uids of -1 as well.
10596 (destroy_loop_vec_info): Delete.
10597 (vect_destroy_datarefs): Likewise.
10598 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
10599 (vec_info::vec_info): New function.
10600 (vec_info::~vec_info): Likewise.
10601 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
10602 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
10603 destroy_loop_vec_info.
10604 * tree-vect-loop.c (new_loop_vec_info): Replace with...
10605 (_loop_vec_info::_loop_vec_info): ...this.
10606 (destroy_loop_vec_info): Replace with...
10607 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
10608 the stmt_vec_infos. Leave handling of vec_info information to its
10609 destructor. Remove explicit vector releases.
10610 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
10611 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
10612 * tree-vect-slp.c (new_bb_vec_info): Replace with...
10613 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
10614 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
10615 (destroy_bb_vec_info): Replace with...
10616 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
10617 information to its destructor.
10618 (vect_slp_analyze_bb_1): Use new and delete instead of
10619 new_bb_vec_info and destroy_bb_vec_info.
10620 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
10623 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10625 * tree-data-ref.h (subscript): Add access_fn field.
10626 (data_dependence_relation): Add could_be_independent_p.
10627 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
10628 (same_access_functions): Move to tree-data-ref.c.
10629 * tree-data-ref.c (ref_contains_union_access_p): New function.
10630 (access_fn_component_p): Likewise.
10631 (access_fn_components_comparable_p): Likewise.
10632 (dr_analyze_indices): Add a reference to access_fn_component_p.
10633 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
10635 (constant_access_functions): Likewise.
10636 (add_other_self_distances): Likewise.
10637 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
10638 (initialize_data_dependence_relation): Use XCNEW and remove
10639 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
10640 of access functions that have the same type. Allow the
10641 subsequence to end with different bases in some circumstances.
10642 Record the chosen access functions in SUB_ACCESS_FN.
10643 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
10644 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
10645 (subscript_dependence_tester_1): Likewise dra and drb.
10646 (build_classic_dist_vector): Update calls accordingly.
10647 (subscript_dependence_tester): Likewise.
10648 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
10649 DDR_COULD_BE_INDEPENDENT_P.
10650 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
10651 comp_alias_ddrs instead of may_alias_ddrs.
10652 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
10654 (vect_analyze_data_ref_dependence): Use it if
10655 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
10656 distance vectors if that fails.
10657 (dependence_distance_ge_vf): New function.
10658 (vect_prune_runtime_alias_test_list): Use it. Don't clear
10659 LOOP_VINFO_MAY_ALIAS_DDRS.
10661 2017-08-04 Richard Biener <rguenther@suse.de>
10663 PR middle-end/81705
10664 * fold-const.c (fold_binary_loc): Properly restrict
10665 minus_var0 && minus_var1 case when associating undefined overflow
10668 2017-08-04 Tom de Vries <tom@codesourcery.com>
10670 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
10672 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10674 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10675 Don't start diagnostic messages with a capital letter.
10676 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10678 (rs6000_invalid_builtin): Likewise.
10679 (rs6000_trampoline_init): Likewise.
10681 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10684 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
10685 after setting changeable df flags.
10687 2017-08-03 Richard Biener <rguenther@suse.de>
10689 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
10690 up if the use is in USE - X.
10692 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
10694 * toplev.c (dumpfile.h): New include.
10695 (internal_error_reentered): New static function. Use it...
10696 (internal_error_function): ...here to handle reentered internal_error.
10698 2017-08-03 Richard Biener <rguenther@suse.de>
10700 PR middle-end/81148
10701 * fold-const.c (split_tree): Add minus_var and minus_con
10702 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
10703 here but always use minus_*.
10704 (associate_trees): Assert we never associate with MINUS_EXPR
10705 and NULL first operand. Do not recurse for PLUS_EXPR operands
10706 when associating as MINUS_EXPR either.
10707 (fold_binary_loc): Track minus_var and minus_con.
10709 2017-08-03 Tom de Vries <tom@codesourcery.com>
10712 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
10713 ACCEL_COMPILER, apply finish_options on
10714 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
10716 2017-08-03 Tom de Vries <tom@codesourcery.com>
10719 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
10720 function_entry_patch_area_size > 0.
10722 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10725 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
10726 instead of 10??LU, perform unit multiplication in wide_int,
10727 don't change alloc_object_size_limit if the limit is larger
10730 PR tree-optimization/81655
10731 PR tree-optimization/81588
10732 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
10733 the case when ranges[i].low and high are 1 for unsigned type with
10736 PR middle-end/81052
10737 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
10738 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
10740 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10742 * tree-vrp.h: Add include guard.
10744 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
10747 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
10748 (ud2): New insn pattern.
10749 * config/i386/i386.c (ix86_expand_epilogue):
10750 For naked functions, generate ud2 instead of trap insn.
10752 2017-08-02 Marek Polacek <polacek@redhat.com>
10755 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
10757 2017-08-02 Tom de Vries <tom@codesourcery.com>
10758 Cesar Philippidis <cesar@codesourcery.com>
10760 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
10761 Add missing edge probabilities.
10763 2017-08-02 Tamar Christina <tamar.christina@arm.com>
10765 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
10766 Correct endianness.
10768 2017-08-02 Jakub Jelinek <jakub@redhat.com>
10770 PR middle-end/79499
10771 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
10772 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
10773 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
10774 prologue_seq sequences - if any.
10776 2017-08-02 Richard Biener <rguenther@suse.de>
10778 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
10779 via vectors if supported, integer extracts via punning if supported
10780 or otherwise vector extracts.
10782 2017-08-02 Richard Biener <rguenther@suse.de>
10784 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
10786 (bitmap_insert_into_set): ... this.
10788 2017-08-02 Richard Biener <rguenther@suse.de>
10790 PR tree-optimization/81633
10792 2015-08-17 Alan Hayward <alan.hayward@arm.com>
10794 PR tree-optimization/71752
10795 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
10797 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
10799 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
10800 (machine_function::call_ms2sysv_pad_out): Remove field.
10801 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
10802 (ix86_compute_frame_layout): Likewise.
10804 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
10807 * config/i386/i386.c (ix86_set_func_type): Disallow naked
10808 attribute with interrupt attribute.
10810 2017-08-01 Andrew Pinski <apinski@cavium.com>
10812 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
10813 BIT_INSERT_EXPR's operand 1
10814 to see if the types precision matches.
10816 2017-08-01 Martin Liska <mliska@suse.cz>
10818 PR middle-end/70140
10819 * builtins.c (expand_builtin_memcpy_args): Remove.
10820 (expand_builtin_memcpy): Call newly added function
10821 expand_builtin_memory_copy_args.
10822 (expand_builtin_memcpy_with_bounds): Likewise.
10823 (expand_builtin_mempcpy): Remove last argument.
10824 (expand_builtin_mempcpy_with_bounds): Likewise.
10825 (expand_builtin_memory_copy_args): New function created from
10826 expand_builtin_mempcpy_args with small modifications.
10827 (expand_builtin_mempcpy_args): Remove.
10828 (expand_builtin_stpcpy): Remove unused argument.
10829 (expand_builtin): Likewise.
10830 (expand_builtin_with_bounds): Likewise.
10832 2017-08-01 Martin Liska <mliska@suse.cz>
10835 Make mempcpy more optimal (PR middle-end/70140).
10837 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10840 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10841 __builtin_vec_cmpne verify both arguments are compatible vectors
10842 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
10843 verify arg1_type is a pointer or array type. For __builtin_vec_st,
10844 move computation of aligned to after checking the argument types.
10848 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10849 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
10852 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
10853 Doug Rupp <rupp@adacore.com>
10854 Olivier Hainque <hainque@adacore.com>
10856 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
10857 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
10859 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
10860 for TARGET_OS_CPP_BUILTIN.
10861 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
10862 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
10864 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
10865 passing required abi options to the assembler for EABI configurations.
10866 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
10867 of .text.hot and .text.unlikely sections for kernel modules when
10868 using ARM style exceptions.
10869 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
10870 options. Add EXTRA_CC1_SPEC.
10871 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
10873 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
10875 (ARM_TARGET2_DWARF_FORMAT): Define.
10876 * config/arm/t-vxworks: Adjust multilib control to removal of the
10877 Diab command line options.
10879 2017-08-01 Martin Liska <mliska@suse.cz>
10881 PR gcov-profile/81561
10882 * gcov.c (unblock): Make unblocking safe as we need to preserve
10883 index correspondence of blocks and block_lists.
10885 2017-08-01 Richard Biener <rguenther@suse.de>
10887 PR tree-optimization/81181
10888 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10889 (compute_antic): ... end of iteration here.
10891 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
10893 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
10894 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
10895 (ftree-slp-vectorize): Likewise.
10896 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
10897 can no longer be set independent of flag_tree_loop_vectorize.
10898 * omp-general.c (emp_max_vf): Likewise.
10899 * opts.c (enable_fdo_optimizations): Remove references to
10900 flag_tree_vectorize, these are now implicit.
10901 (common_handle_option): Remove handling for OPT_ftree_vectorize,
10902 and leave it for the options machinery.
10904 2017-08-01 Martin Liska <mliska@suse.cz>
10906 PR middle-end/70140
10907 * builtins.c (expand_builtin_memcpy_args): Remove.
10908 (expand_builtin_memcpy): Call newly added function
10909 expand_builtin_memory_copy_args.
10910 (expand_builtin_memcpy_with_bounds): Likewise.
10911 (expand_builtin_mempcpy): Remove last argument.
10912 (expand_builtin_mempcpy_with_bounds): Likewise.
10913 (expand_builtin_memory_copy_args): New function created from
10914 expand_builtin_mempcpy_args with small modifications.
10915 (expand_builtin_mempcpy_args): Remove.
10916 (expand_builtin_stpcpy): Remove unused argument.
10917 (expand_builtin): Likewise.
10918 (expand_builtin_with_bounds): Likewise.
10920 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10923 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
10924 print "ds:" only for immediates in generic address space.
10926 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10929 * config/i386/i386.c (ix86_funciton_naked): New prototype.
10930 (ix86_function_ok_for_sibcall): Return false for naked functions.
10932 2017-08-01 Richard Biener <rguenther@suse.de>
10934 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
10935 (compute_antic): Seed worklist with exit block predecessors.
10936 * cfganal.c (dfs_find_deadend): For a cycle return the source
10937 of the edge closing it.
10939 2017-08-01 Tamar Christina <tamar.christina@arm.com>
10941 * config/aarch64/aarch64.c
10942 (aarch64_can_const_movi_rtx_p): Move 0 check.
10944 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10946 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
10947 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
10949 * match.pd: Ditto in address comparison pattern.
10951 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10953 PR tree-optimization/81627
10954 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
10955 closed ssa form for store-store chain.
10957 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10959 PR tree-optimization/81620
10960 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
10961 for store-store chain.
10963 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10965 PR tree-optimization/81588
10966 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
10967 ranges[i].in_p, invert comparison code ccode. For >/>=,
10968 swap rhs1 and rhs2 and comparison code unconditionally,
10969 for </<= don't do that. Don't swap rhs1/rhs2 again if
10970 ranges[i].in_p, instead invert comparison code ccode if
10971 opcode or oe->rank is BIT_IOR_EXPR.
10974 * optabs.def (vec_extract_optab, vec_init_optab): Change from
10975 a direct optab to conversion optab.
10976 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
10977 with GET_MODE_INNER as last argument instead of optab_handler.
10978 * expmed.c (extract_bit_field_1): Likewise. Use vector from
10979 vector extraction if possible and optab is available.
10980 * expr.c (store_constructor): Use convert_optab_handler instead
10981 of optab_handler. Use vector initialization from smaller
10982 vectors if possible and optab is available.
10983 * tree-vect-stmts.c (vectorizable_load): Likewise.
10984 * doc/md.texi (vec_extract, vec_init): Document that the optabs
10985 now have two modes.
10986 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
10987 of vec_init from half-sized vectors with the same element mode.
10988 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
10989 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
10990 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
10991 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
10992 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
10993 after mode in gen_vec_extract* calls.
10994 (vec_extract<mode>): Renamed to ...
10995 (vec_extract<mode><ssescalarmodelower>): ... this.
10996 (vec_extract<mode><ssehalfvecmodelower>): New expander.
10997 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
10998 element mode after mode in gen_vec_init* calls.
10999 (VEC_INIT_HALF_MODE): New mode iterator.
11000 (vec_init<mode>): Renamed to ...
11001 (vec_init<mode><ssescalarmodelower>): ... this.
11002 (vec_init<mode><ssehalfvecmodelower>): New expander.
11003 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11004 (vec_extractv2sfsf): ... this.
11005 (vec_initv2sf): Renamed to ...
11006 (vec_initv2sfsf): ... this.
11007 (vec_extractv2si): Renamed to ...
11008 (vec_extractv2sisi): ... this.
11009 (vec_initv2si): Renamed to ...
11010 (vec_initv2sisi): ... this.
11011 (vec_extractv4hi): Renamed to ...
11012 (vec_extractv4hihi): ... this.
11013 (vec_initv4hi): Renamed to ...
11014 (vec_initv4hihi): ... this.
11015 (vec_extractv8qi): Renamed to ...
11016 (vec_extractv8qiqi): ... this.
11017 (vec_initv8qi): Renamed to ...
11018 (vec_initv8qiqi): ... this.
11019 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
11020 (vec_init<mode>): Renamed to ...
11021 (vec_init<mode><VEC_base_l>): ... this.
11022 (vec_extract<mode>): Renamed to ...
11023 (vec_extract<mode><VEC_base_l>): ... this.
11024 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
11025 (vec_initv2sfsf): ... this.
11026 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
11027 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11028 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
11029 element mode after mode in gen_vec_init* calls.
11030 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11031 (vec_init<mode><Vel>): ... this.
11032 (vec_extract<mode>): Renamed to ...
11033 (vec_extract<mode><Vel>): ... this.
11034 * config/aarch64/iterators.md (Vel): New mode attribute.
11035 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11036 Add element mode after mode in gen_vec_extract* calls.
11037 * config/s390/vector.md (non_vec_l): New mode attribute.
11038 (vec_extract<mode>): Renamed to ...
11039 (vec_extract<mode><non_vec_l>): ... this.
11040 (vec_init<mode>): Renamed to ...
11041 (vec_init<mode><non_vec_l>): ... this.
11042 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11043 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11045 * config/arm/iterators.md (V_elem_l): New mode attribute.
11046 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11047 (vec_extract<mode><V_elem_l>): ... this.
11048 (vec_extractv2di): Renamed to ...
11049 (vec_extractv2didi): ... this.
11050 (vec_init<mode>): Renamed to ...
11051 (vec_init<mode><V_elem_l>): ... this.
11052 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11053 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11054 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11055 Add element mode after gen_vec_extract* calls.
11056 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11057 (vec_init<mode><unitmode>): ... this.
11058 (vec_extract<mode>): Renamed to ...
11059 (vec_extract<mode><unitmode>): ... this.
11060 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11061 (vec_init<mode><unitmode>): ... this.
11062 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11063 (vec_initv2sfsf): ... this.
11064 (vec_extractv2sf): Renamed to ...
11065 (vec_extractv2sfsf): ... this.
11066 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11067 Add element mode after gen_vec_extract* calls.
11068 * config/mips/mips.md (unitmode): New mode iterator.
11069 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11070 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11071 * config/spu/spu.md (inner_l): New mode attribute.
11072 (vec_init<mode>): Renamed to ...
11073 (vec_init<mode><inner_l>): ... this.
11074 (vec_extract<mode>): Renamed to ...
11075 (vec_extract<mode><inner_l>): ... this.
11076 * config/sparc/sparc.md (veltmode): New mode iterator.
11077 (vec_init<VMALL:mode>): Renamed to ...
11078 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
11079 * config/ia64/vect.md (vec_initv2si): Renamed to ...
11080 (vec_initv2sisi): ... this.
11081 (vec_initv2sf): Renamed to ...
11082 (vec_initv2sfsf): ... this.
11083 (vec_extractv2sf): Renamed to ...
11084 (vec_extractv2sfsf): ... this.
11085 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
11086 (vec_init<mode>): Renamed to ...
11087 (vec_init<mode><VEC_base_l>): ... this.
11088 (vec_extract<mode>): Renamed to ...
11089 (vec_extract<mode><VEC_base_l>): ... this.
11090 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
11091 (vec_initv2sfsf): ... this.
11092 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
11093 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11094 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
11095 gen_vec_init* calls.
11097 2017-08-01 Richard Biener <rguenther@suse.de>
11099 PR tree-optimization/81297
11100 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
11101 TREE_OVERFLOW from INTEGER_CSTs.
11103 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
11105 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
11107 2017-07-31 Carl Love <cel@us.ibm.com>
11109 * config/rs6000/rs6000-c: Add support for built-in functions
11110 vector signed char vec_xl_be (signed long long, signed char *);
11111 vector unsigned char vec_xl_be (signed long long, unsigned char *);
11112 vector signed int vec_xl_be (signed long long, signed int *);
11113 vector unsigned int vec_xl_be (signed long long, unsigned int *);
11114 vector signed long long vec_xl_be (signed long long, signed long long *);
11115 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
11116 vector signed short vec_xl_be (signed long long, signed short *);
11117 vector unsigned short vec_xl_be (signed long long, unsigned short *);
11118 vector double vec_xl_be (signed long long, double *);
11119 vector float vec_xl_be (signed long long, float *);
11120 * config/rs6000/altivec.h (vec_xl_be): Add #define.
11121 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
11122 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
11124 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
11125 (altivec_expand_builtin): Add switch statement to call
11126 altivec_expand_xl_be for each builtin.
11127 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
11128 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
11129 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
11130 __builtin_vsx_le_be_v16qi.
11131 * doc/extend.texi: Update the built-in documentation file for the
11132 new built-in functions.
11134 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11137 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
11139 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11141 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11143 * config.gcc: Add z14.
11144 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
11145 CPU model numbers for z13s and z14.
11146 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
11148 * config/s390/s390-opts.h (enum processor_type): Rename
11149 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11150 * config/s390/s390.c (processor_table): Add field for CPU name to
11151 be passed to Binutils.
11152 (s390_asm_output_machine_for_arch): Use the new field in
11153 processor_table for Binutils.
11154 (s390_expand_builtin): Replace arch12 with z14.
11155 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11156 (s390_get_sched_attrmask): Likewise.
11157 (s390_get_unit_mask): Likewise.
11158 * config/s390/s390.opt: Add z14 to processor_type enum.
11160 2017-07-31 Martin Jambor <mjambor@suse.cz>
11163 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
11164 regardless of optimization level.
11166 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
11167 Martin Liska <mliska@suse.cz>
11169 * predict.def: Remove old comment and adjust probability.
11170 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
11171 PREDICT statements.
11173 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11176 * config/i386/i386.c (ix86_function_naked): New function.
11177 (ix86_can_use_return_insn_p): Return false for naked functions.
11178 (ix86_expand_prologue): Skip prologue for naked functions.
11179 (ix86_expand_epilogue): Skip epilogue for naked functions
11180 and emit trap instruction.
11181 (ix86_warn_func_return): New function.
11182 (ix86_attribute_table): Add "naked" attribute specification.
11183 (TARGET_WARN_FUNC_RETURN): Define.
11184 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
11186 2017-07-31 Martin Liska <mliska@suse.cz>
11188 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
11189 (dump_gimple_bb_header): Always dump BB info.
11190 (pp_cfg_jump): Do not append info about BB when dumping a jump.
11192 2017-07-31 Martin Liska <mliska@suse.cz>
11195 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
11196 also with current_function_decl non-null equality.
11198 2017-07-31 Jakub Jelinek <jakub@redhat.com>
11201 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
11202 change type to the element type, instead add eltype variable and
11203 use it where we are interested in the element type.
11205 PR tree-optimization/81603
11206 * ipa-polymorphic-call.c
11207 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
11208 offset arithmetic in offset_int, bail out if the resulting bit offset
11209 doesn't fit into shwi.
11211 2017-07-31 Martin Liska <mliska@suse.cz>
11213 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
11214 (gimplify_save_expr): Fix comment.
11216 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
11219 * config/i386/i386.c (ix86_function_arg): Update arguments for
11221 (ix86_compute_frame_layout): Set the initial stack offset to
11222 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
11223 INCOMING_FRAME_SP_OFFSET.
11224 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
11225 stack before exception handler returns.
11226 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
11227 the 'ERROR_CODE' for exception handler.
11229 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
11231 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
11232 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
11233 (ASM_OUTPUT_REG_POP): Ditto.
11234 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
11235 instead of asm_fprintf to output pure string.
11237 2017-07-29 Jakub Jelinek <jakub@redhat.com>
11239 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
11240 to imported_module_or_decl hook.
11241 (debug_nothing_tree_tree_tree_bool): Remove.
11242 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
11243 * debug.c (do_nothing_debug_hooks): Use
11244 debug_nothing_tree_tree_tree_bool_bool instead of
11245 debug_nothing_tree_tree_tree_bool.
11246 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11247 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
11248 * sdbout.c (sdb_debug_hooks): Likewise.
11249 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
11250 (gen_namespace_die): Add DW_AT_export_symbols attribute if
11252 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
11253 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
11254 attribute, don't add anything.
11256 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11258 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
11259 (fold_build2_stat_loc): Likewise.
11260 (fold_build3_stat_loc): Likewise.
11261 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
11262 (fold_build1_loc): Remove macro.
11263 (fold_build2_loc): Likewise.
11264 (fold_build3_loc): Likewise.
11266 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11268 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
11269 (gimple_build_debug_bind_source_stat): Likewise.
11270 * gimple.h (gimple_build_debug_bind): Remove macro.
11271 (gimple_build_debug_bind_source): Likewise.
11273 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11275 * bitmap.c (bitmap_alloc): Adjust.
11276 (bitmap_gc_alloc): Likewise.
11277 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
11279 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11281 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
11282 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
11283 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
11284 (bitmap_gc_alloc_stat): Likewise.
11285 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
11287 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11289 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
11290 * rtl.h (shallow_copy_rtx): Remove macro.
11292 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11294 * emit-rtl.c (gen_raw_REG): Adjust.
11295 * gengenrtl.c (gendef): Likewise.
11296 * rtl.c (rtx_alloc_stat): Remove _stat from name.
11297 * rtl.h (rtx_alloc): Remove macro.
11299 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11301 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
11302 (build_tree_list_stat): Likewise.
11303 * tree.h (build_tree_list): Remove macro.
11304 (build_tree_list_vec): Likewise.
11306 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11308 * tree.c (make_vector_stat): Remove _stat from name.
11309 (build_vector_stat): Likewise.
11310 * tree.h (make_vector_stat): Remove macro.
11311 (build_vector_stat): Likewise.
11313 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11315 * tree.h (build_var_debug_value): Remove prototype.
11317 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11319 * tree.c (tree_cons_stat): Remove _stat from name.
11320 * tree.h (tree_cons): Remove macro.
11322 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11324 * tree.c (build_vl_exp_stat): Remove _stat from name.
11325 * tree.h (build_vl_exp): Remove macro.
11327 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11329 * tree.c (build_decl_stat): Remove _stat from name.
11330 * tree.h (build_decl): Remove macro.
11332 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11334 * gimple.c (gimple_build_with_ops_stat): Adjust.
11335 (gimple_alloc_stat): Remove _stat from name.
11336 * gimple.h (gimple_alloc): Remove macro.
11338 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11340 * tree.c (make_tree_vec_stat): Remove _stat from name.
11341 (grow_tree_vec_stat): Likewise.
11342 * tree.h (make_tree_vec_stat): Adjust prototype.
11343 (grow_tree_vec_stat): Likewise.
11344 (make_tree_vec): Remove macro.
11345 (grow_tree_vec): Likewise.
11347 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11349 * fold-const.c (fold_build1_stat_loc): Adjust.
11350 (fold_build2_stat_loc): Likewise.
11351 (fold_build3_stat_loc): Likewise.
11352 * tree.c (build0_stat): Remove _stat from name.
11353 (build1_stat): Likewise.
11354 (build2_stat): Likewise.
11355 (build3_stat): Likewise.
11356 (build4_stat): Likewise.
11357 (build5_stat): Likewise.
11358 * tree.h (build1_loc): Remove macro, and rename _stat function
11360 (build2_loc): Likewise.
11361 (build3_loc): Likewise.
11362 (build4_loc): Likewise.
11363 (build5_loc): Likewise.
11365 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11367 * tree.c (make_int_cst_stat): Remove _stat from name.
11368 * tree.h (make_int_cst_stat): Adjust prototype.
11369 (make_int_cst): Remove macro.
11371 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11373 * tree.c (make_tre_binfo_stat): Remove _stat from name.
11374 * tree.h (make_tree_binfo_stat): Adjust prototype.
11375 (make_tree_binfo): Remove.
11377 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11379 * tree.c (copy_node_stat): Rename to copy_node.
11380 (build_distinct_type_copy): Adjust.
11381 * tree.h (copy_node_stat): Adjust prototype.
11382 (copy_node): Remove macro.
11384 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11386 * tree.c (make_node_stat): rename to make_node.
11387 (build_tree_list_stat): Adjust.
11388 (build0_stat): Likewise.
11389 (build2_stat): Likewise.
11390 (build3_stat): Likewise.
11391 (build4_stat): Likewise.
11392 (build5_stat): Likewise.
11393 (build_decl_stat): Likewise.
11394 * tree.h (make_node_stat): Adjust prototype.
11395 (make_node): remove macro.
11397 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
11399 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
11400 (PPC_FEATURE2_SCV): Likewise.
11401 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
11403 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11405 * config/aarch64/aarch64.c
11406 (aarch64_internal_mov_immediate): Add new special pattern.
11407 * config/aarch64/aarch64.md (*movdi_aarch64):
11408 Add reg/32bit const mov case.
11410 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11411 Richard Sandiford <richard.sandiford@linaro.org>
11413 * config/aarch64/aarch64.md (mov<mode>): Generalize.
11414 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11415 Add integer and movi cases.
11416 (movi-split-hf-df-sf split, fp16): New.
11417 (enabled): Added TARGET_FP_F16INST.
11418 * config/aarch64/iterators.md (GPF_HF): New.
11419 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
11421 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11423 * config/aarch64/aarch64.c
11424 (aarch64_simd_container_mode): Add prototype.
11425 (aarch64_expand_mov_immediate): Add HI support.
11426 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
11427 (aarch64_can_const_movi_rtx_p): New.
11428 (aarch64_preferred_reload_class):
11429 Remove restrictions of using FP registers for certain SIMD operations.
11430 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
11431 (aarch64_valid_floating_const): Add integer move validation.
11432 (aarch64_simd_imm_scalar_p): Remove.
11433 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
11434 (aarch64_legitimate_constant_p): Expand list of supported cases.
11435 * config/aarch64/aarch64-protos.h
11436 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
11437 (aarch64_reinterpret_float_as_int): New.
11438 (aarch64_simd_imm_scalar_p): Remove.
11439 * config/aarch64/constraints.md (Uvi): New.
11440 (Dd): Split into Ds and new Dd.
11441 * config/aarch64/aarch64.md (*movsi_aarch64):
11443 (*movdi_aarch64): Add SIMD mov case.
11445 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11447 * tree-predcom.c: (struct chain): Handle store-store chain in which
11448 stores for elimination only store loop invariant values.
11449 (execute_pred_commoning_chain): Ditto.
11450 (prepare_initializers_chain_store_elim): Ditto.
11451 (prepare_finalizers): Ditto.
11452 (is_inv_store_elimination_chain): New function.
11453 (initialize_root_vars_store_elim_1): New function.
11455 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11457 * tree-predcom.c: Revise general description of the pass.
11458 (enum chain_type): New enum type for store elimination.
11459 (struct chain): New field supporting store elimination.
11460 (struct component): Ditto.
11461 (dump_chain): Dump store-stores chain.
11462 (release_chain): Release resources.
11463 (split_data_refs_to_components): Compute and create component
11464 contains only stores for elimination.
11465 (get_chain_last_ref_at): New function.
11466 (make_invariant_chain): Initialization.
11467 (make_rooted_chain): Specify chain type in parameter and record it.
11468 (add_looparound_copies): Skip for store-stores chain.
11469 (determine_roots_comp): Compute type of chain and pass it to
11471 (initialize_root_vars_store_elim_2): New function.
11472 (finalize_eliminated_stores): New function.
11473 (remove_stmt): Handle store for elimination.
11474 (execute_pred_commoning_chain): Execute predictive commoning on
11475 store-store chains.
11476 (determine_unroll_factor): Skip unroll for store-stores chain.
11477 (prepare_initializers_chain_store_elim): New function.
11478 (prepare_initializers_chain): Hanlde store-store chain.
11479 (prepare_finalizers_chain, prepare_finalizers): New function.
11480 (tree_predictive_commoning_loop): Return integer value indicating
11481 if loop is unrolled or lcssa form is corrupted.
11482 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
11484 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11486 * tree-predcom.c (initialize_root): Delete.
11487 (execute_pred_commoning_chain): Initialize root vars and replace
11488 reference of non-combined chain directly, rather than call above
11491 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11493 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
11494 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
11496 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11498 * tree-predcom.c (struct chain): New field init_seq.
11499 (release_chain): Release init_seq.
11500 (prepare_initializers_chain): Record intialization stmts in above
11502 (insert_init_seqs): New function.
11503 (tree_predictive_commoning_loop): Call insert_init_seqs.
11505 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11507 * tree-predcom.c (determine_roots_comp): Skip trivial components.
11509 2017-07-28 Richard Biener <rguenther@suse.de>
11511 * match.pd: Remove superfluous :c.
11512 * genmatch.c (simplify::id): Add member.
11513 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
11515 (current_id): New global.
11516 (dt_node::parent): Move from ...
11517 (dt_operand::parent): ... here. Add for_id member.
11518 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
11519 (decision_tree::find_node): Relax order requirement when
11520 merging DT_TRUE nodes to ones inbetween the current simplify
11521 and the one we try to merge with. Add diagnostic whenever
11522 we need to enforce pattern order by not merging.
11523 (decision_tree::insert): Set current_id.
11524 (decision_tree::print_node): Dump parent node and for_id.
11525 (parser::last_id): Add member.
11526 (parser::push_simplify): Assign unique id.
11527 (parser::parser): Initialize last_id.
11529 2017-07-28 Martin Liska <mliska@suse.cz>
11532 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
11533 gimple_build_debug_bind.
11535 2017-07-28 Richard Biener <rguenther@suse.de>
11537 PR tree-optimization/81502
11538 * match.pd: Add pattern combining BIT_INSERT_EXPR with
11540 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
11542 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
11543 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
11544 for BIT_FIELD_REF args.
11545 * fold-const.c (make_bit_field_ref): Likewise.
11546 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
11548 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11551 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
11552 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
11553 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
11554 Or it into SANITIZER_UNDEFINED.
11555 * ubsan.c: Include gimple-fold.h and varasm.h.
11556 (ubsan_expand_ptr_ifn): New function.
11557 (instrument_pointer_overflow): New function.
11558 (maybe_instrument_pointer_overflow): New function.
11559 (instrument_object_size): Formatting fix.
11560 (pass_ubsan::execute): Call instrument_pointer_overflow
11561 and maybe_instrument_pointer_overflow.
11562 * internal-fn.c (expand_UBSAN_PTR): New function.
11563 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
11564 * sanitizer.def (__ubsan_handle_pointer_overflow,
11565 __ubsan_handle_pointer_overflow_abort): New builtins.
11566 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
11567 * internal-fn.def (UBSAN_PTR): New internal function.
11568 * opts.c (sanitizer_opts): Add pointer-overflow.
11569 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
11570 * fold-const.c (build_range_check): Compute pointer range check in
11571 integral type if pointer arithmetics would be needed. Formatting
11574 2017-07-28 Martin Liska <mliska@suse.cz>
11577 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
11578 parameters that are of a variable-length.
11580 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
11582 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
11584 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
11585 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
11586 (CRT_CALL_STATIC_FUNCTION): Likewise.
11587 (ASM_DEFAULT_SPEC): New define.
11588 (ASM_SPEC32): Likewise.
11589 (ASM_SPEC64): Likewise.
11590 (ASM_SPEC_COMMON): Likewise.
11591 (ASM_SPEC): Likewise.
11592 (INVALID_64BIT): Likewise.
11593 (LINK_OS_DEFAULT_SPEC): Likewise.
11594 (LINK_OS_SPEC32): Likewise.
11595 (LINK_OS_SPEC64): Likewise.
11596 (POWERPC_LINUX): Likewise.
11597 (PTRDIFF_TYPE): Likewise.
11598 (RESTORE_FP_PREFIX): Likewise.
11599 (RESTORE_FP_SUFFIX): Likewise.
11600 (SAVE_FP_PREFIX): Likewise.
11601 (SAVE_FP_SUFFIX): Likewise.
11602 (SIZE_TYPE): Likewise.
11603 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
11604 (TARGET_64BIT): Likewise.
11605 (TARGET_64BIT): Likewise.
11606 (TARGET_AIX): Likewise.
11607 (WCHAR_TYPE_SIZE): Likewise.
11608 (WCHAR_TYPE): Undefine.
11609 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
11610 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
11611 (CPP_OS_RTEMS_SPEC): Delete.
11612 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
11613 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
11615 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
11617 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11619 PR tree-optimization/81578
11620 * tree-parloops.c (build_new_reduction): Bail out if
11621 reduction_code isn't one of the standard OpenMP reductions.
11622 Move the details printing after that decision.
11624 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11626 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
11627 related to reload_in_progress.
11628 (splat_input_operand): Likewise.
11629 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
11631 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
11633 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
11634 (TARGET_INSTANTIATE_DECLS): Likewise.
11635 (legitimate_indexed_address_p): Delete reload_in_progress code.
11636 (rs6000_debug_legitimate_address_p): Likewise.
11637 (rs6000_eliminate_indexed_memrefs): Likewise.
11638 (rs6000_emit_le_vsx_store): Likewise.
11639 (rs6000_emit_move_si_sf_subreg): Likewise.
11640 (rs6000_emit_move): Likewise.
11641 (register_to_reg_type): Likewise.
11642 (rs6000_pre_atomic_barrier): Likewise.
11643 (rs6000_machopic_legitimize_pic_address): Likewise.
11644 (rs6000_allocate_stack_temp): Likewise.
11645 (rs6000_address_for_fpconvert): Likewise.
11646 (rs6000_address_for_altivec): Likewise.
11647 (rs6000_secondary_memory_needed_rtx): Delete function.
11648 (rs6000_check_sdmode): Likewise.
11649 (rs6000_alloc_sdmode_stack_slot): Likewise.
11650 (rs6000_instantiate_decls): Likewise.
11651 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
11652 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
11653 Delete reload_in_progress.
11654 (*vec_reload_and_plus_<mptrsize>): Likewise.
11655 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
11656 (vsx_div_v2di): Likewise.
11657 (vsx_udiv_v2di): Likewise.
11659 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11661 * config/rs6000/rs6000.opt (mlra): Replace with stub.
11662 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
11663 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
11664 (rs6000_debug_reg_global): Delete print of LRA status.
11665 (rs6000_option_override_internal): Delete dead LRA related code.
11666 (rs6000_lra_p): Delete function.
11667 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
11669 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
11671 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
11672 * config/riscv/rtems.h: New file.
11674 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11675 Sudakshina Das <sudi.das@arm.com>
11677 * config/aarch64/aarch64.md
11678 (define_split for and<mode>3nr_compare): Move
11679 non aarch64_logical_operand to a register.
11680 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
11681 register immediate operand to a register.
11682 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
11684 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11686 PR middle-end/81564
11687 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
11689 2017-07-27 Richard Biener <rguenther@suse.de>
11691 PR tree-optimization/81573
11692 PR tree-optimization/81494
11693 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
11694 multi defuse cycle case.
11696 2017-07-27 Richard Biener <rguenther@suse.de>
11698 PR tree-optimization/81571
11699 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
11702 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
11704 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
11705 earlier and only if MASK_FPU is set. Adjust formatting.
11707 2017-07-27 Martin Liska <mliska@suse.cz>
11709 * opt-functions.awk: Add validation of value of Init.
11710 * optc-gen.awk: Pass new argument.
11712 2017-07-27 Martin Liska <mliska@suse.cz>
11714 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11715 Fix wrong condition.
11717 2017-07-27 Martin Liska <mliska@suse.cz>
11719 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
11720 BBs and edges seen by autoFDO.
11722 2017-07-27 Richard Biener <rguenther@suse.de>
11724 PR tree-optimization/81502
11725 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
11726 with incompatible but same sized type.
11727 (execute_update_addresses_taken): Likewise.
11729 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
11731 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
11732 flag_tree_loop_vectorize rather than flag_tree_vectorize.
11734 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11737 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
11738 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
11739 Change s_operand to memory_operand.
11741 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
11743 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11744 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
11745 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
11746 Emit instructions rather than returning an expression. Handle TFmode
11747 and KFmode by casting to TImode.
11748 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
11749 (rs6000_emit_le_vsx_store): Likewise.
11750 * config/rs6000/vsx.md (VSX_TI): New iterator.
11751 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
11752 (*vsx_le_undo_permute_<mode>): Likewise.
11753 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
11754 emit the split sequence.
11755 (*vsx_le_perm_store_<mode>): Likewise.
11757 2017-07-27 Jakub Jelinek <jakub@redhat.com>
11759 PR tree-optimization/81555
11760 PR tree-optimization/81556
11761 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
11762 if true, force CHANGED for the recursive invocation.
11763 (reassociate_bb): Remember original length of ops array, pass
11764 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
11766 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
11767 attributes for noipa attribute. For naked attribute use
11768 lookup_attribute first before lookup_attribute_spec.
11769 * final.c (rest_of_handle_final): Disable IPA RA for functions with
11771 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
11772 for functions with noipa attribute.
11773 (cgraph_externally_visible_p): Return true for functions with noipa
11775 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
11776 for functions with noipa attribute.
11777 * doc/extend.texi: Document noipa function attribute.
11778 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
11779 also for functions with noipa attribute.
11780 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
11782 2017-07-26 Andrew Pinski <apinski@cavium.com>
11784 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
11785 vec_unalign_load_cost and vec_unalign_store_cost.
11787 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11789 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
11790 -mvsx-small-integer option.
11791 (ISA_3_0_MASKS_IEEE): Likewise.
11792 (OTHER_VSX_VECTOR_MASKS): Likewise.
11793 (POWERPC_MASKS): Likewise.
11794 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
11795 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
11796 code, only testing for DImode being allowed in non-VSX floating
11798 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
11799 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
11801 (rs6000_option_override_internal): Delete -mvsx-small-integer.
11802 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
11803 TARGET_P8_VECTOR test.
11804 (rs6000_secondary_reload_simple_move): Likewise.
11805 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
11806 since TARGET_P9_VECTOR was already tested.
11807 (rs6000_opt_masks): Remove -mvsx-small-integer.
11808 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
11809 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11811 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
11812 test for TARGET_VEXTRACTUB was used, and that uses
11814 (p9 extract splitter): Likewise.
11815 (vsx_extract_<mode>_di_p9): Likewise.
11816 (vsx_extract_<mode>_store_p9): Likewise.
11817 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
11818 for TARGET_P9_VECTOR was used. Delete code that is now dead with
11819 the elimination of TARGET_VSX_SMALL_INTEGER.
11820 (vsx_extract_<mode>_p8): Likewise.
11821 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
11822 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
11823 (vsx_set_<mode>_p9): Likewise.
11824 (vsx_set_v4sf_p9): Likewise.
11825 (vsx_set_v4sf_p9_zero): Likewise.
11826 (vsx_insert_extract_v4sf_p9): Likewise.
11827 (vsx_insert_extract_v4sf_p9_2): Likewise.
11828 * config/rs6000/rs6000.md (sign extend splitter): Change
11829 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
11830 (floatsi<mode>2_lfiwax_mem): Likewise.
11831 (floatunssi<mode>2_lfiwzx_mem): Likewise.
11832 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
11833 since a test for TARGET_P9_VECTOR was used.
11834 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11835 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
11836 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11837 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11838 TARGET_P8_VECTOR test.
11839 (fix_trunc<mode>si2_stfiwx): Likewise.
11840 (fix_trunc<mode>si2_internal): Likewise.
11841 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
11842 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11844 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11845 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11846 TARGET_P8_VECTOR test.
11847 (fixuns_trunc<mode>si2_stfiwx): Likewise.
11848 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
11849 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11851 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11852 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
11853 since a test for TARGET_P9_VECTOR was used.
11854 (splitter for loading small constants): Likewise.
11856 2017-07-26 Andrew Pinski <apinski@cavium.com>
11858 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
11861 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
11864 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
11865 (fp_valid_at): Likewise.
11867 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11869 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
11870 (qdf24xx_addrcost_table): Likewise.
11871 (cortexa57_tunings): Update to use generic_branch_cost.
11872 (cortexa72_tunings): Likewise.
11873 (cortexa73_tunings): Likewise.
11874 (qdf24xx_tunings): Likewise.
11876 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11878 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
11879 (thunderx2t99_branch_cost): Likewise.
11880 (cortexa35_tunings): Update to use generic_branch_cost.
11881 (cortexa53_tunings): Likewise.
11882 (cortexa57_tunings): Likewise.
11883 (cortexa72_tunings): Likewise.
11884 (cortexa73_tunings): Likewise.
11885 (thunderx2t99_tunings): Likewise.
11887 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11889 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
11890 (sparc_option_override): Honour MASK_FSMULD.
11891 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
11892 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
11893 * config/sparc/sparc.opt (mfsmuld): New option.
11894 * doc/invoke.texi (mfsmuld): Document option.
11896 2017-07-26 Marek Polacek <polacek@redhat.com>
11898 PR middle-end/70992
11899 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
11901 2017-07-26 Richard Biener <rguenther@suse.de>
11903 * gimple-match-head.c (do_valueize): Return OP if valueize
11905 (get_def): New helper to get at the def stmt of a SSA name
11906 if valueize allows.
11907 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
11908 do_valueize to get at the def stmt.
11909 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
11911 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
11913 PR middle-end/46932
11914 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
11916 2017-07-26 Martin Liska <mliska@suse.cz>
11919 * function.c (expand_function_start): Make expansion of
11920 nonlocal_goto_save_area after parm_birth_insn.
11922 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11924 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
11925 from all CPU target flags enable members.
11927 2017-07-26 Richard Biener <rguenther@suse.de>
11929 * genmatch.c (dt_simplify::gen): Make iterator vars const.
11930 (decision_tree::gen): Make 'type' const.
11931 (write_predicate): Likewise.
11933 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11935 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
11936 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
11937 (rs6000_option_override_internal): Likewise.
11938 (rs6000_expand_vector_set): Likewise.
11939 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
11940 (TARGET_UPPER_REGS_SF): Likewise.
11941 (TARGET_UPPER_REGS_DI): Likewise.
11942 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
11943 (TARGET_DIRECT_MOVE_64BIT): Likewise.
11944 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
11945 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11946 (Splitters for DI constants in Altivec registers): Likewise.
11947 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
11948 (vsx_set_v4sf_p9): Likewise.
11949 (vsx_set_v4sf_p9_zero): Likewise.
11950 (vsx_insert_extract_v4sf_p9): Likewise.
11951 (vsx_insert_extract_v4sf_p9_2): Likewise.
11953 2017-07-25 Carl Love <cel@us.ibm.com>
11955 * doc/extend.texi: Update the built-in documentation file for the
11956 existing built-in functions
11957 vector signed char vec_cnttz (vector signed char);
11958 vector unsigned char vec_cnttz (vector unsigned char);
11959 vector signed short vec_cnttz (vector signed short);
11960 vector unsigned short vec_cnttz (vector unsigned short);
11961 vector signed int vec_cnttz (vector signed int);
11962 vector unsigned int vec_cnttz (vector unsigned int);
11963 vector signed long long vec_cnttz (vector signed long long);
11964 vector unsigned long long vec_cnttz (vector unsigned long long);
11966 2017-07-25 Andrew Pinski <apinski@cavium.com>
11968 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
11969 accesses where the use is for the first operand of a BIT_INSERT.
11971 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11974 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11975 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11977 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11979 * config/i386/gstabs.h: Delete.
11980 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
11982 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
11984 * config/i386/i386.c (ix86_decompose_address): Do not check for
11985 register RTX when looking at index_reg or base_reg.
11986 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
11988 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
11990 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
11991 to update EH info here.
11993 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11995 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
11997 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11999 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12001 2017-07-25 Torsten Duwe <duwe@suse.de>
12003 * common.opt: Introduce -fpatchable-function-entry
12004 command line option, and its variables function_entry_patch_area_size
12005 and function_entry_patch_area_start.
12006 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12007 including a two-value parser.
12008 * target.def (print_patchable_function_entry): New target hook.
12009 * targhooks.h (default_print_patchable_function_entry): New function.
12010 * targhooks.c (default_print_patchable_function_entry): Likewise.
12011 * toplev.c (process_options): Switch off IPA-RA if
12012 patchable function entries are being generated.
12013 * varasm.c (assemble_start_function): Look at the
12014 patchable-function-entry command line switch and current
12015 function attributes and maybe generate NOP instructions by
12016 calling the print_patchable_function_entry hook.
12017 * doc/extend.texi: Document patchable_function_entry attribute.
12018 * doc/invoke.texi: Document -fpatchable_function_entry
12019 command line option.
12020 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12022 * doc/tm.texi: Re-generate.
12024 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12027 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12028 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12030 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12032 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12034 2017-07-25 Richard Biener <rguenther@suse.de>
12036 PR tree-optimization/81455
12037 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12038 not walk in cycles when looking for guards.
12040 2017-07-25 Richard Biener <rguenther@suse.de>
12042 PR tree-optimization/81529
12043 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12044 when optimizing backedge uses.
12046 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12048 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12050 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12051 to dl_section_ref. On AIX, append an expression to subtract
12052 the size of the section length to dl_section_ref.
12054 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12056 * configure.ac: If any of the config.* scripts fail, exit 1.
12057 * configure: Regenerate.
12059 2017-07-25 Richard Biener <rguenther@suse.de>
12061 PR middle-end/81546
12062 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12063 of immediate uses, be more verbose on errors.
12065 2017-07-25 Richard Biener <rguenther@suse.de>
12067 PR tree-optimization/81510
12068 * tree-vect-loop.c (vect_is_simple_reduction): When the
12069 reduction stmt is not inside the loop bail out.
12071 2017-07-25 Richard Biener <rguenther@suse.de>
12073 PR tree-optimization/81303
12074 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12075 profitability check against LOOP_VINFO_NITERSM1.
12077 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12079 * domwalk.c (cmp_bb_postorder): Simplify.
12080 (sort_bbs_postorder): New function. Use it...
12081 (dom_walker::walk): ...here to optimize common cases.
12083 2017-07-25 Martin Liska <mliska@suse.cz>
12086 * ipa-visibility.c (function_and_variable_visibility): Make the
12087 redirection just on target that supports aliasing.
12088 Fix GNU coding style.
12090 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
12093 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
12095 * config/arm/rtems.h: Likewise.
12096 * config/bfin/rtems.h: Likewise.
12097 * config/i386/rtemself.h: Likewise.
12098 * config/lm32/rtems.h: Likewise.
12099 * config/m32c/rtems.h: Likewise.
12100 * config/m68k/rtemself.h: Likewise.
12101 * config/microblaze/rtems.h: Likewise.
12102 * config/mips/rtems.h: Likewise.
12103 * config/moxie/rtems.h: Likewise.
12104 * config/nios2/rtems.h: Likewise.
12105 * config/powerpcspe/rtems.h: Likewise.
12106 * config/rs6000/rtems.h: Likewise.
12107 * config/rtems.h: Likewise.
12108 * config/sh/rtems.h: Likewise.
12109 * config/sh/rtemself.h: Likewise.
12110 * config/sparc/rtemself.h: Likewise.
12112 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12115 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
12116 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
12117 * tree-ssa-structalias.c (alias_get_name): Same.
12119 2017-07-25 Bin Cheng <bin.cheng@arm.com>
12122 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
12123 instructions if no du chain is found.
12125 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12127 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
12129 2017-07-25 Richard Biener <rguenther@suse.de>
12131 PR middle-end/81505
12132 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
12135 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12137 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
12138 upper-regs options.
12139 (ISA_2_7_MASKS_SERVER): Likewise.
12140 (ISA_3_0_MASKS_IEEE): Likewise.
12141 (OTHER_P8_VECTOR_MASKS): Likewise.
12142 (OTHER_VSX_VECTOR_MASKS): Likewise.
12143 (POWERPC_MASKS): Likewise.
12144 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
12145 duplicate list of options.
12146 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
12147 explicit -mupper-regs options.
12148 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
12149 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
12150 alias for -mupper-regs-df.
12151 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12152 (rs6000_init_hard_regno_mode_ok): Likewise.
12153 (rs6000_option_override_internal): Likewise.
12154 (rs6000_opt_masks): Likewise.
12155 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
12156 options in terms of whether -mvsx or -mpower8-vector was used.
12157 (TARGET_UPPER_REGS_DI): Likewise.
12158 (TARGET_UPPER_REGS_SF): Likewise.
12159 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
12160 -mupper-regs-* options.
12162 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12164 * passes.c (emergency_dump_function): Print some empty lines and a
12165 header before the RTL dump.
12167 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12169 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
12171 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
12174 * config/aarch64/aarch64.c (aarch64_classify_symbol):
12175 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
12177 2017-07-24 Carl Love <cel@us.ibm.com>
12179 * config/rs6000/rs6000-c.c: Add support for built-in functions
12180 vector float vec_extract_fp32_from_shorth (vector unsigned short);
12181 vector float vec_extract_fp32_from_shortl (vector unsigned short);
12182 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
12183 vec_extract_fp_from_shortl): Add defines for the two builtins.
12184 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
12185 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
12187 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
12188 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
12189 * doc/extend.texi: Update the built-in documentation file for the
12190 new built-in function.
12192 2017-07-24 Jakub Jelinek <jakub@redhat.com>
12195 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
12197 * doc/generic.texi: Likewise.
12198 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12199 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12201 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
12203 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
12204 (aarch64_mls_elt_merge<mode>): Likewise.
12206 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
12208 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12209 having __cxa_atexit.
12211 2017-07-23 Michael Collison <michael.collison@arm.com>
12213 * config/arm/arm.c (arm_option_override): Deprecate
12214 use of -mstructure-size-boundary.
12215 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
12216 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
12218 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12221 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12222 Reduce cost estimate for direct moves.
12224 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
12227 * config/i386/i386.c (ix86_option_override_internal): Disable
12228 BMI, BMI2 and TBM instructions for -m16.
12230 2017-07-21 Carl Love <cel@us.ibm.com>
12232 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12233 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12234 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12235 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12236 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12237 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12238 VMULOSW): New enum "unspec" values.
12239 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12240 altivec_vmulosw): New patterns.
12241 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12242 VMULOSW): Add definitions.
12244 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
12246 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
12247 (qdf24xx): Likewise.
12248 * config/aarch64/aarch64-options-extensions.def (rdma); New.
12249 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
12250 (AARCH64_FL_V8_1): Renumber.
12251 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
12252 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
12253 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
12254 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
12255 rdma to feature modifiers list.
12257 2017-07-21 Yury Gribov <tetra2005@gmail.com>
12259 PR middle-end/56727
12260 * ipa-visibility (function_and_variable_visibility): Convert
12261 recursive PLT call to direct call if appropriate.
12263 2017-07-21 Andrew Pinski <apinski@cavium.com>
12265 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
12266 operand 1 to see if the types precision matches.
12267 * fold-const.c (operand_equal_p): Likewise.
12269 2017-07-21 Richard Biener <rguenther@suse.de>
12271 PR tree-optimization/81303
12272 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
12273 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
12274 (vect_peeling_hash_get_lowest_cost): Adjust.
12275 (vect_enhance_data_refs_alignment): Likewise. Use
12276 vect_get_peeling_costs_all_drs to compute the penalty for no
12277 peeling to match up costs.
12279 2017-07-21 Richard Biener <rguenther@suse.de>
12281 PR tree-optimization/81500
12282 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
12283 we didn't identify a reduction path.
12285 2017-07-21 Tom de Vries <tom@codesourcery.com>
12286 Cesar Philippidis <cesar@codesourcery.com>
12288 PR gcov-profile/81442
12289 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
12292 2017-07-21 Tom de Vries <tom@codesourcery.com>
12295 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
12297 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
12298 nvptx_override_options_after_change.
12300 2017-07-21 Ulrich Drepper <drepper@redhat.com>
12302 * dwarf2out.c (output_file_names): Avoid double testing for
12303 dwarf_version >= 5.
12305 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
12307 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
12309 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12311 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
12313 (try_crossjump_to_edge): Do not punt on partitioned functions.
12315 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12317 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
12318 Put all BBs reachable only via paths crossing cold region to cold
12320 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
12322 2016-07-21 Richard Biener <rguenther@suse.de>
12324 PR tree-optimization/81303
12325 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
12326 into account prologue and epilogue iterations when raising
12327 min_profitable_iters to sth at least covering one vector iteration.
12329 2017-07-21 Tamar Christina <tamar.christina@arm.com>
12331 * config/arm/arm.c (arm_test_cpu_arch_dat):
12334 2017-07-20 Nathan Sidwell <nathan@acm.org>
12336 Remove TYPE_METHODS.
12337 * tree.h (TYPE_METHODS): Delete.
12338 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
12339 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
12340 (dbxout_type_methods): Scan TYPE_FIELDS.
12341 (dbxout_type): Don't check TYPE_METHODS here.
12342 * function.c (use_register_for_decl): Always ignore register for
12343 class types when not optimizing.
12344 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
12345 * tree.c (free_lang_data_in_type): Stitch out member functions and
12346 templates from TYPE_FIELDS.
12347 (build_distinct_type_copy, verify_type_variant,
12348 verify_type): Member fns are on TYPE_FIELDS.
12349 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
12350 * tree-pretty-print.c (dump_generic_node): Likewise.
12352 2017-07-20 Jakub Jelinek <jakub@redhat.com>
12355 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
12356 V2TImode and V4TImode.
12357 (ix86_expand_vector_extract): Likewise.
12358 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
12359 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
12360 (ssescalarmode): Handle V4TImode and V2TImode.
12361 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
12362 (*vec_extractv2ti, *vec_extractv4ti): New insns.
12363 (VEXTRACTI128_MODE): New mode iterator.
12364 (splitter for *vec_extractv?ti first element): New.
12365 (VEC_INIT_MODE): New mode iterator.
12366 (vec_init<mode>): Consolidate 3 expanders into one using
12367 VEC_INIT_MODE mode iterator.
12369 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12371 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
12372 non_spilled_static_chain_regno_p.
12374 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12376 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
12378 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
12380 * bb-reorder.c (connect_traces): Allow copying of blocks within
12383 2017-07-20 Richard Biener <rguenther@suse.de>
12385 * gimple.h (gimple_phi_result): Add gphi * overload.
12386 (gimple_phi_result_ptr): Likewise.
12387 (gimple_phi_arg): Likewise. Adjust index assert to only
12388 allow actual argument accesses rather than all slots available
12390 (gimple_phi_arg_def): Add gphi * overload.
12391 * tree-phinodes.c (make_phi_node): Initialize only actual
12393 (resize_phi_node): Clear memory not covered by old node,
12394 do not initialize excess argument slots.
12395 (reserve_phi_args_for_new_edge): Initialize new argument slot
12398 2017-07-20 Bin Cheng <bin.cheng@arm.com>
12400 PR tree-optimization/81388
12402 2016-07-21 Bin Cheng <bin.cheng@arm.com>
12404 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
12405 by removing computation of may_be_zero.
12407 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12408 Tom de Vries <tom@codesourcery.com>
12410 PR middle-end/81030
12411 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
12412 when gimple level profile disagrees with what RTL expander did.
12414 2017-07-20 Richard Biener <rguenther@suse.de>
12416 PR tree-optimization/61171
12417 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
12418 (vect_analyze_stmt): Add slp instance parameter.
12419 (vectorizable_reduction): Likewise.
12420 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
12421 (vect_is_simple_reduction): Deal with chains not detected
12422 as SLP reduction chain, specifically not properly associated
12423 chains containing a mix of plus/minus.
12424 (get_reduction_op): Remove.
12425 (get_initial_defs_for_reduction): Simplify, pass in whether
12426 this is a reduction chain, pass in the SLP node for the PHIs.
12427 (vect_create_epilog_for_reduction): Get the SLP instance as
12429 (vectorizable_reduction): Get the SLP instance as arg.
12430 During analysis remember the SLP node with the PHIs in the
12431 instance. Simplify getting at the vectorized reduction PHIs.
12432 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
12433 through SLP instance.
12434 (vect_slp_analyze_operations): Likewise.
12435 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
12436 (vect_transform_stmt): Likewise.
12438 2017-07-20 Tom de Vries <tom@codesourcery.com>
12440 PR tree-optimization/81489
12441 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
12442 read of phi arg location to before loop that modifies phi.
12444 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12446 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
12449 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12451 PR middle-end/81331
12452 * except.c (execute): Fix ordering issue.
12454 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12456 PR rtl-optimization/81423
12457 * combine.c (make_compound_operation_int): Don't try to optimize
12458 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
12460 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12462 PR rtl-optimization/81423
12463 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
12464 with a constant that is -1 in the truncated to mode.
12466 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12468 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
12469 (determine_unlikely_bbs): ... here.
12470 * predict.h (propagate_unlikely_bbs_forward): Declare.
12471 * cfgexpand.c (pass_expand::execute): Use it.
12472 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
12474 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
12475 propagate_unlikely_bbs_forward.
12477 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12479 PR middle-end/81331
12480 * except.c (maybe_add_nop_after_section_switch): New function.
12483 2017-07-19 Tom de Vries <tom@codesourcery.com>
12485 * gimple.h (gimple_phi_set_arg): Make assert more strict.
12487 2017-07-19 Tom de Vries <tom@codesourcery.com>
12489 * gimple.h (gimple_phi_arg): Make assert more strict.
12491 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
12493 * config.gcc (powerpc*-*-*): Add mmintrin.h.
12494 * config/rs6000/mmintrin.h: New file.
12495 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
12497 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12499 PR tree-optimization/81346
12500 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
12502 2017-07-19 Tom de Vries <tom@codesourcery.com>
12504 * config/nvptx/nvptx.md (VECIM): Add V2DI.
12506 2017-07-19 Tom de Vries <tom@codesourcery.com>
12508 * config/nvptx/nvptx-modes.def: Add V2DImode.
12509 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
12510 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
12511 (nvptx_output_mov_insn): Handle lack of mov.b128.
12512 (nvptx_print_operand): Handle 'H' and 'L' codes.
12513 (nvptx_vector_mode_supported): Allow V2DImode.
12514 (nvptx_preferred_simd_mode): New function.
12515 (nvptx_data_alignment): New function.
12516 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
12517 nvptx_preferred_simd_mode.
12518 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
12520 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
12522 2017-07-19 Tom de Vries <tom@codesourcery.com>
12524 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
12525 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
12526 (nvptx_vector_mode_supported): New function. Allow V2SImode.
12527 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
12528 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
12529 (mov<VECIM>_insn): New define_insn.
12530 (define_expand "mov<VECIM>): New define_expand.
12532 2017-07-19 Tom de Vries <tom@codesourcery.com>
12534 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
12536 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12538 PR tree-optimization/81346
12539 * fold-const.h (fold_div_compare, range_check_type): Declare.
12540 * fold-const.c (range_check_type): New function.
12541 (build_range_check): Use range_check_type.
12542 (fold_div_compare): No longer static, rewritten into
12543 a match.pd helper function.
12544 (fold_comparison): Don't call fold_div_compare here.
12545 * match.pd (X / C1 op C2): New optimization using fold_div_compare
12546 as helper function.
12548 2017-07-19 Nathan Sidwell <nathan@acm.org>
12550 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
12551 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
12552 * tree.c (find_decls_types_r, verify_type): Use
12553 TYPE_{MIN,MAX}_VALUE_RAW.
12554 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
12555 (hash_tree): Likewise.
12556 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12558 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12561 2017-07-18 Tom de Vries <tom@codesourcery.com>
12563 PR middle-end/81464
12564 * omp-expand.c (expand_omp_for_static_chunk): Handle
12565 equal-argument loop exit phi.
12567 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
12570 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
12571 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
12572 operand 2 predicate.
12573 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
12574 operand 2 predicate.
12575 (ror,rol -> rorx splitters): Use const_int_operand as
12576 operand 2 predicate.
12578 2017-06-18 Richard Biener <rguenther@suse.de>
12580 PR tree-optimization/81410
12581 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
12582 the gap in the ! slp_perm SLP case after each group.
12584 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12586 PR middle-end/81463
12587 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
12590 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12592 PR middle-end/81462
12593 * predict.c (set_even_probabilities): Cleanup; do not affect
12594 probabilities that are already known.
12595 (combine_predictions_for_bb): Call even when count is set.
12597 2017-07-18 Nathan Sidwell <nathan@acm.org>
12599 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
12602 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12605 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
12606 optimization for loop niter analysis.
12608 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
12611 * config/avr/avr.c (avr_optimize_casesi): Don't use
12612 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
12614 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12616 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
12617 body_cost_vec from _vect_peel_extended_info.
12618 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
12619 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
12622 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12624 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
12626 2017-07-18 Richard Biener <rguenther@suse.de>
12628 PR tree-optimization/80620
12629 PR tree-optimization/81403
12630 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
12631 info when re-using a VN table entry.
12633 2017-07-18 Richard Biener <rguenther@suse.de>
12635 PR tree-optimization/81418
12636 * tree-vect-loop.c (vectorizable_reduction): Properly compute
12637 vectype_in. Verify that with lane-reducing reduction operations
12638 we have a single def-use cycle.
12640 2017-07-17 Carl Love <cel@us.ibm.com>
12642 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
12644 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12645 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12646 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12647 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12648 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12649 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12650 VMULOSW): New enum "unspec" values.
12651 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12652 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12653 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12654 altivec_vmulosw): New patterns.
12655 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12656 VMULOSW): Add definitions.
12657 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
12659 * config/alpha/alpha.c: Include predict.h.
12661 2017-07-17 Yury Gribov <tetra2005@gmail.com>
12663 * tree-vrp.c (compare_assert_loc): Fix comparison function
12664 to return predictable results.
12666 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12668 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
12670 (subdi3): Likewise.
12671 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
12672 * doc/invoke.texi (mexpand-adddi): Update text.
12674 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12676 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
12677 that also clobbers the CC register. The old expand code is moved
12679 (*arc_clzsi2): ... here.
12680 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
12681 the CC register. The old expand code is moved to ...
12682 (arc_ctzsi2): ... here.
12684 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12686 * config/arc/arc.opt (mindexed-loads): Use initial value
12687 TARGET_INDEXED_LOADS_DEFAULT.
12688 (mauto-modify-reg): Use initial value
12689 TARGET_AUTO_MODIFY_REG_DEFAULT.
12690 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12691 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12692 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12693 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12695 2017-07-17 Martin Liska <mliska@suse.cz>
12698 * opts.c (finish_options): Do not allow -fgnu-tm
12699 w/ -fsanitize={kernel-,}address. Say sorry.
12701 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12704 * tree-loop-distribution.c (classify_partition): Only assert on
12705 numer of iterations.
12706 (merge_dep_scc_partitions): Delete prameter. Update function call.
12707 (distribute_loop): Remove code handling loop with unknown niters.
12708 (pass_loop_distribution::execute): Skip loop with unknown niters.
12710 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12713 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
12714 function sort_partitions_by_post_order.
12716 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12718 PR tree-optimization/81374
12719 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
12720 the max index of basic blocks, rather than number of basic blocks.
12722 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12724 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
12726 (arc_legitimate_pic_operand_p): Likewise.
12727 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
12729 (arc_needs_pcl_p): Likewise.
12730 (arc_legitimate_pc_offset_p): Likewise.
12731 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
12732 function is also used in constrains.md.
12733 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
12734 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
12735 PLUS. Only return true/false in known cases, otherwise assert.
12736 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
12737 is already called in arc_legitimate_constant_p.
12738 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
12740 (LEGITIMATE_PIC_OPERAND_P): Use
12741 arc_raw_symbolic_reference_mentioned_p function.
12742 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
12747 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12748 Andrew Burgess <andrew.burgess@embecosm.com>
12750 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
12751 (arc_return_address_register): New function.
12752 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
12753 (arc_handle_fndecl_attribute): Add naked attribute.
12754 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12755 (TARGET_WARN_FUNC_RETURN): Likewise.
12756 (arc_allocate_stack_slots_for_args): New function.
12757 (arc_warn_func_return): Likewise.
12758 (machine_function): Change type fn_type.
12759 (arc_compute_function_type): Consider new naked function type,
12760 change function return type.
12761 (arc_must_save_register): Adapt to handle new
12762 arc_compute_function_type's return type.
12763 (arc_expand_prologue): Likewise.
12764 (arc_expand_epilogue): Likewise.
12765 (arc_return_address_regs): Delete.
12766 (arc_return_address_register): New function.
12767 (arc_epilogue_uses): Use above function.
12768 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
12769 (arc_function_type): Change encoding, add naked type.
12770 (ARC_INTERRUPT_P): Change to handle the new encoding.
12771 (ARC_FAST_INTERRUPT_P): Likewise.
12772 (ARC_NORMAL_P): Define.
12773 (ARC_NAKED_P): Likewise.
12774 (arc_compute_function_type): Delete prototype.
12775 * config/arc/arc.md (in_ret_delay_slot): Use
12776 arc_return_address_register function.
12777 (simple_return): Likewise.
12778 (p_return_i): Likewise.
12780 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12782 PR tree-optimization/81428
12783 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
12784 can't be built for those types.
12786 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12788 Remove stuff dead since r239246.
12790 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
12791 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
12792 (avr_inform_devices): Remove dead stuff.
12794 2017-07-17 Tamar Christina <tamar.christina@arm.com>
12796 * config/arm/arm_neon.h: Fix softp typo.
12798 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12800 PR tree-optimization/81365
12801 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
12802 aggregate moves onto bb predecessor edges, make sure there are no
12803 loads that could alias the lhs in between the start of bb and the
12806 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12809 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
12810 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
12811 [LSHIFTRT, outer_code = TRUNCATE]: Same.
12813 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12815 PR tree-optimization/81396
12816 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
12817 (init_symbolic_number): Initialize it to 1.
12818 (perform_symbolic_merge): Add n_ops from both operands into the new
12820 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
12821 without base_addr as useless if they need more than one operation.
12822 (bswap_replace): Handle !bswap case for NULL base_addr.
12824 2017-07-17 Tom de Vries <tom@codesourcery.com>
12827 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
12830 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
12832 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
12833 conditional builtin define __FIX_LEON3FT_B2BST.
12835 2017-07-17 Daniel Cederman <cederman@gaisler.com>
12837 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
12838 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
12841 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12843 PR rtl-optimization/81424
12844 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
12845 to remove potential trapping from operands if -fnon-call-exceptions.
12847 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12849 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
12850 profile_proability for scalling.
12851 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
12853 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12855 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
12857 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12859 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
12860 fixpoint arithmetics.
12862 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12864 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
12865 fixpoint arithmetics.
12867 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12869 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
12870 fixpoint arithmetics.
12872 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12874 * profile-count.h (profile_probability::from_reg_br_prob_note,
12875 profile_probability::to_reg_br_prob_note): New functions.
12876 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
12877 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
12878 * predict.c (probability_reliable_p): Update.
12879 (edge_probability_reliable_p): Update.
12880 (br_prob_note_reliable_p): Update.
12881 (invert_br_probabilities): Update.
12882 (add_reg_br_prob_note): New function.
12883 (combine_predictions_for_insn): Update.
12884 * asan.c (asan_clear_shadow): Update.
12885 * cfgbuild.c (compute_outgoing_frequencies): Update.
12886 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12887 (update_br_prob_note): Update.
12888 (rtl_verify_edges): Update.
12889 (purge_dead_edges): Update.
12890 (fixup_reorder_chain): Update.
12891 * emit-rtl.c (try_split): Update.
12892 * ifcvt.c (cond_exec_process_insns): Update.
12893 (cond_exec_process_if_block): Update.
12894 (dead_or_predicable): Update.
12895 * internal-fn.c (expand_addsub_overflow): Update.
12896 (expand_neg_overflow): Update.
12897 (expand_mul_overflow): Update.
12898 * loop-doloop.c (doloop_modify): Update.
12899 * loop-unroll.c (compare_and_jump_seq): Update.
12900 * optabs.c (emit_cmp_and_jump_insn_1): Update.
12901 * predict.h: Update.
12902 * reorg.c (mostly_true_jump): Update.
12904 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
12905 * config/alpha/alpha.c (emit_unlikely_jump): Update.
12906 * config/arc/arc.c: (emit_unlikely_jump): Update.
12907 * config/arm/arm.c: (emit_unlikely_jump): Update.
12908 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
12909 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
12910 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
12911 (ix86_print_operand): Update.
12912 (ix86_split_fp_branch): Update.
12913 (predict_jump): Update.
12914 * config/ia64/ia64.c (ia64_print_operand): Update.
12915 * config/mmix/mmix.c (mmix_print_operand): Update.
12916 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
12917 (rs6000_expand_split_stack_prologue): Update.
12918 * config/rs6000/rs6000.c: Update.
12919 * config/s390/s390.c (s390_expand_vec_strlen): Update.
12920 (s390_expand_vec_movstr): Update.
12921 (s390_expand_cs_tdsi): Update.
12922 (s390_expand_split_stack_prologue): Update.
12923 * config/sh/sh.c (sh_print_operand): Update.
12924 (expand_cbranchsi4): Update.
12925 (expand_cbranchdi4): Update.
12926 * config/sparc/sparc.c (output_v9branch): Update.
12927 * config/spu/spu.c (get_branch_target): Update.
12928 (ea_load_store_inline): Update.
12929 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
12930 * config/tilepro/tilepro.c: Update.
12932 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12934 * gimplify.c (mostly_copy_tree_r): Revert latest change.
12935 (gimplify_save_expr): Likewise.
12937 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12939 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
12941 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12943 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
12945 * timevar.def (TV_IPA_FNSUMMARY): Define.
12947 2017-07-16 Daniel Cederman <cederman@gaisler.com>
12949 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
12950 to back store errata sensitive sequence from being generated.
12951 (sqrtdf2_fix): Likewise.
12953 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12955 * tree-ssa-threadupdate.c (compute_path_counts,
12956 update_joiner_offpath_counts): Use profile_probability.
12958 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
12961 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12963 * config/arm/arm-c.c (arm_cpu_builtins): Define
12964 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12966 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
12968 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12969 array entries to represent __ieee128 versions of the
12970 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
12971 scalar_extract_sig, and scalar_insert_exp built-in functions.
12972 (altivec_resolve_overloaded_builtin): Add special case handling
12973 for the __builtin_scalar_insert_exp function, as represented by
12974 the P9V_BUILTIN_VEC_VSIEDP constant.
12975 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
12976 exponent support for __ieee128 argument.
12977 (VSESQP): Add scalar extract signature support for __ieee128
12979 (VSTDCNQP): Add scalar test negative support for __ieee128
12981 (VSIEQP): Add scalar insert exponent support for __int128 argument
12982 with __ieee128 result.
12983 (VSIEQPF): Add scalar insert exponent support for __ieee128
12984 argument with __ieee128 result.
12985 (VSTDCQP): Add scalar test data class support for __ieee128
12987 (VSTDCNQP): Add overload support for scalar test negative with
12988 __ieee128 argument.
12989 (VSTDCQP): Add overload support for scalar test data class
12990 __ieee128 argument.
12991 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
12992 UNSPEC_VSX_SXSIGDP.
12993 (UNSPEC_VSX_SIEXPQP): New constant.
12994 (xsxexpqp): New insn for VSX scalar extract exponent quad
12996 (xsxsigqp): New insn for VSX scalar extract significand quad
12998 (xsiexpqpf): New insn for VSX scalar insert exponent quad
12999 precision with floating point argument.
13000 (xststdcqp): New expand for VSX scalar test data class quad
13002 (xststdcnegqp): New expand for VSX scalar test negative quad
13004 (xststdcqp): New insn to match expansions for VSX scalar test data
13005 class quad precision and VSX scalar test negative quad precision.
13006 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13007 special case operand checking to enforce that second operand of
13008 VSX scalar test data class with quad precision argument is a 7-bit
13010 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13011 prototypes and descriptions of __ieee128 versions of
13012 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13013 scalar_test_data_class, and scalar_test_neg built-in functions.
13015 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13017 PR tree-optimization/81162
13018 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13019 replace a negate with an add.
13021 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13023 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13025 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13027 * config/arm/arm-c.c (arm_cpu_builtins): Define
13028 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13030 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13032 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13033 (armv8-r): Set ARM Cortex-R52 as default CPU.
13034 * config/arm/arm-tables.opt: Regenerate.
13035 * config/arm/arm-tune.md: Regenerate.
13036 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13038 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13039 extension for -mcpu=cortex-r52.
13041 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13043 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13044 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13045 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13046 (fp-armv8): Define it as FP_ARMv8 only.
13047 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13048 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13050 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13051 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13052 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13053 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13054 than TARGET_FPU_ARMV8.
13055 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13056 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13057 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13059 * config/arm/neon.md (neon_vrint): Likewise.
13060 (neon_vcvt): Likewise.
13061 (neon_<fmaxmin_op><mode>): Likewise.
13062 (<fmaxmin><mode>3): Likewise.
13063 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13064 * config/arm/predicates.md (arm_cond_move_operator): Check against
13065 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13067 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13069 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13070 to top of function.
13072 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13074 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13075 loop in comment with memset.
13077 2017-07-14 Martin Liska <mliska@suse.cz>
13079 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
13080 * dwarf2out.c (is_java): Remove the function.
13081 (output_pubname): Remove usage of the function.
13082 (lower_bound_default): Remove usage of DW_LANG_Java.
13083 (gen_compile_unit_die): Likewise.
13084 * gcc.c: Remove compiler defaults for .java and .zip files.
13085 * gimple-expr.c (remove_suffix): Change as there's no longer
13086 extension than 4-letter one.
13087 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
13088 (gimplify_save_expr): Likewise.
13089 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
13090 as it's possible even for other languages than Java.
13091 * langhooks.h (struct lang_hooks): Remove Java from a comment.
13092 * lto-opts.c (lto_write_options): Remove reference to Java.
13093 * opts.c (strip_off_ending): Update file extension handling.
13094 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
13095 * tree-eh.c (lower_resx): Likewise.
13096 * tree.c (free_lang_data_in_type): Remove dead code.
13097 (find_decls_types_r): Likewise.
13098 (build_common_builtin_nodes): Remove Java from a comment.
13099 (verify_type): Remove dead code.
13100 * varasm.c (assemble_external): Remove Java from a comment.
13102 2017-07-14 Martin Liska <mliska@suse.cz>
13104 * opts.c (finish_options): Add quotes.
13105 (common_handle_option): Likewise.
13107 2017-07-14 Martin Liska <mliska@suse.cz>
13109 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
13110 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
13111 Remove N_SO_PASCAL.
13112 * dwarf2out.c (lower_bound_default): Do not handle
13114 (gen_compile_unit_die): Likewise.
13115 * gcc.c: Remove default extension binding for GNU Pascal.
13116 * stmt.c: Remove Pascal language from a comment.
13117 * xcoffout.c: Likewise.
13119 2017-07-13 David Malcolm <dmalcolm@redhat.com>
13122 * diagnostic-show-locus.c (fixit_cmp): New function.
13123 (layout::layout): Sort m_fixit_hints.
13124 (column_range::column_range): Assert that the values are valid.
13125 (struct char_span): New struct.
13126 (correction::overwrite): New method.
13127 (struct source_line): New struct.
13128 (line_corrections::add_hint): Add assertions. Reimplement memcpy
13129 calls in terms of classes source_line and char_span, and
13130 correction::overwrite.
13131 (selftest::test_overlapped_fixit_printing_2): New function.
13132 (selftest::diagnostic_show_locus_c_tests): Call it.
13134 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
13136 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
13137 early if there is no lhs.
13139 2017-07-13 Martin Liska <mliska@suse.cz>
13141 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
13142 (gen_reference_type_die): Likewise.
13143 * stor-layout.c: Remove Pascal-related comment.
13145 2017-07-13 Martin Liska <mliska@suse.cz>
13147 * opts.c (finish_options): Add quotes to error messages.
13148 (parse_sanitizer_options): Likewise.
13150 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
13152 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
13154 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
13156 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
13158 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
13160 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
13162 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
13164 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
13167 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
13168 provides the hardware capability bits, define the macro
13169 __BUILTIN_CPU_SUPPORTS__.
13170 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
13171 if GLIBC does not provide the hardware capability bits. Add a
13172 gcc_unreachable call if the built-in cpu function is neither
13173 __builtin_cpu_is nor __builtin_cpu_supports.
13174 (rs6000_get_function_versions_dispatcher): Change the warning
13175 that an old GLIBC is used which does not export the capability
13176 bits to be an error.
13177 * doc/extend.texi (target_clones attribute): Document the
13178 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
13179 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
13180 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
13181 the macros defined by GCC if the newer GLIBC is available.
13183 2017-07-12 Jeff Law <law@redhat.com>
13185 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
13186 remaining includes slightly.
13187 * config/riscv/riscv-builtins.c: Include profile-count.h.
13189 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13192 * config/avr/avr.c (avr_set_current_function): In diagnostic
13193 messages: Quote keywords and (parts of) identifiers.
13194 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
13197 2017-07-12 Carl Love <cel@us.ibm.com>
13199 * config/rs6000/rs6000-c.c: Add support for built-in functions
13200 vector bool char vec_revb (vector bool char);
13201 vector bool short vec_revb (vector short char);
13202 vector bool int vec_revb (vector bool int);
13203 vector bool long long vec_revb (vector bool long long);
13204 * doc/extend.texi: Update the built-in documentation file for the
13205 new built-in functions.
13207 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13209 * config/s390/s390.md: Remove movcc splitter.
13211 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13213 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
13214 load/store on condition.
13216 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13219 * config/avr/avr.c (avr_encode_section_info)
13220 [progmem && !TREE_READONLY]: Error if progmem object needs
13223 2017-07-11 Michael Collison <michael.collison@arm.com>
13225 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
13228 2017-07-11 Carl Love <cel@us.ibm.com>
13230 * config/rs6000/rs6000-c.c: Add support for builtins
13231 vector unsigned int vec_parity_lsbb (vector signed int);
13232 vector unsigned int vec_parity_lsbb (vector unsigned int);
13233 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
13234 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
13235 vector unsigned long long vec_parity_lsbb (vector signed long long);
13236 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
13237 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
13238 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
13239 * doc/extend.texi: Update the built-in documentation file for the
13240 new built-in functions.
13242 2017-07-11 David Malcolm <dmalcolm@redhat.com>
13244 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
13245 (layout::m_primary_loc): New field.
13246 (layout::layout): Initialize new field. Move location filtering
13247 logic from here to...
13248 (layout::maybe_add_location_range): ...this new method. Add
13249 support for filtering to just the lines already specified by other
13251 (layout::will_show_line_p): New method.
13252 (gcc_rich_location::add_location_if_nearby): New method.
13253 (selftest::test_add_location_if_nearby): New test function.
13254 (selftest::diagnostic_show_locus_c_tests): Call it.
13255 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
13258 2017-07-11 Tom de Vries <tom@codesourcery.com>
13260 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
13261 (bb_first_real_insn): New function.
13262 (nvptx_single): Add extra initialization of broadcasted condition
13265 2017-07-11 Nathan Sidwell <nathan@acm.org>
13267 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
13269 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
13271 * doc/extend.texi (AVR Function Attributes): Remove weblink to
13272 Binutils doc as TEXI will mess them up.
13273 * doc/invoke.texi (AVR Options): Same here.
13275 2017-07-11 Daniel Cederman <cederman@gaisler.com>
13277 * config/sparc/sparc.opt (mfix-ut700): New option.
13278 (mfix-gr712rc): Likewise.
13279 (sparc_fix_b2bst): New variable.
13280 * doc/invoke.texi (SPARC options): Document them.
13281 (ARM options): Fix warnings.
13282 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
13283 instructions to prevent sequences that can trigger the store-store
13284 errata for certain LEON3FT processors.
13285 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
13286 (sparc_option_override): Set sparc_fix_b2bst appropriately.
13287 * config/sparc/sparc.md (fix_b2bst): New attribute.
13288 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
13290 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
13293 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
13295 (*rsqrtsf2_sse): Ditto.
13297 (div<mode>3): Macroize insn from divdf3 and divsf3
13298 using MODEF mode iterator.
13300 2017-07-10 Martin Sebor <msebor@redhat.com>
13302 PR tree-optimization/80397
13303 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
13304 instead of testing for equality to INTEGER_TYPE.
13306 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
13308 * config.gcc: Remove uclibc from arc target spec.
13310 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
13312 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
13314 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13317 * lto-wrapper.c (remove_option): New function.
13318 (merge_and_complain): Merge PIC/PIE options more realistically.
13320 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13322 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
13326 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
13327 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
13328 * config.in: Regenerate.
13329 * configure: Regenerate.
13330 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
13331 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
13332 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
13333 (TARGET_GASISR_PROLOGUES): ...target mask.
13334 * common/config/avr/avr-common.c
13335 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
13336 Set -mgas-isr-prologues.
13337 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
13338 INSERT_PASS_BEFORE for it.
13339 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
13340 * config/avr/avr.c (avr_option_override)
13341 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
13342 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
13343 (avr_attribute_table) <no_gccisr>: Add new function attribute.
13344 (avr_set_current_function) <is_no_gccisr>: Init machine field.
13345 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
13347 (make_avr_pass_pre_proep): New function.
13348 (emit_push_sfr) <treg>: Add argument to function and use it
13349 instead of TMP_REG.
13350 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
13351 and set machine->gasisr.yes.
13352 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
13353 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
13354 __gcc_isr.n_pushed to .L__stack_usage.
13355 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
13356 (avr_asm_final_postscan_insn): ...this new static function.
13357 * config/avr/avr.h (machine_function)
13358 <is_no_gccisr, use_L__stack_usage>: New fields.
13359 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
13360 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
13361 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
13362 (gasisr, *gasisr): New expander and insn.
13363 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13364 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
13365 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
13367 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
13369 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
13372 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13374 Move jump-tables out of .text again.
13377 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
13378 (ASM_OUTPUT_ADDR_VEC): New function.
13379 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
13380 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
13381 INSN_ADDRESSes as asm comment.
13382 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
13383 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
13384 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
13385 * config/avr/avr.md (*tablejump): Adjust comment.
13386 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
13387 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
13389 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
13390 (avr_output_addr_vec): New proto.
13391 (avr_log_t) <insn_addresses>: New field.
13393 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
13396 * config/i386/i386.c (ix86_function_arg_advance): Set
13397 outgoing_args_on_stack to true if there are outgoing arguments
13399 (ix86_function_arg): Likewise.
13400 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
13401 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
13402 * config/i386/i386.h (machine_function): Add
13403 outgoing_args_on_stack.
13405 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
13407 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13408 supporting pthreds.
13409 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
13411 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
13413 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
13414 (REAL_H): Remove $(MACHMODE_H).
13415 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
13417 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
13418 $(MACHMODE_H) and double-int.h.
13419 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
13421 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
13424 2017-07-07 Andrew Pinski <apinski@cavium.com>
13426 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
13427 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
13429 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13431 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
13432 Add warning if GCC was not configured to link against a GLIBC that
13433 exports the hardware capability bits.
13434 (make_resolver_func): Make resolver function private and not a
13435 COMDAT function. Create the name with clone_function_name instead
13436 of make_unique_name.
13439 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
13440 correct operand in doing the split.
13442 2017-07-07 Carl Love <cel@us.ibm.com>
13444 * config/rs6000/rs6000-c: Add support for built-in function
13445 vector unsigned short vec_pack_to_short_fp32 (vector float,
13447 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
13448 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
13449 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
13450 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
13451 (convert_4f32_8i16): Add define_expand.
13452 * doc/extend.texi: Update the built-in documentation file for the
13453 new built-in function.
13455 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13457 * config/sparc/m8.md: New file.
13458 * config/sparc/sparc.md: Include m8.md.
13460 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13462 * config/sparc/sparc.opt: New option -mvis4b.
13463 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
13464 (sparc_option_override): Handle VIS4B.
13465 (enum sparc_builtins): Define
13466 SPARC_BUILTIN_DICTUNPACK{8,16,32},
13467 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
13468 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
13469 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
13470 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
13471 (check_constant_argument): New function.
13472 (sparc_vis_init_builtins): Define builtins
13473 __builtin_vis_dictunpack{8,16,32},
13474 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
13475 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
13476 __builtin_vis_fpcmpde{8,16,32}shl and
13477 __builtin_vis_fpcmpur{8,16,32}shl.
13478 (sparc_expand_builtin): Check that the constant operands to
13479 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
13480 constant and in range.
13481 * config/sparc/sparc-c.c (sparc_target_macros): Handle
13483 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
13484 (SPARC_IMM5_P): Likewise.
13485 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
13486 (enabled): Handle vis4b.
13487 (UNSPEC_DICTUNPACK): New unspec.
13488 (UNSPEC_FPCMPSHL): Likewise.
13489 (UNSPEC_FPUCMPSHL): Likewise.
13490 (UNSPEC_FPCMPDESHL): Likewise.
13491 (UNSPEC_FPCMPURSHL): Likewise.
13492 (cpu_feature): New CPU feature `vis4b'.
13493 (dictunpack{8,16,32}): New insns.
13494 (FPCSMODE): New mode iterator.
13495 (fpcscond): New code iterator.
13496 (fpcsucond): Likewise.
13497 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
13498 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
13499 (fpcmpde{8,16,32}{si,di}shl): Likewise.
13500 (fpcmpur{8,16,32}{si,di}shl): Likewise.
13501 * config/sparc/constraints.md: Define constraints `q' for unsigned
13502 2-bit integer constants and `t' for unsigned 5-bit integer
13504 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
13506 (imm5_operand_dictunpack16): Likewise.
13507 (imm5_operand_dictunpack32): Likewise.
13508 (imm2_operand): Likewise.
13509 * doc/invoke.texi (SPARC Options): Document -mvis4b.
13510 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
13511 ditunpack* and fpcmp*shl builtins.
13513 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13515 * config.gcc: Handle m8 in --with-{cpu,tune} options.
13516 * config.in: Add HAVE_AS_SPARC6 define.
13517 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
13519 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
13521 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
13522 (CPP_CPU_SPEC): Handle m8.
13523 (ASM_CPU_SPEC): Likewise.
13524 * config/sparc/sparc-opts.h (enum processor_type): Add
13526 * config/sparc/sparc.c (m8_costs): New struct.
13527 (sparc_option_override): Handle TARGET_CPU_m8.
13528 (sparc32_initialize_trampoline): Likewise.
13529 (sparc64_initialize_trampoline): Likewise.
13530 (sparc_issue_rate): Likewise.
13531 (sparc_register_move_cost): Likewise.
13532 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
13533 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
13534 (ASM_CPU64_DEFAULT_SPEC): Likewise.
13535 (CPP_CPU_SPEC): Handle M8.
13536 (ASM_CPU_SPEC): Likewise.
13537 (AS_M8_FLAG): Define.
13538 * config/sparc/sparc.md: Add m8 to the cpu attribute.
13539 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
13540 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
13542 * configure: Regenerate.
13543 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
13546 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13548 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
13550 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
13551 ("*movdi_insn_sp32"): Do not set v3pipe.
13552 ("*movsi_insn"): Likewise.
13553 ("*movdi_insn_sp64"): Likewise.
13554 ("*movsf_insn"): Likewise.
13555 ("*movdf_insn_sp32"): Likewise.
13556 ("*movdf_insn_sp64"): Likewise.
13557 ("*zero_extendsidi2_insn_sp64"): Likewise.
13558 ("*sign_extendsidi2_insn"): Likewise.
13559 ("*mov<VM32:mode>_insn"): Likewise.
13560 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13561 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13562 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13563 ("<vlop:code><VL:mode>3"): Likewise.
13564 ("*not_<vlop:code><VL:mode>3"): Likewise.
13565 ("*nand<VL:mode>_vis"): Likewise.
13566 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
13567 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
13568 ("one_cmpl<VL:mode>2"): Likewise.
13569 ("faligndata<VM64:mode>_vis"): Likewise.
13570 ("alignaddrsi_vis"): Likewise.
13571 ("alignaddrdi_vis"): Likweise.
13572 ("alignaddrlsi_vis"): Likewise.
13573 ("alignaddrldi_vis"): Likewise.
13574 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13575 ("bmaskdi_vis"): Likewise.
13576 ("bmasksi_vis"): Likewise.
13577 ("bshuffle<VM64:mode>_vis"): Likewise.
13578 ("cmask8<P:mode>_vis"): Likewise.
13579 ("cmask16<P:mode>_vis"): Likewise.
13580 ("cmask32<P:mode>_vis"): Likewise.
13581 ("pdistn<P:mode>_vis"): Likewise.
13582 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13584 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13586 * config/sparc/sparc.md ("subtype"): New insn attribute.
13587 ("*wrgsr_sp64"): Set insn subtype.
13588 ("*rdgsr_sp64"): Likewise.
13589 ("alignaddrsi_vis"): Likewise.
13590 ("alignaddrdi_vis"): Likewise.
13591 ("alignaddrlsi_vis"): Likewise.
13592 ("alignaddrldi_vis"): Likewise.
13593 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13594 ("fexpand_vis"): Likewise.
13595 ("fpmerge_vis"): Likewise.
13596 ("faligndata<VM64:mode>_vis"): Likewise.
13597 ("bshuffle<VM64:mode>_vis"): Likewise.
13598 ("cmask8<P:mode>_vis"): Likewise.
13599 ("cmask16<P:mode>_vis"): Likewise.
13600 ("cmask32<P:mode>_vis"): Likewise.
13601 ("fchksm16_vis"): Likewise.
13602 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
13603 ("fmean16_vis"): Likewise.
13604 ("fp<plusminus_insn>64_vis"): Likewise.
13605 ("<plusminus_insn>v8qi3"): Likewise.
13606 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13607 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
13608 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
13609 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
13610 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
13611 ("*movqi_insn"): Likewise.
13612 ("*movhi_insn"): Likewise.
13613 ("*movsi_insn"): Likewise.
13614 ("movsi_pic_gotdata_op"): Likewise.
13615 ("*movdi_insn_sp32"): Likewise.
13616 ("*movdi_insn_sp64"): Likewise.
13617 ("movdi_pic_gotdata_op"): Likewise.
13618 ("*movsf_insn"): Likewise.
13619 ("*movdf_insn_sp32"): Likewise.
13620 ("*movdf_insn_sp64"): Likewise.
13621 ("*zero_extendhisi2_insn"): Likewise.
13622 ("*zero_extendqihi2_insn"): Likewise.
13623 ("*zero_extendqisi2_insn"): Likewise.
13624 ("*zero_extendqidi2_insn"): Likewise.
13625 ("*zero_extendhidi2_insn"): Likewise.
13626 ("*zero_extendsidi2_insn_sp64"): Likewise.
13627 ("ldfsr"): Likewise.
13628 ("prefetch_64"): Likewise.
13629 ("prefetch_32"): Likewise.
13630 ("tie_ld32"): Likewise.
13631 ("tie_ld64"): Likewise.
13632 ("*tldo_ldub_sp32"): Likewise.
13633 ("*tldo_ldub1_sp32"): Likewise.
13634 ("*tldo_ldub2_sp32"): Likewise.
13635 ("*tldo_ldub_sp64"): Likewise.
13636 ("*tldo_ldub1_sp64"): Likewise.
13637 ("*tldo_ldub2_sp64"): Likewise.
13638 ("*tldo_ldub3_sp64"): Likewise.
13639 ("*tldo_lduh_sp32"): Likewise.
13640 ("*tldo_lduh1_sp32"): Likewise.
13641 ("*tldo_lduh_sp64"): Likewise.
13642 ("*tldo_lduh1_sp64"): Likewise.
13643 ("*tldo_lduh2_sp64"): Likewise.
13644 ("*tldo_lduw_sp32"): Likewise.
13645 ("*tldo_lduw_sp64"): Likewise.
13646 ("*tldo_lduw1_sp64"): Likewise.
13647 ("*tldo_ldx_sp64"): Likewise.
13648 ("*mov<VM32:mode>_insn"): Likewise.
13649 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13650 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13652 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13654 * config/sparc/sparc.md ("type"): New insn type viscmp.
13655 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
13657 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
13658 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
13659 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13660 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
13662 ("n7_vis_logical_11cycle"): Likewise.
13663 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
13664 * config/sparc/niagara2.md ("niag3_vis": Likewise.
13665 * config/sparc/niagara.md ("niag_vis"): Likewise.
13666 * config/sparc/ultra3.md ("us3_fga"): Likewise.
13667 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
13669 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13671 * config/sparc/sparc.md: New instruction type `bmask'.
13672 (bmaskdi_vis): Use the `bmask' type.
13673 (bmasksi_vis): Likewise.
13674 * config/sparc/ultra3.md (us3_array): Likewise.
13675 * config/sparc/niagara7.md (n7_array): Likewise.
13676 * config/sparc/niagara4.md (n4_array): Likewise.
13677 * config/sparc/niagara2.md (niag2_vis): Likewise.
13678 (niag3_vis): Likewise.
13679 * config/sparc/niagara.md (niag_vis): Likewise.
13681 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13683 * ipa-comdats.c: Remove optimize check from gate.
13684 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
13685 for functions not optimized.
13686 (ipa_fn_summary_read): Skip optimize check.
13687 (ipa_fn_summary_write): Likewise.
13688 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
13690 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
13692 (can_inline_edge_p): Check flag_pcc_struct_return for match.
13693 (check_callers): Give up on caller which is not optimized.
13694 (inline_small_functions): Likewise.
13695 (ipa_inline): Do not give up when not optimizing.
13696 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
13697 away unoptimizes cdtors.
13698 (whole_program_function_and_variable_visibility): Do
13699 ipa_discover_readonly_nonaddressable_vars in LTO mode.
13700 * ipa.c (process_references): Do not check optimize.
13701 (symbol_table::remove_unreachable_nodes): Update optimize check.
13702 (set_writeonly_bit): Update optimize check.
13703 (pass_ipa_cdtor_merge::gate): Do not check optimize.
13704 (pass_ipa_single_use::gate): Remove.
13706 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13708 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
13709 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
13710 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
13711 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
13712 permute_load, permute_store, adjust_extract, adjust_splat,
13713 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
13714 replace_swap_with_copy, dump_swap_insn_table,
13715 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
13716 recombine_lvx_pattern, recombine_stvx_pattern,
13717 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
13718 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
13719 to file rs6000-p8swap.c.
13720 * config/rs6000/rs6000-p8swap.c: New file.
13721 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
13722 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
13723 and rs6000*-*-* targets.
13725 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13727 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
13729 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13731 * lto-wrapper.c (merge_and_complain): Do not merge
13732 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
13733 fsigned_zeros, ftrapping_math, fwrapv.
13734 (append_compiler_options): Do not track these options.
13735 (append_linker_options): Likewie
13737 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13739 * cgraphunit.c (cgraph_node::finalize_function): When
13740 !flag_toplevel_reorde set no_reorder flag.
13741 (varpool_node::finalize_decl): Likewise.
13742 (symbol_table::compile): Drop no toplevel reorder path.
13744 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13746 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
13747 edges; zero probability is not better than uninitialized.
13749 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13751 * asan.h (asan_sanitize_allocas_p): Declare.
13752 * asan.c (asan_sanitize_allocas_p): New function.
13753 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
13754 (handle_builtin_alloca): Likewise.
13755 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
13756 if !asan_sanitize_allocas_p.
13757 * params.def (asan-instrument-allocas): Add new option.
13758 * params.h (ASAN_PROTECT_ALLOCAS): Define.
13759 * opts.c (common_handle_option): Disable allocas sanitization for
13762 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13764 * asan.c: Include gimple-fold.h.
13765 (get_last_alloca_addr): New function.
13766 (handle_builtin_stackrestore): Likewise.
13767 (handle_builtin_alloca): Likewise.
13768 (asan_emit_allocas_unpoison): Likewise.
13769 (get_mem_refs_of_builtin_call): Add new parameter, remove const
13770 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
13771 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
13772 (instrument_builtin_call): Pass gimple iterator to
13773 get_mem_refs_of_builtin_call.
13774 (last_alloca_addr): New global.
13775 * asan.h (asan_emit_allocas_unpoison): Declare.
13776 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
13777 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
13778 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
13779 if function calls alloca.
13780 * gimple-fold.c (replace_call_with_value): Remove static keyword.
13781 * gimple-fold.h (replace_call_with_value): Declare.
13782 * internal-fn.c: Include asan.h.
13783 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
13784 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
13786 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13788 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
13789 (C_SELFTEST_FLAGS): New.
13790 (CPP_SELFTEST_FLAGS): New.
13791 (SELFTEST_DEPS): New, from deps of s-selftest.
13792 (C_SELFTEST_DEPS): New, from deps of s-selftest.
13793 (CPP_SELFTEST_DEPS): New.
13794 (selftest): Add dependency on s-selftest-c++.
13795 (s-selftest): Rename to...
13796 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
13797 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
13798 than SELFTEST_FLAGS.
13799 (selftest-gdb): Rename to...
13800 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
13802 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
13803 (selftest-valgrind): Rename to...
13804 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
13806 (selftest-valgrind): Reintroduce as an alias for
13807 selftest-c-valgrind.
13808 (s-selftest-c++): New.
13809 (selftest-c++-gdb): New.
13810 (selftest-c++-valgrind): New.
13812 2017-07-06 Olivier Hainque <hainque@adacore.com>
13814 * gcc.c (process_command): When deciding if undefined variables
13815 should be ignored when processing specs, accept "gcc -v" as well.
13817 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13819 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
13820 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
13822 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13824 * config/arm/arm-cpus.in (armv8-r): Add new entry.
13825 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
13826 * config/arm/arm-tables.opt: Regenerate.
13827 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
13829 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
13831 2017-07-06 Carl Love <cel@us.ibm.com>
13833 * ChangeLog: Clean up from mid air collision
13835 2017-07-06 Carl Love <cel@us.ibm.com>
13837 * config/rs6000/rs6000-c.c: Add support for built-in functions
13838 vector signed int vec_subc (vector signed int, vector signed int);
13839 vector signed __int128 vec_subc (vector signed __int128,
13840 vector signed __int128);
13841 vector unsigned __int128 vec_subc (vector unsigned __int128,
13842 vector unsigned __int128);
13843 vector signed int vec_sube (vector signed int, vector signed int,
13844 vector signed int);
13845 vector unsigned int vec_sube (vector unsigned int,
13846 vector unsigned int,
13847 vector unsigned int);
13848 vector signed __int128 vec_sube (vector signed __int128,
13849 vector signed __int128,
13850 vector signed__int128);
13851 vector unsigned __int128 vec_sube (vector unsigned __int128,
13852 vector unsigned __int128,
13853 vector unsigned __int128);
13854 vector signed int vec_subec (vector signed int, vector signed int,
13855 vector signed int);
13856 vector unsigned int vec_subec (vector unsigned int,
13857 vector unsigned int,
13858 vector unsigned int);
13859 vector signed __int128 vec_subec (vector signed __int128,
13860 vector signed __int128,
13861 vector signed__int128);
13862 vector unsigned __int128 vec_subec (vector unsigned __int128,
13863 vector unsigned __int128,
13864 vector unsigned __int128);
13865 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
13866 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
13867 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
13868 BU_ALTIVEC_OVERLOAD_X definitions.
13869 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
13870 * doc/extend.texi: Update the built-in documentation file for the new
13871 built-in functions.
13873 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13876 * diagnostic-show-locus.c (layout::layout): Use start and finish
13877 spelling location for the start and finish of each range.
13878 * genmatch.c (linemap_client_expand_location_to_spelling_point):
13879 Add unused aspect param.
13880 * input.c (expand_location_1): Add "aspect" param, and use it
13881 to access the correct part of the location.
13882 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
13884 (expand_location_to_spelling_point): Likewise.
13885 (linemap_client_expand_location_to_spelling_point): Add "aspect"
13886 param, and pass it to expand_location_1.
13888 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
13890 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
13891 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
13892 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
13893 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
13894 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
13895 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
13896 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
13897 _mm_maskz_getmant_ss): New intrinsics.
13898 (__builtin_ia32_getexpss128_mask): Changed to ...
13899 __builtin_ia32_getexpss128_round ... this.
13900 (__builtin_ia32_getexpsd128_mask): Changed to ...
13901 __builtin_ia32_getexpsd128_round ... this.
13902 * config/i386/i386-builtin-types.def
13903 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
13904 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
13905 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
13906 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
13907 __builtin_ia32_getmantss_mask_round): New builtins.
13908 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
13909 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
13910 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
13911 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
13912 * config/i386/sse.md
13913 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
13914 avx512f_sgetexp<mode><mask_scalar_name>
13915 <round_saeonly_scalar_name> ... this.
13916 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
13917 %0, %1, %2<round_saeonly_op3>}): Changed to ...
13918 vgetexp<ssescalarmodesuffix>
13919 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13920 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
13921 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
13922 avx512f_vgetmant<mode><mask_scalar_name>
13923 <round_saeonly_scalar_name> ... this.
13924 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
13925 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
13926 vgetmant<ssescalarmodesuffix>
13927 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
13928 %0<mask_scalar_operand4>, %1, %2
13929 <round_saeonly_scalar_mask_op4>, %3} ... this.
13930 * config/i386/subst.md (mask_scalar_operand4,
13931 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
13932 round_saeonly_scalar_nimm_predicate): New subst attributes.
13934 2017-07-06 Julia Koval <julia.koval@intel.com>
13936 * config/i386/i386.c (ix86_erase_embedded_rounding):
13937 Remove code for old rounding pattern.
13939 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
13941 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
13943 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
13945 * doc/sourcebuild.texi (Test Directives, Variants of
13946 dg-require-support): Add documentation for dg-require-stack-check.
13948 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
13950 * config/i386/subst.md (mask_scalar, round_scalar,
13951 round_saeonly_scalar): New meta-templates.
13952 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
13953 round_scalar_mask_operand3, round_scalar_mask_op3,
13954 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
13955 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
13956 round_saeonly_scalar_constraint,
13957 round_saeonly_scalar_prefix): New subst attribute.
13958 * config/i386/sse.md
13959 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
13960 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
13961 <round_scalar_name> ... this.
13962 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
13963 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
13964 <round_scalar_name> ... this.
13965 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
13966 <sse>_vm<code><mode>3<mask_scalar_name>
13967 <round_saeonly_scalar_name> ... this.
13968 (v<plusminus_mnemonic><ssescalarmodesuffix>
13969 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13970 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13971 v<plusminus_mnemonic><ssescalarmodesuffix>
13972 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13973 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13974 (v<multdiv_mnemonic><ssescalarmodesuffix>
13975 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13976 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13977 v<multdiv_mnemonic><ssescalarmodesuffix>
13978 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13979 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13980 (v<maxmin_float><ssescalarmodesuffix>
13981 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
13982 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
13983 v<maxmin_float><ssescalarmodesuffix>
13984 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13985 %0<mask_scalar_operand3>, %1, %<iptr>2
13986 <round_saeonly_scalar_mask_op3>} ... this.
13988 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
13990 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
13991 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
13993 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
13994 Alan Hayward <alan.hayward@arm.com>
13995 David Sherwood <david.sherwood@arm.com>
13997 * combine.c (simplify_if_then_else): Remove "enum" before
13999 * compare-elim.c (can_eliminate_compare): Likewise.
14000 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14002 (aarch64_lookup_simd_builtin_type): Likewise.
14003 (aarch64_simd_builtin_type): Likewise.
14004 (aarch64_init_simd_builtin_types): Likewise.
14005 (aarch64_simd_expand_args): Likewise.
14006 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14008 (aarch64_reverse_mask): Likewise.
14009 (aarch64_simd_emit_reg_reg_move): Likewise.
14010 (aarch64_gen_adjusted_ldpstp): Likewise.
14011 (aarch64_ccmp_mode_to_code): Likewise.
14012 (aarch64_operands_ok_for_ldpstp): Likewise.
14013 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14014 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14016 (aarch64_min_divisions_for_recip_mul): Likewise.
14017 (aarch64_reassociation_width): Likewise.
14018 (aarch64_get_condition_code_1): Likewise.
14019 (aarch64_simd_emit_reg_reg_move): Likewise.
14020 (aarch64_simd_attr_length_rglist): Likewise.
14021 (aarch64_reverse_mask): Likewise.
14022 (aarch64_operands_ok_for_ldpstp): Likewise.
14023 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14024 (aarch64_gen_adjusted_ldpstp): Likewise.
14025 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14027 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14028 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14029 (arm_lookup_simd_builtin_type): Likewise.
14030 (arm_simd_builtin_type): Likewise.
14031 (arm_init_simd_builtin_types): Likewise.
14032 (arm_expand_builtin_args): Likewise.
14033 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14034 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14035 (ft32_setup_incoming_varargs): Likewise.
14036 (ft32_function_arg): Likewise.
14037 (ft32_function_arg_advance): Likewise.
14038 (ft32_pass_by_reference): Likewise.
14039 (ft32_arg_partial_bytes): Likewise.
14040 (ft32_valid_pointer_mode): Likewise.
14041 (ft32_addr_space_pointer_mode): Likewise.
14042 (ft32_addr_space_legitimate_address_p): Likewise.
14043 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14045 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14046 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14047 (iamcu_alignment): Likewise.
14048 (canonicalize_vector_int_perm): Likewise.
14049 (ix86_noce_conversion_profitable_p): Likewise.
14050 (ix86_mpx_bound_mode): Likewise.
14051 (ix86_operands_ok_for_move_multiple): Likewise.
14052 * config/microblaze/microblaze-protos.h
14053 (microblaze_expand_conditional_branch_reg): Likewise.
14054 * config/microblaze/microblaze.c
14055 (microblaze_expand_conditional_branch_reg): Likewise.
14056 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14058 (rs6000_reassociation_width): Likewise.
14059 (rs6000_invalid_binary_op): Likewise.
14060 (fusion_p9_p): Likewise.
14061 (emit_fusion_p9_load): Likewise.
14062 (emit_fusion_p9_store): Likewise.
14063 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14065 (riscv_hard_regno_mode_ok_p): Likewise.
14066 (riscv_address_insns): Likewise.
14067 (riscv_split_symbol): Likewise.
14068 (riscv_legitimize_move): Likewise.
14069 (riscv_function_value): Likewise.
14070 (riscv_hard_regno_nregs): Likewise.
14071 (riscv_expand_builtin): Likewise.
14072 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14073 (riscv_build_integer): Likewise.
14074 (riscv_split_integer): Likewise.
14075 (riscv_legitimate_constant_p): Likewise.
14076 (riscv_cannot_force_const_mem): Likewise.
14077 (riscv_regno_mode_ok_for_base_p): Likewise.
14078 (riscv_valid_base_register_p): Likewise.
14079 (riscv_valid_offset_p): Likewise.
14080 (riscv_valid_lo_sum_p): Likewise.
14081 (riscv_classify_address): Likewise.
14082 (riscv_legitimate_address_p): Likewise.
14083 (riscv_address_insns): Likewise.
14084 (riscv_load_store_insns): Likewise.
14085 (riscv_force_binary): Likewise.
14086 (riscv_split_symbol): Likewise.
14087 (riscv_force_address): Likewise.
14088 (riscv_legitimize_address): Likewise.
14089 (riscv_move_integer): Likewise.
14090 (riscv_legitimize_const_move): Likewise.
14091 (riscv_legitimize_move): Likewise.
14092 (riscv_address_cost): Likewise.
14093 (riscv_subword): Likewise.
14094 (riscv_output_move): Likewise.
14095 (riscv_canonicalize_int_order_test): Likewise.
14096 (riscv_emit_int_order_test): Likewise.
14097 (riscv_function_arg_boundary): Likewise.
14098 (riscv_pass_mode_in_fpr_p): Likewise.
14099 (riscv_pass_fpr_single): Likewise.
14100 (riscv_pass_fpr_pair): Likewise.
14101 (riscv_get_arg_info): Likewise.
14102 (riscv_function_arg): Likewise.
14103 (riscv_function_arg_advance): Likewise.
14104 (riscv_arg_partial_bytes): Likewise.
14105 (riscv_function_value): Likewise.
14106 (riscv_pass_by_reference): Likewise.
14107 (riscv_setup_incoming_varargs): Likewise.
14108 (riscv_print_operand): Likewise.
14109 (riscv_elf_select_rtx_section): Likewise.
14110 (riscv_save_restore_reg): Likewise.
14111 (riscv_for_each_saved_reg): Likewise.
14112 (riscv_register_move_cost): Likewise.
14113 (riscv_hard_regno_mode_ok_p): Likewise.
14114 (riscv_hard_regno_nregs): Likewise.
14115 (riscv_class_max_nregs): Likewise.
14116 (riscv_memory_move_cost): Likewise.
14117 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
14118 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
14119 (rl78_addr_space_address_mode): Likewise.
14120 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14122 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14123 (rs6000_reassociation_width): Likewise.
14124 (rs6000_invalid_binary_op): Likewise.
14125 (fusion_p9_p): Likewise.
14126 (emit_fusion_p9_load): Likewise.
14127 (emit_fusion_p9_store): Likewise.
14128 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
14129 (ok_for_simple_move_operands): Likewise.
14130 (ok_for_simple_move_strict_operands): Likewise.
14131 (ok_for_simple_arith_logic_operands): Likewise.
14132 (visium_legitimize_reload_address): Likewise.
14133 (visium_select_cc_mode): Likewise.
14134 (output_cbranch): Likewise.
14135 (visium_split_double_move): Likewise.
14136 (visium_expand_copysign): Likewise.
14137 (visium_expand_int_cstore): Likewise.
14138 (visium_expand_fp_cstore): Likewise.
14139 * config/visium/visium.c (visium_pass_by_reference): Likewise.
14140 (visium_function_arg): Likewise.
14141 (visium_function_arg_advance): Likewise.
14142 (visium_libcall_value): Likewise.
14143 (visium_setup_incoming_varargs): Likewise.
14144 (visium_legitimate_constant_p): Likewise.
14145 (visium_legitimate_address_p): Likewise.
14146 (visium_legitimize_address): Likewise.
14147 (visium_secondary_reload): Likewise.
14148 (visium_register_move_cost): Likewise.
14149 (visium_memory_move_cost): Likewise.
14150 (prepare_move_operands): Likewise.
14151 (ok_for_simple_move_operands): Likewise.
14152 (ok_for_simple_move_strict_operands): Likewise.
14153 (ok_for_simple_arith_logic_operands): Likewise.
14154 (visium_function_value_1): Likewise.
14155 (rtx_ok_for_offset_p): Likewise.
14156 (visium_legitimize_reload_address): Likewise.
14157 (visium_split_double_move): Likewise.
14158 (visium_expand_copysign): Likewise.
14159 (visium_expand_int_cstore): Likewise.
14160 (visium_expand_fp_cstore): Likewise.
14161 (visium_split_cstore): Likewise.
14162 (visium_select_cc_mode): Likewise.
14163 (visium_split_cbranch): Likewise.
14164 (output_cbranch): Likewise.
14165 (visium_print_operand_address): Likewise.
14166 * expmed.c (flip_storage_order): Likewise.
14167 * expmed.h (emit_cstore): Likewise.
14168 (flip_storage_order): Likewise.
14169 * genrecog.c (validate_pattern): Likewise.
14170 * hsa-gen.c (gen_hsa_addr): Likewise.
14171 * internal-fn.c (expand_arith_overflow): Likewise.
14172 * ira-color.c (allocno_copy_cost_saving): Likewise.
14173 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14174 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
14175 (process_invariant_for_inheritance): Likewise.
14176 * lra-eliminations.c (move_plus_up): Likewise.
14177 * omp-low.c (lower_oacc_reductions): Likewise.
14178 * simplify-rtx.c (simplify_subreg): Likewise.
14179 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
14180 (TARGET_CHKP_BOUND_MODE): Likewise..
14181 * targhooks.c (default_chkp_bound_mode): Likewise.
14182 (default_setup_incoming_vararg_bounds): Likewise.
14183 * targhooks.h (default_chkp_bound_mode): Likewise.
14184 (default_setup_incoming_vararg_bounds): Likewise.
14185 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
14186 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
14187 (have_whole_vector_shift): Likewise.
14188 * tree-vect-stmts.c (vectorizable_load): Likewise.
14189 * doc/tm.texi: Regenerate.
14191 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14193 Graceful degrade if Binutils PR21472 is not available.
14196 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
14197 .rodata in flash test fails.
14198 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
14199 * confgure: Regenerate.
14200 * config.in: Regenerate.
14201 * config/avr/avr.c (avr_asm_named_section)
14202 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
14203 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
14204 (avr_asm_init_sections): Same.
14206 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14208 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
14209 (fma<VH:mode>4_intrinsic): Likewise.
14210 (*fmsub<VCVTF:mode>4): Likewise.
14211 (*fmsub<VH:mode>4_intrinsic): Likewise.
14213 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14216 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
14217 Don't depend on "optimize > 0".
14218 (out_movhi_r_mr, out_movqi_mr_r): Same.
14219 (out_movhi_mr_r, out_movqi_r_mr): Same.
14220 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
14221 io_address_operand on "optimize > 0".
14223 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14225 * tree-loop-distribution.c: Add general explanantion on the pass.
14226 (generate_loops_for_partition): Mark distributed loop.
14227 (pg_add_dependence_edges): New parameter. Handle alias data
14228 dependence specially and record it in the parameter if asked.
14229 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
14230 (init_partition_graph_vertices, add_partition_graph_edge): New.
14231 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
14232 (free_partition_graph_vdata, build_partition_graph): New.
14233 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
14234 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
14235 (data_ref_segment_size, latch_dominated_by_data_ref): New.
14236 (compute_alias_check_pairs, version_loop_by_alias_check): New.
14237 (version_for_distribution_p, finalize_partitions): New.
14238 (distribute_loop): Handle alias data dependence specially. Factor
14239 out loop fusion code as functions and call these functions.
14241 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14243 * tree-loop-distribution.c (classify_partition): New parameter and
14244 better handle reduction statement.
14245 (rdg_build_partitions): Revise comment.
14246 (distribute_loop): Compute statements in all partitions and pass it
14247 to classify_partition.
14249 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14251 * tree-loop-distribution.c (enum partition_type): New.
14252 (struct partition): New field type.
14253 (partition_merge_into): Add parameter. Update partition type.
14254 (data_dep_in_cycle_p, update_type_for_merge): New functions.
14255 (build_rdg_partition_for_vertex): Compute partition type.
14256 (rdg_build_partitions): Dump partition type.
14257 (distribute_loop): Update calls to partition_merge_into.
14259 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14261 * tree-loop-distribution.c (struct ddr_hasher): New.
14262 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
14264 (classify_partition): Call get_data_dependence.
14265 (pg_add_dependence_edges): Ditto.
14266 (distribute_loop): Release data dependence hash table.
14268 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14270 * tree-loop-distribution.c (ref_base_address): Delete.
14271 (similar_memory_accesses): Rename ...
14272 (share_memory_accesses): ... to this. Check if partitions access
14273 the same memory reference.
14274 (distribute_loop): Call share_memory_accesses.
14276 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14278 * tree-loop-distribution.c (struct partition): New field recording
14279 its data reference.
14280 (partition_alloc, partition_free): Init and release data refs.
14281 (partition_merge_into): Merge data refs.
14282 (build_rdg_partition_for_vertex): Collect data refs for partition.
14283 (pg_add_dependence_edges): Change parameters from vector to bitmap.
14285 (distribute_loop): Remve data refs from vertice data of partition
14288 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14290 * tree-loop-distribution.c (params.h): Include header file.
14291 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
14292 (datarefs_vec): New global var.
14293 (create_rdg_vertices): Use datarefs_vec directly.
14294 (free_rdg): Don't free data references.
14295 (build_rdg): Update use. Don't free data references.
14296 (distribute_loop): Compute global variable for data references.
14297 Bail out if there are too many data references.
14299 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14301 * tree-loop-distribution.c (loop_nest): New global var.
14302 (build_rdg): Use loop directly, rather than loop nest.
14303 (pg_add_dependence_edges): Remove loop nest parameter. Use global
14305 (distribute_loop): Compute global variable loop nest. Update use.
14307 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14309 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
14310 (partition_merge_into): New parameter. Dump reason for fusion.
14311 (distribute_loop): Update use of partition_merge_into.
14313 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14315 * tree-loop-distribution.c (bb_top_order_index): New.
14316 (bb_top_order_index_size, bb_top_order_cmp): New.
14317 (stmts_from_loop): Use topological order.
14318 (pass_loop_distribution::execute): Compute and release topological
14319 order for basic blocks.
14321 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14323 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
14326 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14328 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
14329 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
14330 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
14331 * internal-fn.def (LOOP_DIST_ALIAS): New.
14332 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
14333 (fold_loop_internal_call): ... this.
14334 (vect_loop_dist_alias_call): New function.
14335 (set_uid_loop_bbs): Call fold_loop_internal_call.
14336 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
14339 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14342 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
14343 Require dead FLAGS_REG at the beginning of a peephole.
14345 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14348 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
14349 arguments in the call to __builtin_ia32_sbb_u32.
14350 (_subborrow_u64): Swap _X and _Y arguments in the call to
14351 __builtin_ia32_sbb_u64.
14353 2017-07-04 Jakub Jelinek <jakub@redhat.com>
14356 * tree-vrp.c (compare_assert_loc): Turn into a function template
14357 with stable template parameter. Only test if a->e is NULL,
14358 !a->e == !b->e has been verified already. Use e == NULL or
14359 e != NULL instead of e or ! e tests. If stable is true, don't use
14360 iterative_hash_expr, on the other side allow a or b or both NULL
14361 and sort the NULLs last.
14362 (process_assert_insertions): Sort using compare_assert_loc<false>
14363 instead of compare_assert_loc, later sort using
14364 compare_assert_loc<true> before calling process_assert_insertions_for
14365 in a loop. Use break instead of continue once seen NULL pointer.
14367 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14369 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14370 Cortex-R7 and Cortex-R8 processors.
14372 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14374 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
14375 uninitialized while src is not.
14377 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
14379 * common/config/arm/arm-common.c: Adjust include path for
14381 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
14382 (arm-cpu.h): Create in build directory. Adjust dependency rules.
14383 (arm-cpu-data.h): Likewise.
14384 (arm-cpu-cdata.h): Likewise.
14385 * config/arm/arm-cpu.h: Delete.
14386 * config/arm/arm-cpu-cdata.h: Delete.
14387 * config/arm/arm-cpu-data.h: Delete.
14389 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
14391 * config/arm/arm-cpus.in (cortex-a55): New.
14392 (cortex-a75): Likewise.
14393 (cortex-a75.cortex-a55): Likewise.
14394 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
14396 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
14397 * config/arm/arm-cpu-cdata.h: Regenerate.
14398 * config/arm/arm-cpu-data.h: Regenerate.
14399 * config/arm/arm-cpu.h: Regenerate.
14400 * config/arm/arm-tables.opt: Regenerate.
14401 * config/arm/arm-tune.md: Regenerate.
14403 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14405 * haifa-sched.c (sched_create_recovery_edges): Update profile.
14407 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14409 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
14412 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
14414 PR tree-optimization/81292
14415 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
14416 full_string_p, also call adjust_related_strinfos if the adjustment
14417 is simple, otherwise invalidate related strinfos.
14419 2017-07-04 Martin Liska <mliska@suse.cz>
14422 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
14423 newly created variable as DECL_IGNORED_P.
14425 2017-07-04 Martin Liska <mliska@suse.cz>
14428 * ipa-inline.c (inline_small_functions):
14429 Use xstrdup_for_dump.
14431 2017-07-04 Tom de Vries <tom@codesourcery.com>
14433 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
14435 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14438 * config/darwin.c (darwin_function_switched_text_sections):
14441 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
14443 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
14445 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14447 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
14449 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14451 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
14452 min_profitable_iters, and th as inclusive lower bounds.
14453 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
14454 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
14455 for min_profitable_iters and min_profitable_estimate.
14456 (vect_transform_loop): Treat th as an inclusive lower bound.
14457 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
14459 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14462 * config/darwin.c (darwin_function_switched_text_sections):
14463 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
14464 in two pieces, and suppress the use of buf.
14466 2017-07-03 Nathan Sidwell <nathan@acm.org>
14468 * hash-table.h (hash_table_mod1): Fix indentation.
14470 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14472 PR middle-end/81290
14473 * predict.c (force_edge_cold): Be more careful about propagation
14475 * profile-count.h (profile_probability::guessed,
14476 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
14478 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
14480 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
14482 * doc/invoke.texi (rcpc architecture extension): Document it.
14484 2017-07-03 Richard Biener <rguenther@suse.de>
14486 PR tree-optimization/60510
14487 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
14488 the scalar reduction PHI and use it.
14489 (vectorizable_reduction): Properly guard the single_defuse_cycle
14490 path for non-SLP reduction chains where we cannot use it.
14491 Rework reduc_def/index and vector type deduction. Rework
14492 vector operand gathering during reduction op code-gen.
14493 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
14494 chains dissolve the chain and leave it to non-SLP reduction
14497 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14499 * tree-data-ref.h (dr_alignment): Declare.
14500 * tree-data-ref.c (dr_alignment): New function.
14501 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
14502 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14504 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
14506 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14508 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
14509 and base_misalignment fields.
14510 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
14511 * tree-data-ref.c: Include builtins.h.
14512 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
14513 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
14514 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
14515 * tree-vect-data-refs.c: Include tree-cfg.h.
14516 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
14517 fields instead of calculating an alignment here.
14518 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
14519 innermost_loop_behavior fields.
14521 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14523 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
14525 (DR_STEP_ALIGNMENT): New macro.
14526 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
14527 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
14528 (create_data_ref): Print it.
14529 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
14530 to tell whether the step preserves vector (mis)alignment.
14531 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14532 Move the check for an integer step and generalise to all INTEGER_CST.
14533 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
14534 Print the outer step alignment.
14536 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14538 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
14539 with offset_alignment.
14540 (DR_ALIGNED_TO): Delete.
14541 (DR_OFFSET_ALIGNMENT): New macro.
14542 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
14543 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
14544 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
14545 (create_data_ref): Likewise.
14546 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14547 (vect_analyze_data_refs): Likewise.
14548 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
14549 creating dummy innermost behavior.
14551 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14553 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
14554 with a "innermost_loop_behavior *" and refeence tree.
14555 * tree-data-ref.c (dr_analyze_innermost): Likewise.
14556 (create_data_ref): Update call accordingly.
14557 * tree-predcom.c (find_looparound_phi): Likewise.
14559 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14561 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
14562 fields with dr_wrt_vec_loop.
14563 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
14564 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
14565 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
14566 (vect_dr_behavior): New function.
14567 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14568 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
14569 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
14570 track whether the step preserves the misalignment.
14571 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14572 Use vect_dr_behavior.
14573 (vect_setup_realignment): Update call accordingly.
14574 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
14575 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
14576 call to vect_create_addr_base_for_vector_ref.
14577 (vect_create_cond_for_align_checks): Likewise.
14578 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
14579 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
14580 (vect_recog_mask_conversion_pattern): Likewise.
14581 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
14582 (new_stmt_vec_info): Remove redundant zeroing.
14584 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14586 * common/config/arm/arm-common.c (arm_be8_option): New function.
14587 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
14588 (ISA_ARMv6): Add isa_bit_be8.
14589 * config/arm/arm.h (arm_be8_option): Add prototype.
14590 (BE8_SPEC_FUNCTION): New define.
14591 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
14592 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
14593 (mlittle-endian): Similarly.
14594 (mbe8, mbe32): New options.
14595 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
14596 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
14598 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14600 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
14602 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14604 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
14605 (cleanup_tree_cfg_bb): Use it.
14606 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
14608 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
14610 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14613 * loop-doloop.c (add_test): Update profile.
14615 2017-07-03 Martin Liska <mliska@suse.cz>
14618 * sanopt.c (rewrite_usage_of_param): New function.
14619 (sanitize_rewrite_addressable_params): Likewise.
14620 (pass_sanopt::execute): Call rewrite_usage_of_param.
14622 2017-07-03 Richard Biener <rguenther@suse.de>
14624 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
14625 back to using VIEW_CONVERT_EXPR.
14627 2017-07-03 Martin Liska <mliska@suse.cz>
14630 * doc/extend.texi: Document when a resolver function is
14631 generated for target_clones.
14633 2017-07-03 Martin Liska <mliska@suse.cz>
14635 * asan.c (asan_emit_stack_protection): Unpoison just red zones
14636 and shadow memory of auto variables which are subject of
14637 use-after-scope sanitization.
14638 (asan_expand_mark_ifn): Add do set only when is_poison.
14640 2016-07-03 Richard Biener <rguenther@suse.de>
14642 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
14644 (vect_force_simple_reduction): Record reduction def -> phi mapping.
14645 (vectorizable_reduction): Perform reduction PHI creation when
14646 visiting a reduction PHI and adjust and simplify code generation
14647 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
14648 (vect_transform_loop): Visit reduction PHIs.
14649 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
14650 defs into the SLP tree.
14651 (vect_build_slp_tree): Reduction defs terminate the recursion.
14652 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
14654 (vect_get_vec_defs_for_stmt_copy): Export.
14655 (vect_get_vec_defs): Likewise.
14656 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
14658 (vect_get_vec_defs_for_stmt_copy): Declare.
14659 (vect_get_vec_defs): Likewise.
14661 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14663 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
14664 parameter with a "loop" parameter and use it instead of the
14665 loop containing DR_STMT. Don't check simple_iv when doing
14666 BB analysis. Describe the two analysis modes in the comment.
14668 2017-07-03 Tom de Vries <tom@codesourcery.com>
14670 PR tree-optimization/69468
14671 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
14672 (find_same_succ_bb): Handle ignore_edge_flags.
14674 2017-07-03 Tom de Vries <tom@codesourcery.com>
14676 PR tree-optimization/81192
14677 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
14679 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
14681 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
14683 2017-07-03 Tom de Vries <tom@codesourcery.com>
14685 PR tree-optimization/81192
14686 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
14687 BB_SAME_SUCC (bb) == NULL.
14689 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14691 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
14694 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14696 * dumpfile.c: Include profile-count.h
14697 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
14699 (insert_cond_bb): Update profile.
14700 * tree-cfg.h (insert_cond_bb): Update prototype.
14701 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
14702 * tree-dump.c: Do not include tree-cfg.
14704 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14706 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
14708 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14710 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
14713 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14715 * tree-complex.c (expand_complex_div_wide): update profile.
14717 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14718 Alan Hayward <alan.hayward@arm.com>
14719 David Sherwood <david.sherwood@arm.com>
14721 * Makefile.in (MACHMODE_H): Remove insn-modes.h
14722 (CORETYPES_H): New define.
14723 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
14724 (insn-modes-inline.h, s-modes-inline-h): New rules.
14725 (generated_files): Add insn-modes-inline.h.
14726 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
14727 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
14728 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
14729 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
14730 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
14731 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
14732 (build/gencodes.o, build/genconditions.o): Likewise.
14733 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
14734 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
14735 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
14736 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
14737 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
14738 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
14739 * coretypes.h: Include everything up to real.h for generators.
14740 Include insn-modes.h first. Include wide-int-print.h after
14741 wide-int.h. Include insn-modes-inline.h and then machmode.h.
14742 * machmode.h: Don't include insn-modes.h here.
14743 * function-tests.c: Remove includes of signop.h, machmode.h,
14744 double-int.h and wide-int.h.
14746 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
14748 * optc-save-gen.awk: Likewise.
14749 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
14750 * godump.c: Remove include of wide-int-print.h.
14751 * pretty-print.h: Likewise.
14752 * wide-int-print.cc: Likewise.
14753 * wide-int.cc: Likewise.
14754 * hash-map-tests.c: Remove include of signop.h.
14755 * hash-set-tests.c: Likewise.
14756 * rtl-tests.c: Likewise.
14757 * mkconfig.sh: Remove include of machmode.h.
14758 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
14760 (emit_insn_modes_inline_h): ...this new function. Emit the code
14761 into an insn-modes-inline.h header file, adding appropriate
14762 include guards and end comments.
14763 (emit_insn_modes_c_header): Remove include of machmode.h.
14764 (emit_min_insn_modes_c_header): Include coretypes.h rather than
14766 (main): Handle -i flag and call emit_insn_modes_inline_h when
14769 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14771 * tree-ssa-strlen.c (strinfo): Rename the length field to
14772 nonzero_chars. Add a full_string_p field.
14773 (compare_nonzero_chars, zero_length_string_p): New functions.
14774 (get_addr_stridx): Add an offset_out parameter.
14775 Use compare_nonzero_chars.
14776 (get_stridx): Update accordingly. Use compare_nonzero_chars.
14777 (new_strinfo): Update after above changes to strinfo.
14778 (set_endptr_and_length): Set full_string_p.
14779 (get_string_length): Update after above changes to strinfo.
14780 (unshare_strinfo): Update call to new_strinfo.
14781 (maybe_invalidate): Likewise.
14782 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
14783 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
14784 as a uhwi instead of an shwi. Update after above changes to
14785 strinfo and new_strinfo.
14786 (zero_length_string): Assert that chainsi contains full strings.
14787 Use zero_length_string_p. Update call to new_strinfo.
14788 (adjust_related_strinfos): Update after above changes to strinfo.
14789 Copy full_string_p from origsi.
14790 (adjust_last_stmt): Use zero_length_string_p.
14791 (handle_builtin_strlen): Update after above changes to strinfo and
14792 new_strinfo. Install the lhs as the string length if the previous
14793 entry didn't describe a full string.
14794 (handle_builtin_strchr): Update after above changes to strinfo
14796 (handle_builtin_strcpy): Likewise.
14797 (handle_builtin_strcat): Likewise.
14798 (handle_builtin_malloc): Likewise.
14799 (handle_pointer_plus): Likewise.
14800 (handle_builtin_memcpy): Likewise. Track nonzero characters
14801 that aren't necessarily followed by a nul terminator.
14802 (handle_char_store): Likewise.
14804 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14806 PR tree-optimization/80769
14807 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
14808 for malloc and calloc. Document the new invariant that all related
14809 strinfos have delayed lengths or none do.
14810 (verify_related_strinfos): Move earlier in file.
14811 (set_endptr_and_length): New function, split out from...
14812 (get_string_length): ...here. Also set the lengths of related
14814 (zero_length_string): Assert that chainsi has known (rather than
14816 (adjust_related_strinfos): Likewise.
14818 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14820 PR tree-optimization/81136
14821 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
14822 assert that two references with the same misalignment have the same
14823 compile-time misalignment if those compile-time misalignments
14826 2017-07-01 Andi Kleen <ak@linux.intel.com>
14828 * print-tree.c (print_node): Print all attributes.
14830 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14832 * cfg.c (scale_bbs_frequencies): New function.
14833 * cfg.h (scale_bbs_frequencies): Declare it.
14834 * cfgloopanal.c (single_likely_exit): Cleanup.
14835 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
14837 (scale_loop_profile): Likewise.
14838 (loop_version): Likewise.
14839 (create_empty_loop_on_edge): Update.
14840 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
14841 scale_loop_frequencies, scale_loop_profile, loopify,
14842 loop_version): Update prototypes.
14843 * modulo-sched.c (sms_schedule): Update.
14844 * predict.c (unlikely_executed_edge_p): Also check probability.
14845 (probably_never_executed_edge_p): Fix typo.
14846 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14847 * tree-parloops.c (gen_parallel_loop): Update.
14848 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
14849 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14850 * tree-ssa-loop-split.c (split_loop): Update.
14851 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14852 * tree-vect-loop-manip.c (vect_do_peeling): Update.
14853 (vect_loop_versioning): Update.
14854 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14856 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14858 * trans-mem.c (split_bb_make_tm_edge): Update profile.
14860 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14862 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
14863 to keep profile consistent.
14865 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14867 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
14868 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
14869 * profile-count.h (max_safe_multiplier): Make unsigned.
14870 (profile_count::guessed_zero): New.
14872 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14874 * bb-reorder.c (fix_up_crossing_landing_pad,
14875 fix_crossing_conditional_branches): Use make_single_succ_edge
14876 to keep profile consistent.
14878 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14880 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
14883 2017-07-01 Jakub Jelinek <jakub@redhat.com>
14886 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
14887 the right scopes, make sure cond_jump isn't preserved between multiple
14888 iterations. Search for fallthru edge whenever there are 3+ edges and
14889 use find_fallthru_edge for it.
14891 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14893 Patch by Alexander Monakov <amonakov@ispras.ru>
14894 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
14895 probabilities consistently.
14897 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14899 * pa.c (pa_expand_compare_and_swap_loop): Update call of
14900 emit_cmp_and_jump_insns.
14902 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14905 * tree-inline.c (expand_call_inline): Combine profile statuses.
14907 2017-06-30 Andrew Pinski <apinski@cavium.com>
14909 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
14910 fold_stmt returned true.
14912 2017-06-30 Nathan Sidwell <nathan@acm.org>
14914 * ggc.h (empty_string): Delete.
14915 * cfgexpand.c (expand_asm_stmt): Use plain "".
14916 * optabs.c (expand_asm_memory_barrier): Likewise.
14917 * stringpool.c (empty_string): Delete.
14918 (digit_vector, digit_string): Delete.
14919 (ggc_alloc_string): Use plain "", don't optimize single digit
14920 strings. Use ggc_alloc_atomic.
14922 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
14924 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
14925 comparison set and one other set, use the cost of the non-comparison
14928 2017-06-30 Nathan Sidwell <nathan@acm.org>
14930 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
14933 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
14935 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
14936 loops. Remove now unneeded calls to gimple_switch_set_label() that
14937 just set removed labels to NULL_TREE.
14939 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
14941 * tree-ssanames.c (set_range_info_raw): Abstract from ...
14942 (set_range_info): ...here. Only call set_range_info_raw if domain
14944 (set_nonzero_bits): Call set_range_info_raw.
14945 * tree-ssanames.h (set_range_info_raw): New.
14947 2017-06-30 Jakub Jelinek <jakub@redhat.com>
14950 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
14951 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
14952 of nonimmediate_operand and <store_mask_constraint> instead of m
14953 for the input operand. For V8FI iterator, always split if input
14954 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
14955 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
14956 <store_mask_predicate> instead of register_operand and
14957 <store_mask_constraint> instead of v for the input operand. Make
14958 sure both operands aren't MEMs for if not <mask_applied>.
14960 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
14962 * lto-wrapper.c (copy_file) Close both file descriptors before
14965 2017-06-30 Martin Liska <mliska@suse.cz>
14968 * multiple_target.c (create_dispatcher_calls): Make ifunc
14969 also for function that don't have calls or are not referenced.
14971 2017-06-30 Richard Biener <rguenther@suse.de>
14973 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
14974 analyze the first scalar stmt. Move vector type computation
14975 for the BB case here from ...
14976 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
14977 live operation processing in the SLP case properly.
14979 2017-06-30 Richard Biener <rguenther@suse.de>
14981 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
14983 2017-06-30 Martin Liska <mliska@suse.cz>
14986 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
14987 before BUILT_IN_UNWIND_RESUME when ASAN is used.
14989 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
14991 * doc/invoke.texi (AArch64): Add missing options and remove redundant
14994 2017-06-30 Richard Biener <rguenther@suse.de>
14996 PR tree-optimization/81249
14997 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
14998 condition reduction result to original scalar type.
15000 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15002 * profile-count.h (enum profile_quality): Fix typos and whitespace
15005 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15007 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15008 type for branch probabilities.
15010 2017-06-29 Julian Brown <julian@codesourcery.com>
15011 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15013 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15014 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15015 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15016 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15018 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15020 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15021 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15022 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15023 CC usage from generic code to here.
15024 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15025 CC usage into the target macros.
15027 2017-06-29 Maya Rashish <coypu@sdf.org>
15029 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15032 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15034 * arm/arm-builtins.c: Include profile-count.h
15035 * except.c (sjlj_emit_function_enter): Use
15036 profile_probability::unlikely.
15038 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15040 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15041 and tocrel_offset be pointer args rather than implicitly using
15043 (legitimate_constant_pool_address_p, rs6000_emit_move,
15044 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15045 tocrel_offset and use in toc_relative_expr_p call.
15046 (print_operand, print_operand_address): Use static tocrel_base_oac
15047 and tocrel_offset_oac.
15048 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15051 2017-06-29 Maya Rashish <coypu@sdf.org>
15053 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15055 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15057 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15058 objects, take into account only the alignment of 'op0' and 'mode1' if
15061 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15063 * ccmp.c (ccmp_tree_comparison_p): New function.
15064 (ccmp_candidate_p): Update to use above function.
15065 (get_compare_parts): New function.
15066 (expand_ccmp_next): Update to use new functions.
15067 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15069 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15070 take mode as argument.
15071 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15072 * expr.c (expand_expr_real_1): Pass mode as argument.
15074 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15076 * combine.c (combine_instructions): Print insns to dump_file, together
15079 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15081 * asan.c (asan_emit_stack_protection): Update.
15082 (create_cond_insert_point): Update.
15083 * auto-profile.c (afdo_propagate_circuit): Update.
15084 * basic-block.h (struct edge_def): Turn probability to
15085 profile_probability.
15086 (EDGE_FREQUENCY): Update.
15087 * bb-reorder.c (find_traces_1_round): Update.
15088 (better_edge_p): Update.
15089 (sanitize_hot_paths): Update.
15090 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
15091 (make_single_succ_edge): Update.
15092 (check_bb_profile): Update.
15093 (dump_edge_info): Update.
15094 (update_bb_profile_for_threading): Update.
15095 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
15096 probabilitycount to 0.
15097 * cfgbuild.c (compute_outgoing_frequencies): Update.
15098 * cfgcleanup.c (try_forward_edges): Update.
15099 (outgoing_edges_match): Update.
15100 (try_crossjump_to_edge): Update.
15101 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
15102 (expand_gimple_tailcall): Update.
15103 (construct_init_block): Use make_single_succ_edge.
15104 (construct_exit_block): Use make_single_succ_edge.
15105 * cfghooks.c (verify_flow_info): Update.
15106 (redirect_edge_succ_nodup): Update.
15107 (split_edge): Update.
15108 (account_profile_record): Update.
15109 * cfgloopanal.c (single_likely_exit): Update.
15110 * cfgloopmanip.c (scale_loop_profile): Update.
15111 (set_zero_probability): Remove.
15112 (duplicate_loop_to_header_edge): Update.
15113 * cfgloopmanip.h (loop_version): Update prototype.
15114 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
15115 (force_nonfallthru_and_redirect): Update.
15116 (update_br_prob_note): Update.
15117 (rtl_verify_edges): Update.
15118 (purge_dead_edges): Update.
15119 (rtl_lv_add_condition_to_bb): Update.
15120 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
15121 * cgraphunit.c (init_lowered_empty_function): Update.
15122 (cgraph_node::expand_thunk): Update.
15123 * cilk-common.c: Include profile-count.h
15124 * dojump.c (inv): Remove.
15125 (jumpifnot): Update.
15126 (jumpifnot_1): Update.
15127 (do_jump_1): Update.
15129 (do_jump_by_parts_greater_rtx): Update.
15130 (do_compare_rtx_and_jump): Update.
15131 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
15132 do_jump_1. do_compare_rtx_and_jump): Update prototype.
15133 * dwarf2cfi.c: Include profile-count.h
15134 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
15135 (sjlj_emit_dispatch_table): Likewise.
15136 * explow.c: Include profile-count.h
15137 * expmed.c (emit_store_flag_force): Update.
15138 (do_cmp_and_jump): Update.
15139 * expr.c (compare_by_pieces_d::generate): Update.
15140 (compare_by_pieces_d::finish_mode): Update.
15141 (emit_block_move_via_loop): Update.
15142 (store_expr_with_bounds): Update.
15143 (store_constructor): Update.
15144 (expand_expr_real_2): Update.
15145 (expand_expr_real_1): Update.
15146 * expr.h (try_casesi, try_tablejump): Update prototypes.
15147 * gimple-pretty-print.c (dump_probability): Update.
15148 (dump_profile): New.
15149 (dump_gimple_label): Update.
15150 (dump_gimple_bb_header): Update.
15151 * graph.c (draw_cfg_node_succ_edges): Update.
15152 * hsa-gen.c (convert_switch_statements): Update.
15153 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
15154 (find_if_case_1): Update.
15155 (find_if_case_2): Update.
15156 * internal-fn.c (expand_arith_overflow_result_store): Update.
15157 (expand_addsub_overflow): Update.
15158 (expand_neg_overflow): Update.
15159 (expand_mul_overflow): Update.
15160 (expand_vector_ubsan_overflow): Update.
15161 * ipa-cp.c (good_cloning_opportunity_p): Update.
15162 * ipa-split.c (split_function): Use make_single_succ_edge.
15163 * ipa-utils.c (ipa_merge_profiles): Update.
15164 * loop-doloop.c (add_test): Update.
15165 (doloop_modify): Update.
15166 * loop-unroll.c (compare_and_jump_seq): Update.
15167 (unroll_loop_runtime_iterations): Update.
15168 * lra-constraints.c (lra_inheritance): Update.
15169 * lto-streamer-in.c (input_cfg): Update.
15170 * lto-streamer-out.c (output_cfg): Update.
15171 * mcf.c (adjust_cfg_counts): Update.
15172 * modulo-sched.c (sms_schedule): Update.
15173 * omp-expand.c (expand_omp_for_init_counts): Update.
15174 (extract_omp_for_update_vars): Update.
15175 (expand_omp_ordered_sink): Update.
15176 (expand_omp_for_ordered_loops): Update.
15177 (expand_omp_for_generic): Update.
15178 (expand_omp_for_static_nochunk): Update.
15179 (expand_omp_for_static_chunk): Update.
15180 (expand_cilk_for): Update.
15181 (expand_omp_simd): Update.
15182 (expand_omp_taskloop_for_outer): Update.
15183 (expand_omp_taskloop_for_inner): Update.
15184 * omp-simd-clone.c (simd_clone_adjust): Update.
15185 * optabs.c (expand_doubleword_shift): Update.
15186 (expand_abs): Update.
15187 (emit_cmp_and_jump_insn_1): Update.
15188 (expand_compare_and_swap_loop): Update.
15189 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
15190 * predict.c (predictable_edge_p): Update.
15191 (edge_probability_reliable_p): Update.
15192 (set_even_probabilities): Update.
15193 (combine_predictions_for_insn): Update.
15194 (combine_predictions_for_bb): Update.
15195 (propagate_freq): Update.
15196 (estimate_bb_frequencies): Update.
15197 (force_edge_cold): Update.
15198 * profile-count.c (profile_count::dump): Add missing space into dump.
15199 (profile_count::debug): Add newline.
15200 (profile_count::differs_from_p): Explicitly convert to unsigned.
15201 (profile_count::stream_in): Update.
15202 (profile_probability::dump): New member function.
15203 (profile_probability::debug): New member function.
15204 (profile_probability::differs_from_p): New member function.
15205 (profile_probability::differs_lot_from_p): New member function.
15206 (profile_probability::stream_in): New member function.
15207 (profile_probability::stream_out): New member function.
15208 * profile-count.h (profile_count_quality): Rename to ...
15209 (profile_quality): ... this one.
15210 (profile_probability): New.
15211 (profile_count): Update.
15212 * profile.c (compute_branch_probabilities): Update.
15213 * recog.c (peep2_attempt): Update.
15214 * sched-ebb.c (schedule_ebbs): Update.
15215 * sched-rgn.c (find_single_block_region): Update.
15216 (compute_dom_prob_ps): Update.
15217 (schedule_region): Update.
15218 * sel-sched-ir.c (compute_succs_info): Update.
15219 * stmt.c (struct case_node): Update.
15220 (do_jump_if_equal): Update.
15221 (get_outgoing_edge_probs): Update.
15222 (conditional_probability): Update.
15223 (emit_case_dispatch_table): Update.
15224 (expand_case): Update.
15225 (expand_sjlj_dispatch_table): Update.
15226 (emit_case_nodes): Update.
15227 * targhooks.c: Update.
15228 * tracer.c (better_p): Update.
15229 (find_best_successor): Update.
15230 * trans-mem.c (expand_transaction): Update.
15231 * tree-call-cdce.c: Update.
15232 * tree-cfg.c (gimple_split_edge): Upate.
15233 (move_sese_region_to_fn): Upate.
15234 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
15235 * tree-eh.c (lower_resx): Upate.
15236 (cleanup_empty_eh_move_lp): Upate.
15237 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15238 * tree-inline.c (copy_edges_for_bb): Update.
15239 (copy_cfg_body): Update.
15240 * tree-parloops.c (gen_parallel_loop): Update.
15241 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
15242 (gimple_gen_time_profiler): Update.
15243 * tree-ssa-dce.c (remove_dead_stmt): Update.
15244 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15245 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
15246 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
15247 (unloop_loops): Update.
15248 (try_peel_loop): Update.
15249 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15250 * tree-ssa-loop-split.c (connect_loops): Update.
15251 (split_loop): Update.
15252 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15253 (hoist_guard): Update.
15254 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
15255 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
15256 (value_replacement): Update.
15257 * tree-ssa-reassoc.c (branch_fixup): Update.
15258 * tree-ssa-tail-merge.c (replace_block_by): Update.
15259 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
15260 (create_edge_and_update_destination_phis): Update.
15261 (compute_path_counts): Update.
15262 (recompute_probabilities): Update.
15263 (update_joiner_offpath_counts): Update.
15264 (freqs_to_counts_path): Update.
15265 (duplicate_thread_path): Update.
15266 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
15267 (struct switch_conv_info): Update.
15268 (gen_inbound_check): Update.
15269 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15270 (vect_do_peeling): Update.
15271 (vect_loop_versioning): Update.
15272 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15273 (optimize_mask_stores): Update.
15274 * ubsan.c (ubsan_expand_null_ifn): Update.
15275 * value-prof.c (gimple_divmod_fixed_value): Update.
15276 (gimple_divmod_fixed_value_transform): Update.
15277 (gimple_mod_pow2): Update.
15278 (gimple_mod_pow2_value_transform): Update.
15279 (gimple_mod_subtract): Update.
15280 (gimple_mod_subtract_transform): Update.
15281 (gimple_ic): Update.
15282 (gimple_stringop_fixed_value): Update.
15283 (gimple_stringops_transform): Update.
15284 * value-prof.h: Update.
15286 2017-06-29 Carl Love <cel@us.ibm.com>
15288 * config/rs6000/rs6000-c.c: Add support for built-in functions
15289 vector signed int vec_signed (vector float);
15290 vector signed long long vec_signed (vector double);
15291 vector signed int vec_signed2 (vector double, vector double);
15292 vector signed int vec_signede (vector double);
15293 vector signed int vec_signedo (vector double);
15294 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
15295 instruction generator.
15296 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15297 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
15298 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
15300 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
15301 vunsignede_v2df): Add define_expands.
15302 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
15303 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
15304 VEC_UNSIGNEDO): Add definitions.
15305 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15306 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
15307 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
15308 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
15309 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
15310 * config/rs6000/altivec.h (vec_signed, vec_signed2,
15311 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
15312 vec_unsignede, vec_unsignedo): Add builtin defines.
15313 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
15315 * doc/extend.texi: Update the built-in documentation file for the
15316 new built-in functions.
15318 2017-06-29 Richard Biener <rguenther@suse.de>
15320 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
15321 reduction chains to LOOP_VINFO_REDUCTIONS.
15322 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
15323 SLP reductions after processing reduction chains.
15325 2017-06-29 Nathan Sidwell <nathan@acm.org>
15327 * builtins.c (fold_builtin_FUNCTION): Use
15328 lang_hooks.decl_printable_name.
15330 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
15332 PR middle-end/81194
15333 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
15334 with only one label.
15335 * stmt.c (expand_case): Assert NCASES is greater than one.
15337 2017-06-29 Richard Biener <rguenther@suse.de>
15339 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
15341 (group_case_labels): Likewise.
15342 (find_taken_edge): Push sanity checking on val to workers...
15343 (find_taken_edge_cond_expr): ... here
15344 (find_taken_edge_switch_expr): ... and here, handle cases
15345 with just a default label.
15346 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
15347 (group_case_labels): Likewise.
15348 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
15349 group_case_labels does anything cleanup the CFG again.
15351 2017-06-29 Bin Cheng <bin.cheng@arm.com>
15353 PR tree-optimization/81196
15354 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
15355 exit condition comparing two IVs.
15357 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
15359 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
15360 profile to the dummy entry at the end of the list of architectures.
15361 * config/arm/arm-cpu-cdata.h: Regenerated.
15363 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15364 Michael Collison <michael.collison@arm.com>
15367 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
15369 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
15370 (*aarch64_reg_<mode>3_minus_mask): New pattern.
15371 (*aarch64_<optab>_reg_di3_mask2): New pattern.
15372 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
15373 of shift when the shift amount is masked with constant equal to
15374 the size of the mode.
15375 * config/aarch64/predicates.md (subreg_lowpart_operator): New
15378 2017-06-29 Martin Liska <mliska@suse.cz>
15380 * config/i386/i386.opt: Change range from [1,5] to [0,5].
15382 2017-06-29 Yury Gribov <tetra2005@gmail.com>
15385 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
15388 (edge_growth_cache_entry::edge_growth_cache_entry): New
15390 (reset_edge_growth_cache): Update to use constructor.
15392 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15394 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
15395 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15396 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15398 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
15400 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
15401 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
15403 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
15405 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
15406 (*-linux-uclibc*): Add t-uclibc tmake_file.
15407 * config/t-musl: New.
15408 * config/t-uclibc: New.
15410 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
15412 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
15414 (gen_comm_data): Emit architectural setting of arch_prof.
15415 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
15417 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
15418 (armv8-m.base, armv8-m.main): Likewise.
15419 * arm-protos.h (arm_build_target): Add profile field.
15420 (arch_option): Likewise.
15421 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
15423 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
15424 arm_active_target.profile.
15426 2017-06-28 Richard Biener <rguenther@suse.de>
15428 PR middle-end/81227
15429 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
15430 TYPE_OVERFLOW_WRAPS.
15431 * match.pd (negate_expr_p): Likewise.
15432 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
15433 fold_build2, not fold_binary.
15435 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15437 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
15438 Convert memory address to Pmode.
15439 (aarch64_print_operand): Assert MEM operands are always Pmode.
15441 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15444 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
15445 Remove redundant if.
15446 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
15447 * config/arm/aarch-common-protos.h
15448 (aarch_forward_to_shift_is_not_shifted_re): Remove.
15449 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
15451 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
15454 * multiple_target.c (create_dispatcher_calls): Set the default
15455 clone to be static, not public.
15457 2017-06-28 Richard Biener <rguenther@suse.de>
15459 * tree-vect-loop.c (vectorizable_reduction): Move special
15460 cond reduction IV var creation ...
15461 (vect_create_epilog_for_reduction): ... here. Remove induction_index
15462 parameter. Use STMT_VINFO_VECTYPE.
15463 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
15466 2017-06-28 Martin Liska <mliska@suse.cz>
15469 * ipa-visibility.c (non_local_p): Handle visibility.
15471 2017-06-28 Martin Liska <mliska@suse.cz>
15474 * common.opt: Add IntegerRange to various options.
15475 * opt-functions.awk (integer_range_info): New function.
15476 * optc-gen.awk: Add integer_range_info to cl_options struct.
15477 * opts-common.c (decode_cmdline_option): Handle
15478 CL_ERR_INT_RANGE_ARG.
15479 (cmdline_handle_error): Likewise.
15480 * opts.c (print_filtered_help): Show valid interval in
15481 when --help is provided.
15482 * opts.h (struct cl_option): Add range_min and range_max fields.
15483 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
15485 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
15487 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
15488 (x * C EQ/NE y * C): New transformation.
15490 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
15492 * genmultilib (combination_space): Accept '+' in option names.
15494 2017-06-28 Martin Liska <mliska@suse.cz>
15497 * asan.c (instrument_derefs): Bail out inner references
15498 that are hard register variables.
15500 2017-06-28 Jakub Jelinek <jakub@redhat.com>
15503 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
15504 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
15506 2017-06-28 Richard Biener <rguenther@suse.de>
15508 * tree-vectorizer.h (vect_get_vec_defs): Remove.
15509 (vect_get_slp_defs): Adjust.
15510 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
15512 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
15514 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
15515 get_initial_defs_for_reduction instead of vect_get_vec_defs.
15516 (vectorizable_reduction): Adjust.
15517 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
15519 (vect_get_slp_defs): Likewise.
15520 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
15521 (vectorizable_bswap): Adjust.
15522 (vectorizable_call): Likewise.
15523 (vectorizable_conversion): Likewise.
15524 (vectorizable_assignment): Likewise.
15525 (vectorizable_shift): Likewise.
15526 (vectorizable_operation): Likewise.
15527 (vectorizable_store): Likewise.
15528 (vectorizable_condition): Likewise.
15529 (vectorizable_comparison): Likewise.
15531 2017-06-28 Michael Collison <michael.collison@arm.com>
15534 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
15536 (arm_gen_movmemqi): Removed unused variable 'i'.
15537 Convert 'for' loop into 'while' loop.
15538 (arm_expand_prologue): Remove last unnecessary set of insn.
15539 (thumb_pop): Remove unused variable 'pushed_words'.
15540 (thumb_exit): Remove last unnecessary set of regs_to_pop.
15542 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15544 * config/s390/predicates.md: Use s390_rel_address_ok_p.
15545 * config/s390/s390-protos.h: Add prototype of
15546 s390_rel_address_ok_p.
15547 * config/s390/s390.c (s390_got_symbol): New function.
15548 (s390_rel_address_ok_p): New function.
15549 (legitimize_pic_address): Use s390_rel_address_ok_p.
15550 (s390_load_got): Use s390_got_symbol.
15551 (s390_option_override): Issue error if
15552 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
15553 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
15555 * config/s390/s390.opt: New option mpic-data-is-text-relative.
15557 2017-06-27 Andrew Pinski <apinski@cavium.com>
15559 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
15560 (X * copysign (1.0, X)): New pattern.
15561 (X * copysign (1.0, -X)): New pattern.
15562 (copysign (-1.0, CST)): New pattern.
15564 2017-06-27 Joseph Myers <joseph@codesourcery.com>
15566 * genmultilib (combination_space): Remove variable.
15567 Validate reuse rules against regular expression for any sequence
15568 of multilib options in any order.
15570 2017-06-27 Michael Collison <michael.collison@arm.com>
15572 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
15573 call aarch64_split_simd_combine.
15574 * (aarch64_combine_internal<mode>): Delete pattern.
15575 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
15576 Allow register and subreg operands.
15578 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15580 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
15581 specific need, just fallback on defaults.
15582 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
15584 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15585 Olivier Hainque <hainque@adacore.com>
15587 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
15589 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
15590 targets. Pick a default if no particular attempt applied.
15591 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
15594 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15596 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
15597 (x86_64-wrs-vxworks7): Likewise.
15599 2017-06-27 Marek Polacek <polacek@redhat.com>
15602 * ubsan.c (instrument_null): Check get_base_address's result for null.
15604 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15606 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
15608 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15610 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
15611 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
15612 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
15613 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
15614 New function types.
15615 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
15616 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
15617 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
15618 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
15619 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
15620 BUILT_IN_FEUPDATEENV): New builtins.
15621 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
15622 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
15623 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
15624 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
15626 (builtin_structptr_types): Adjust size.
15627 * tree.c (builtin_structptr_types): Add four entries.
15629 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15630 Olivier Hainque <hainque@adacore.com>
15632 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
15633 (TLS_SYM): New local macro, forcing reference to __tls__ on
15634 link command lines for VxWorks 7 RTPs, triggering initialization
15636 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
15637 OS features TLS support, true for RTPs on VxWorks 7.
15638 * config/vxworks.c (vxworks_override_options): Setup emutls
15641 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15643 * predict.c (test_prediction_value_range): Use -1U instead of -1
15644 to avoid narrowing conversion warning.
15645 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
15646 to avoid narrowing conversion warning.
15647 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
15649 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
15651 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15653 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
15654 64bit configurations.
15655 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
15656 (SIZE_TYPE): Likewise.
15657 * config/vxworks.c (vxworks_emutls_var_fields): Use
15658 long_unsigned_type_node instead of unsigned_type_node as the offset
15659 field type, which is "pointer" mode in emutls.c.
15661 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15664 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
15666 PR middle-end/81207
15667 * gimple-fold.c (replace_call_with_call_and_fold): Handle
15668 gimple_vuse copying separately from gimple_vdef copying.
15670 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15672 * value-prof.c (free_hist): Remove call to memset and the enclosing if
15675 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15676 Olivier Hainque <hainque@adacore.com>
15678 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
15679 for all vxworks7 targets.
15680 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
15681 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
15682 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
15683 variations for VX6/VX7 and 32/64bits later on in ...
15684 (VXWORKS_LIB_SPEC): Leverage new macros.
15685 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
15686 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
15688 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15690 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
15691 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
15693 2017-06-26 Carl Love <cel@us.ibm.com>
15695 * config/rs6000/rs6000-c.c: Add support for built-in functions
15696 vector bool char vec_reve (vector bool char);
15697 vector signed char vec_reve (vector signed char);
15698 vector unsigned char vec_reve (vector unsigned char);
15699 vector bool int vec_reve (vector bool int);
15700 vector signed int vec_reve (vector signed int);
15701 vector unsigned int vec_reve (vector unsigned int);
15702 vector bool long long vec_reve (vector bool long long);
15703 vector signed long long vec_reve (vector signed long long);
15704 vector unsigned long long vec_reve (vector unsigned long long);
15705 vector bool short vec_reve (vector bool short);
15706 vector signed short vec_reve (vector signed short);
15707 vector double vec_reve (vector double);
15708 vector float vec_reve (vector float);
15709 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
15710 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
15711 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
15712 (altivec_vreve): New pattern.
15713 * config/rs6000/altivec.h (vec_reve): New define.
15714 * doc/extend.texi (vec_rev): Update the built-in documentation file
15715 for the new built-in functions.
15717 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15719 PR tree-optimization/71815
15720 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
15722 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
15724 (slsr_process_phi): Likewise.
15725 (replace_uncond_cands_and_profitable_phis): Don't replace a
15726 multiply candidate with a stride of 1 (copy or cast).
15727 (phi_incr_cost): Call uses_consumed_by_stmt rather than
15729 (lowest_cost_path): Likewise.
15730 (total_savings): Likewise.
15732 2017-06-26 Richard Biener <rguenther@suse.de>
15735 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
15736 Use def_builtin_pure for all gather builtins.
15738 2017-06-26 Richard Biener <rguenther@suse.de>
15740 PR tree-optimization/81203
15741 * tree-tailcall.c (find_tail_calls): Do not move stmts into
15742 non-dominating BBs.
15744 2017-06-26 Marek Polacek <polacek@redhat.com>
15747 * doc/invoke.texi: Document -Wmultistatement-macros.
15749 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
15751 * doc/sourcebuild.texi (ARM-specific attributes): Document new
15752 arm_neon_ok_no_float_abi effective target.
15754 2017-06-26 Richard Biener <rguenther@suse.de>
15756 PR tree-optimization/80928
15757 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
15758 (copy_bbs): Set BB_DUPLICATED flag early.
15759 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
15761 (execute_on_shrinking_pred): Likewise.
15762 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
15763 BB_DUPLICATED blocks.
15764 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
15765 iterate over all PHIs considering removal of *gsi.
15767 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
15769 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
15772 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15774 * config/rs6000/rs6000-string.c: (expand_block_clear,
15775 do_load_for_compare, select_block_compare_mode,
15776 compute_current_alignment, expand_block_compare,
15777 expand_strncmp_align_check, expand_strn_compare,
15778 expand_block_move, rs6000_output_load_multiple)
15779 Move functions related to string/block move/compare
15780 to a separate file.
15781 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
15782 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
15783 for this function which is now used in two files.
15784 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
15785 * config.gcc: Add rs6000-string.o to extra_objs for
15786 targets powerpc*-*-* and rs6000*-*-*.
15788 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
15791 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
15792 32-bit, since indexed is not valid for DImode.
15793 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
15794 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
15795 (define_peephole2 for Altivec d-form load): Add 32-bit support.
15796 (define_peephole2 for Altivec d-form store): Likewise.
15799 * multiple_target.c (create_dispatcher_calls): Only create the
15800 dispatcher call if the function is the default clone of a
15801 versioned function.
15803 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
15805 PR middle-end/80902
15806 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
15807 a call, force the call to not be a tail call.
15809 2017-06-23 Jeff Law <law@redhat.com>
15811 * doc/contrib.texi: Add entry for Steven Pemberton's work on
15814 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
15816 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
15817 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
15818 handling for early expansion of vector shifts (sl,sr,sra,rl).
15819 (builtin_function_type): Add vector shift right instructions
15820 to the unsigned argument list.
15822 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
15824 rtl-optimizatoin/79286
15825 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
15826 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
15827 trap. PIC register plus a const unspec without offset can never trap.
15829 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
15831 * tree.h (builtin_structptr_type): New type.
15832 (builtin_structptr_types): Declare new array.
15833 * tree.c (builtin_structptr_types): New array.
15834 (free_lang_data, build_common_tree_nodes): Use it.
15836 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
15839 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
15842 2017-06-22 Matt Turner <mattst88@gmail.com>
15844 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
15845 Lake models to skylake case. Assume skylake for unknown
15846 models with clflushopt.
15848 2017-06-22 Jeff Law <law@redhat.com>
15850 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
15851 frame sizes that do not satisfy aarch64_uimm12_shift.
15853 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
15855 * profile-count.h (apply_probability,
15856 apply_scale, probability_in): Fix checks for zero.
15858 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15860 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
15861 * doc/cppdiropts.texi (-I @var{dir}): Document it.
15863 2016-06-22 Richard Biener <rguenther@suse.de>
15865 * tree-vect-loop.c (vect_model_reduction_cost): Handle
15866 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
15867 REDUC_MAX_EXPR support.
15868 (vectorizable_reduction): Likewise.
15869 (vect_create_epilog_for_reduction): Likewise.
15871 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
15873 * match.pd (A / (1 << B) -> A >> B): New.
15874 * generic-match-head.c: Include optabs-tree.h.
15875 * gimple-match-head.c: Likewise.
15876 * optabs-tree.h (target_supports_op_p): New.
15877 * optabs-tree.c (target_supports_op_p): New.
15879 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15881 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
15882 $gcc_cv_ld --help output.
15883 (gcc_cv_ld_demangle): Likewise.
15884 (gcc_cv_ld_eh_frame_hdr): Likewise.
15885 (gcc_cv_ld_pie): Likewise.
15886 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
15887 (gcc_cv_ld_buildid): Likewise.
15888 (gcc_cv_ld_sysroot): Likewise.
15889 (ld_bndplt_support): Likewise.
15890 (ld_pushpopstate_support): Likewise.
15891 * configure: Regenerate.
15892 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
15894 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15897 * config/i386/sse.md (round<mode>2): Renumber match_dup and
15898 operands indexes to avoid gap between operands and match_dups.
15900 2017-06-21 Andrew Pinski <apinski@cavium.com>
15902 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
15903 Increment Arith_shift and Arith_shift_reg by 1.
15904 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
15906 * config/aarch64/aarch64.c (thunderx_tunings): Enable
15907 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
15908 (aarch64_strip_extend): Add new argument and test for it.
15909 (aarch64_cheap_mult_shift_p): New function.
15910 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
15911 add a cost if it is true.
15912 Update calls to aarch64_strip_extend.
15913 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
15915 2017-06-21 Andrew Pinski <apinski@cavium.com>
15917 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
15919 (thunderxt88): Likewise.
15920 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
15921 (thunderx_prefetch_tune): New variable.
15922 (thunderx2t99_prefetch_tune): Update for the correct values.
15923 (thunderxt88_tunings): New variable.
15924 (thunderx_tunings): Use thunderx_prefetch_tune instead of
15925 generic_prefetch_tune.
15926 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
15928 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15930 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
15931 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
15932 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
15933 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
15934 (aarch64_atomic_cas<mode>, GPI): Likewise.
15936 2017-06-21 Martin Liska <mliska@suse.cz>
15938 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
15939 statements on cold and hot labels.
15940 * predict.c (tree_estimate_probability_bb): Remove the
15941 prediction from this place.
15943 2017-06-21 Martin Liska <mliska@suse.cz>
15945 PR tree-optimization/79489
15946 * gimplify.c (maybe_add_early_return_predict_stmt): New
15948 (gimplify_return_expr): Call the function.
15949 * predict.c (tree_estimate_probability_bb): Remove handling
15951 * predict.def: Update comment about early return predictor.
15952 * gimple-predict.h (is_gimple_predict): New function.
15953 * predict.def: Change default value of early return to 66.
15954 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
15956 * passes.def: Put pass_strip_predict_hints to the beginning of
15959 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
15961 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
15962 FUNCTION_DECL declarations.
15963 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
15965 (dwaf2out_decl): Likewise.
15966 * godump.c (go_early_global_decl): Skip call to the real debug hook
15967 for FUNCTION_DECL declarations.
15968 * passes.c (rest_of_decl_compilation): Skip call to the
15969 early_global_decl debug hook for FUNCTION_DECL declarations, unless
15970 -fdump-go-spec is passed.
15972 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15974 * config/i386/i386.c (struct builtin_isa): New field pure_p.
15975 Reorder for compactness.
15976 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
15977 (def_builtin_pure, def_builtin_pure2): New functions.
15978 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
15980 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15982 * match.pd (nop_convert): New predicate.
15983 ((A +- CST1) +- CST2): Allow some NOP conversions.
15985 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15988 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
15989 with ctors/dtors if GOVD_SHARED is set.
15991 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15993 * config/aarch64/aarch64.md (movti_aarch64):
15994 Emit mov rather than orr.
15995 (movtf_aarch64): Likewise.
15996 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
15997 Emit mov rather than orr.
15999 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16001 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16002 Swap alternatives, make integer dup more expensive.
16004 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16006 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16007 Return true for non-tls symbols.
16009 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16011 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16012 (cortex-a75): Likewise.
16013 (cortex-a75.cortex-a55): Likewise.
16014 * config/aarch64/aarch64-tune.md: Regenerate.
16015 * doc/invoke.texi (-mtune): Document new values for -mtune.
16017 2017-06-21 Tom de Vries <tom@codesourcery.com>
16019 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16020 stack_size feature.
16021 (Effective-Target Keywords, Other attributes): Suggest using
16022 dg-add-options stack_size feature to get stack limit in stack_size
16023 effective target documentation.
16025 2017-06-21 Julian Brown <julian@codesourcery.com>
16026 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16028 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16029 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16030 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16032 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16033 attribute type list for neon_multiply.
16034 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16035 attribute type list for neon_multiply.
16036 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16037 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16038 attribute type list for neon_multiply.
16039 * config/arm/types.md (crypto_pmull): Add.
16040 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16041 attribute type list.
16043 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16045 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16048 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16050 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16051 to make sure not to dereference a NULL cost_classes_ptr pointer.
16053 2017-06-20 Carl Love <cel@us.ibm.com>
16055 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16056 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16057 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16058 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16059 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16060 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16061 VMULOSW): New enum "unspec" values.
16062 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16063 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16064 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16065 altivec_vmulosw): New patterns.
16066 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16067 VMULOSW): Add definitions.
16069 2017-06-20 Julia Koval <julia.koval@intel.com>
16071 * config/i386/i386.c: Fix rounding expand for new pattern.
16072 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16074 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16076 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16077 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
16079 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16081 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
16084 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16086 * config/aarch64/aarch64-cores.def: Rearrange to sort by
16087 architecture, then by implementer ID.
16088 * config/aarch64/aarch64-tune.md: Regenerate.
16090 2017-06-20 Richard Biener <rguenther@suse.de>
16092 PR middle-end/81097
16093 * fold-const.c (split_tree): Fold to type before negating.
16095 2017-06-20 David Malcolm <dmalcolm@redhat.com>
16097 * diagnostic-show-locus.c
16098 (selftest::test_fixit_deletion_affecting_newline): New function.
16099 (selftest::diagnostic_show_locus_c_tests): Call it.
16101 2017-06-20 Andreas Schwab <schwab@suse.de>
16104 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
16107 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16109 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
16110 __ARM_FEATURE_COPROC according to support.
16112 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16114 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16115 Rewritten to avoid overflow for > 32-bit pointers.
16118 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
16119 by removing enum keyword.
16120 (ubsan_type_descriptor): Likewise. Formatting fix.
16123 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
16124 splitter): Require TARGET_SSE2 in the condition.
16126 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
16129 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
16130 for doing vector set of SFmode on ISA 3.0.
16131 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
16132 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
16134 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
16135 SFmode value into a V4SF variable that was extracted from another
16136 V4SF variable without converting the element to double precision
16137 and back to single precision vector format.
16138 (vsx_insert_extract_v4sf_p9_2): Likewise.
16140 2017-06-19 Jakub Jelinek <jakub@redhat.com>
16142 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
16143 in UWHI to avoid undefined overflow.
16146 * ubsan.h (enum ubsan_encode_value_phase): New.
16147 (ubsan_encode_value): Change second argument to
16148 enum ubsan_encode_value_phase with default value of
16149 UBSAN_ENCODE_VALUE_GENERIC.
16150 * ubsan.c (ubsan_encode_value): Change second argument to
16151 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
16152 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
16153 create_tmp_var_raw instead of create_tmp_var and use a
16155 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
16156 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
16157 ubsan_encode_value callers.
16160 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
16161 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
16162 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
16164 2017-06-19 Richard Biener <rguenther@suse.de>
16166 PR middle-end/81118
16167 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
16168 estimates if we changed anything.
16170 2017-06-19 Richard Biener <rguenther@suse.de>
16172 PR tree-optimization/80887
16173 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
16174 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
16175 simplified lookups, then reset mprts_hook.
16176 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
16178 (try_to_simplify): Likewise.
16180 2017-06-19 Martin Liska <mliska@suse.cz>
16183 * gimplify.c (gimplify_switch_expr):
16184 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
16186 2017-06-19 Martin Liska <mliska@suse.cz>
16188 * doc/install.texi: Document that PGO runs in 4 stages.
16190 2017-06-19 Martin Liska <mliska@suse.cz>
16193 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
16194 to dispatcher function name.
16195 * multiple_target.c (replace_function_decl): New function.
16196 (create_dispatcher_calls): Redirect both edges and references.
16198 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
16200 * profile-count.c (profile_count::dump): Dump quality.
16201 (profile_count::differs_from_p): Update for unsigned val.
16202 * profile-count.h (profile_count_quality): New enum.
16203 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
16205 2017-06-19 Richard Biener <rguenther@suse.de>
16207 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
16208 struct function as arg.
16209 (estimate_numbers_of_iterations): Export overload with loop arg.
16210 (free_numbers_of_iterations_estimates_loop): Use an overload of
16211 free_numbers_of_iterations_estimates instead.
16212 * tree-cfg.c (remove_bb): Adjust.
16213 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
16214 * tree-parloops.c (gen_parallel_loop): Likewise.
16215 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
16217 (tree_unroll_loops_completely): Likewise.
16218 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
16219 Use an overload instead and export.
16220 (estimated_loop_iterations): Adjust.
16221 (max_loop_iterations): Likewise.
16222 (likely_max_loop_iterations): Likewise.
16223 (estimate_numbers_of_iterations): Take struct function as arg
16225 (loop_exits_before_overflow): Adjust.
16226 (free_numbers_of_iterations_estimates_loop): Use an overload.
16227 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
16228 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
16230 2017-06-19 Richard Biener <rguenther@suse.de>
16233 * ipa-prop.c (find_constructor_constant_at_offset): Handle
16234 RANGE_EXPR conservatively.
16236 2017-06-16 Carl Love <cel@us.ibm.com>
16238 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16239 definitions for vec_float, vec_float2, vec_floato,
16240 vec_floate built-ins.
16241 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
16242 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
16244 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
16245 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
16246 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
16247 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
16248 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
16249 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
16250 vec_floato): Add builtin defines.
16251 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
16252 Update the built-in documentation file for the new built-in
16255 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16257 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
16258 (mthumb): Mark as the negative of -marm.
16260 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16262 * doc/invoke.texi (ARM Options, -mcpu): Document supported
16264 (ARM Options, -mtune): Document that this accepts the same
16265 extension options as -mcpu.
16266 (ARM Options, -mfpu): Document addition of -mfpu=auto.
16268 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16270 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
16271 permitted extensions.
16273 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16275 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
16276 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
16277 (armv8-m.main): Add option +nodsp.
16278 * config/arm/arm-cpu-cdata.h: Regenerated.
16280 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16282 * config/arm/t-fuchsia: New file.
16283 * config.gcc (arm*-*-fuchsia*): Use it.
16285 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16287 * config/arm/t-symbian: Rewrite for new option infrastructure.
16289 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16291 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
16292 (MULTILIB_REQUIRED): Likewise.
16294 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16296 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
16297 (MULTILIB_RESUE): Likewise.
16298 (MULTILIB_MATCHES): Likewise.
16299 (MULTLIB_REQUIRED): Likewise.
16301 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16303 * config/arm/t-rtems: Rewrite for new option framework.
16305 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16307 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
16308 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
16309 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
16310 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
16311 * config/arm/t-multilib: ... here.
16312 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
16313 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
16314 armv7-a and armv8*-a when A-profile libraries have not been built.
16315 * config/arm/t-rmprofile: Rewrite.
16317 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16319 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
16320 with a backslash. Remove the backslash after substituting unescaped
16322 * doc/fragments.texi (MULTILIB_REUSE): Document it.
16324 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16326 * config.gcc: (arm*-*-*): When building a-profile libraries, force
16327 the driver to pass through the default setting of -mfloat-abi.
16328 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
16330 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
16331 (all_feat_combs): New rule.
16332 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
16334 * config/arm/t-aprofile: Rewrite.
16336 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16338 * config/arm/arm.h (FPUTYPE_AUTO): Define.
16339 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
16340 fpu is not specified by the user/command-line.
16341 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
16342 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
16343 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
16344 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
16345 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
16346 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
16348 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16350 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
16351 * config/arm/t-arm-elf: Rewritten.
16353 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16355 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
16356 have some floating-point instructions.
16357 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
16358 (TARGET_MAYBE_HARD_FLOAT): New macro.
16359 * config/arm/arm-builtins.c (arm_init_builtins): Use
16360 TARGET_MAYBE_HARD_FLOAT.
16361 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
16363 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16365 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
16366 (configargs.h): Include it.
16367 (arm_print_hint_for_fpu_option): New function.
16368 (arm_parse_fpu_option): New function.
16369 (candidate_extension): New class.
16370 (arm_canon_for_multilib): New function.
16371 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
16372 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
16373 (ARCH_CANONICAL_SPECS): New macro.
16374 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
16376 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16378 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
16379 are set after handling multilib fragments. Set target_cpu_default2
16382 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16384 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
16386 (arm*-*-*): Set target_cpu_default2 to a quoted string.
16387 * config/arm/parsecpu.awk (check_cpu): Validate any extension
16389 (check_arch): Likewise.
16390 * config/arm/arm.c (arm_configure_build_target): Handle
16391 TARGET_CPU_DEFAULT being a string constant. Scan any feature
16392 options in the default.
16394 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16396 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
16397 when an option is an alias of another.
16398 * config/arm/parsecpu.awk (optalias): New parser token.
16399 (gen_comm_data): Mark non-alias options as such. Emit entries
16400 for extension aliases.
16401 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
16402 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
16403 (armv6kz, armv6zk, armv6t2): Likewise.
16404 (armv7): Make vfpv3-d16 an alias.
16405 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
16407 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
16408 Sort in canonical order.
16409 (armv8-a): Sort in canonical order.
16410 (armv8.1-a, armv8.2-a): Likewise.
16411 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
16413 (cortex-a9): Sort in canonical order.
16414 * config/arm/arm.c (selftests.h): Include it.
16415 (arm_test_cpu_arch_data): New function.
16416 (arm_run_self_tests): New function.
16417 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
16418 (targetm): Move declaration to the end of the file.
16419 * arm-cpu-cdata.h: Regenerated.
16421 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16423 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
16424 call to target_mode_check describing the type of option passed.
16425 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
16426 (arm_target_thumb_only): Use arm_parse_arch_option_name or
16427 arm_parse_cpu_option_name to match parameters against list of
16429 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
16430 arm_arch_core_flags data structure.
16431 * config/arm/arm-cpu_cdata.h: Regenerated.
16433 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16435 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
16437 (arm_print_hint_for_cpu_option): Likewise.
16438 (arm_print_hint_for_arch_option): Likewise.
16439 (arm_parse_cpu_option_name): Likewise.
16440 (arm_parse_arch_option_name): Likewise.
16441 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
16442 of entries in the all_fpus list.
16443 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
16444 (arm_parse_cpu_option_name): Declare.
16445 (arm_parse_arch_option_name): Declare.
16446 (arm_parse_option_features): Declare.
16447 (arm_intialize_isa): Declare.
16448 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
16450 (gen_comm_data): ... here. Make definitions non-static.
16451 * config/arm/arm-cpu-data.h: Regenerated.
16452 * config/arm/arm-cpu-cdata.h: Regenerated.
16454 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16456 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
16457 (cpu_arch_extension): New structure.
16458 (cpu_arch_option, arch_option, cpu_option): New structures.
16459 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
16460 architecture types.
16461 (gen_data): Generate new format data tables.
16462 * config/arm/arm.c (cpu_tune): New structure.
16463 (cpu_option, processors): Delete.
16464 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
16465 (arm_print_hint_for_cpu_option): ... this and ...
16466 (arm_print_hint_for_arch_option): ... this.
16467 (arm_parse_arch_cpu_name): Delete. Replace with ...
16468 (arm_parse_cpu_option_name): ... this and ...
16469 (arm_parse_arch_option_name): ... this.
16470 (arm_unrecognized_feature): Change type of target parameter to
16472 (arm_parse_arch_cpu_features): Delete. Replace with ...
16473 (arm_parse_option_features): ... this.
16474 (arm_configure_build_target): Rework to use new configuration data
16476 (arm_print_tune_info): Rework for new configuration data tables.
16477 * config/arm/arm-cpu-data.h: Regenerated.
16478 * config/arm/arm-cpu.h: Regenerated.
16480 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16482 * Makefile.in (OBJS): Move sbitmap.o from here ...
16483 (OBJS-libcommon): ... to here.
16485 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16487 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
16488 (ISA_ALL_CRYPTO): New macro.
16489 (ISA_ALL_SIMD): New macro
16490 (ISA_ALL_FP): New macro.
16491 * config/arm/arm.c (fpu_bitlist): Update initializer.
16492 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
16494 (arm9e): Add fpu. Add option for nofp
16495 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
16496 (arm926ej-s, arm1026ej-s): Likewise.
16497 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
16498 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
16499 neon-fp16, neon-vfpv4, nofp and nosimd.
16500 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
16501 (cortex-a8): Add fpu. Add option for nofp.
16502 (cortex-a9): Add fpu. Add options for nosimd and nofp.
16503 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
16504 (cortex-r4f): Add fpu.
16505 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
16506 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
16508 (cortex-r8): Likewise.
16509 (cortex-m4): Add fpu. Add option for nofp.
16510 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
16511 (cortex-a17.cortex-a7): Likewise.
16512 (cortex-a32): Add fpu. Add options for crypto and nofp.
16513 (cortex-a35, cortex-a53): Likewise.
16514 (cortex-a57): Add fpu. Add option for crypto.
16515 (cortex-a72, cortex-a73): Likewise.
16516 (exynos-m1): Likewise.
16517 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
16518 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
16519 (cortex-m33): Add fpu. Add option for nofp.
16520 * config/arm/arm-cpu-cdata.h: Regenerated
16521 * config/arm/arm-cpu-data.h: Regenerated.
16523 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16525 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
16526 (armv5te, armv5tej): Likewise.
16527 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
16528 (armv7): Add options fp and vfpv3-d16.
16529 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
16530 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
16532 (armv7ve): Likewise.
16533 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
16534 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
16535 (armv8-a): Add nocrypto option.
16536 (armv8.1-a, armv8.2-a): Likewise.
16537 (armv8-m.main): add options fp, fp.dp and nofp.
16539 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16541 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
16543 (armv8-a+crc): Delete.
16544 (armv8.1-a): Add options simd, crypto and nofp.
16545 (armv8.2-a): Add options fp16, simd, crypto and nofp.
16546 (armv8.2-a+fp16): Delete.
16547 (armv8-m.main): Add option dsp.
16548 (armv8-m.main+dsp): Delete.
16549 (cortex-a8): Add fpu. Add nofp option.
16550 (cortex-a9): Add fpu. Add nofp and nosimd options.
16551 * config/arm/parsecpu.awk (gen_data): Generate option tables and
16552 link to main cpu and architecture data structures.
16553 (gen_comm_data): Only put isa attributes from the main architecture
16555 (option): New statement for architecture and CPU entries.
16556 * arm.c (struct cpu_option): New structure.
16557 (struct processors): Add entry for options.
16558 (arm_unrecognized_feature): New function.
16559 (arm_parse_arch_cpu_name): Ignore any characters after the first
16561 (arm_parse_arch_cpu_feature): New function.
16562 (arm_configure_build_target): Separate out any CPU and architecture
16563 features and parse separately. Don't error out if -mfpu=auto is
16564 used with only an architecture string.
16565 (arm_print_asm_arch_directives): New function.
16566 (arm_file_start): Call it.
16567 * config/arm/arm-cpu-cdata.h: Regenerated.
16568 * config/arm/arm-cpu-data.h: Likewise.
16569 * config/arm/arm-tables.opt: Likewise.
16571 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16573 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
16574 assembler when it is not -mfpu=auto.
16576 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16578 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
16579 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
16580 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
16581 (ASM_CPU_SPEC): Rewrite.
16582 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
16583 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
16584 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
16585 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
16586 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
16587 copied string is NUL-terminated. Also strip any characters prefixed
16589 (arm_rewrite_selected_arch): New function.
16590 (arm_rewrite_march): New function.
16592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16594 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
16595 (x_arm_cpu_string, x_arm_tune_string): Likewise.
16596 (march, mcpu, mtune): Convert to string-based options.
16597 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
16598 (arm_parse_arch_cpu_name): New function.
16599 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
16600 identify selected architecture or CPU.
16601 (arm_option_save): New function.
16602 (TARGET_OPTION_SAVE): Redefine.
16603 (arm_option_restore): Restore string options.
16604 (arm_option_print): Print string options.
16606 2017-06-16 Martin Sebor <msebor@redhat.com>
16608 PR tree-optimization/80933
16609 PR tree-optimization/80934
16610 * builtins.c (fold_builtin_3): Do not handle bcmp here.
16611 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
16612 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
16613 (gimple_fold_builtin): Call them.
16615 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16617 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
16618 as unlikely; update profile.
16620 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16622 * predict.c (force_edge_cold): Handle declaring edges impossible
16625 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16627 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
16629 (try_unroll_loop_completely): Fix reporting.
16631 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16633 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
16635 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
16638 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
16639 if given a non-constant argument for an intrinsic which requires a
16642 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16644 * profile.c (compare_freqs): New function.
16645 (branch_prob): Sort edge list.
16646 (find_spanning_tree): Assume that the list is priority sorted.
16648 2017-06-16 Richard Biener <rguenther@suse.de>
16650 PR tree-optimization/81090
16651 * passes.def (pass_record_bounds): Remove.
16652 * tree-pass.h (make_pass_record_bounds): Likewise.
16653 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
16654 make_pass_record_bounds): Likewise.
16655 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
16656 not free niter estimates at the beginning but at the end.
16657 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
16659 2017-06-16 Richard Biener <rguenther@suse.de>
16661 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
16662 initializer to workaround ICE in host GCC 4.8.
16664 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16666 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
16668 (clone_inlined_nodes): Update.
16670 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16672 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
16673 prefetch settings, and enable prefetching by default at -O3.
16675 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16677 * config/aarch64/aarch64.c (aarch64_override_options_internal):
16678 Set flag_prefetch_loop_arrays according to tuning data.
16680 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16682 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
16683 New tune structure.
16684 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
16685 [Unrelated to main purpose of the patch] Place the pointer field last
16686 to enable type checking errors when tune structure are wrongly merged.
16687 * config/aarch64/aarch64.c (generic_prefetch_tune,)
16688 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
16689 (thunderx2t99_prefetch_tune): New tune constants.
16690 (tune_params *_tunings): Update all tunings (no functional change).
16691 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
16692 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
16693 from tunings structures.
16695 2017-06-16 Jakub Jelinek <jakub@redhat.com>
16698 * ubsan.c (instrument_null): Add T argument, use it instead
16699 of computing it based on IS_LHS.
16700 (instrument_object_size): Likewise.
16701 (pass_ubsan::execute): Adjust instrument_null and
16702 instrument_object_size callers to pass gimple_get_lhs or
16703 gimple_assign_rhs1 result to it. Use instrument_null instead of
16704 calling get_base_address and instrument_mem_ref. Handle
16705 aggregate call arguments for object-size sanitization.
16707 2017-06-16 Yury Gribov <tetra2005@gmail.com>
16709 PR tree-optimization/81089
16710 * tree-vrp.c (is_masked_range_test): Validate operands of
16713 2017-06-15 Martin Sebor <msebor@redhat.com>
16716 * dumpfile.c (dump_register): Avoid calling memset to initialize
16717 a class with a default ctor.
16718 * gcc.c (struct compiler): Remove const qualification.
16719 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
16720 * hash-table.h: Ditto.
16721 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
16723 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
16724 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
16726 * params.h (struct param_info): Make struct members non-const.
16727 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
16728 with default initialization.
16729 * vec.h (vec_copy_construct, vec_default_construct): New helper
16731 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
16732 with vec_copy_construct.
16733 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
16734 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
16735 * doc/invoke.texi (-Wclass-memaccess): Document.
16737 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16739 * emit-rtl.h (is_leaf): Update comment about local
16740 register allocator.
16742 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16745 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
16746 for a variable to have a section before checking if the section has a
16748 Set section to.persistent if persistent attribute is set.
16749 Warn if .persistent attribute is used on an automatic variable.
16751 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
16753 PR rtl-optimization/80474
16754 * reorg.c (update_block): Do not ignore instructions in a delay slot.
16756 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
16758 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
16761 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
16763 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
16764 (casesi): Emit bounds checking as RTL.
16765 (casesi_internal_mips16_<mode>): Remove bounds checking.
16767 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
16769 * config/xtensa/xtensa.c (xtensa_option_override): Append
16770 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
16771 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
16772 xtensa_doloop_hooks): Define unconditionally.
16773 (xtensa_reorg_loops): Only call reorg_loops in the presence of
16775 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
16776 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
16777 for it in xtensa_option_override.
16778 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
16779 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
16781 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
16783 * doc/cppopts.texi: Document '-' special value to -MF.
16785 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
16787 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
16788 (cortex_a53_fconst): Likewise.
16789 (cortex_a53_fpmul): Likewise.
16790 (cortex_a53_f_load_64): Likewise.
16791 (cortex_a53_f_load_many): Likewise.
16792 (cortex_a53_advsimd_alu): Likewise.
16793 (cortex_a53_advsimd_alu_q): Likewise.
16794 (cortex_a53_advsimd_mul): Likewise.
16795 (cortex_a53_advsimd_mul_q): Likewise.
16796 (fpmac bypass): Add new bypass for fpmac-fpmac case.
16797 Add missing fmul, r2f_cvt and fconst cases.
16799 2017-06-14 Richard Biener <rguenther@suse.de>
16801 PR middle-end/81088
16802 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
16804 (fold_binary_loc): When associating do not treat pre-existing
16805 TREE_OVERFLOW on literal constants as a reason to allow
16806 TREE_OVERFLOW on associated literal constants.
16808 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16810 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
16811 (MASK_FEATURES): New macro.
16812 * config/sparc/sparc.c (sparc_option_override): Remove the special
16813 handling of -mfpu and generalize it to all MASK_FEATURES switches.
16815 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16817 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
16818 a division of 0 if non-call exceptions are enabled.
16820 2017-06-14 Andrew Pinski <apinski@cavium.com>
16821 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16824 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
16825 Improve vector initialization code gen for only variable case.
16827 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16829 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
16831 2017-06-14 Richard Biener <rguenther@suse.de>
16833 PR tree-optimization/81083
16834 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
16837 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16839 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
16840 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
16841 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
16842 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
16843 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
16844 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
16846 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16848 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
16849 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
16851 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16853 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
16855 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16857 * config/rs6000/t-rtems: Don't handle SPE.
16859 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16861 * config/rs6000/t-linux: Don't handle SPE.
16863 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16865 * config/rs6000/eabispe.h: Delete file.
16867 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16869 * config/rs6000/t-spe: Delete file.
16871 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16873 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
16874 (rs6000_legitimate_offset_address_p): Return false for anything in
16875 V2SImode or V2SFmode.
16877 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16879 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
16880 except V2SF and V2SI. Rearrange the vector modes, and add comments.
16881 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
16883 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
16884 (rs6000_legitimate_offset_address_p): Ditto.
16885 (rs6000_emit_move): Ditto.
16886 (rs6000_init_builtins): Remove V4HI_type_node.
16888 2017-06-13 Martin Liska <mliska@suse.cz>
16891 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
16892 (gate_asan): Likewise.
16893 * asan.h (asan_no_sanitize_address_p): Remove the function.
16894 (sanitize_flags_p): New function.
16895 * builtins.def: Fix coding style.
16896 * common.opt: Use renamed enum value.
16897 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
16898 * doc/extend.texi: Document no_sanitize attribute.
16899 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
16900 to SANITIZE_UNDEFINED_NONDEFAULT.
16901 * gcc.c (sanitize_spec_function): Use the renamed enum value.
16902 * gimple-fold.c (optimize_atomic_compare_exchange_p):
16903 Use sanitize_flags_p.
16904 * gimplify.c (gimplify_function_tree): Likewise.
16905 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
16906 * opts.c (parse_no_sanitize_attribute): New function.
16907 (common_handle_option): Use renamed enum value.
16908 * opts.h (parse_no_sanitize_attribute): Declare.
16909 * tree.c (sanitize_flags_p): New function.
16910 * tree.h: Declared here.
16911 * tsan.c: Use sanitize_flags_p.
16912 * ubsan.c (ubsan_expand_null_ifn): Likewise.
16913 (instrument_mem_ref): Likewise.
16914 (instrument_bool_enum_load): Likewise.
16915 (do_ubsan_in_current_function): Remove the function.
16916 (pass_ubsan::execute): Use sanitize_flags_p.
16917 * ubsan.h: Remove do_ubsan_in_current_function
16918 * tree-cfg.c (print_no_sanitize_attr_value): New function.
16919 (dump_function_to_file): Use it here.
16921 2017-06-13 Martin Jambor <mjambor@suse.cz>
16923 PR tree-optimization/80803
16924 PR tree-optimization/81063
16925 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
16926 (propagate_subaccesses_across_link): Enqueue subtree whenever
16927 necessary instead of relying on the caller.
16929 2017-06-13 Martin Jambor <mjambor@suse.cz>
16931 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
16932 that have a first_link.
16933 (sort_and_splice_var_accesses): Do not check first_link before
16935 (subtree_mark_written_and_enqueue): Likewise.
16936 (propagate_all_subaccesses): Likewise and do not stop at first
16937 parent with a first_link.
16939 2017-06-13 Martin Jambor <mjambor@suse.cz>
16941 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
16944 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16946 * match.pd: New pattern.
16948 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16950 * tree-vrp.c (is_masked_range_test): New function.
16951 (register_edge_assert_for): Determine ranges for
16954 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16956 PR tree-optimization/67328
16957 * fold-const.c (maskable_range_p): New function.
16958 (build_range_check): Generate bittests if possible.
16960 2017-06-13 Martin Liska <mliska@suse.cz>
16962 * gimple-pretty-print.c (dump_probability): Add new argument.
16963 (dump_edge_probability): Dump both probability and count.
16964 (dump_gimple_label): Likewise.
16965 (dump_gimple_bb_header): Likewise.
16967 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
16970 * config/avr/avr-devices.c: Fix indentation.
16971 * config/avr/gen-avr-mmcu-specs.c: Dito.
16973 2017-06-13 Richard Biener <rguenther@suse.de>
16975 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
16976 instead get vector type from stmt_info.
16977 (vectorizable_reduction): Adjust. Remove dead code.
16979 2017-06-13 Richard Biener <rguenther@suse.de>
16981 PR middle-end/81065
16982 * fold-const.c (extract_muldiv_1): Remove bogus distribution
16983 case of C * (x * C2 + C3).
16984 (fold_addr_of_array_ref_difference): Properly fold index difference.
16986 2017-06-12 David S. Miller <davem@davemloft.net>
16989 * config/sparc/sparc.md (return expander): Emit frame blockage if
16990 function uses alloca.
16992 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16994 * combine.c (make_field_assignment): Check len rather than the mode
16995 precision when calling force_to_mode.
16997 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
16999 Support multilibs and devices that see flash in RAM address range.
17002 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17003 (avr_mcu_t) <flash_pm_offset>: New field.
17004 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17005 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17006 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17007 (AVR_TINY_PM_OFFSET): Remove macro.
17008 * config/avr/avr.opt (-mshort-calls): New option.
17009 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17010 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17011 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17012 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17013 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17014 instead of avr_arch->have_jmp_call.
17015 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17016 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17017 avr_arch->flash_pm_offset to define.
17018 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17019 new field flash_pm_offset. Add entry for avrxmega3.
17020 (avr_texinfo): Add entry for avrxmega3.
17021 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17022 attiny212, attiny214,
17023 attiny412, attiny414, attiny416, attiny417,
17024 attiny814, attiny816, attiny817,
17025 attiny1614, attiny1616, attiny1617,
17026 attiny3214, attiny3216, attiny3217.
17027 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17028 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17029 (avr_print_operand_address) [AVR_TINY]: Same.
17030 (avr_asm_init_sections) <readonly_data_section>: Only patch
17031 callback if avr_arch->flash_pm_offset = 0.
17032 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17033 for rodata if avr_arch->flash_pm_offset != 0.
17034 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17035 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17036 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17037 (m_options): Append opt_rcall.
17038 (m_dirnames): Append dir_rcall.
17039 * config/avr/t-multilib: Regenerate.
17041 * configure.ac [target=avr]: Check whether avrxmega3 default
17042 linker description file works as needed.
17043 * configure: Regenerate.
17044 * doc/avr-mmcu.texi: Regenerate.
17045 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17046 <__AVR_ARCH__>: Document avrxmega3 and 103.
17047 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17048 <__AVR_SHORT_CALLS__>: Document it.
17049 <__AVR_PM_BASE_ADDRESS__>: Document it.
17050 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17051 (AVR Variable Attributes) <progmem>: Document this is
17052 not needed for avrxmega3.
17053 (AVR Named Address Spaces) <__flash>: Dito.
17055 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17057 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17059 2017-06-12 Doug Rupp <rupp@adacore.com>
17061 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17062 Append vxworks-stdint.h to the tm_file list.
17063 * config/vxworks-stdint.h: New file.
17065 2017-06-12 Martin Liska <mliska@suse.cz>
17067 PR tree-optimization/81041
17068 * tree-profile.c (gimple_gen_ic_func_profiler):
17069 Create an extra BB in profile-generate
17070 (gimple_gen_time_profiler): Likewise.
17072 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17074 PR tree-optimization/81003
17075 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17076 (update_range_test): Use it instead of force_gimple_operand_gsi.
17078 2017-06-12 Richard Biener <rguenther@suse.de>
17080 PR tree-optimization/81053
17081 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
17082 with backedge value not defined in loop. Simplify def stmt
17085 2017-06-11 Tom de Vries <tom@codesourcery.com>
17088 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
17090 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
17091 nvptx_cannot_force_const_mem.
17093 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17095 * opts.c (finish_options): Move test for flag_split_stack after
17096 it has been initialized.
17098 2017-06-11 Jason Merrill <jason@redhat.com>
17100 * tree.h (id_equal): New.
17101 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
17102 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
17103 instead of strcmp of IDENTIFIER_POINTER.
17105 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17107 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
17108 (mark_all_inlined_calls_cdtor): Fix formating.
17109 (inline_transform): Rescale profile before inlining.
17111 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17113 * cgraph.h (cgraph_edge::clone): Update prototype.
17114 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
17115 (cgraph_node::create_clone): Update.
17116 (cgraph_node::create_version_clone): Update.
17117 * tree-inline.c (copy_bb): Update.
17118 (expand_call_inline): Update.
17120 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17122 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
17123 factored out from ...
17124 (rs6000_emit_prologue): ... here.
17126 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17128 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
17129 factored out from ...
17130 (rs6000_emit_prologue): ... here.
17132 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17134 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
17136 (handle_missing_profiles): Fix computation of tp_first_run.
17137 (counts_to_freqs): Do not touch freqs when count is 0.
17139 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17141 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
17144 2017-06-10 Tom de Vries <tom@codesourcery.com>
17146 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17147 attributes): Document signal effective target.
17149 2017-06-10 Tom de Vries <tom@codesourcery.com>
17151 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17152 Document effective target stack_size.
17154 2017-06-09 David Malcolm <dmalcolm@redhat.com>
17156 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
17157 to the edit_context if they can be auto-applied.
17159 2017-06-9 Ian Lance Taylor <iant@golang.org>
17161 * opts.c (finish_options): If -fsplit-stack, disable implicit
17162 -forder-blocks-and-partition.
17163 * doc/invoke.texi (Optimize Options): Document that when using
17164 -fsplit-stack -forder-blocks-and-partition is not implicitly
17167 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17169 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
17170 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
17171 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
17172 * builtins.def (abort, trap, unreachable): Declare cold.
17173 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
17174 * tree-core.h (ECF_COLD): New.
17175 * tree.c (set_call_expr_flags): Handle ECF_COLD.
17176 (build_common_builtin_nodes): Mark unreachable and abort as cold.
17178 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17180 * predict.c (unlikely_executed_stmt_p): Cleanup.
17182 2017-06-09 Richard Biener <rguenther@suse.de>
17184 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
17185 model if the ref is always written to.
17187 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17189 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
17191 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17193 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
17196 2017-06-09 Tom de Vries <tom@codesourcery.com>
17199 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
17200 "target cannot support label values" when encountering LABEL_REF.
17202 2017-06-09 Martin Liska <mliska@suse.cz>
17204 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
17205 (gimple_gen_ic_func_profiler): Emit direct comparison
17206 of __gcov_indirect_call_callee with NULL.
17207 (gimple_gen_time_profiler): Change probability from
17208 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
17210 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17212 * profile.c (edge_gcov_counts): Turn to pointer.
17213 (compute_branch_probabilities, compute_branch_probabilities): Update.
17214 (branch_prob): Do not clear edge_gcov_count.
17215 * profile.h (edge_gcov_counts): Turn to pointer.
17216 (edge_gcov_count): Update.
17218 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17220 * gimple.h (gimple_check_failed): Mark cold.
17222 2017-06-09 Richard Biener <rguenther@suse.de>
17224 PR tree-optimization/66623
17225 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
17226 refactor check_reduction into two parts, properly computing
17227 whether we have to check reduction validity for outer loop
17230 2017-06-09 Richard Biener <rguenther@suse.de>
17232 PR tree-optimization/79483
17233 * graphite-scop-detection.c (order): New global.
17234 (get_order): Compute bb to order mapping that satisfies code
17235 generation constraints.
17236 (cmp_pbbs): New helper.
17237 (build_scops): Start domwalk at entry block, sort generated
17240 2017-06-09 Richard Biener <rguenther@suse.de>
17242 PR middle-end/81007
17243 * ipa-polymorphic-call.c
17244 (ipa_polymorphic_call_context::restrict_to_inner_class):
17245 Skip FIELD_DECLs with error_mark_node type.
17246 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
17249 2017-06-09 Martin Liska <mliska@suse.cz>
17251 * predict.c (struct branch_predictor): New struct.
17252 (test_prediction_value_range): New test.
17253 (predict_c_tests): New function.
17254 * selftest-run-tests.c (selftest::run_tests): Run the function.
17255 * selftest.h: Declare new tests.
17257 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
17260 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
17261 gen_add3_insn did not fail.
17262 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
17263 r0, construct that number in a temporary reg and add that reg to r0.
17264 If asked to put the result in r0 as well, fail.
17266 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
17268 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17269 for early expansion of vec_eqv.
17271 2017-06-08 Jakub Jelinek <jakub@redhat.com>
17273 PR middle-end/81005
17274 * ubsan.c (instrument_null): Avoid pointless code temporary.
17275 (pass_ubsan::execute): Instrument aggregate arguments of calls.
17277 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
17281 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
17284 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
17287 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17289 * predict.c (unlikely_executed_edge_p): Move ahead.
17290 (probably_never_executed_edge_p): Use it.
17292 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17294 PR middle-end/79988
17295 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
17296 gimple_call_builtin_p call.
17298 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17300 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
17301 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
17302 rtl_check_failed_type2, rtl_check_failed_code1,
17303 rtl_check_failed_code2, rtl_check_failed_code_mode,
17304 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
17305 rtvec_check_failed_bounds, rtl_check_failed_flag,
17306 _fatal_insn_not_found, _fatal_insn): Likewise.
17307 * tree.h (tree_contains_struct_check_failed,
17308 tree_check_failed, tree_not_check_failed,
17309 tree_class_check_failed, tree_range_check_failed,
17310 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
17311 tree_vec_elt_check_failed, phi_node_elt_check_failed,
17312 tree_operand_check_failed, omp_clause_check_failed,
17313 omp_clause_operand_check_failed, omp_clause_range_check_failed):
17316 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17318 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
17319 flag_branch_probabilities.
17320 * ipa-inline.c (edge_badness): Likewise.
17321 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17322 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
17323 * predict.c (maybe_hot_frequency_p): Likewise.
17324 (probably_never_executed): Likewise.
17325 * sched-ebb.c (schedule_ebbs): Likewise.
17326 * sched-rgn.c (find_single_block_region): Likewise.
17327 * tracer.c (tail_duplicate): Likewise.
17329 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17331 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
17332 longer requires x_flag_profile_use.
17334 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17336 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
17337 instead of flag_reorder_blocks_and_partition.
17338 * dbxout.c (dbxout_function_end): Likewise.
17339 * dwarf2out.c (gen_subprogram_die): Likewise.
17340 * haifa-sched.c (sched_create_recovery_edges): Likewise.
17341 * hw-doloop.c (reorg_loops): Likewise.
17342 * varasm.c (assemble_start_function,
17343 assemble_end_function): Likewise.
17344 (decide_function_section): Do not check for
17345 flag_reorder_blocks_and_partition.
17347 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17349 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17351 (chkp_get_hard_register_fake_addr_expr): Ditto.
17352 (chkp_build_addr_expr): Add check for hard reg case.
17353 (chkp_parse_array_and_component_ref): Ditto.
17354 (chkp_find_bounds_1): Ditto.
17355 (chkp_process_stmt): Don't generate bounds store for
17358 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17360 * predict.c (maybe_hot_bb_p): Do not check profile status.
17361 (maybe_hot_edge_p): Likewise.
17362 (probably_never_executed): Check for zero counts even if profile
17364 (unlikely_executed_edge_p): New function.
17365 (unlikely_executed_stmt_p): New function.
17366 (unlikely_executed_bb_p): New function.
17367 (set_even_probabilities): Use unlikely predicates.
17368 (combine_predictions_for_bb): Likewise.
17369 (predict_paths_for_bb): Likewise.
17370 (predict_paths_leading_to_edge): Likewise.
17371 (determine_unlikely_bbs): New function.
17372 (estimate_bb_frequencies): Use it.
17373 (compute_function_frequency): Use zero counts even if profile is
17375 * profile-count.h: Fix typo.
17377 2017-08-08 Julia Koval <julia.koval@intel.com>
17379 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
17380 _mm512_mask_cvtsepi16_storeu_epi8,
17381 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
17382 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
17383 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
17384 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
17385 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
17386 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
17387 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
17388 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
17389 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
17390 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
17391 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
17392 __builtin_ia32_pmovuswb256mem_mask,
17393 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
17394 __builtin_ia32_pmovwb512mem_mask): New builtins.
17396 2017-08-08 Julia Koval <julia.koval@intel.com>
17398 PR target/73350,80862
17399 * config/i386/subst.md (round): Fix round pattern.
17400 * config/i386/i386.c (ix86_erase_embedded_rounding):
17401 Fix erasing rounding for the fixed pattern.
17403 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17405 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
17407 2017-06-08 Martin Liska <mliska@suse.cz>
17409 PR gcov-profile/80911
17410 * gcov.c (block_info::block_info): New constructor.
17412 2017-06-07 Carl Love <cel@us.ibm.com>
17414 * config/rs6000/rs6000-c: The return type of the following
17415 built-in functions was implemented as int not long long. Fix sign
17416 of return value for the unsigned version of vec_mulo and vec_mule.
17417 vector unsigned long long vec_bperm (vector unsigned long long,
17418 vector unsigned char)
17419 vector signed long long vec_mule (vector signed int,
17421 vector unsigned long long vec_mule (vector unsigned int,
17422 vector unsigned int)
17423 vector signed long long vec_mulo (vector signed int,
17425 vector unsigned long long vec_mulo (vector unsigned int,
17426 vector unsigned int)
17427 * doc/extend.texi: Fix the documentation for the built-in
17430 2017-06-07 Carl Love <cel@us.ibm.com>
17433 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
17436 2017-06-07 Carl Love <cel@us.ibm.com>
17438 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
17439 support, Generate doublehv for signed int/float for BE case only.
17441 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
17443 * doc/invoke.texi (mcx16): Rewrite.
17445 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17447 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
17448 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
17449 *mov<mode>_softfloat, and an anonymous splitter): Use
17450 nonimmediate_operand instead of rs6000_nonimmediate_operand.
17452 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17454 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
17456 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17457 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
17458 (rs6000_debug_reg_global): Adjust.
17459 (rs6000_init_hard_regno_mode_ok): Adjust.
17460 (rs6000_dbx_register_number): Adjust.
17461 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
17462 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17463 Remove SPE_ACC and SPEFSCR.
17464 (REG_ALLOC_ORDER): Ditto.
17465 (FRAME_POINTER_REGNUM): Change to 111.
17466 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
17467 (REG_CLASS_NAMES): Ditto.
17468 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
17469 (REGISTER_NAMES): Ditto.
17470 (ADDITIONAL_REG_NAMES): Ditto.
17471 (rs6000_reg_names): Ditto.
17472 * config/rs6000/rs6000.md: Renumber some register number
17475 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17477 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
17479 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17480 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
17482 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
17483 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17484 Delete the SPE high registers.
17485 (REG_ALLOC_ORDER): Ditto.
17486 (enum reg_class): Remove SPE_HIGH_REGS.
17487 (REG_CLASS_NAMES): Ditto.
17488 (REG_CLASS_CONTENTS): Delete the SPE high registers.
17489 (REGISTER_NAMES): Ditto.
17490 (rs6000_reg_names): Ditto.
17491 * doc/tm.texi.in: Remove SPE as example.
17492 * doc/tm.texi: Regenerate.
17494 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17496 * config/rs6000/8540.md (ppc8540_brinc): Delete.
17497 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
17498 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
17499 * config/rs6000/rs6000.md (type): Remove "brinc".
17501 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17503 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
17504 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
17505 * config/rs6000/linuxspe.h: Delete file.
17506 * config/rs6000/rs6000.md: Don't include spe.md.
17507 * config/rs6000/spe.h: Delete file.
17508 * config/rs6000/spe.md: Delete file.
17509 * config/rs6000/t-rs6000: Remove spe.md.
17511 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17513 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
17514 (reg_or_none500mem_operand): Delete.
17515 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
17516 instead of reg_or_none500mem_operand.
17518 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17520 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17521 handling of SPE flags.
17522 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
17524 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17526 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17528 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
17529 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
17530 paired_divv2sf3): Similar.
17531 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
17532 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
17533 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
17535 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
17536 Rename the paired_* instruction patterns.
17537 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
17539 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
17540 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
17541 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
17542 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
17543 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
17544 PAIRED_VECTOR_MODE.
17545 (struct machine_function): Delete field spe_insn_chain_scanned_p.
17546 (spe_func_has_64bit_regs_p): Delete.
17547 (spe_expand_predicate_builtin): Delete.
17548 (spe_expand_evsel_builtin): Delete.
17549 (TARGET_DWARF_REGISTER_SPAN): Do not define.
17550 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
17551 (invalid_e500_subreg): Delete.
17552 (rs6000_legitimize_address): Always force_reg op2 as well, for
17553 paired single memory accesses.
17554 (rs6000_member_type_forces_blk): Delete.
17555 (rs6000_spe_function_arg): Delete.
17556 (rs6000_expand_unop_builtin): Delete SPE handling.
17557 (rs6000_expand_binop_builtin): Ditto.
17558 (spe_expand_stv_builtin): Delete.
17559 (bdesc_2arg_spe): Delete.
17560 (spe_expand_builtin): Delete.
17561 (spe_expand_predicate_builtin): Delete.
17562 (spe_expand_evsel_builtin): Delete.
17563 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
17564 (spe_init_builtins): Delete.
17565 (spe_func_has_64bit_regs_p): Delete.
17566 (savres_routine_name): Delete "info" parameter. Adjust callers.
17567 (rs6000_emit_stack_reset): Ditto.
17568 (rs6000_dwarf_register_span): Delete.
17569 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
17570 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
17571 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
17573 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
17575 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
17576 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
17577 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
17578 mulv2sf3, divv2sf3): Delete expanders.
17580 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17582 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
17584 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17586 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
17587 * config/rs6000/rs6000.c: Ditto.
17589 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17591 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
17592 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
17593 comparison_operator.
17595 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17597 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
17598 * config/rs6000/rs6000.opt: Ditto.
17599 * config/rs6000/t-rtems: Ditto.
17601 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17603 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
17604 TARGET_E500_SINGLE by 0, simplify.
17605 * config/rs6000/rs6000.c: Ditto.
17606 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
17607 (spe_build_register_parallel): Delete.
17608 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
17609 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
17610 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
17611 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
17612 (E500_CONVERT): Delete.
17613 * config/rs6000/spe.md: Remove many patterns and all define_constants.
17615 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17617 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
17618 * config/rs6000/dfp.md: Ditto.
17619 (negdd2, *negdd2_fpr): Merge.
17620 (absdd2, *absdd2_fpr): Merge.
17621 (negtd2, *negtd2_fpr): Merge.
17622 (abstd2, *abstd2_fpr): Merge.
17623 * config/rs6000/e500.h: Delete file.
17624 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
17625 TARGET_FPRS by 1 and simplify.
17626 * config/rs6000/rs6000-c.c: Ditto.
17627 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
17628 TARGET_DF_SPE by 0.
17629 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
17631 * config/rs6000/rs6000.md: Ditto.
17632 (floatdidf2, *floatdidf2_fpr): Merge.
17633 (move_from_CR_gt_bit): Delete.
17634 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
17635 (E500_CR_IOR_COMPARE): Delete.
17636 (All patterns that require !TARGET_FPRS): Delete.
17637 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
17639 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17641 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
17643 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17645 * graphds.c (add_edge): Intitialize edge's attached data.
17646 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
17647 pointer parameter. Call pointed function on each edge during
17648 graph traversing. Skip traversing the edge when the function
17650 (graphds_dfs, graphds_scc): Ditto.
17651 (for_each_edge): New parameter. Pass the new parameter to callback
17653 * graphds.h (skip_edge_callback): New function pointer type.
17654 (graphds_dfs, graphds_scc): New function pointer parameter.
17655 (graphds_edge_callback, for_each_edge): New parameter.
17657 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17659 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
17660 out code checking if runtime alias check is possible to below ...
17661 Call the new function.
17662 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
17663 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
17665 2017-06-07 Marek Polacek <polacek@redhat.com>
17668 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
17669 TYPE_OVERFLOW_WRAPS check.
17671 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17673 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
17674 if versioning is required.
17675 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
17676 peeling with the check for versioning.
17678 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17680 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
17681 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
17682 Set true to new parameter if new ssa variable is defined.
17683 (vect_gen_vector_loop_niters): Refactor. Set range information
17684 for the new vector loop bound variable.
17685 (vect_do_peeling): Ditto.
17687 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17689 * tree-affine.c (ssa.h): Include header file.
17690 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
17691 has wrapping overflow behavior.
17693 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17695 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
17697 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17699 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
17700 (tree_to_aff_combination): ... here.
17702 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17704 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
17705 reg_pressure model function.
17706 (ivopts_global_cost_for_size): Delete.
17707 (determine_set_costs, iv_ca_recount_cost): Call new model function
17708 ivopts_estimate_reg_pressure.
17710 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17712 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
17713 expensive than udiv. Remove floating point cases from mod.
17715 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17717 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
17718 Increase idiv cost.
17720 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17722 * config/aarch64/aarch64.md
17723 (copysignsf3): Fix mask generation.
17725 2017-06-07 Jakub Jelinek <jakub@redhat.com>
17727 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
17729 (class dump_manager): Add register_dumps method.
17730 * dumpfile.c: Include langhooks.h.
17731 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
17732 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
17733 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
17734 (dump_manager::dump_register): Start with 512 entries instead of 32.
17735 (dump_manager::register_dumps): New method.
17736 * toplev.c (general_init): Instead of invoking register_dumps
17737 langhook, invoke register_dumps method on the dump manager.
17738 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
17741 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
17743 * doc/md.texi: Clarify the restrictions on a define_insn condition.
17744 Say that # requires an associated define_split to exist, and that
17745 the define_split must be suitable for use after register allocation.
17747 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17749 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
17750 (compute_outgoing_frequencies): Also initialize zero counts.
17751 (find_many_sub_basic_blocks): Do not produce uninitialized profile
17752 around loops; preserve more of profile when nothing changes.
17754 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
17756 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
17758 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
17759 * config/arm/arm-cpu-cdata.h: Regenerate.
17760 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
17761 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
17762 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
17763 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
17764 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
17766 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
17768 * config/arm/t-rmprofile: Likewise.
17769 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
17771 2017-06-06 David S. Miller <davem@davemloft.net>
17774 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
17775 blockage if function uses alloca.
17777 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17779 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
17780 New "uid" fields to hold pretty-print IDs of group and ref.
17781 Memory references are now identified as <group_id>:<ref_id>
17782 instead of using [random] addresses.
17783 (dump_mem_details): Simplify, no functional change.
17784 (dump_mem_ref): Simplify and make output more concise.
17785 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
17786 (find_or_create_group): Initialize group uid.
17787 (record_ref): Initialize ref uid. Improve debug output.
17788 (prune_group_by_reuse, should_issue_prefetch_p,)
17789 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
17790 (mark_nontemporal_store, determine_loop_nest_reuse):
17791 Improve debug output.
17793 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17795 * dbgcnt.def (prefetch): New debug counter.
17796 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
17797 (schedule_prefetches): Stop issueing prefetches if debug counter
17800 2017-06-06 Tom de Vries <tom@codesourcery.com>
17802 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
17803 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
17805 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17807 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
17808 Use aarch64_reg_or_zero predicate for operand 4.
17809 (aarch64_compare_and_swap<mode> define_insn_and_split):
17810 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
17811 (aarch64_store_exclusive<mode>): Likewise for operand 2.
17813 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17815 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
17816 (arm_compute_save_core_reg_mask): This.
17817 (thumb1_compute_save_reg_mask): Rename into ...
17818 (thumb1_compute_save_core_reg_mask): This.
17819 (arm_compute_save_reg0_reg12_mask): Adapt comment.
17820 (arm_compute_frame_layout): Likewise.
17822 2017-06-06 Richard Biener <rguenther@suse.de>
17824 PR tree-optimization/80974
17825 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
17826 keep or clear leaders SSA info.
17828 2017-06-06 Tom de Vries <tom@codesourcery.com>
17830 * config/nvptx/nvptx.c (split_mode_p): New function.
17831 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
17833 2017-06-06 Tom de Vries <tom@codesourcery.com>
17835 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
17837 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17840 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
17843 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17845 * shrink-wrap.c (handle_simple_exit): Update profile.
17846 (try_shrink_wrapping): Upate profile.
17848 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17850 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
17851 (tree_guess_outgoing_edge_probabilities): New.
17852 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
17853 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
17855 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17857 * ipa-split.c (split_function): Initialize return bb profile.
17859 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17861 * profile.c (compute_branch_probabilities): Also initialize
17862 EXIT_BLOCK profile.
17864 2017-06-06 Richard Biener <rguenther@suse.de>
17866 PR tree-optimization/80928
17867 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
17868 (vect_analyze_loop_operations): Properly guard analysis for
17870 (vect_transform_loop): Likewise.
17871 (vect_analyze_loop_2): Also reset SLP type on PHIs.
17872 (vect_model_induction_cost): Do not cost for pure SLP.
17873 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
17874 of induction in inner loop vectorization.
17875 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
17876 (vect_get_and_check_slp_defs): Handle vect_induction_def.
17877 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
17879 (vect_analyze_slp_cost_1): Cost induction.
17880 (vect_detect_hybrid_slp_stmts): Handle PHIs.
17881 (vect_get_slp_vect_defs): Likewise.
17882 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
17883 (vect_transform_stmt): Handle SLP reductions.
17884 * tree-vectorizer.h (vectorizable_induction): Adjust.
17886 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17888 * config/rs6000/rs6000.c (make_resolver_func): Update
17889 init_lowered_empty_function call.
17891 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
17893 * doc/invoke.texi: Document the -fprofile-abs-path option.
17894 * common.opt (fprofile-abs-path): New option.
17895 * gcov-io.h (gcov_write_filename): Declare.
17896 * gcov-io.c (gcov_write_filename): New function.
17897 * coverage.c (coverage_begin_function): Use gcov_write_filename.
17898 * profile.c (output_location): Likewise.
17900 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17902 * shring-wrap.c: Revert accidental commit.
17904 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
17906 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
17908 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17910 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
17912 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
17913 profile in callgraph edge.
17914 * profile-count.h (apply_probability): If THIS is 0, then result is 0
17915 (apply_scale): Likewise.
17916 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
17917 Also scale profile when inlining function with zero profile.
17918 (initialize_cfun): Update exit block profile even when it is zero.
17919 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
17920 when profile is read.
17922 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17924 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
17925 (CLONE_*): New constants to define the processors we can generate
17926 code for with the target_clone attribute.
17927 (rs6000_clone_map): New array to identify which clone processors
17928 the current program is running on.
17929 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
17930 target_clone attribute.
17931 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
17932 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
17933 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
17934 (cpu_expand_builtin): Add support for target_clone attribute.
17935 (rs6000_valid_attribute_p): Allow "default" attribute.
17936 (get_decl_name): New debug function to simplify printing the
17937 current function name in debugging statements.
17938 (rs6000_clone_priority): New functions to support the target_clone
17939 attribute, and be able to generate code to switch between ISA 2.05
17940 through ISA 3.0 (power6 through power9).
17941 (rs6000_compare_version_priority): Likewise.
17942 (rs6000_get_function_versions_dispatcher): Likewise.
17943 (make_resolver_func): Likewise.
17944 (add_condition_to_bb): Likewise.
17945 (dispatch_function_versions): Likewise.
17946 (rs6000_generate_version_dispatcher_body): Likewise.
17947 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
17948 (fusion_gpr_load_p): Fix a spacing issue.
17949 * doc/extend.texi (Common Function Attributes): Document that the
17950 PowerPC supports the target_clone attribute.
17952 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17954 * config/arm/arm.h: explain F symbol found in description of ARM
17955 register allocation in its legend.
17957 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17959 * config/mips/frame-header-opt.c: Include profile-count.h.
17960 * config/riscv/riscv.c: Include profile-count.h
17962 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17964 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
17966 (sm_set_flag_if_changed): Add bbs field.
17967 (execute_sm_if_changed_flag_set): Pass BBS.
17968 (execute_sm): Update.
17970 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17972 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
17975 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17977 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
17978 (peephole2): New peephole2 to emit the above.
17979 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
17981 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17983 * config/aarch64/aarch64.c (define_peephole2 above
17984 *sub_<shift>_<mode>): New peephole.
17986 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17988 * config/i386/i386.c (make_resolver_func): Update.
17989 * Makefile.in: Add profile-count.h and profile-count.o
17990 * auto-profile.c (afdo_indirect_call): Update to new API.
17991 (afdo_set_bb_count): Update.
17992 (afdo_propagate_edge): Update.
17993 (afdo_propagate_circuit): Update.
17994 (afdo_calculate_branch_prob): Update.
17995 (afdo_annotate_cfg): Update.
17996 * basic-block.h: Include profile-count.h
17997 (struct edge_def): Turn count to profile_count.
17998 (struct basic_block_def): Likewie.
17999 (REG_BR_PROB_BASE): Move to profile-count.h
18000 (RDIV): Move to profile-count.h
18001 * bb-reorder.c (max_entry_count): Turn to profile_count.
18002 (find_traces): Update.
18003 (rotate_loop):Update.
18004 (connect_traces):Update.
18005 (sanitize_hot_paths):Update.
18006 * bt-load.c (migrate_btr_defs): Update.
18007 * cfg.c (RDIV): Remove.
18008 (init_flow): Use alloc_block.
18009 (alloc_block): Uninitialize count.
18010 (unchecked_make_edge): Uninitialize count.
18011 (check_bb_profile): Update.
18012 (dump_edge_info): Update.
18013 (dump_bb_info): Update.
18014 (update_bb_profile_for_threading): Update.
18015 (scale_bbs_frequencies_int): Update.
18016 (scale_bbs_frequencies_gcov_type): Update.
18017 (scale_bbs_frequencies_profile_count): New.
18018 * cfg.h (update_bb_profile_for_threading): Update.
18019 (scale_bbs_frequencies_profile_count): Declare.
18020 * cfgbuild.c (compute_outgoing_frequencies): Update.
18021 (find_many_sub_basic_blocks): Update.
18022 * cfgcleanup.c (try_forward_edges): Update.
18023 (try_crossjump_to_edge): Update.
18024 * cfgexpand.c (expand_gimple_tailcall): Update.
18025 (construct_exit_block): Update.
18026 * cfghooks.c (verify_flow_info): Update.
18027 (dump_bb_for_graph): Update.
18028 (split_edge): Update.
18029 (make_forwarder_block): Update.
18030 (duplicate_block): Update.
18031 (account_profile_record): Update.
18032 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18033 (get_estimated_loop_iterations): Update.
18034 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18035 (single_likely_exit): Update.
18036 * cfgloopmanip.c (scale_loop_profile): Update.
18038 (set_zero_probability): Update.
18039 (lv_adjust_loop_entry_edge): Update.
18040 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18041 (purge_dead_edges): Update.
18042 (rtl_account_profile_record): Update.
18043 * cgraph.c (cgraph_node::create): Uninitialize count.
18044 (symbol_table::create_edge): Uninitialize count.
18045 (cgraph_update_edges_for_call_stmt_node): Update.
18046 (cgraph_edge::dump_edge_flags): Update.
18047 (cgraph_node::dump): Update.
18048 (cgraph_edge::maybe_hot_p): Update.
18049 * cgraph.h: Include profile-count.h
18050 (create_clone), create_edge, create_indirect_edge): Update.
18051 (cgraph_node): Turn count to profile_count.
18052 (cgraph_edge0: Likewise.
18053 (make_speculative, clone): Update.
18054 (create_edge): Update.
18055 (init_lowered_empty_function): Update.
18056 * cgraphclones.c (cgraph_edge::clone): Update.
18057 (duplicate_thunk_for_node): Update.
18058 (cgraph_node::create_clone): Update.
18059 * cgraphunit.c (cgraph_node::analyze): Update.
18060 (cgraph_node::expand_thunk): Update.
18061 * final.c (dump_basic_block_info): Update.
18062 * gimple-streamer-in.c (input_bb): Update.
18063 * gimple-streamer-out.c (output_bb): Update.
18064 * graphite.c (print_global_statistics): Update.
18065 (print_graphite_scop_statistics): Update.
18066 * hsa-brig.c: Include basic-block.h.
18067 * hsa-dump.c: Include basic-block.h.
18068 * hsa-gen.c (T sum_slice): Update.
18069 (convert_switch_statements):Update.
18070 * hsa-regalloc.c: Include basic-block.h.
18071 * ipa-chkp.c (chkp_produce_thunks): Update.
18072 * ipa-cp.c (struct caller_statistics): Update.
18073 (init_caller_stats): Update.
18074 (gather_caller_stats): Update.
18075 (ipcp_cloning_candidate_p): Update.
18076 (good_cloning_opportunity_p): Update.
18077 (get_info_about_necessary_edges): Update.
18078 (dump_profile_updates): Update.
18079 (update_profiling_info): Update.
18080 (update_specialized_profile): Update.
18081 (perhaps_add_new_callers): Update.
18082 (decide_about_value): Update.
18083 (ipa_cp_c_finalize): Update.
18084 * ipa-devirt.c (struct odr_type_warn_count): Update.
18085 (struct decl_warn_count): Update.
18086 (struct final_warning_record): Update.
18087 (possible_polymorphic_call_targets): Update.
18088 (ipa_devirt): Update.
18089 * ipa-fnsummary.c (redirect_to_unreachable): Update.
18090 * ipa-icf.c (sem_function::merge): Update.
18091 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
18092 * ipa-inline.c (compute_uninlined_call_time): Update.
18093 (compute_inlined_call_time): Update.
18094 (want_inline_small_function_p): Update.
18095 (want_inline_self_recursive_call_p): Update.
18096 (edge_badness): Update.
18097 (lookup_recursive_calls): Update.
18098 (recursive_inlining): Update.
18099 (inline_small_functions): Update.
18100 (dump_overall_stats): Update.
18101 (dump_inline_stats): Update.
18102 * ipa-profile.c (ipa_profile_generate_summary): Update.
18103 (ipa_propagate_frequency): Update.
18104 (ipa_profile): Update.
18105 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18106 * ipa-utils.c (ipa_merge_profiles): Update.
18107 * loop-doloop.c (doloop_modify): Update.
18108 * loop-unroll.c (report_unroll): Update.
18109 (unroll_loop_runtime_iterations): Update.
18110 * lto-cgraph.c (lto_output_edge): Update.
18111 (lto_output_node): Update.
18112 (input_node): Update.
18113 (input_edge): Update.
18114 (merge_profile_summaries): Update.
18115 * lto-streamer-in.c (input_cfg): Update.
18116 * lto-streamer-out.c (output_cfg): Update.
18117 * mcf.c (create_fixup_graph): Update.
18118 (adjust_cfg_counts): Update.
18119 (sum_edge_counts): Update.
18120 * modulo-sched.c (sms_schedule): Update.
18121 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
18122 * predict.c (maybe_hot_count_p): Update.
18123 (probably_never_executed): Update.
18124 (dump_prediction): Update.
18125 (combine_predictions_for_bb): Update.
18126 (propagate_freq): Update.
18127 (handle_missing_profiles): Update.
18128 (counts_to_freqs): Update.
18129 (rebuild_frequencies): Update.
18130 (force_edge_cold): Update.
18131 * predict.h: Include profile-count.h
18132 (maybe_hot_count_p, counts_to_freqs): UPdate.
18133 * print-rtl-function.c: Do not include cfg.h
18134 * print-rtl.c: Include basic-block.h
18135 * profile-count.c: New file.
18136 * profile-count.h: New file.
18137 * profile.c (is_edge_inconsistent): Update.
18138 (correct_negative_edge_counts): Update.
18139 (is_inconsistent): Update.
18140 (set_bb_counts): Update.
18141 (read_profile_edge_counts): Update.
18142 (compute_frequency_overlap): Update.
18143 (compute_branch_probabilities): Update; Initialize and deinitialize
18145 (branch_prob): Update.
18146 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
18147 (edge_gcov_count): New.
18148 (bb_gcov_count): New.
18149 * shrink-wrap.c (try_shrink_wrapping): Update.
18150 * tracer.c (better_p): Update.
18151 * trans-mem.c (expand_transaction): Update.
18152 (ipa_tm_insert_irr_call): Update.
18153 (ipa_tm_insert_gettmclone_call): Update.
18154 * tree-call-cdce.c: Update.
18155 * tree-cfg.c (gimple_duplicate_sese_region): Update.
18156 (gimple_duplicate_sese_tail): Update.
18157 (gimple_account_profile_record): Update.
18158 (execute_fixup_cfg): Update.
18159 * tree-inline.c (copy_bb): Update.
18160 (copy_edges_for_bb): Update.
18161 (initialize_cfun): Update.
18162 (freqs_to_counts): Update.
18163 (copy_cfg_body): Update.
18164 (expand_call_inline): Update.
18165 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18166 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
18167 (try_unroll_loop_completely): Update.
18168 (try_peel_loop): Update.
18169 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18170 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
18171 * tree-ssa-loop-split.c (connect_loops): Update.
18172 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
18173 * tree-ssa-reassoc.c (branch_fixup): Update.
18174 * tree-ssa-tail-merge.c (replace_block_by): Update.
18175 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
18176 (compute_path_counts): Update.
18177 (update_profile): Update.
18178 (recompute_probabilities): Update.
18179 (update_joiner_offpath_counts): Update.
18180 (estimated_freqs_path): Update.
18181 (freqs_to_counts_path): Update.
18182 (clear_counts_path): Update.
18183 (ssa_fix_duplicate_block_edges): Update.
18184 (duplicate_thread_path): Update.
18185 * tree-switch-conversion.c (case_bit_test_cmp): Update.
18186 (struct switch_conv_info): Update.
18187 * tree-tailcall.c (decrease_profile): Update.
18188 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18189 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18190 * value-prof.c (check_counter): Update.
18191 (gimple_divmod_fixed_value): Update.
18192 (gimple_mod_pow2): Update.
18193 (gimple_mod_subtract): Update.
18194 (gimple_ic_transform): Update.
18195 (gimple_stringop_fixed_value): Update.
18196 * value-prof.h (gimple_ic): Update.
18198 2017-06-02 Carl Love <cel@us.ibm.com>
18200 * config/rs6000/rs6000-c: Add support for built-in functions
18201 vector double vec_doublee (vector signed int);
18202 vector double vec_doublee (vector unsigned int);
18203 vector double vec_doublee (vector float);
18204 vector double vec_doubleh (vector signed int);
18205 vector double vec_doubleh (vector unsigned int);
18206 vector double vec_doubleh (vector float);
18207 vector double vec_doublel (vector signed int);
18208 vector double vec_doublel (vector unsigned int);
18209 vector double vec_doublel (vector float);
18210 vector double vec_doubleo (vector signed int);
18211 vector double vec_doubleo (vector unsigned int);
18212 vector double vec_doubleo (vector float);.
18213 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
18214 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
18216 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
18217 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
18218 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
18220 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
18221 vec_doublel, vec_doubleh.
18222 * doc/extend.texi: Update the built-in documentation file for the
18223 new built-in functions.
18225 2017-06-02 David Malcolm <dmalcolm@redhat.com>
18228 * ipa-inline-analysis.c (free_growth_caches): Set
18229 edge_removal_hook_holder to NULL after removing it.
18231 2017-06-02 Sudakshina Das <sudi.das@arm.com>
18233 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
18234 comparision with zero.
18236 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18237 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18238 for early expansion of vec_min and vec_max builtins.
18239 (builtin_function_type): Add min/max unsigned variants to those
18240 identified as having unsigned arguments.
18242 2017-06-02 Olivier Hainque <hainque@adacore.com>
18244 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
18246 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18248 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
18249 Use VALL_F16 iterator rather than VALL.
18251 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18253 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
18254 Emit CBNZ inside loop when doing a strong exchange and comparing
18255 against zero. Generate the CC flags after the loop.
18257 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
18259 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
18260 (dl_section_ref): New.
18261 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
18262 On AIX, append an expression to subtract the size of the
18263 section length to dl_section_ref.
18265 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18267 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18268 for early expansion of vector absolute builtins.
18270 2017-06-02 Richard Biener <rguenther@suse.de>
18272 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
18273 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
18275 2017-06-02 Richard Biener <rguenther@suse.de>
18277 PR tree-optimization/80948
18278 * tree-tailcall.c (find_tail_calls): Track stmts to move in
18279 stmt order as well.
18281 2017-06-02 Richard Biener <rguenther@suse.de>
18283 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
18285 * tree-vect-stmts.c (process_use): Do not mark backedge defs
18286 for inductions as relevant.
18288 2017-06-02 Richard Biener <rguenther@suse.de>
18290 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
18291 (vectorizable_induction): ... this. Remove dead code.
18293 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
18295 * builtins. (expand_builtin_alloca): Remove second parameter and
18296 infer its value from the first parameter instead.
18297 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
18299 2017-06-02 Jakub Jelinek <jakub@redhat.com>
18301 PR rtl-optimization/80903
18302 * loop-doloop.c (add_test): Unshare sequence.
18304 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
18306 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
18308 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
18310 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
18312 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
18313 xlogue_layout::get_instance, logue_layout::xlogue_layout,
18314 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
18315 (xlogue_layout::get_stub_rtx): Make static.
18316 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
18317 (xlogue_layout::compute_stub_managed_regs): Rename to...
18318 (xlogue_layout::count_stub_managed_regs): ...this.
18319 (xlogue_layout::is_stub_managed_reg): New function.
18320 (xlogue_layout::m_stub_names): Rename to...
18321 (xlogue_layout::s_stub_names): ...this, make static.
18322 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
18323 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
18324 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
18325 xlogue_layout::s_stub_names): Instantiate statics.
18326 (stub_managed_regs): Remove.
18327 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
18328 (disable_call_ms2sysv_xlogues): Rename to...
18329 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
18330 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
18332 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
18333 change after reload_completed.
18334 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
18336 (ix86_expand_prologue): Likewise.
18337 (ix86_expand_epilogue): Likewise.
18338 (ix86_expand_split_stack_prologue): Likewise.
18339 (ix86_compute_frame_layout): Remove frame parameter ...
18340 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
18341 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
18343 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
18344 (ix86_frame): Move from here ...
18345 * config/i386/i386.h (ix86_frame): ... to here.
18346 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
18347 complete ix86_frame data structure instead. Remove some_ld_name.
18349 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18351 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
18352 symbols that hold a DECL_VALUE_EXPR.
18354 2017-06-01 Martin Jambor <mjambor@suse.cz>
18356 PR tree-optimization/80898
18357 * tree-sra.c (process_subtree_disqualification): Removed.
18358 (disqualify_candidate): Do not acll
18359 process_subtree_disqualification.
18360 (subtree_mark_written_and_enqueue): New function.
18361 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
18362 RHS has been disqualified and re-queue LHS if necessary. Apart
18363 from that, ignore disqualified RHS.
18365 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18367 * config/s390/s390.c (s390_emit_epilogue): Disable early return
18368 address fetch for z10 or later.
18370 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18372 * config/arc/arc.md (tst_movb): Add guard when splitting.
18374 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18376 * config/arc/arc.c (arc_can_eliminate): Test against
18377 arc_frame_pointer_needed.
18379 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18381 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
18382 to prevent store reordering.
18383 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
18384 (type): Add block type.
18385 (stack_tie): Define special instruction to be used in
18388 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18390 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
18391 constraint. It is not valid for the pattern.
18392 (noncommutative_binary_comparison): Likewise.
18394 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18396 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
18399 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18401 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
18402 be used by the reg-alloc.
18404 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18406 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
18407 reg-alloc when having mul64 or mul32x16 instructions.
18408 (mulsidi3): Likewise.
18409 (umulsidi3): Likewise.
18410 (mulsi32x16): New pattern.
18411 (mulsi64): Likewise.
18412 (mulsidi64): Likewise.
18413 (umulsidi64): Likewise.
18414 (MUL32x16_REG): Define.
18415 (mul64_600): Use MUL32x16_REG.
18416 (mac64_600): Likewise.
18417 (umul64_600): Likewise.
18418 (umac64_600): Likewise.
18420 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18422 * config/arc/arc.md (mulsi3_700): Make it commutative.
18424 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
18426 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
18428 (*sign_extendsidi2_insn): Likewise for movstosw.
18430 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18432 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
18433 the type of the input discriminant value. Convert the
18434 discriminant value of signedness vary.
18436 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
18438 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
18439 Add to -Wall section.
18441 2017-06-01 Richard Biener <rguenther@suse.de>
18443 PR middle-end/66313
18444 * fold-const.c (fold_plusminus_mult_expr): If the factored
18445 factor may be zero use a wrapping type for the inner operation.
18446 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
18447 and handle moved defs.
18448 (process_assignment): Properly guard the unary op case. Return a
18449 tri-state indicating that moving the stmt before the call may allow
18450 to continue. Pass through to_move.
18451 (find_tail_calls): Handle moving unrelated defs before
18454 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
18457 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
18458 splitter result in the canonical way.
18460 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
18462 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
18463 also for 32bit target. Update insn attributes.
18464 (zero-extendsidi2 splitter): Allow all registers for operand 1.
18466 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
18468 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
18469 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
18470 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
18471 (_mm_maskz_min_ss): New intrinsics.
18473 2017-05-31 Martin Liska <mliska@suse.cz>
18475 * tree-vect-loop.c (vect_create_epilog_for_reduction):
18476 Change comment style to one we normally use.
18477 (vectorizable_reduction): Likewise.
18478 (vectorizable_induction): Likewise.
18479 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18480 (vectorizable_call): Likewise.
18481 (vectorizable_simd_clone_call): Likewise.
18482 (vectorizable_conversion): Likewise.
18483 (vectorizable_assignment): Likewise.
18484 (vectorizable_shift): Likewise.
18485 (vectorizable_operation): Likewise.
18486 (vectorizable_store): Likewise.
18487 (vectorizable_load): Likewise.
18488 * tree-vectorizer.h: Likewise.
18490 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
18492 * passes.c (emergency_dump_function): New.
18493 * tree-pass.h (emergency_dump_function): Declare.
18494 * plugin.c (plugins_internal_error_function): Remove.
18495 * plugin.h (plugins_internal_error_function): Remove declaration.
18496 * toplev.c (internal_error_function): New static function. Use it...
18497 (general_init): ...here.
18499 2017-05-31 Graham Markall <graham.markall@embecosm.com>
18501 * config/arc/arc.c (arc_print_operand): Handle constant operands.
18502 (arc_rtx_costs): Add costs for new patterns.
18503 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
18504 * config/arc/predicates.md: Add _1_2_3_operand predicate.
18506 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18508 * tree-ssa-strlen.c (get_next_strinfo): New function.
18509 (get_stridx_plus_constant): Use it.
18510 (zero_length_string): Likewise.
18511 (adjust_related_strinfos): Likewise.
18512 (adjust_last_stmt): Likewise.
18514 2017-05-31 Richard Biener <rguenther@suse.de>
18517 * config/i386/i386.c (ix86_expand_builtin): Remove assert
18518 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
18520 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18522 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
18523 loop_vinfo argument and use of dependence distance vectors.
18524 Check instead whether the two references differ only in their
18525 initial value and assume that they have the same alignment if the
18526 difference is a multiple of the vector alignment.
18527 (vect_analyze_data_refs_alignment): Update call accordingly.
18529 2017-05-31 Martin Liska <mliska@suse.cz>
18532 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
18534 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18536 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
18537 (create_intersect_range_checks): Move from ...
18538 * tree-data-ref.c (create_intersect_range_checks_index)
18539 (create_intersect_range_checks): ... to here.
18540 (create_runtime_alias_checks): New function factored from ...
18541 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
18542 here. Call above function.
18543 * tree-data-ref.h (create_runtime_alias_checks): New function.
18545 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18547 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
18548 segment length for dr_b and compute it in wide_int.
18550 2017-05-31 Richard Biener <rguenther@suse.de>
18552 PR tree-optimization/80906
18553 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
18554 and pass through iv_map.
18555 (copy_bb_and_scalar_dependences): Adjust.
18556 (translate_pending_phi_nodes): Likewise.
18557 (copy_loop_close_phi_args): Handle code-generating IVs instead
18560 2017-05-30 David Malcolm <dmalcolm@redhat.com>
18562 * diagnostic-color.c (color_dict): Add "type-diff".
18563 (parse_gcc_colors): Update comment.
18564 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
18565 -fdiagnostics-show-template-tree and -fno-elide-type.
18566 (GCC_COLORS): Add type-diff to example.
18568 (-fdiagnostics-show-template-tree): New.
18569 (-fno-elide-type): New.
18570 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
18571 the pp_format_decoder callback. Call any m_format_postprocessor's
18573 (pretty_printer::pretty_printer): Initialize
18574 m_format_postprocessor.
18575 (pretty_printer::~pretty_printer): Delete any
18576 m_format_postprocessor.
18577 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
18578 (class format_postprocessor): New class.
18579 (struct pretty_printer::format_decoder): Document the new parameters.
18580 (struct pretty_printer::m_format_postprocessor): New field.
18581 * tree-diagnostic.c (default_tree_printer): Update for new
18582 bool and const char ** params.
18583 * tree-diagnostic.h (default_tree_printer): Likewise.
18585 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
18587 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
18588 (lwa_operand): Delete rs6000_gen_cell_microcode test.
18589 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18590 rs6000_gen_cell_microcode code.
18591 (rs6000_final_prescan_insn): Delete.
18592 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
18593 "warn-cell-microcode" entries.
18594 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
18595 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
18596 throughout. Change cc_reg_not_micro_cr0_operand to
18597 cc_reg_not_cr0_operand throughout.
18598 (*extendhi<mode>2_noload): Delete.
18599 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
18600 (mwarn-cell-microcode): Delete.
18601 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
18602 -mgen-cell-microcode and -mwarn-cell-microcode.
18604 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
18607 * config/i386/constraints.md (Yd): New constraint.
18609 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
18610 and (?Yd, r) alternatives. Update insn attributes.
18611 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
18612 and (?*Yd, r) alternatives. Update insn attributes.
18613 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
18615 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
18617 * gimplify.c (gimplify_modify_expr): Don't create a
18618 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
18621 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
18623 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
18625 2017-05-30 Richard Biener <rguenther@suse.de>
18627 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
18628 and reduc_def fields.
18629 (STMT_VINFO_REDUC_TYPE): New define.
18630 (STMT_VINFO_REDUC_DEF): Likewise.
18631 (vect_force_simple_reduction): Adjust prototype.
18632 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
18633 (vect_is_simple_reduction): Remove check_reduction argument.
18634 (vect_force_simple_reduction): Adjust and set
18635 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18636 (vectorizable_reduction): Do not re-do reduction analysis
18637 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18638 * tree-parloops.c (gather_scalar_reductions): Adjust.
18640 2017-05-30 Richard Biener <rguenther@suse.de>
18642 PR middle-end/80901
18643 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
18646 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
18648 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
18649 Introduce unknown_misalignment parameter and remove vf.
18650 (vect_peeling_hash_get_lowest_cost):
18651 Pass unknown_misalignment parameter.
18652 (vect_enhance_data_refs_alignment):
18653 Fix unsupportable data ref treatment.
18655 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18657 * tree-vect-data-refs.c (vect_get_data_access_cost):
18658 Workaround for SLP handling.
18659 (vect_enhance_data_refs_alignment):
18660 Compute costs for doing no peeling at all, compare to the best
18661 peeling costs so far and avoid peeling if cheaper.
18663 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18665 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
18666 Return peeling info and set costs to zero for unlimited cost
18668 (vect_enhance_data_refs_alignment): Also inspect all datarefs
18669 with unknown misalignment. Compute and costs for unknown
18670 misalignment, compare them to the costs for known misalignment
18671 and choose the cheapest for peeling.
18673 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18675 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
18676 (vect_get_peeling_costs_all_drs): Create function.
18677 (vect_peeling_hash_get_lowest_cost):
18678 Use vect_get_peeling_costs_all_drs.
18679 (vect_peeling_supportable): Create function.
18680 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
18682 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18684 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
18685 DR_HAS_NEGATIVE_STEP.
18686 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
18687 (vect_enhance_data_refs_alignment): Use.
18688 (vect_duplicate_ssa_name_ptr_info): Use.
18689 * tree-vectorizer.h (dr_misalignment): Use.
18690 (known_alignment_for_access_p): Use.
18692 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18695 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
18697 (has_section_name): New function.
18699 2017-05-30 Martin Liska <mliska@suse.cz>
18702 * auto-profile.c (get_function_decl_from_block): Fix
18705 2017-05-30 Richard Biener <rguenther@suse.de>
18707 PR middle-end/80876
18708 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
18710 2017-05-30 Martin Liska <mliska@suse.cz>
18712 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
18713 * dumpfile.h (struct dump_file_info): Remove ctors.
18715 2017-05-30 Martin Liska <mliska@suse.cz>
18717 * predict.def: Fix GNU coding style.
18719 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18721 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18722 Mark 'to' argument with ATTRIBUTE_UNUSED.
18724 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18726 * config/xtensa/xtensa.c (xtensa_emit_call): Use
18727 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
18728 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
18731 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
18733 * doc/install.texi (Options specification): Restore entry of
18734 --enable-sjlj-exceptions.
18736 2017-05-27 Michael Eager <eager@eagercon.com>
18739 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
18741 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
18743 * config/microblaze/microblaze.h
18744 (FIXED_REGISTERS): Update in macro.
18745 (CALL_USED_REGISTERS): Update in macro.
18747 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
18749 * doc/install.texi: Add links to macOS binary distributions.
18751 2017-05-27 Jakub Jelinek <jakub@redhat.com>
18755 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18757 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18759 2017-05-26 Martin Liska <mliska@suse.cz>
18761 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
18763 2017-05-26 Martin Liska <mliska@suse.cz>
18765 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
18766 always leading ';; '.
18767 (dump_bb_info): Likewise.
18768 (brief_dump_cfg): Likewise.
18769 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
18770 * dumpfile.c: Remove usage of TDF_VERBOSE.
18771 * dumpfile.h (enum dump_kind): Likewise.
18772 (dump_gimple_bb_header): Do not use TDF_COMMENT.
18773 * print-tree.c (debug_verbose): Remove.
18774 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
18775 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
18776 * tree-diagnostic.c (default_tree_printer): Replace
18777 TDF_DIAGNOSTIC with TDF_SLIM.
18779 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18781 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
18782 in parameter loop, rather than loop_vinfo.
18783 (create_intersect_range_checks): Ditto.
18784 (vect_create_cond_for_alias_checks): Update call to above functions.
18786 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18788 PR tree-optimization/80815
18789 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
18790 for merging runtime alias checks. Handle negative DR_STEPs.
18792 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18794 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
18796 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
18797 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
18798 out code pruning runtime alias checks.
18799 * tree-data-ref.c (prune_runtime_alias_test_list): New function
18800 factored out from above.
18801 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18803 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18805 (prune_runtime_alias_test_list): New decalaration.
18807 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18809 * tree-vect-data-refs.c (compare_tree): Rename and move ...
18810 * tree-data-ref.c (data_ref_compare_tree): ... to here.
18811 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
18812 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
18813 (operator==, comp_dr_with_seg_len_pair): Ditto.
18814 (vect_prune_runtime_alias_test_list): Ditto.
18816 2017-05-26 Martin Liska <mliska@suse.cz>
18819 * params.def: Bound partial-inlining-entry-probability param.
18821 2017-05-26 Marek Polacek <polacek@redhat.com>
18824 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
18827 2017-05-26 Richard Biener <rguenther@suse.de>
18829 PR tree-optimization/80842
18830 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
18833 2017-05-26 Richard Biener <rguenther@suse.de>
18835 PR tree-optimization/80844
18836 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
18838 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
18840 * doc/md.texi (Machine Constraints): Update x86 family
18841 machine constraints section to match 'config/i386/constraints.md'.
18843 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
18845 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
18847 2017-05-25 Nathan Sidwell <nathan@acm.org>
18849 * doc/invoke.texi (--enable-languages): Update documentation.
18851 2017-05-25 Martin Liska <mliska@suse.cz>
18853 * dumpfile.c: Add TDF_FOLDING.
18854 * dumpfile.h (enum dump_kind): Likewise.
18855 * genmatch.c (dt_simplify::gen_1): Use it.
18857 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18859 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
18861 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18863 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18864 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
18866 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18868 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
18869 * match.pd (X == C): Rewrite it here.
18870 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
18871 with_certain_nonzero_bits2): New predicates.
18872 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
18874 2017-05-24 Nathan Sidwell <nathan@acm.org>
18876 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
18879 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
18882 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18884 * config/powerpcspe: New port. Files are copied from the rs6000
18885 port, with "rs6000" in filenames replaced by "powerpcspe".
18887 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
18889 PR rtl-optimization/80754
18890 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
18892 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
18894 * config/sparc/sparc.md (length): Return the correct value for -mflat
18895 sibcalls to match output_sibcall.
18897 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18901 * config/rs6000/rs6000.c (struct machine_function): Add new field
18903 (rs6000_get_separate_components): Init that field, use it.
18904 (rs6000_components_for_bb): Use the field.
18906 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
18908 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
18910 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
18912 PR middle-end/80823
18913 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
18915 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18918 * config/s390/s390.c (s390_check_qrst_address): Check incoming
18919 address against address_operand predicate.
18920 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
18922 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
18924 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
18925 parameters passed indirectly.
18927 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
18929 * config/i386/i386.md (*movdi_internal): Remove SSE4
18930 alternative 18 (?r, *v). Update insn attributes.
18931 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
18932 Update insn attributes.
18933 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
18934 Update insn attributes.
18935 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
18936 alternative 1 (r, v). Remove isa attribute.
18937 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
18938 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
18939 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18941 2017-05-23 Tom de Vries <tom@codesourcery.com>
18943 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
18946 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18948 * cgraphunit.c (symbol_table::process_new_functions): Update.
18949 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
18950 (inline_generate_summary): Rename to ...
18951 (ipa_fn_summary_generate): ... this one.
18952 (inline_read_summary): Rename to ...
18953 (ipa_fn_summary_read): ... this one.
18954 (inline_write_summary): Rename to ...
18955 (ipa_fn_summary_write): ... this one.
18956 (inline_free_summary): Rename to ...
18957 (ipa_free_fn_summary): ... this one.
18958 (pass_data_local_fn_summary, pass_local_fn_summary,
18959 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
18960 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
18961 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
18962 make_pass_ipa_fn_summary): New.
18963 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
18964 inline_write_summary, inline_free_summary): Remove.
18965 (ipa_free_fn_summary) : New.
18966 * ipa-inline.c (ipa_inline): Update.
18967 (pass_ipa_inline): Do not generate summaries.
18968 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
18970 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
18971 and add pass_ipa_fn_summary.
18972 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
18974 (make_pass_inline_parameters): Remove.
18976 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
18978 * omp-low.c (struct omp_context): Remove "default_kind" member.
18981 * omp-offload.c (execute_oacc_device_lower): Remove the
18982 parallelism dimensions function attributes for unparallelized
18983 OpenACC kernels constructs.
18985 2017-05-23 Martin Liska <mliska@suse.cz>
18987 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
18989 (cgraph_edge::make_speculative): Likewise.
18990 (cgraph_edge::resolve_speculation): Likewise.
18991 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
18992 (cgraph_node::dump): Likewise.
18993 * cgraph.h: Likewise.
18994 * cgraphunit.c (analyze_functions): Likewise.
18995 (symbol_table::compile): Likewise.
18996 * ipa-cp.c (print_all_lattices): Likewise.
18997 (determine_versionability): Likewise.
18998 (initialize_node_lattices): Likewise.
18999 (ipcp_verify_propagated_values): Likewise.
19000 (estimate_local_effects): Likewise.
19001 (update_profiling_info): Likewise.
19002 (create_specialized_node): Likewise.
19003 (perhaps_add_new_callers): Likewise.
19004 (decide_about_value): Likewise.
19005 (decide_whether_version_node): Likewise.
19006 (identify_dead_nodes): Likewise.
19007 (ipcp_store_bits_results): Likewise.
19008 * ipa-devirt.c (dump_targets): Likewise.
19009 (ipa_devirt): Likewise.
19010 * ipa-icf.c (sem_item::dump): Likewise.
19011 (sem_function::equals): Likewise.
19012 (sem_variable::equals): Likewise.
19013 (sem_item_optimizer::read_section): Likewise.
19014 (sem_item_optimizer::execute): Likewise.
19015 (congruence_class::dump): Likewise.
19016 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19017 (dump_inline_summary): Likewise.
19018 (estimate_node_size_and_time): Likewise.
19019 (inline_analyze_function): Likewise.
19020 * ipa-inline-transform.c (inline_call): Likewise.
19021 * ipa-inline.c (report_inline_failed_reason): Likewise.
19022 (want_early_inline_function_p): Likewise.
19023 (edge_badness): Likewise.
19024 (update_edge_key): Likewise.
19025 (inline_small_functions): Likewise.
19026 * ipa-profile.c (ipa_profile): Likewise.
19027 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19028 (ipa_make_edge_direct_to_target): Likewise.
19029 (remove_described_reference): Likewise.
19030 (ipa_impossible_devirt_target): Likewise.
19031 (propagate_controlled_uses): Likewise.
19032 (ipa_print_node_params): Likewise.
19033 (ipcp_transform_function): Likewise.
19034 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19035 (propagate_pure_const): Likewise.
19036 * ipa-reference.c (generate_summary): Likewise.
19037 (read_write_all_from_decl): Likewise.
19038 (propagate): Likewise.
19039 (ipa_reference_read_optimization_summary): Likewise.
19040 * ipa-utils.c (ipa_merge_profiles): Likewise.
19041 * ipa.c (walk_polymorphic_call_targets): Likewise.
19042 (symbol_table::remove_unreachable_nodes): Likewise.
19043 (ipa_single_use): Likewise.
19044 * passes.c (execute_todo): Likewise.
19045 * predict.c (drop_profile): Likewise.
19046 * symtab.c (symtab_node::get_dump_name): New function.
19047 (symtab_node::dump_name): Likewise.
19048 (symtab_node::dump_asm_name): Likewise.
19049 (symtab_node::dump_references): Likewise.
19050 (symtab_node::dump_referring): Likewise.
19051 (symtab_node::dump_base): Likewise.
19052 (symtab_node::debug_symtab): Likewise.
19053 * tree-sra.c (convert_callers_for_node): Likewise.
19054 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19055 * value-prof.c (init_node_map): Likewise.
19057 2017-05-23 Martin Liska <mliska@suse.cz>
19059 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19060 and symtab_node::debug_symtab to symbol_table::debug.
19061 * cgraphunit.c (analyze_functions): Use the renamed function.
19062 (symbol_table::compile): Likewise.
19063 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19064 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19065 * passes.c (execute_todo): Likewise.
19066 * symtab.c (symbol_table::dump): New function.
19067 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19069 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19071 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19072 that nonconst implies exec.
19074 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19076 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19077 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19078 (inline_edge_summary_vec): Turn into ...
19079 (ipa_call_summaries): ... this one.
19080 (redirect_to_unreachable, edge_set_predicate,
19081 evaluate_properties_for_edge, inline_summary_alloc,
19082 reset_ipa_call_summary, reset_inline_summary,
19083 inline_summary_t::duplicate): Update.
19084 (inline_edge_duplication_hook): Turn to ...
19085 (ipa_call_summary_t::duplicate): ... this one.
19086 (inline_edge_removal_hook): Turn to ...
19087 (ipa_call_summary_t::remove): ... this one.
19088 (dump_inline_edge_summary): Turn to ...
19089 (dump_ipa_call_summary): ... this one.
19090 (estimate_function_body_sizes): Update.
19091 (inline_update_callee_summaries): Update.
19092 (remap_edge_change_prob): Update.
19093 (remap_edge_summaries): Update.
19094 (inline_merge_summary): Update.
19095 (do_estimate_edge_time): Update.
19096 (inline_generate_summary): Update.
19097 (inline_read_section): Update.
19098 (inline_read_summary): Update.
19099 (inline_free_summary): Update.
19100 * ipa-inline.c (can_inline_edge_p): Update.
19101 (compute_inlined_call_time): Update.
19102 (want_inline_small_function_p): Update.
19103 (edge_badness): Update.
19104 (early_inliner): Update.
19105 * ipa-inline.h (inline_edge_summary): Turn to ...
19106 (ipa_call_summary): ... this one.
19107 (ipa_call_summary_t): New class.
19108 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19109 (ipa_call_summaries): New.
19110 (inline_edge_summary): Remove.
19111 (estimate_edge_growth): Update.
19112 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19113 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19114 * ipa-split.c (execute_split_functions): Update.
19115 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19117 2017-05-23 Tom de Vries <tom@codesourcery.com>
19119 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19120 attributes): Document rdrand effective target.
19122 2017-05-23 Tom de Vries <tom@codesourcery.com>
19124 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19125 attributes): Sort alphabetically.
19127 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
19129 * config/avr/genmultilib.awk: Use gsub instead of gensub.
19131 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19134 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
19135 V2DF/V2DI splat into two separate patterns, one that handles
19136 registers, and the other that only handles memory. Drop support
19137 for splatting from a GPR on ISA 2.07 and then splitting the
19138 splat into direct move and splat.
19139 (vsx_splat_<mode>_reg): Likewise.
19140 (vsx_splat_<mode>_mem): Likewise.
19142 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
19144 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
19146 2017-05-22 Jakub Jelinek <jakub@redhat.com>
19148 PR middle-end/80809
19149 * omp-low.c (finish_taskreg_remap): New function.
19150 (finish_taskreg_scan): If unit size of ctx->record_type
19151 is non-constant, unshare the size expression and replace
19152 decls in it with possible outer var refs.
19154 PR middle-end/80809
19155 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
19156 GOVD_SHARED rather than GOVD_PRIVATE with it.
19157 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
19158 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
19160 PR middle-end/80853
19161 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
19162 as last argument to build_outer_var_ref for pointer bases of array
19163 section reductions.
19165 2017-05-19 Martin Sebor <msebor@redhat.com>
19167 * print-tree.c (print_node): Print DECL_READ_P flag.
19169 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19171 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
19172 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
19173 * cgraph.c: Likewise.
19174 * cgraphunit.c: Likewise.
19175 * gengtype.c: Likewise.
19176 * ipa-cp.c: Likewise.
19177 * ipa-devirt.c: Likewise.
19178 * ipa-icf.c: Likewise.
19179 * ipa-predicate.c: Likewise.
19180 * ipa-profile.c: Likewise.
19181 * ipa-prop.c: Likewise.
19182 * ipa-split.c: Likewise.
19184 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
19185 edge_predicate_pool, dump_inline_hints,
19186 inline_summary::account_size_time, redirect_to_unreachable,
19187 edge_set_predicate, set_hint_predicate,
19188 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
19189 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
19190 inline_summary_t::remove, remap_hint_predicate_after_duplication,
19191 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
19192 ipa_call_summary_t::remove, initialize_growth_caches,
19193 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
19194 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
19195 mark_modified, unmodified_parm_1, unmodified_parm,
19196 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
19197 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
19198 compute_bb_predicates, will_be_nonconstant_expr_predicate,
19199 will_be_nonconstant_predicate, record_modified_bb_info,
19200 get_minimal_bb, record_modified, param_change_prob,
19201 phi_result_unknown_predicate, predicate_for_phi_result,
19202 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
19203 estimate_function_body_sizes, compute_inline_parameters,
19204 compute_inline_parameters_for_curren, pass_data_inline_parameters,
19205 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
19206 inline_update_callee_summaries, remap_edge_change_prob,
19207 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
19208 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
19209 inline_analyze_function, inline_summary_t::insert,
19210 inline_generate_summary, read_ipa_call_summary, inline_read_section,
19211 inline_read_summary, write_ipa_call_summary, inline_write_summary,
19212 inline_free_summary): Move to ipa-fnsummary.h
19213 (predicate_t): Remove.
19214 * ipa-fnsummary.c: New file.
19215 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
19216 (enum inline_hints_vals, inline_hints, agg_position_info,
19217 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
19218 inline_summaries, ipa_call_summary, ipa_call_summary_t,
19219 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
19220 dump_inline_summary, dump_inline_hints, inline_generate_summary,
19221 inline_read_summary, inline_write_summary, inline_free_summary,
19222 inline_analyze_function, initialize_inline_failed,
19223 inline_merge_summary, inline_update_overall_summary,
19224 compute_inline_parameters): Move to ipa-fnsummary.h
19225 * ipa-fnsummary.h: New file.
19226 * ipa-inline-transform.h: Include ipa-inline.h.
19227 * ipa-inline.c: LIkewise.
19229 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19231 * ipa-inline.c (edge_badness): Use inlined_time instead of
19232 inline_summaries->get.
19234 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19236 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
19238 2017-05-22 Nathan Sidwell <nathan@acm.org>
19240 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
19241 (fdump-lang): Document 'raw' option.
19242 * dumpfile.h (TDI_tu): Delete.
19243 * dumpfile.c (dump_files): Remove translation-unit.
19244 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
19246 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
19248 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
19249 command option from $(AWK) call.
19250 * config/avr/genmultilib.awk: Simplify and rewrite so that it
19251 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
19252 [FORMAT]: Remove handling of variable.
19253 * config/avr/t-multilib: Regenerate.
19255 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19257 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
19259 (dump_inline_summary): Do not print self_time.
19260 (estimate_function_body_sizes): Do not set self_time.
19261 (compute_inline_parameters): Likewise.
19262 (inline_read_section, inline_write_summary): Do not stream self_time.
19263 * ipa-inline.h (inline_summary): Drop self_time.
19265 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19267 * ipa-inline-analysis.c (account_size_time): Rename to ...
19268 (inline_summary::account_size_time): ... this one.
19269 (reset_ipa_call_summary): Turn to ...
19270 (ipa_call_summary::reset): ... this one.
19271 (reset_inline_summary): Turn to ...
19272 (inline_summary::reset): ... this one.
19273 (inline_summary_t::remove): Update.
19274 (inline_summary_t::duplicate): Update.
19275 (ipa_call_summary_t::remove): Update.
19276 (dump_inline_summary): Update.
19277 (estimate_function_body_sizes): Update.
19278 (compute_inline_parameters): Update.
19279 (estimate_node_size_and_time): Update.
19280 (inline_merge_summary): Update.
19281 (inline_update_overall_summary): Update.
19282 (inline_read_section): Update.
19283 (inline_write_summary): Update.
19284 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
19285 add account_size_time and reset member functions.
19286 (ipa_call_summary): Add reset function.
19287 * ipa-predicate.h (predicate::operator &): Constify.
19289 2017-05-22 Richard Biener <rguenther@suse.de>
19291 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
19293 2017-05-19 Jason Merrill <jason@redhat.com>
19295 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
19297 2017-05-19 Marek Polacek <polacek@redhat.com>
19300 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
19301 TYPE_OVERFLOW_WRAPS checks.
19303 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
19305 * tree-core.h (enum omp_clause_default_kind): Add
19306 "OMP_CLAUSE_DEFAULT_PRESENT".
19307 * tree-pretty-print.c (dump_omp_clause): Handle it.
19308 * gimplify.c (enum gimplify_omp_var_data): Add
19309 "GOVD_MAP_FORCE_PRESENT".
19310 (gimplify_adjust_omp_clauses_1): Map it to
19311 "GOMP_MAP_FORCE_PRESENT".
19312 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
19314 * gimplify.c (oacc_default_clause): Clarify.
19316 2017-05-19 Nathan Sidwell <nathan@acm.org>
19318 LANG_HOOK_REGISTER_DUMPS
19319 * toplev.c (general_init): Call register dump lang hook.
19320 * doc/invoke.texi: Document -fdump-lang option family.
19321 * dumpfile.c (dump_files): Remove class dump here.
19322 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
19323 * dumpfile.h (tree_dump_index): Remove TDI_class.
19324 * langhooks-def.h (lhd_register_dumps): Declare.
19325 (LANG_HOOKS_REGISTER_DUMPS): Define.
19326 (LANG_HOOKS_INITIALIZER): Add it.
19327 * langhooks.c (lhd_register_dumps): Define.
19328 * langhooks.h (struct lang_hooks): Add register_dumps.
19330 2017-05-19 Nathan Sidwell <nathan@acm.org>
19332 * context.h (context::set_passes): New.
19333 * context.c (context::context): Do not create pass manager.
19334 * toplev.c (general_init): Create pass manager here.
19336 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
19338 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
19339 use this splitter if two add or or instructions would also work for
19340 the constant we want to generate.
19342 2017-05-19 Richard Biener <rguenther@suse.de>
19345 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
19346 predicate evaluation.
19348 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19350 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
19352 * ipa-inline.c (want_inline_small_function_p): Do not cast to
19355 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19357 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19358 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19359 (inline_edge_summary_vec): Turn into ...
19360 (ipa_call_summaries): ... this one.
19361 (redirect_to_unreachable, edge_set_predicate,
19362 evaluate_properties_for_edge, inline_summary_alloc,
19363 reset_ipa_call_summary, reset_inline_summary,
19364 inline_summary_t::duplicate): Update.
19365 (inline_edge_duplication_hook): Turn to ...
19366 (ipa_call_summary_t::duplicate): ... this one.
19367 (inline_edge_removal_hook): Turn to ...
19368 (ipa_call_summary_t::remove): ... this one.
19369 (dump_inline_edge_summary): Turn to ...
19370 (dump_ipa_call_summary): ... this one.
19371 (estimate_function_body_sizes): Update.
19372 (inline_update_callee_summaries): Update.
19373 (remap_edge_change_prob): Update.
19374 (remap_edge_summaries): Update.
19375 (inline_merge_summary): Update.
19376 (do_estimate_edge_time): Update.
19377 (inline_generate_summary): Update.
19378 (inline_read_section): Update.
19379 (inline_read_summary): Update.
19380 (inline_free_summary): Update.
19381 * ipa-inline.c (can_inline_edge_p): Update.
19382 (compute_inlined_call_time): Update.
19383 (want_inline_small_function_p): Update.
19384 (edge_badness): Update.
19385 (early_inliner): Update.
19386 * ipa-inline.h (inline_edge_summary): Turn to ...
19387 (ipa_call_summary): ... this one.
19388 (ipa_call_summary_t): New class.
19389 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19390 (ipa_call_summaries): New.
19391 (inline_edge_summary): Remove.
19392 (estimate_edge_growth): Update.
19393 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19394 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19395 * ipa-split.c (execute_split_functions): Update.
19396 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19398 2017-05-19 Richard Biener <rguenther@suse.de>
19400 PR middle-end/80764
19401 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
19403 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
19405 * config/rs6000/rs6000.c (struct machine_function): Add field
19406 fpr_is_wrapped_separately.
19407 (rs6000_get_separate_components): Use 64 components. Handle the
19408 new FPR components.
19409 (rs6000_components_for_bb): Handle the FPR components.
19410 (rs6000_emit_prologue_components): Handle the FPR components.
19411 (rs6000_emit_epilogue_components): Handle the FPR components.
19412 (rs6000_set_handled_components): Handle the FPR components.
19413 (rs6000_emit_prologue): Don't output prologue code for those FPRs
19414 that are already separately shrink-wrapped.
19415 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
19416 that are already separately shrink-wrapped.
19418 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19421 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
19424 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
19425 (define_peephole2 for Altivec d-form load): Add peepholes to catch
19426 cases where the register allocator uses a move and an offsettable
19427 memory operation to/from a FPR register on ISA 2.06/2.07.
19428 (define_peephole2 for Altivec d-form store): Likewise.
19430 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
19433 * config/i386/mmx.md (*mov<mode>_internal): Enable
19434 alternatives 11, 12, 13 and 14 also for 32bit targets.
19435 Remove alternatives 15, 16, 17 and 18.
19436 * config/i386/sse.md (vec_concatv2di): Change
19437 alternative (!x, *y) to (x, ?!*Yn).
19439 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
19441 * dumpfile.h (enum dump_kind): Remove stray comma.
19443 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19445 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
19446 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
19447 predicate::num_conditions
19448 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
19449 (CHANGED): turn into predicate::changed.
19450 (agg_position_info): Move to ipa-predicate.h
19451 (add_condition, predicate::add_clause, predicate::operator &=,
19452 predicate::or_with, predicate::evaluate, predicate::probability,
19453 dump_condition, dump_clause, predicate::dump,
19454 predicate::remap_after_duplication, predicate::remap_after_inlining,
19455 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
19456 (evaluate_conditions_for_known_args): Update.
19457 (set_cond_stmt_execution_predicate): Update.
19458 * ipa-inline.h: Include ipa-predicate.h
19459 (condition, inline_param_summary, conditions, agg_position_info,
19460 predicate): Move to ipa-predicate.h
19461 * ipa-predicate.c: New file.
19462 * ipa-predicate.h: New file.
19464 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
19466 * final.c (leaf_function_p): Check we are not in a sequence.
19468 2017-05-18 Martin Liska <mliska@suse.cz>
19470 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
19471 * dumpfile.c (dump_register): Use new enum dump_kind.
19472 (get_dump_file_name): Likewise.
19473 (dump_enable_all): Likewise.
19474 (dump_switch_p_1): Likewise.
19475 (enable_rtl_dump_file): Remove usage of TDF_RTL.
19476 * dumpfile.h (enum dump_kind): New enum type.
19477 (struct dump_file_info): Create constructor and
19478 format fields and comments.
19479 * passes.c (pass_manager::register_one_dump_file):
19481 * statistics.c (statistics_early_init): Likewise.
19482 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
19483 TDF_TREE with TDF_SLIM.
19484 (gather_memory_references_ref): Likewise.
19486 2017-05-18 Martin Liska <mliska@suse.cz>
19488 * vec.h (struct vnull): Use it.
19490 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19492 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
19493 (true_predicate, false_predicate, true_predicate_p,
19494 false_predicate_p): Remove.
19495 (single_cond_predicate, not_inlined_predicate): Turn to member function
19497 (add_condition): Update.
19498 (add_clause): Turn to...
19499 (predicate::add_clause): ... this one; update; allow passing NULL
19501 (and_predicates): Turn to ...
19502 (predicate::operator &=): ... this one.
19503 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
19504 (or_predicates): Turn to ...
19505 (predicate::or_with): ... this one.
19506 (evaluate_predicate): Turn to ...
19507 (predicate::evaluate): ... this one.
19508 (predicate_probability): Turn to ...
19509 (predicate::probability): ... this one.
19510 (dump_condition): Update.
19511 (dump_predicate): Turn to ...
19512 (predicate::dump): ... this one.
19513 (account_size_time): Update.
19514 (edge_set_predicate): Update.
19515 (set_hint_predicate): UPdate.
19516 (evaluate_conditions_for_known_args): Update.
19517 (evaluate_properties_for_edge): Update.
19518 (remap_predicate_after_duplication): Turn to...
19519 (predicate::remap_after_duplication): ... this one.
19520 (remap_hint_predicate_after_duplication): Update.
19521 (inline_summary_t::duplicate): UPdate.
19522 (dump_inline_edge_summary): Update.
19523 (dump_inline_summary): Update.
19524 (set_cond_stmt_execution_predicate): Update.
19525 (set_switch_stmt_execution_predicate): Update.
19526 (compute_bb_predicates): Update.
19527 (will_be_nonconstant_expr_predicate): Update.
19528 (will_be_nonconstant_predicate): Update.
19529 (phi_result_unknown_predicate): Update.
19530 (predicate_for_phi_result): Update.
19531 (array_index_predicate): Update.
19532 (estimate_function_body_sizes): Update.
19533 (estimate_node_size_and_time): Update.
19534 (estimate_ipcp_clone_size_and_time): Update.
19535 (remap_predicate): Rename to ...
19536 (predicate::remap_after_inlining): ... this one.
19537 (remap_hint_predicate): Update.
19538 (inline_merge_summary): Update.
19539 (inline_update_overall_summary): Update.
19540 (estimate_size_after_inlining): Update.
19541 (read_predicate): Rename to ...
19542 (predicate::stream_in): ... this one.
19543 (read_inline_edge_summary): Update.
19544 (write_predicate): Rename to ...
19545 (predicate::stream_out): ... this one.
19546 (write_inline_edge_summary): Update.
19547 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
19548 (clause_t): Turn to uint32_t
19549 (predicate): Turn to class; implement constructor and operators
19551 (size_time_entry): Update.
19552 (inline_summary): Update.
19553 (inline_edge_summary): Update.
19555 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
19557 * fold-const.c (fold_binary_loc): Move transformation...
19558 * match.pd (C - X CMP X): ... here.
19560 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
19562 * config/sparc/sparc.c (sparc_option_override): Set function
19563 alignment for -mcpu=niagara7 to 64 to match the I$ line.
19564 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
19566 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
19568 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
19570 2017-05-18 Marek Polacek <polacek@redhat.com>
19573 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
19574 (pass_ubsan::execute): Call gimple_assign_single_p instead of
19575 gimple_assign_load_p.
19577 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
19579 PR middle-end/80692
19580 * real.c (do_compare): Give decimal_do_compare preference over
19581 comparing just the signs.
19583 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
19585 * doc/md.texi (Canonicalization of Instructions): Describe the
19586 canonical form of instructions that inherently set a condition
19589 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
19591 PR middle-end/80775
19592 * tree-cfg.c: Move deletion of unreachable case statements to after
19593 the merging of consecutive case labels.
19595 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
19597 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
19598 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
19599 restoring of callee-saved registers.
19601 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
19603 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
19604 * config/visium/visium.c (single_set_and_flags): Likewise.
19605 * config/visium/visium.md (Substitutions): Likewise.
19607 2017-05-17 Martin Liska <mliska@suse.cz>
19609 * cfg.c: Introduce dump_flags_t type and
19610 use it instead of int type.
19612 * cfghooks.c: Likewise.
19613 * cfghooks.h (struct cfg_hooks): Likewise.
19614 * cfgrtl.c: Likewise.
19615 * cfgrtl.h: Likewise.
19616 * cgraph.c (cgraph_node::get_body): Likewise.
19617 * coretypes.h: Likewise.
19618 * domwalk.c: Likewise.
19619 * domwalk.h: Likewise.
19620 * dumpfile.c (struct dump_option_value_info): Likewise.
19621 (dump_enable_all): Likewise.
19622 (dump_switch_p_1): Likewise.
19623 (opt_info_switch_p): Likewise.
19624 * dumpfile.h (enum tree_dump_index): Likewise.
19625 (struct dump_file_info): Likewise.
19626 * genemit.c: Likewise.
19627 * generic-match-head.c: Likewise.
19628 * gengtype.c (open_base_files): Likewise.
19629 * gimple-pretty-print.c: Likewise.
19630 * gimple-pretty-print.h: Likewise.
19631 * graph.c (print_graph_cfg): Likewise.
19632 * graphite-scop-detection.c (dot_all_sese): Likewise.
19633 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
19634 * loop-unroll.c (report_unroll): Likewise.
19635 * passes.c (pass_manager::register_one_dump_file): Likewise.
19636 * print-tree.c: Likewise.
19637 * statistics.c: Likewise.
19638 * tree-cfg.c: Likewise.
19639 * tree-cfg.h: Likewise.
19640 * tree-dfa.c: Likewise.
19641 * tree-dfa.h: Likewise.
19642 * tree-dump.c (dump_function): Likewise.
19643 * tree-dump.h (struct dump_info): Likewise.
19644 * tree-pretty-print.c: Likewise.
19645 * tree-pretty-print.h: Likewise.
19646 * tree-ssa-live.c: Likewise.
19647 * tree-ssa-live.h: Likewise.
19648 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
19649 * tree-vect-loop.c: Likewise.
19650 * tree-vect-slp.c: Likewise.
19652 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
19653 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19655 PR tree-optimization/80457
19656 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
19657 of all arguments to a statement as scalar_to_vec operations.
19658 (vectorizable_call): Adjust call to vect_model_simple_cost for
19660 (vectorizable_conversion): Likewise.
19661 (vectorizable_assignment): Likewise.
19662 (vectorizable_shift): Likewise.
19663 (vectorizable_operation): Likewise.
19664 (vectorizable_comparison): Likewise.
19665 (vect_is_simple_cond): Record the def types for operands.
19666 (vectorizable_condition): Likewise, call vect_model_simple_cost.
19667 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
19668 for statement argument count.
19670 2017-05-16 Carl Love <cel@us.ibm.com>
19672 * config/rs6000/rs6000-c: Add support for built-in functions
19673 vector unsigned long long vec_bperm (vector unsigned long long,
19674 vector unsigned char)
19675 vector signed long long vec_mule (vector signed int,
19677 vector unsigned long long vec_mule (vector unsigned int,
19678 vector unsigned int)
19679 vector signed long long vec_mulo (vector signed int,
19681 vector unsigned long long vec_mulo (vector unsigned int,
19682 vector unsigned int)
19683 vector signed char vec_sldw (vector signed char,
19684 vector signed char,
19686 vector unsigned char vec_sldw (vector unsigned char,
19687 vector unsigned char,
19689 vector signed short vec_sldw (vector signed short,
19690 vector signed short,
19692 vector unsigned short vec_sldw (vector unsigned short,
19693 vector unsigned short,
19695 vector signed int vec_sldw (vector signed int,
19698 vector unsigned int vec_sldw (vector unsigned int,
19699 vector unsigned int,
19701 vector signed long long vec_sldw (vector signed long long,
19702 vector signed long long,
19704 vector unsigned long long vec_sldw (vector unsigned long long,
19705 vector unsigned long long,
19707 * config/rs6000/rs6000-c: Add support for built-in functions
19708 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
19709 * config/rs6000/altivec.h: Add defintion for vec_sldw.
19710 * doc/extend.texi: Update the built-in documentation for the
19711 new built-in functions.
19713 2017-05-16 Marek Polacek <polacek@redhat.com>
19717 * tree.c (save_expr): Don't fold the expression.
19719 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
19721 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
19722 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
19723 and (?*y,m). Update insn attributes.
19725 2017-05-16 Martin Liska <mliska@suse.cz>
19727 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
19728 flags argument of print_gimple_stmt, print_gimple_expr,
19729 print_generic_stmt and print_generic_expr.
19730 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
19731 * coretypes.h: Likewise.
19732 * except.c (dump_eh_tree): Likewise.
19733 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
19734 * gimple-pretty-print.h: Likewise.
19735 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
19736 (backprop::push_to_worklist): Likewise.
19737 (backprop::pop_from_worklist): Likewise.
19738 (backprop::process_use): Likewise.
19739 (backprop::intersect_uses): Likewise.
19740 (note_replacement): Likewise.
19741 * gimple-ssa-store-merging.c
19742 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
19743 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
19744 (pass_store_merging::execute): Likewise.
19745 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
19746 (ssa_base_cand_dump_callback): Likewise.
19747 (dump_incr_vec): Likewise.
19748 (replace_refs): Likewise.
19749 (replace_mult_candidate): Likewise.
19750 (create_add_on_incoming_edge): Likewise.
19751 (create_phi_basis): Likewise.
19752 (insert_initializers): Likewise.
19753 (all_phi_incrs_profitable): Likewise.
19754 (introduce_cast_before_cand): Likewise.
19755 (replace_one_candidate): Likewise.
19756 * gimplify.c (gimplify_expr): Likewise.
19757 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
19758 (set_rename): Likewise.
19759 (rename_uses): Likewise.
19760 (copy_loop_phi_nodes): Likewise.
19761 (add_close_phis_to_merge_points): Likewise.
19762 (copy_loop_close_phi_args): Likewise.
19763 (copy_cond_phi_args): Likewise.
19764 (graphite_copy_stmts_from_block): Likewise.
19765 (translate_pending_phi_nodes): Likewise.
19766 * graphite-poly.c (print_pdr): Likewise.
19767 (dump_gbb_cases): Likewise.
19768 (dump_gbb_conditions): Likewise.
19769 (print_scop_params): Likewise.
19770 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
19771 (build_cross_bb_scalars_use): Likewise.
19772 (gather_bbs::before_dom_children): Likewise.
19773 * hsa-dump.c (dump_hsa_immed): Likewise.
19774 * ipa-cp.c (print_ipcp_constant_value): Likewise.
19775 (get_replacement_map): Likewise.
19776 * ipa-inline-analysis.c (dump_condition): Likewise.
19777 (estimate_function_body_sizes): Likewise.
19778 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
19779 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
19780 * ipa-prop.c (ipa_dump_param): Likewise.
19781 (ipa_print_node_jump_functions_for_edge): Likewise.
19782 (ipa_modify_call_arguments): Likewise.
19783 (ipa_modify_expr): Likewise.
19784 (ipa_dump_param_adjustments): Likewise.
19785 (ipa_dump_agg_replacement_values): Likewise.
19786 (ipcp_modif_dom_walker::before_dom_children): Likewise.
19787 * ipa-pure-const.c (check_stmt): Likewise.
19788 (pass_nothrow::execute): Likewise.
19789 * ipa-split.c (execute_split_functions): Likewise.
19790 * omp-offload.c (dump_oacc_loop_part): Likewise.
19791 (dump_oacc_loop): Likewise.
19792 * trans-mem.c (tm_log_emit): Likewise.
19793 (tm_memopt_accumulate_memops): Likewise.
19794 (dump_tm_memopt_set): Likewise.
19795 (dump_tm_memopt_transform): Likewise.
19796 * tree-cfg.c (gimple_verify_flow_info): Likewise.
19797 (print_loop): Likewise.
19798 * tree-chkp-opt.c (chkp_print_addr): Likewise.
19799 (chkp_gather_checks_info): Likewise.
19800 (chkp_get_check_result): Likewise.
19801 (chkp_remove_check_if_pass): Likewise.
19802 (chkp_use_outer_bounds_if_possible): Likewise.
19803 (chkp_reduce_bounds_lifetime): Likewise.
19804 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
19805 (chkp_mark_completed_bounds): Likewise.
19806 (chkp_register_incomplete_bounds): Likewise.
19807 (chkp_mark_invalid_bounds): Likewise.
19808 (chkp_maybe_copy_and_register_bounds): Likewise.
19809 (chkp_build_returned_bound): Likewise.
19810 (chkp_get_bound_for_parm): Likewise.
19811 (chkp_build_bndldx): Likewise.
19812 (chkp_get_bounds_by_definition): Likewise.
19813 (chkp_generate_extern_var_bounds): Likewise.
19814 (chkp_get_bounds_for_decl_addr): Likewise.
19815 * tree-chrec.c (chrec_apply): Likewise.
19816 * tree-data-ref.c (dump_data_reference): Likewise.
19817 (dump_subscript): Likewise.
19818 (dump_data_dependence_relation): Likewise.
19819 (analyze_overlapping_iterations): Likewise.
19820 * tree-inline.c (expand_call_inline): Likewise.
19821 (tree_function_versioning): Likewise.
19822 * tree-into-ssa.c (dump_defs_stack): Likewise.
19823 (dump_currdefs): Likewise.
19824 (dump_names_replaced_by): Likewise.
19825 (dump_update_ssa): Likewise.
19826 (update_ssa): Likewise.
19827 * tree-object-size.c (pass_object_sizes::execute): Likewise.
19828 * tree-parloops.c (build_new_reduction): Likewise.
19829 (try_create_reduction_list): Likewise.
19830 (ref_conflicts_with_region): Likewise.
19831 (oacc_entry_exit_ok_1): Likewise.
19832 (oacc_entry_exit_single_gang): Likewise.
19833 * tree-pretty-print.h: Likewise.
19834 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
19835 (get_scalar_evolution): Likewise.
19836 (add_to_evolution): Likewise.
19837 (get_loop_exit_condition): Likewise.
19838 (analyze_evolution_in_loop): Likewise.
19839 (analyze_initial_condition): Likewise.
19840 (analyze_scalar_evolution): Likewise.
19841 (instantiate_scev): Likewise.
19842 (number_of_latch_executions): Likewise.
19843 (gather_chrec_stats): Likewise.
19844 (final_value_replacement_loop): Likewise.
19845 (scev_const_prop): Likewise.
19846 * tree-sra.c (dump_access): Likewise.
19847 (disqualify_candidate): Likewise.
19848 (create_access): Likewise.
19849 (reject): Likewise.
19850 (maybe_add_sra_candidate): Likewise.
19851 (create_access_replacement): Likewise.
19852 (analyze_access_subtree): Likewise.
19853 (analyze_all_variable_accesses): Likewise.
19854 (sra_modify_assign): Likewise.
19855 (initialize_constant_pool_replacements): Likewise.
19856 (find_param_candidates): Likewise.
19857 (decide_one_param_reduction): Likewise.
19858 (replace_removed_params_ssa_names): Likewise.
19859 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
19860 * tree-ssa-copy.c (dump_copy_of): Likewise.
19861 (copy_prop_visit_cond_stmt): Likewise.
19862 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
19863 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
19864 (record_equivalences_from_stmt): Likewise.
19865 * tree-ssa-dse.c (compute_trims): Likewise.
19866 (delete_dead_call): Likewise.
19867 (delete_dead_assignment): Likewise.
19868 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
19869 (forward_propagate_into_cond): Likewise.
19870 (pass_forwprop::execute): Likewise.
19871 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
19872 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
19874 (move_computations_worker): Likewise.
19875 (execute_sm): Likewise.
19876 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
19877 (remove_exits_and_undefined_stmts): Likewise.
19878 (remove_redundant_iv_tests): Likewise.
19879 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
19880 (adjust_iv_update_pos): Likewise.
19881 * tree-ssa-math-opts.c (bswap_replace): Likewise.
19882 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
19883 (value_replacement): Likewise.
19884 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
19885 * tree-ssa-pre.c (print_pre_expr): Likewise.
19886 (get_representative_for): Likewise.
19887 (create_expression_by_pieces): Likewise.
19888 (insert_into_preds_of_block): Likewise.
19889 (eliminate_insert): Likewise.
19890 (eliminate_dom_walker::before_dom_children): Likewise.
19891 (eliminate): Likewise.
19892 (remove_dead_inserted_code): Likewise.
19893 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
19894 * tree-ssa-reassoc.c (get_rank): Likewise.
19895 (eliminate_duplicate_pair): Likewise.
19896 (eliminate_plus_minus_pair): Likewise.
19897 (eliminate_not_pairs): Likewise.
19898 (undistribute_ops_list): Likewise.
19899 (eliminate_redundant_comparison): Likewise.
19900 (update_range_test): Likewise.
19901 (optimize_range_tests_var_bound): Likewise.
19902 (optimize_vec_cond_expr): Likewise.
19903 (rewrite_expr_tree): Likewise.
19904 (rewrite_expr_tree_parallel): Likewise.
19905 (linearize_expr): Likewise.
19906 (break_up_subtract): Likewise.
19907 (linearize_expr_tree): Likewise.
19908 (attempt_builtin_powi): Likewise.
19909 (attempt_builtin_copysign): Likewise.
19910 (transform_stmt_to_copy): Likewise.
19911 (transform_stmt_to_multiply): Likewise.
19912 (dump_ops_vector): Likewise.
19913 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
19914 (print_scc): Likewise.
19915 (set_ssa_val_to): Likewise.
19916 (visit_reference_op_store): Likewise.
19917 (visit_use): Likewise.
19918 (sccvn_dom_walker::before_dom_children): Likewise.
19919 (run_scc_vn): Likewise.
19920 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
19922 (expr_hash_elt::print): Likewise.
19923 (const_and_copies::pop_to_marker): Likewise.
19924 (const_and_copies::record_const_or_copy_raw): Likewise.
19925 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
19926 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
19927 (dump_predicates): Likewise.
19928 (find_uninit_use): Likewise.
19929 (warn_uninitialized_phi): Likewise.
19930 (pass_late_warn_uninitialized::execute): Likewise.
19931 * tree-ssa.c (verify_vssa): Likewise.
19932 (verify_ssa): Likewise.
19933 (maybe_optimize_var): Likewise.
19934 * tree-vrp.c (dump_value_range): Likewise.
19935 (dump_all_value_ranges): Likewise.
19936 (dump_asserts_for): Likewise.
19937 (register_edge_assert_for_2): Likewise.
19938 (vrp_visit_cond_stmt): Likewise.
19939 (vrp_visit_switch_stmt): Likewise.
19940 (vrp_visit_stmt): Likewise.
19941 (vrp_visit_phi_node): Likewise.
19942 (simplify_cond_using_ranges_1): Likewise.
19943 (fold_predicate_in): Likewise.
19944 (evrp_dom_walker::before_dom_children): Likewise.
19945 (evrp_dom_walker::push_value_range): Likewise.
19946 (evrp_dom_walker::pop_value_range): Likewise.
19947 (execute_early_vrp): Likewise.
19949 2017-05-16 Richard Biener <rguenther@suse.de>
19951 * dwarf2out.c (loc_list_from_tree_1): Do not create
19952 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
19954 2017-05-16 Richard Biener <rguenther@suse.de>
19956 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
19958 (note_variable_value_in_expr): If we resolved the decl ref
19959 do not push to the stack.
19961 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
19963 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
19964 operations in fast-math mode.
19965 (vaddq_f16): Likewise.
19966 (vmul_f16): Likewise.
19967 (vmulq_f16): Likewise.
19968 (vsub_f16): Likewise.
19969 (vsubq_f16): Likewise.
19970 * config/arm/neon.md (add<mode>3): New.
19972 (fma:<VH:mode>3): New. Also remove outdated comment.
19975 2017-05-16 Martin Liska <mliska@suse.cz>
19979 * ipa-devirt.c (warn_types_mismatch): Fix typo.
19980 (odr_types_equivalent_p): Likewise.
19982 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
19984 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
19986 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
19989 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
19990 non-interunit SSE move alternatives with '?'.
19991 (zero-extendsidi peephole2): New peephole to skip intermediate
19992 general register in SSE zero-extend sequence.
19994 2017-05-15 Jeff Law <law@redhat.com>
19996 * reorg.c (relax_delay_slots): Create a new variable to hold
19997 the temporary target rather than clobbering TARGET_LABEL.
19999 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20000 missing argument to extract_bit_field call.
20001 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20003 2017-05-15 Martin Liska <mliska@suse.cz>
20006 * gcc.c (process_command): Do not allow empty argument of -o option.
20008 2017-05-15 Renlin Li <renlin.li@arm.com>
20010 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20011 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20012 * config/aarch64/constraints.md (Usf): Add long call check.
20013 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20014 (call_value): Likewise.
20015 (sibcall): Likewise.
20016 (sibcall_value): Likewise.
20018 (call_value_insn): New.
20019 (sibcall_insn): Update rtx pattern.
20020 (sibcall_value_insn): Likewise.
20021 (call_internal): Remove.
20022 (call_value_internal): Likewise.
20023 (sibcall_internal): Likewise.
20024 (sibcall_value_internal): Likewise.
20025 (call_reg): Likewise.
20026 (call_symbol): Likewise.
20027 (call_value_reg): Likewise.
20028 (call_value_symbol): Likewise.
20030 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20033 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20035 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20037 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20038 compatible with CCGOCmode and with CCZmode.
20040 2017-05-14 Martin Sebor <msebor@redhat.com>
20042 PR middle-end/77671
20043 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20044 (gimple_fold_builtin_snprintf): Same.
20045 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20046 (gimple_fold_builtin_snprintf): Same.
20047 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20048 of character types.
20049 (is_call_safe): New function.
20050 (try_substitute_return_value): Call it.
20051 (try_simplify_call): New function.
20052 (pass_sprintf_length::handle_gimple_call): Call it.
20054 2017-05-14 Martin Sebor <msebor@redhat.com>
20056 PR middle-end/80669
20057 * builtins.c (expand_builtin_stpncpy): Simplify.
20059 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20061 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20062 * config/i386/i386.h
20063 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20064 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20065 (struct machine_function): Add new members call_ms2sysv,
20066 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20067 (struct machine_frame_state): New fields sp_realigned and
20068 sp_realigned_offset.
20069 * config/i386/i386.c
20070 (enum xlogue_stub): New enum.
20071 (enum xlogue_stub_sets): New enum.
20072 (class xlogue_layout): New class.
20073 (struct ix86_frame): New fields stack_realign_allocate_offset,
20074 stack_realign_offset and outlined_save_offset. Modify comments to
20075 detail stack layout when using out-of-line stubs.
20076 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20077 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
20078 -mcall-ms2sysv-xlogues.
20079 (stub_managed_regs): New static variable.
20080 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
20081 registers managed by out-of-line stub.
20082 (disable_call_ms2sysv_xlogues): New function.
20083 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
20084 m->call_ms2sysv when appropriate and compute frame layout for
20086 (sp_valid_at, fp_valid_at): New inline functions.
20087 (choose_basereg): New function.
20088 (choose_baseaddr): Add align parameter, use choose_basereg and modify
20090 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
20091 Use align parameter of choose_baseaddr to generated aligned SSE movs
20093 (pro_epilogue_adjust_stack): Modify to track
20094 machine_frame_state::sp_realigned.
20095 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
20096 (ix86_nsaved_sseregs): Likewise.
20097 (ix86_emit_save_regs): Likewise.
20098 (ix86_emit_save_regs_using_mov): Likewise.
20099 (ix86_emit_save_sse_regs_using_mov): Likewise.
20100 (get_scratch_register_on_entry): Likewise.
20101 (gen_frame_set): New function.
20102 (gen_frame_load): Likewise.
20103 (gen_frame_store): Likewise.
20104 (emit_outlined_ms2sysv_save): Likewise.
20105 (emit_outlined_ms2sysv_restore): Likewise.
20106 (ix86_expand_prologue): Modify stack re-alignment code and call
20107 emit_outlined_ms2sysv_save when appropriate.
20108 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
20109 parameter rtx_insn *insn, which allows the function to be used to only
20110 generate the notes.
20111 (ix86_expand_epilogue): Modify validity checks of frame and stack
20112 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
20113 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
20114 * config/i386/predicates.md
20115 (save_multiple): New predicate.
20116 (restore_multiple): Likewise.
20117 * config/i386/sse.md
20118 (save_multiple<mode>): New pattern.
20119 (save_multiple_realign<mode>): Likewise.
20120 (restore_multiple<mode>): Likewise.
20121 (restore_multiple_and_return<mode>): Likewise.
20122 (restore_multiple_leave_return<mode>): Likewise.
20123 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
20125 2017-05-14 Julia Koval <julia.koval@intel.com>
20127 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
20128 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
20129 (__builtin_ia32_xsetbv): New builtins.
20130 * config/i386/i386.c (ix86_expand_special_args_builtin):
20132 (ix86_expand_builtin): Special expand for new intrinsics.
20133 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
20134 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
20135 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
20137 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20139 * cfganal.c (inverted_post_order_compute): Change argument type
20141 * cfganal.h (inverted_post_order_compute): Adjust prototype.
20142 * df-core.c (rest_of_handle_df_initialize): Adjust.
20143 (rest_of_handle_df_finish): Likewise.
20144 (df_analyze_1): Likewise.
20145 (df_analyze): Likewise.
20146 (loop_inverted_post_order_compute): Change argument to be a vec *.
20147 (df_analyze_loop): Adjust.
20148 (df_get_n_blocks): Likewise.
20149 (df_get_postorder): Likewise.
20150 * df.h (struct df_d): Change field to be a vec.
20151 * lcm.c (compute_laterin): Adjust.
20152 (compute_available): Likewise.
20153 * lra-lives.c (lra_create_live_ranges_1): Likewise.
20154 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
20155 * tree-ssa-pre.c (compute_antic): Likewise.
20157 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20159 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
20160 (depth_first_search::depth_first_search): Change structure init
20161 function to this constructor.
20162 (depth_first_search::add_bb): Rename function to this member.
20163 (depth_first_search::execute): Likewise.
20164 (flow_dfs_compute_reverse_finish): Adjust.
20166 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20168 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
20169 (longest_simple_path): Likewise.
20170 * shrink-wrap.c (spread_components): Likewise.
20171 (disqualify_problematic_components): Likewise.
20172 (emit_common_heads_for_components): Likewise.
20173 (emit_common_tails_for_components): Likewise.
20174 (insert_prologue_epilogue_for_components): Likewise.
20176 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20178 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
20181 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20183 * df-core.c (df_set_blocks): Start using auto_bitmap.
20184 (df_compact_blocks): Likewise.
20185 * df-problems.c (df_rd_confluence_n): Likewise.
20186 * df-scan.c (df_insn_rescan_all): Likewise.
20187 (df_process_deferred_rescans): Likewise.
20188 (df_update_entry_block_defs): Likewise.
20189 (df_update_exit_block_uses): Likewise.
20190 (df_entry_block_bitmap_verify): Likewise.
20191 (df_exit_block_bitmap_verify): Likewise.
20192 (df_scan_verify): Likewise.
20193 * lra-constraints.c (lra_constraints): Likewise.
20194 (undo_optional_reloads): Likewise.
20195 (lra_undo_inheritance): Likewise.
20196 * lra-remat.c (calculate_gen_cands): Likewise.
20197 (do_remat): Likewise.
20198 * lra-spills.c (assign_spill_hard_regs): Likewise.
20199 (spill_pseudos): Likewise.
20200 * tree-ssa-pre.c (bitmap_set_and): Likewise.
20201 (bitmap_set_subtract_values): Likewise.
20203 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20205 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
20206 management with auto_bitmap.
20207 (fix_inter_tick): Likewise.
20208 (fix_recovery_deps): Likewise.
20209 * ira.c (add_store_equivs): Likewise.
20210 (find_moveable_pseudos): Likewise.
20211 (split_live_ranges_for_shrink_wrap): Likewise.
20212 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
20213 (rtx_reuse_manager::seen_def_p): Likewise.
20214 (rtx_reuse_manager::set_seen_def): Likewise.
20215 * print-rtl.h (class rtx_reuse_manager): Likewise.
20217 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20219 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
20221 (migrate_btr_def): Likewise.
20222 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
20223 * df-core.c (loop_post_order_compute): Likewise.
20224 (loop_inverted_post_order_compute): Likewise.
20225 * hsa-common.h: Likewise.
20226 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
20227 * init-regs.c (initialize_uninitialized_regs): Likewise.
20228 * ipa-inline.c (resolve_noninline_speculation): Likewise.
20229 (inline_small_functions): Likewise.
20230 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
20231 * ira.c (combine_and_move_insns): Likewise.
20232 (build_insn_chain): Likewise.
20233 * loop-invariant.c (find_invariants): Likewise.
20234 * lower-subreg.c (propagate_pseudo_copies): Likewise.
20235 * predict.c (tree_predict_by_opcode): Likewise.
20236 (predict_paths_leading_to): Likewise.
20237 (predict_paths_leading_to_edge): Likewise.
20238 (estimate_loops_at_level): Likewise.
20239 (estimate_loops): Likewise.
20240 * shrink-wrap.c (try_shrink_wrapping): Likewise.
20241 (spread_components): Likewise.
20242 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
20243 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
20244 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
20245 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
20246 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
20247 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
20248 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
20249 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
20250 (mark_threaded_blocks): Likewise.
20251 (thread_through_all_blocks): Likewise.
20252 * tree-ssa.c (verify_ssa): Likewise.
20253 (execute_update_addresses_taken): Likewise.
20254 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
20256 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20258 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
20260 (post_order_compute): Likewise.
20261 (inverted_post_order_compute): Likewise.
20262 (pre_and_rev_post_order_compute_fn): Likewise.
20264 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20266 * genrecog.c (int_set::int_set): Explicitly construct our
20267 auto_vec base class.
20268 * vec.h (auto_vec::auto_vec): New constructor.
20270 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20272 * bitmap.h (class auto_bitmap): New constructor taking
20273 bitmap_obstack * argument.
20275 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20277 * bitmap.h (class auto_bitmap): Change type of m_bits to
20278 bitmap_head, and adjust ctor / dtor and member operators.
20280 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
20282 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
20283 when returned register mode doesn't match original mode.
20285 2017-05-12 Jeff Law <law@redhat.com>
20286 Jakub Jelinek <jakub@redhat.com>
20288 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
20289 we look for cc setter after the compare-elim changes.
20290 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
20291 within the vector to match what compare-elim now expects.
20292 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
20293 (xorsi3_flags, one_cmplsi2_flags): Likewise.
20295 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
20296 after the compare-elim changes.
20297 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
20298 the vector to match what compare-elim now expects.
20299 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
20300 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
20301 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
20302 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
20303 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
20305 * config/visium/visium.c (single_set_and_flags): Fix where
20306 we look for cc setter after the compare-elim changes.
20307 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
20308 with the vector to match what compare-elim now expects.
20309 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
20310 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
20311 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
20312 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
20313 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
20314 (neg<mode>2_insn_set_overflow): Likewise.
20316 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
20318 PR middle-end/79794
20319 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
20320 maybe_expand_insn call, set ops[0].target. If still set after call,
20321 set alt_rtl. Add extra arg to recursive calls.
20322 (extract_bit_field): Add alt_rtl argument. Pass to
20324 * expmed.h (extract_bit_field): Fix prototype.
20325 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
20326 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
20327 to extract_bit_field_calls.
20328 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
20329 Pass alt_rtl to extract_bit_field calls.
20330 * calls.c (store_unaligned_arguments_into_psuedos)
20331 load_register_parameters): Pass extra NULL to extract_bit_field calls.
20332 * optabs.c (maybe_legitimize_operand): Clear op->target when call
20334 * optabs.h (struct expand_operand): Add target bitfield.
20336 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20338 * compare-elim.c (try_eliminate_compare): Canonicalize
20339 operation with embedded compare to
20340 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
20341 (set (reg) (operation)].
20343 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
20345 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20348 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
20349 cost of adding a carry flag for ADC instruction.
20350 [case MINUS]: Ignore the cost of subtracting a carry flag
20351 for SBB instruction.
20353 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
20355 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
20357 * config/rs6000/bmiintrin.h: New file.
20358 * config/rs6000/bmi2intrin.h: New file.
20359 * config/rs6000/x86intrin.h: New file.
20361 2017-05-12 Jeff Law <law@redhat.com>
20363 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
20366 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
20368 PR middle-end/80707
20369 * tree-cfg.c: Remove cfg edges of unreachable case statements.
20371 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20373 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20374 early expansion of vector divide builtins.
20375 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
20376 builtins identified as having unsigned arguments.
20378 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20380 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
20381 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
20382 expansion of vector logical operations (and, andc, or, xor,
20385 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20387 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
20388 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
20390 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20392 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20393 early GIMPLE expansion of vector multiplies.
20395 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20397 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
20398 TARGET_HAVE_MOVT conditional.
20399 (movt splitter): Likewise.
20401 2017-05-12 Richard Biener <rguenther@suse.de>
20403 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
20404 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20405 Fold all stmts not inplace.
20407 2017-05-12 Richard Biener <rguenther@suse.de>
20409 PR tree-optimization/80713
20410 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
20411 inserted_exprs bit for not removed stmts.
20413 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
20415 PR middle-end/69921
20416 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
20417 parallelized" attribute for parallelized OpenACC kernels.
20418 * omp-offload.c (execute_oacc_device_lower): Use it.
20420 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
20421 Set "oacc kernels" attribute.
20422 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
20423 parameter. Adjust all users.
20424 (oacc_fn_attrib_kernels_p): Remove function.
20425 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
20426 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
20427 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
20428 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
20429 assert "oacc kernels" attribute is set.
20431 2017-05-11 Carl Love <cel@us.ibm.com>
20433 * config/rs6000/rs6000-c: Add support for built-in functions
20434 vector unsigned char vec_popcnt (vector signed char)
20435 vector unsigned char vec_popcnt (vector unsigned char)
20436 vector unsigned short vec_popcnt (vector signed short)
20437 vector unsigned short vec_popcnt (vector unsigned short)
20438 vector unsigned int vec_popcnt (vector signed int)
20439 vector unsigned int vec_popcnt (vector unsigned int)
20440 vector unsigned long long vec_popcnt (vector signed long long)
20441 vector unsigned long long vec_popcnt (vector unsigned long long)
20442 vector signed long long vec_slo (vector signed long long,
20443 vector signed char)
20444 vector signed long long vec_slo (vector signed long long,
20445 vector unsigned char)
20446 vector unsigned long long vec_slo (vector unsigned long long,
20447 vector signed char)
20448 vector unsigned long long vec_slo (vector unsigned long long,
20449 vector unsigned char)
20450 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
20451 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
20452 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
20453 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
20454 * doc/extend.texi: Update the built-in documentation file for the
20455 new built-in functions.
20457 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20459 * attribs.h (sorted_attr_string): Move machine independent
20460 functions for target clone support from the i386 port to common
20461 code. Rename ix86_function_versions to common_function_versions.
20462 Rename make_name to make_unique_name.
20463 (common_function_versions): Likewise.
20464 (make_unique_name): Likewise.
20465 (make_dispatcher_decl): Likewise.
20466 (is_function_default_version): Likewise.
20467 * attribs.c (attr_strcmp): Likewise.
20468 (sorted_attr_string): Likewise.
20469 (common_function_versions): Likewise.
20470 (make_unique_name): Likewise.
20471 (make_dispatcher_decl): Likewise.
20472 (is_function_default_version): Likewise.
20473 * config/i386/i386.c (attr_strcmp): Likewise.
20474 (sorted_attr_string): Likewise.
20475 (ix86_function_versions): Likewise.
20476 (make_name): Likewise.
20477 (make_dispatcher_decl): Likewise.
20478 (is_function_default_version): Likewise.
20479 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
20481 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20484 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20485 Account for direct move costs for vec_construct of integer
20488 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
20491 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
20492 (UNSPEC_STX_ATOMIC): Ditto.
20493 (loaddi_via_sse): New insn.
20494 (storedi_via_sse): Ditto.
20495 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
20496 Update corresponding peephole2 patterns.
20497 (atomic_storedi_fpu): Ditto.
20499 2017-05-11 Julia Koval <julia.koval@intel.com>
20501 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
20502 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
20504 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
20505 (__builtin_ia32_rsqrt14ss_mask): New builtins.
20506 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
20508 2017-05-11 Nathan Sidwell <nathan@acm.org>
20510 * graphite-poly.c: Include dumpfile.h.
20512 * dumpfle.h (dump_function): Declare here ...
20513 * tree-dump.h (dump_function): ... not here.
20514 * dumpfile.c: #include tree-cfg.h.
20515 (dump_function): Move here from ...
20516 * tree-dump.c (dump_function): ... here.
20517 * gimplify.c: #include splay-tree.h, not tree-dump.h.
20518 * graphite-poly.c: Don't include tree-dump.h.
20519 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
20520 * print-tree.c: Likewise.
20521 * stor-layout.c: Likewise.
20522 * tree-nested.c: Likewise.
20524 * dumpfile.c (dump_start): Use TDF_FLAGS.
20525 (dump_enable_all): Fix TDF_KIND check thinko.
20527 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20529 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20530 array entries to represent two legal parameterizations of the
20531 overloaded __builtin_cmpb function, as represented by the
20532 P6_OV_BUILTIN_CMPB constant.
20533 (altivec_resolve_overloaded_builtin): Add special case handling
20534 for the __builtin_cmpb function, as represented by the
20535 P6_OV_BUILTIN_CMPB constant.
20536 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
20537 (BU_P6_64BIT_2): New macro.
20538 (BU_P6_OVERLOAD_2): New macro
20539 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
20540 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
20541 (CMPB): Add overload support to represent both 32-bit and 64-bit
20542 compare-bytes function.
20543 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
20544 support for TARGET_CMPB.
20545 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
20546 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20547 documentation of the __builtin_cmpb overloaded built-in function.
20549 2017-05-11 Richard Biener <rguenther@suse.de>
20551 PR tree-optimization/80705
20552 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
20553 bases are not vectorizable.
20555 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20557 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
20558 when counting register pressure.
20560 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20562 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
20563 (iv_ca_more_deps): Renamed to ...
20564 (iv_ca_compare_deps): ... this.
20565 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
20567 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20569 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
20571 (determine_group_iv_costs): ... here.
20572 (find_inv_vars_cb): Record inv var if it's not recorded before.
20574 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20576 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
20577 (get_shiftadd_cost): Ditto.
20579 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20581 * tree-ssa-address.c: Include header file.
20582 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
20584 (add_to_parts): Refactor.
20585 (addr_to_parts): New parameter. Update use of move_hint_to_base.
20586 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
20589 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20591 PR tree-optimization/53090
20592 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
20594 (extract_cond_operands): Detect condition with IV on both sides
20595 and return COMP_IV_EXPR_2.
20596 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
20597 (rewrite_use_compare): Simplify by removing call to function
20598 extract_cond_operands.
20600 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20602 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
20603 (extract_cond_operands): Detect condition comparing against non-
20604 invariant bound and return appropriate enum value.
20605 (find_interesting_uses_cond): Update use of extract_cond_operands.
20606 Handle its return value accordingly.
20607 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
20609 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20611 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
20612 nonlinear iv_use computation in loop invariant sensitive way.
20614 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20616 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
20617 (find_iv_candidates): Call relate_compare_use_with_all_cands.
20619 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20621 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
20622 (dump_cand): Support iv_cand.inv_exprs.
20623 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
20625 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
20628 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20630 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
20632 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
20633 as local function. Include necessary header files.
20634 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
20636 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20638 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
20640 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20642 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
20643 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
20644 RSHIFT_EXPR and BIT_NOT_EXPR.
20646 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20648 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
20649 (adjust_setup_cost): New parameter supporting round up adjustment.
20650 (struct address_cost_data): Delete.
20651 (force_expr_to_var_cost): Don't bound cost with spill_cost.
20652 (split_address_cost, ptr_difference_cost): Delete.
20653 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
20654 (struct ainc_cost_data): New struct.
20655 (get_address_cost_ainc): New function.
20656 (get_address_cost, get_computation_cost): Reimplement.
20657 (determine_group_iv_cost_address): Record inv_expr for all uses of
20659 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
20660 (iv_ca_has_deps): Reimplemented to ...
20661 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
20663 (iv_ca_extend): Call iv_ca_more_deps.
20665 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20667 * tree-ssa-address.c (struct mem_address): Move to header file.
20668 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
20669 * tree-ssa-address.h (struct mem_address): Move from C file.
20670 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
20672 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20674 * tree-affine.h (aff_combination_type): New interface.
20675 (aff_combination_zero_p): Remove static.
20676 (aff_combination_const_p): New interface.
20677 (aff_combination_singleton_var_p): New interfaces.
20679 2017-05-11 Richard Biener <rguenther@suse.de>
20681 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20682 Skip unreachable blocks and destinations.
20683 (eliminate): Move stmt removal and fixup ...
20684 (fini_eliminate): ... here. Skip inserted exprs.
20685 (pass_pre::execute): Move fini_pre after fini_eliminate.
20686 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
20687 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
20688 PRE to get rid of dead code that has invalid SSA form and
20689 split critical edges again.
20691 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20693 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20695 2017-05-11 Richard Biener <rguenther@suse.de>
20697 * passes.c (execute_function_todo): Verify loops if they are
20698 said to be up-to-date.
20699 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
20700 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
20702 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
20705 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
20706 handle calling assemble_external ourself.
20709 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
20710 modes with zero size. Enhance comment.
20712 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20714 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
20715 built-ins for vec_xl and vec_xst with short and char pointer
20718 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20720 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
20721 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
20722 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
20723 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
20724 (_mm_maskz_min_round_ss): New intrinsics.
20725 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
20726 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
20727 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
20728 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
20729 (__builtin_ia32_minss_mask_round): New builtins.
20730 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20731 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20732 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
20734 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
20735 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
20737 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
20740 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20742 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
20743 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
20744 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
20745 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
20746 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
20747 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
20748 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
20749 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20750 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20751 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
20752 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
20753 (__builtin_ia32_mulss_mask_round): New builtins.
20754 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20755 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20756 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
20758 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
20759 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20761 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20764 2017-05-10 Julia Koval <julia.koval@intel.com>
20766 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
20767 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
20768 (_mm256_setr_m128i): New intrinsics.
20770 2017-05-10 Julia Koval <julia.koval@intel.com>
20772 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
20773 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
20774 (_mm_maskz_rcp14_ss): New intrinsics.
20775 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
20776 (__builtin_ia32_rcp14ss_mask): New builtins.
20777 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
20779 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
20781 PR tree-optimization/51513
20782 * tree-cfg.c (gimple_seq_unreachable_p): New function.
20783 (assert_unreachable_fallthru_edge_p): Use it.
20784 (group_case_labels_stmt): Likewise.
20785 * tree-cfg.h: Prototype it.
20786 * stmt.c: Include cfghooks.h and tree-cfg.h.
20787 (emit_case_dispatch_table) <gap_label>: New local variable.
20788 Use it to fill dispatch table gaps.
20789 Test for default_label before updating probabilities.
20790 (expand_case) <default_label>: Remove unneeded initialization.
20791 Test for unreachable default case statement and remove its edge.
20792 Set default_label accordingly.
20793 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
20795 2017-05-10 Carl Love <cel@us.ibm.com>
20797 * config/rs6000/rs6000-c: Add support for built-in functions
20798 vector signed char vec_neg (vector signed char)
20799 vector signed short int vec_neg (vector short int)
20800 vector signed int vec_neg (vector signed int)
20801 vector signed long long vec_neg (vector signed long long)
20802 vector float vec_neg (vector float)
20803 vector double vec_neg (vector double)
20804 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
20806 * config/rs6000/altivec.h: Add define for vec_neg
20807 * doc/extend.texi: Update the built-in documentation for the
20808 new built-in functions.
20810 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20812 PR tree-optimization/77644
20813 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
20815 2017-05-10 Nathan Sidwell <nathan@acm.org>
20817 * dumpfile.h (TDI_lang_all): New.
20818 (TDF_KIND): New. Renumber others
20819 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
20821 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
20823 (get_dump_file_name): Adjust suffix generation.
20824 (dump_enable_all): Use TDF_KIND.
20825 * doc/invoke.texi (-fdump-lang-all): Document.
20827 * dumpfile.h: Tabify.
20829 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
20832 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
20833 Move member access before delete.
20835 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
20837 * tree-inline.c (expand_call_inline): Split block at stmt
20840 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
20843 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
20844 are now unused after splitting mov{sf,sd}_hardfloat.
20845 (f32_lr2): Likewise.
20846 (f32_lm): Likewise.
20847 (f32_lm2): Likewise.
20848 (f32_li): Likewise.
20849 (f32_li2): Likewise.
20850 (f32_lv): Likewise.
20851 (f32_sr): Likewise.
20852 (f32_sr2): Likewise.
20853 (f32_sm): Likewise.
20854 (f32_sm2): Likewise.
20855 (f32_si): Likewise.
20856 (f32_si2): Likewise.
20857 (f32_sv): Likewise.
20858 (f32_dm): Likewise.
20859 (f32_vsx): Likewise.
20860 (f32_av): Likewise.
20861 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
20862 For movsf, order stores so the VSX stores occur before the GPR
20863 store which encourages the register allocator to use a traditional
20864 FPR instead of a GPR. For movsd, order the stores so that the GPR
20865 store comes before the VSX stores to allow the power6 to work.
20866 This is due to the power6 not having a 32-bit integer store
20867 instruction from a FPR.
20868 (movsf_hardfloat): Likewise.
20869 (movsd_hardfloat): Likewise.
20871 2017-05-09 Martin Sebor <msebor@redhat.com>
20873 PR translation/80280
20874 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
20877 PR translation/80280
20878 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
20879 data member added in r247778.
20880 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
20882 2017-05-09 Nathan Sidwell <nathan@acm.org>
20884 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
20886 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
20889 2017-05-09 Marek Polacek <polacek@redhat.com>
20891 * doc/invoke.texi: Fix typo.
20893 2017-05-09 Richard Biener <rguenther@suse.de>
20895 * tree-vrp.c (vrp_val_is_max): Adjust comment.
20896 (vrp_val_is_min): Likewise.
20897 (set_value_range_to_value): Likewise.
20898 (set_value_range_to_nonnegative): Likewise.
20899 (gimple_assign_nonzero_p): Likewise.
20900 (gimple_stmt_nonzero_p): Likewise.
20901 (vrp_int_const_binop): Likewise. Remove unreachable case.
20902 (adjust_range_with_scev): Adjust comments.
20903 (compare_range_with_value): Likewise.
20904 (extract_range_from_phi_node): Likewise.
20905 (test_for_singularity): Likewise.
20907 2017-05-09 Richard Biener <rguenther@suse.de>
20909 * tree-vrp.c (get_single_symbol): Add assert that we don't
20910 get overflowed constants as invariant part.
20911 (compare_values_warnv): Add comment before the TREE_NO_WARNING
20912 checks. Use wi::cmp instead of recursing for integer constants.
20913 (compare_values): Just ignore whether we assumed undefined
20914 overflow instead of failing the compare.
20915 (extract_range_for_var_from_comparison_expr): Add comment before the
20916 TREE_NO_WARNING sets.
20917 (test_for_singularity): Likewise.
20918 (extract_range_from_comparison): Do not disable optimization
20919 when we assumed undefined overflow.
20920 (extract_range_basic): Remove init of unused var.
20922 2017-05-09 Richard Biener <rguenther@suse.de>
20924 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
20925 (extract_range_from_multiplicative_op_1): Adjust.
20926 (extract_range_from_binary_expr_1): Use int_const_binop.
20928 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20931 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
20932 rs6000_store_data_bypass_p in seven define_bypass directives and
20933 in several comments.
20934 * config/rs6000/rs6000-protos.h: Add prototype for
20935 rs6000_store_data_bypass_p function.
20936 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
20937 function implements slightly different (rs6000-specific) semantics
20938 than store_data_bypass_p, returning false rather than aborting
20939 with assertion error when arguments do not satisfy the
20940 requirements of store data bypass.
20941 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
20942 rs6000_store_data_bypass_p.
20944 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
20946 * config/xtensa/xtensa-protos.h
20947 (xtensa_initial_elimination_offset): New declaration.
20948 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20949 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
20950 macro definition, add case for FRAME_POINTER_REGNUM when
20951 FRAME_GROWS_DOWNWARD.
20952 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
20953 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
20954 xtensa_initial_elimination_offset.
20956 2017-05-08 Nathan Sidwell <nathan@acm.org>
20958 * doc/invoke.texi: Alphabetize -fdump options.
20960 2017-05-08 Martin Sebor <msebor@redhat.com>
20962 PR translation/80280
20963 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
20965 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
20967 * target.def (compute_frame_layout): New optional target hook.
20968 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
20969 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
20970 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
20972 * reload1.c (verify_initial_elim_offsets): Likewise.
20973 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
20974 (use_simple_return_p): Call arm_compute_frame_layout if needed.
20975 (arm_get_frame_offsets): Split up into this ...
20976 (arm_compute_frame_layout): ... and this function.
20978 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
20980 * config/aarch64/constraints.md (Usa): New constraint.
20981 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
20983 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
20985 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
20986 with_multilib_list after it has been checked.
20988 2017-05-08 Richard Biener <rguenther@suse.de>
20990 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
20991 (bitmap_set_subtract_values): Likewise.
20993 2017-05-08 Richard Biener <rguenther@suse.de>
20995 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
20996 (gimple_assign_nonzero): ... this and remove strict_overflow_p
20998 (gimple_stmt_nonzero_warnv_p): Rename to ...
20999 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21001 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21002 (extract_range_basic): Adjust, do not disable propagation on
21003 strict overflow sensitive simplification.
21004 (vrp_visit_cond_stmt): Likewise.
21006 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21008 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21009 body size unconditionally.
21011 2017-05-07 Jeff Law <law@redhat.com>
21014 2017-05-06 Jeff Law <law@redhat.com>
21015 PR tree-optimization/78496
21016 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21019 PR tree-optimization/78496
21020 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21021 (simplify_stmt_using_ranges): Call it.
21022 (vrp_dom_walker::before_dom_children): Extract equivalences
21023 from an ASSERT_EXPR with an equality comparison against a
21026 2017-05-06 Jeff Law <law@redhat.com>
21028 PR tree-optimization/78496
21029 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21032 PR tree-optimization/78496
21033 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21034 (simplify_stmt_using_ranges): Call it.
21035 (vrp_dom_walker::before_dom_children): Extract equivalences
21036 from an ASSERT_EXPR with an equality comparison against a
21039 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21041 * lra-constraints.c (lra_copy_reg_equiv): New function.
21042 (split_reg): Use it to copy equivalence information from the
21043 original register to the spill register.
21045 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21047 PR rtl-optimization/75964
21048 * simplify-rtx.c (simplify_const_relational_operation): Remove
21049 invalid handling of comparisons of integer ABS.
21051 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21053 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21054 initialize to zero.
21055 (init_regs): Remove declaration.
21056 (function_arg_advance_32): Initialize error_p as boolean variable.
21058 2017-05-05 Nathan Sidwell <nathan@acm.org>
21060 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21061 lines. Use for (;;).
21063 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21065 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21066 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21067 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21068 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21069 VF=2 that require versioning.
21071 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21073 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21076 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21078 * diagnostic.h (diagnostic_override_option_index): Convert from
21079 macro to inline function.
21081 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21083 * diagnostic.c (last_module_changed_p): New function.
21084 (set_last_module): New function.
21085 (diagnostic_report_current_module): Convert macro usage to
21086 the above functions.
21087 * diagnostic.h (diagnostic_context::last_module): Strengthen
21088 from const line_map * to const line_map_ordinary *.
21089 (diagnostic_last_module_changed): Delete macro.
21090 (diagnostic_set_last_module): Delete macro.
21092 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21094 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
21095 with diagnostic_report_diagnostic.
21096 (diagnostic_n_impl_richloc): Likewise.
21097 * diagnostic.h (report_diagnostic): Delete macro.
21098 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
21099 with diagnostic_report_diagnostic.
21100 * substring-locations.c (format_warning_va): Likewise.
21102 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21104 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
21105 save/restor of format_spec. Move option-printing code to...
21106 (print_option_information): ...this new function, and
21107 reimplement by simply printing to the pretty_printer,
21108 rather than appending to the format string.
21110 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21112 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
21113 handling logic into...
21114 (update_effective_level_from_pragmas): ...this new function.
21116 2017-05-04 Andrew Waterman <andrew@sifive.com>
21118 * config/riscv/riscv.opt (mstrict-align): New option.
21119 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
21120 (SLOW_UNALIGNED_ACCESS): Define.
21121 (riscv_slow_unaligned_access): Declare.
21122 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
21124 (riscv_slow_unaligned_access): New variable.
21125 (rocket_tune_info): Set slow_unaligned_access to true.
21126 (optimize_size_tune_info): Set slow_unaligned_access to false.
21127 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
21128 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
21129 (riscv_option_override): Set riscv_slow_unaligned_access.
21130 * doc/invoke.texi: Add -mstrict-align to RISC-V.
21132 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
21134 * config/riscv/riscv.md: Unify indentation.
21136 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21141 * config/rs6000/rs6000.md (u code attribute): Add FIX and
21143 (extendsi<mode>2): Add support for doing sign extension via
21144 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
21145 don't have ISA 3.0 instructions.
21146 (extendsi<mode>2 splitter): Likewise.
21147 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
21148 generate the normal insns since SImode can now go in vector
21149 registers. Disallow the special UNSPECs needed for previous
21150 machines to hide SImode being used. Add new insns
21151 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
21152 (fix_trunc<mode>si2_stfiwx): Likewise.
21153 (fix_trunc<mode>si2_internal): Likewise.
21154 (fixuns_trunc<mode>si2): Likewise.
21155 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21156 (fctiw<u>z_<mode>_smallint): Likewise.
21157 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
21158 of floating point to 32-bit integer from doing a direct move to
21159 the GPR registers to do a store.
21160 (fctiwz_<mode>): Break long line.
21162 2017-05-05 Bin Cheng <bin.cheng@arm.com>
21164 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
21165 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
21166 (addr_list, addr_offset_valid_p): New.
21167 (split_address_groups): Check offset validity with above function.
21168 (gt-tree-ssa-loop-ivopts.h): Include header file.
21170 2017-05-05 Nathan Sidwell <nathan@acm.org>
21172 * config.gcc (arm*-*-*): Add missing 'fi'.
21174 2017-05-05 Steve Ellcey <sellcey@cavium.com>
21176 * doc/invoke.texi (-fopt-info): Explicitly say order of options
21177 included in -fopt-info does not matter.
21178 * doc/optinfo.texi (-fopt-info): Fix description of default
21179 behavour. Explicitly say order of options included in -fopt-info
21182 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21184 * config.gcc: Allow combinations of aprofile and rmprofile values for
21185 --with-multilib-list.
21186 * config/arm/t-multilib: New file.
21187 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
21188 variables. Remove setting of ISA and floating-point ABI in
21189 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
21190 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
21191 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
21192 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
21194 * config/arm/t-rmprofile: Likewise except for the matches changes.
21195 * doc/install.texi (--with-multilib-list): Document the combination of
21196 aprofile and rmprofile values and warn about pitfalls in doing that.
21198 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21200 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
21201 (movdi_aarch64): Likewise.
21203 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21205 PR tree-optimization/80632
21206 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
21208 (build_arrays): Initialize it for virtual phis.
21209 (fix_phi_nodes): Use it for virtual phis.
21211 PR tree-optimization/80558
21212 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
21213 [x, y] op z into [x op, y op z] for op & or | if conditions
21216 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
21217 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21220 * config/arm/arm.md (use_literal_pool): Remove.
21221 (64-bit immediate split): No longer takes cost into consideration
21222 if arm_disable_literal_pool is enabled.
21223 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
21224 used when arm_disable_literal_pool is enabled.
21225 (arm_max_const_double_inline_cost): Remove use of
21226 arm_disable_literal_pool.
21227 (push_minipool_fix): Add assert.
21228 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
21229 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
21230 (no_literal_pool_sf_immediate): New.
21232 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21234 PR tree-optimization/80613
21235 * tree-ssa-dce.c (propagate_necessity): Remove cases for
21236 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
21238 2017-05-05 Richard Biener <rguenther@suse.de>
21240 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
21242 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
21244 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
21245 of this flag from insn conditions due to removal from r247495.
21247 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21249 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
21251 (arm_early_store_addr_dep_ptr): Likewise.
21252 * config/arm/aarch-common-protos.h
21253 (arm_early_load_addr_dep_ptr): Add prototype.
21254 (arm_early_store_addr_dep_ptr): Likewise.
21255 * config/arm/cortex-a53.md: Add new bypasses.
21257 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21259 * tree.c (next_type_uid): Change type to unsigned.
21260 (type_hash_canon): Decrement back next_type_uid if
21261 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
21262 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
21265 2017-05-04 Martin Sebor <msebor@redhat.com>
21267 * builtins.c: Fix a trivial typo in a comment.
21269 PR middle-end/79234
21270 * builtins.c (check_sizes): Adjust to handle reading past the end.
21271 Avoid printing excessive upper bound of ranges. Use %E to print
21272 tree nodes instead of converting them to %wu.
21273 (expand_builtin_memchr): New function.
21274 (compute_dest_size): Rename...
21275 (compute_objsize): ...to this.
21276 (expand_builtin_memcpy): Adjust.
21277 (expand_builtin_mempcpy): Adjust.
21278 (expand_builtin_strcat): Adjust.
21279 (expand_builtin_strcpy): Adjust.
21280 (check_strncat_sizes): Adjust.
21281 (expand_builtin_strncat): Adjust.
21282 (expand_builtin_strncpy): Adjust and simplify.
21283 (expand_builtin_memset): Adjust.
21284 (expand_builtin_bzero): Adjust.
21285 (expand_builtin_memcmp): Adjust.
21286 (expand_builtin): Handle memcmp.
21287 (maybe_emit_chk_warning): Check strncat just once.
21289 2017-05-04 Martin Sebor <msebor@redhat.com>
21291 PR preprocessor/79214
21292 PR middle-end/79222
21293 PR middle-end/79223
21294 * builtins.c (check_sizes): Add inlining context and issue
21295 warnings even when -Wno-system-headers is set.
21296 (check_strncat_sizes): Same.
21297 (expand_builtin_strncat): Same.
21298 (expand_builtin_memmove): New function.
21299 (expand_builtin_stpncpy): Same.
21300 (expand_builtin): Handle memmove and stpncpy.
21302 2017-05-04 Bin Cheng <bin.cheng@arm.com>
21304 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
21305 which is not used any more.
21307 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21309 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
21311 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21313 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
21314 (cortexa53_tunings): Likewise.
21315 (cortexa57_tunings): Likewise.
21316 (cortexa72_tunings): Likewise.
21317 (cortexa73_tunings): Likewise.
21319 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21321 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
21322 Set loop alignment to 8.
21324 2017-05-04 Martin Sebor <msebor@redhat.com>
21326 PR translation/80280
21327 * builtins.c (expand_builtin_object_size): Add missing quoting to
21328 %D and like directives.
21329 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
21330 (hsa_type_for_tree_type): Same.
21331 (verify_function_arguments): Same.
21332 * symtab.c (symbol_table::change_decl_assembler_name): Same.
21333 * varasm.c (get_section): Same.
21336 2017-05-04 Martin Sebor <msebor@redhat.com>
21338 PR translation/80280
21339 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
21341 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21343 * config/aarch64/aarch64.c (generic_addrcost_table):
21344 Change HI/TI mode setting.
21346 2017-05-04 Martin Jambor <mjambor@suse.cz>
21348 PR tree-optimization/80622
21349 * tree-sra.c (comes_initialized_p): New function.
21350 (build_accesses_from_assign): Only set write lazily when
21351 comes_initialized_p is false.
21352 (analyze_access_subtree): Use comes_initialized_p.
21353 (propagate_subaccesses_across_link): Assert !comes_initialized_p
21354 instead of testing for PARM_DECL.
21356 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21358 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
21359 constraint on operand 0 to allow more general addressing modes.
21360 Adjust output template.
21361 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
21363 * config/aarch64/aarch64-protos.h
21364 (aarch64_address_valid_for_prefetch_p): Declare prototype.
21365 * config/aarch64/constraints.md (Dp): New address constraint.
21366 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
21369 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
21371 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
21372 update use of estimate_ipcp_clone_size_and_time.
21373 (estimate_local_effects): Update use of
21374 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
21375 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
21376 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
21377 Return nonspecialized time.
21379 2017-05-04 Richard Biener <rguenther@suse.de>
21381 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
21382 for the last VUSE which def dominates the PHI. Directly call
21384 (get_continuation_for_phi_1): Remove.
21386 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
21388 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
21389 to explain the use of truncating division. Cap the number of
21390 iterations to the maximum given by nb_iterations_upper_bound,
21393 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21395 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
21396 * configure: Regenerate.
21397 * config.in: Regenerate.
21398 * config/i386/driver-mingw32.c: new file.
21399 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
21400 * config.host: Link driver-mingw32.o on MinGW host.
21401 * doc/install.texi: Document new --enable-mingw-wildcard configure
21404 2017-05-04 Marek Polacek <polacek@redhat.com>
21406 PR tree-optimization/80612
21407 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
21409 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21410 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
21412 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
21413 (movt splitter): Likewise.
21414 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
21415 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
21416 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
21417 block for Thumb-1 with MOVT.
21418 (thumb2_legitimate_address_p): Move code block ...
21419 (can_avoid_literal_pool_for_label_p): ... into this new function.
21420 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
21422 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
21423 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
21424 "M-profile targets with the MOVT instruction".
21426 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21428 * config/arm/arm-builtins.c (arm_init_builtins): Rename
21429 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
21430 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
21432 2017-05-04 Martin Liska <mliska@suse.cz>
21434 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
21435 variable cond_code.
21437 2017-05-04 Richard Biener <rguenther@suse.de>
21439 * tree.c (array_at_struct_end_p): Handle arrays at struct
21440 end with flexarrays more conservatively. Refactor and treat
21441 arrays of arrays or aggregates more strict. Fix
21442 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
21443 * tree.c (array_at_struct_end_p): Adjust prototype.
21444 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
21445 * gimple-fold.c (get_range_strlen): Likewise.
21446 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
21448 2017-05-04 Richard Biener <rguenther@suse.de>
21450 PR tree-optimization/31130
21451 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
21453 (supports_overflow_infinity): Likewise.
21454 (is_negative_overflow_infinity): Likewise.
21455 (is_positive_overflow_infinity): Likewise.
21456 (is_overflow_infinity): Likewise.
21457 (stmt_overflow_infinity): Likewise.
21458 (overflow_infinity_range_p): Likewise.
21459 (usable_range_p): Remove as always returning true.
21460 (make_overflow_infinity): Remove.
21461 (negative_overflow_infinity): Likewise.
21462 (positive_overflow_infinity): Likewise.
21463 (avoid_overflow_infinity): Likewise.
21464 (set_value_range): Adjust accordingly.
21465 (set_value_range_to_nonnegative): Likewise, remove now unused
21466 overflow_infinity arg.
21467 (vrp_operand_equal_p): Adjust.
21468 (update_value_range): Likewise.
21469 (range_int_cst_singleton_p): Likewise.
21470 (operand_less_p): Likewise.
21471 (compare_values_warnv): Likewise.
21472 (extract_range_for_var_from_comparison_expr): Likewise.
21473 (vrp_int_const_binop): Likewise.
21474 (zero_nonzero_bits_from_vr): Likewise.
21475 (extract_range_from_multiplicative_op_1): Likewise.
21476 (extract_range_from_binary_expr_1): Likewise.
21477 (extract_range_from_unary_expr): Likewise.
21478 (extract_range_from_comparison): Likewise.
21479 (extract_range_basic): Likewise.
21480 (adjust_range_with_scev): Likewise.
21481 (compare_ranges): Likewise.
21482 (compare_range_with_value): Likewise.
21483 (dump_value_range): Likewise.
21484 (test_for_singularity): Likewise, remove strict_overflow_p parameter
21486 (simplify_cond_using_ranges): Adjust.
21488 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21490 * brig-builtins.def: Added a builtin for class_f64.
21491 * builtin-types.def: Added a builtin type needed by class_f64.
21493 2017-05-03 Jason Merrill <jason@redhat.com>
21495 * timevar.def: Add TV_CONSTEXPR.
21497 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21499 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
21501 2017-05-03 Martin Jambor <mjambor@suse.cz>
21503 * ipa-prop.c (ipa_update_after_lto_read): Removed.
21504 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
21505 * ipa-cp.c (ipcp_propagate_stage): Do not call
21506 ipa_update_after_lto_read.
21507 * ipa-inline.c (ipa_inline): Likewise.
21509 2017-05-03 Martin Jambor <mjambor@suse.cz>
21511 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
21512 tag. Added a default constructor and a destructor.
21513 (ipa_edge_args_sum_t): New class;
21514 (ipa_edge_args_sum): Declare.
21515 (ipa_edge_args_vector): Remove declaration.
21516 (IPA_EDGE_REF): Use ipa_edge_args_sum.
21517 (ipa_free_edge_args_substructures): Remove declaration.
21518 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
21519 (ipa_edge_args_info_available_for_edge_p): Likewise.
21520 * ipa-prop.c (ipa_edge_args_vector): Removed.
21521 (edge_removal_hook_holder): Likewise.
21522 (edge_duplication_hook_holder): Likewise.
21523 (ipa_edge_args_sum): New variable.
21524 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
21525 ipa_edge_args_vector.
21526 (ipa_free_edge_args_substructures): Likewise.
21527 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
21528 ipa_edge_args_vector.
21529 (ipa_edge_removal_hook): Turned into method
21530 ipa_edge_args_sum_t::remove.
21531 (ipa_edge_duplication_hook): Turned into method
21532 ipa_edge_args_sum_t::duplicate.
21533 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
21534 registering edge hooks.
21535 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
21536 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
21537 ipa_edge_args_sum instead of ipa_edge_args_vector.
21538 * ipa-profile.c (ipa_profile): Likewise.
21540 2017-05-03 Martin Jambor <mjambor@suse.cz>
21542 * symbol-summary.h (function_summary): New method exists.
21543 (function_summary::symtab_removal): Deallocate through release.
21544 (call_summary): New class.
21545 (gt_ggc_mx): New overload.
21546 (gt_pch_nx): Likewise.
21547 (gt_pch_nx): Likewise.
21549 2017-05-03 Jeff Law <law@redhat.com>
21551 PR tree-optimization/78496
21552 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
21553 from simplify_cond_using_ranges. Split off code to walk
21554 backwards through casts into ...
21555 (simplify_cond_using_ranges_2): New function.
21556 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
21557 (execute_vrp): After identifying jump threads, call
21558 simplify_cond_using_ranges_2.
21560 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
21563 * ipa-inline.h (inline_summary): Add ctor.
21564 (create_ggc): Do not use ggc_cleared_alloc.
21566 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
21567 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21569 * gcc.c (handle_braces): Support escaping in switch matching
21571 * doc/invoke.texi (Spec Files): Document it.
21572 Remove superfluous @code markup in items.
21574 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21576 * diagnostic-show-locus.c (struct column_range): New struct.
21577 (get_affected_columns): New function.
21578 (get_printed_columns): New function.
21579 (struct correction): New struct.
21580 (correction::ensure_capacity): New function.
21581 (correction::ensure_terminated): New function.
21582 (struct line_corrections): New struct.
21583 (line_corrections::~line_corrections): New dtor.
21584 (line_corrections::add_hint): New function.
21585 (layout::print_trailing_fixits): Reimplement in terms of the new
21587 (selftest::test_overlapped_fixit_printing): New function.
21588 (selftest::diagnostic_show_locus_c_tests): Call it.
21590 2017-05-03 Nathan Sidwell <nathan@acm.org>
21592 Canonicalize canonical type hashing
21593 * tree.h (type_hash_canon_hash): Declare.
21594 * tree.c (type_hash_list, attribute_hash_list): Move into
21595 type_hash_canon_hash.
21596 (build_type_attribute_qual_variant): Break out hash code calc into
21597 type_hash_canon_hash.
21598 (type_hash_canon_hash): New. Generic type hash computation.
21599 (build_range_type_1, build_array_type_1, build_function_type,
21600 build_method_type_directly, build_offset_type, build_complex_type,
21601 make_vector_type): Call it.
21603 2017-05-03 Richard Biener <rguenther@suse.de>
21605 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
21606 When all DRs have unknown misaligned do not always peel
21607 when there is a store but apply the same costing model as if
21608 there were only loads.
21610 2017-05-03 Richard Biener <rguenther@suse.de>
21613 PR tree-optimization/80492
21614 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21615 compare_base_decls returning dont-know properly.
21617 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
21619 * config/arm/iterators.md (CCSI): New mode iterator.
21620 (arch): New mode attribute.
21621 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
21622 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
21623 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
21624 code iterator for success result mode.
21625 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
21626 the corresponding new insn generators.
21628 2017-05-03 Bin Cheng <bin.cheng@arm.com>
21631 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21632 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21634 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
21636 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
21637 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
21638 (DDR_A): Wrap DDR argument in brackets.
21639 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
21640 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
21641 (DDR_REVERSED_P): Likewise.
21643 2017-05-03 Jakub Jelinek <jakub@redhat.com>
21645 PR tree-optimization/79472
21646 * tree-switch-conversion.c (struct switch_conv_info): Add
21647 contiguous_range and default_case_nonstandard fields.
21648 (collect_switch_conv_info): Compute contiguous_range and
21649 default_case_nonstandard fields, don't clear final_bb if
21650 contiguous_range and only the default case doesn't have the required
21652 (check_all_empty_except_final): Set default_case_nonstandard instead
21653 of failing if contiguous_range and the default case doesn't have empty
21655 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
21656 and only the default case doesn't have the required constants. Skip
21658 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
21659 if default_case_nonstandard.
21660 (build_constructors): Build constant 1 just once. Assert that default
21661 values aren't inserted in between cases if contiguous_range. Skip
21663 (build_arrays): Skip virtual phis.
21664 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
21665 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
21666 Handle virtual phis.
21667 (gen_inbound_check): Handle default_case_nonstandard case.
21668 (process_switch): Adjust check_final_bb caller. Call
21669 gather_default_values with the first non-default case instead of
21670 default case if default_case_nonstandard.
21672 2017-05-02 Nathan Sidwell <nathan@acm.org>
21674 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
21675 check. Fix formatting.
21677 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
21679 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
21680 errors when comparing specialized and unspecialized times.
21682 2017-05-02 David Malcolm <dmalcolm@redhat.com>
21684 * diagnostic-show-locus.c
21685 (layout::should_print_annotation_line_p): Make private.
21686 (layout::print_annotation_line): Make private.
21687 (layout::annotation_line_showed_range_p): Make private.
21688 (layout::show_ruler): Make private.
21689 (layout::print_source_line): Make private. Pass in line and
21690 line_width, rather than calling location_get_source_line. Drop
21692 (layout::print_leading_fixits): New method.
21693 (layout::print_any_fixits): Rename to...
21694 (layout::print_trailing_fixits): ...this, and make private.
21695 Don't print newline fixits.
21696 (diagnostic_show_locus): Move logic for printing one row into...
21697 (layout::print_line): ...this new function. Move the
21698 location_get_source_line call and error-handling from
21699 print_source_line to here. Call print_leading_fixits, and rename
21700 print_any_fixits to print_trailing_fixits.
21701 (selftest::test_fixit_insert_containing_newline): Update now that
21702 newlines are partially supported.
21703 (selftest::test_fixit_insert_containing_newline_2): New test.
21704 (selftest::test_fixit_replace_containing_newline): Update comments.
21705 (selftest::diagnostic_show_locus_c_tests): Call the new test.
21706 * edit-context.c (class added_line): New class.
21707 (class edited_line): Describe newline handling in comment.
21708 (edited_line::actually_edited_p): New method.
21709 (edited_line::print_content): Delete redundant decl.
21710 (edited_line::m_predecessors): New field.
21711 (edited_file::print_content): Call edited_line::print_content.
21712 (edited_file::print_diff): Update to support newlines.
21713 (edited_file::print_diff_hunk): Likewise.
21714 (edited_file::print_run_of_changed_lines): New function.
21715 (edited_file::print_diff_line): Convert to...
21716 (print_diff_line): ...this.
21717 (edited_file::get_effective_line_count): New function.
21718 (edited_line::edited_line): Initialize new field m_predecessors.
21719 (edited_line::~edited_line): Clean up m_predecessors.
21720 (edited_line::apply_fixit): Handle newlines.
21721 (edited_line::get_effective_line_count): New function.
21722 (edited_line::print_content): New function.
21723 (edited_line::print_diff_lines): New function.
21724 (selftest::test_applying_fixits_insert_containing_newline): New
21726 (selftest::test_applying_fixits_replace_containing_newline): New
21728 (selftest::insert_line): New function.
21729 (selftest::test_applying_fixits_multiple_lines): Add example of
21731 (selftest::edit_context_c_tests): Call the new tests.
21733 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21735 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
21736 parameter cand. Update dump information.
21737 (get_computation_cost): Update uses.
21739 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21741 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
21742 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
21743 (get_computation_at, rewrite_use_address): Update use of
21744 get_computation_aff.
21746 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21748 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
21749 (get_computation): Delete.
21750 (get_computation_cost): Implement like get_computation_cost_at.
21751 Use get_computation_at.
21752 (get_computation_cost_at): Delete.
21753 (rewrite_use_nonlinear_expr): Use get_computation_at.
21754 (rewrite_use_compare, remove_unused_ivs): Ditto.
21756 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21758 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
21760 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21762 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
21763 (ivopts_global_cost_for_size): Rename parameter and update uses.
21764 (iv_ca_recount_cost): Update uses.
21765 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21766 candidates seperately in n_invs and n_cands.
21767 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
21769 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21771 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
21772 (find_inv_vars_cb): New.
21773 (find_depends): Renamed to ...
21774 (find_inv_vars): ... this.
21775 (add_candidate_1, force_var_cost): Call find_inv_vars.
21776 (split_address_cost, determine_group_iv_cost_cond): Ditto.
21778 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21780 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
21781 inv_vars. Add inv_exprs.
21782 (struct iv_cand): Rename depends_on to inv_vars.
21783 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
21784 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
21785 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
21786 (dump_cand): Dump inv_vars.
21787 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
21788 (record_invariant, find_depends, add_candidate_1): Ditto.
21789 (set_group_iv_cost, force_var_cost): Ditto.
21790 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
21791 (get_computation_cost_at, get_computation_cost): Ditto.
21792 (determine_group_iv_cost_generic): Ditto.
21793 (determine_group_iv_cost_address): Ditto.
21794 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
21795 (determine_group_iv_costs): Ditto.
21796 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
21797 (iv_ca_set_remove_invariants): Renamed to ...
21798 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
21799 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
21800 (iv_ca_set_add_invariants): Renamed to ...
21801 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
21802 (iv_ca_set_cp): Use iv_ca_set_add_invs.
21803 (iv_ca_has_deps): Support inv_vars and inv_exprs.
21804 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
21805 (create_new_ivs): Remove useless dump.
21807 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21809 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
21811 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
21812 (iv_ca_set_no_cp, create_new_iv): Ditto.
21814 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21816 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21818 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21820 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
21821 function tree_check2.
21823 2017-05-02 Martin Liska <mliska@suse.cz>
21825 * doc/gcov.texi: Add missing preposition.
21826 * gcov.c (function_info::function_info): Properly fill up
21827 all member variables.
21829 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21831 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
21833 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21835 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
21837 2017-05-02 Martin Liska <mliska@suse.cz>
21840 * lto-streamer-in.c (lto_read_tree_1): Remove
21841 LTO_STREAMER_DEBUG.
21842 * lto-streamer.c (struct tree_hash_entry): Likewise.
21843 (struct tree_entry_hasher): Likewise.
21844 (tree_entry_hasher::hash): Likewise.
21845 (tree_entry_hasher::equal): Likewise.
21846 (lto_streamer_init): Likewise.
21847 (lto_orig_address_map): Likewise.
21848 (lto_orig_address_get): Likewise.
21849 (lto_orig_address_remove): Likewise.
21850 * lto-streamer.h: Likewise.
21851 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
21852 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
21854 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
21856 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
21857 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
21858 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
21859 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
21860 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
21861 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
21862 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
21863 (mm_maskz_sub_ss): New intrinsics.
21864 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21865 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21866 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
21867 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
21868 (__builtin_ia32_subss_mask_round): New builtins.
21869 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21870 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21871 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
21873 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
21874 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21876 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21879 2017-05-02 Martin Jambor <mjambor@suse.cz>
21881 PR tree-optimization/78687
21882 * tree-sra.c (access): New field parent.
21883 (process_subtree_disqualification): New function.
21884 (disqualify_candidate): Call it.
21885 (build_accesses_from_assign): Reset write flag if creating an
21887 (build_access_subtree): Fill in parent field and also prpagate
21888 down grp_write flag.
21889 (create_artificial_child_access): New parameter set_grp_write, set
21890 grp_write to its value.
21891 (propagate_subaccesses_across_link): Also propagate grp_write flag
21893 (propagate_all_subaccesses): Push the closest parent back to work
21894 queue if add_access_to_work_queue returned true.
21896 2017-05-02 Richard Biener <rguenther@suse.de>
21898 * common.opt (fstrict-overflow): Alias negative to fwrapv.
21899 * doc/invoke.texi (fstrict-overflow): Remove all traces of
21900 -fstrict-overflow documentation.
21901 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
21902 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
21903 flag_strict_overflow.
21904 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
21905 * lto-opts.c (lto_write_options): Do not stream it.
21906 * lto-wrapper.c (merge_and_complain): Do not handle it.
21907 * opts.c (default_options_table): Do not set -fstrict-overflow.
21908 (finish_options): Likewise do not clear it when sanitizing.
21909 * simplify-rtx.c (simplify_const_relational_operation): Do not
21910 test flag_strict_overflow.
21912 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21914 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
21915 using enabled attribute.
21916 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
21917 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
21918 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
21919 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
21920 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
21921 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
21922 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
21923 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
21924 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
21925 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
21927 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21929 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
21931 2017-05-02 Richard Biener <rguenther@suse.de>
21933 PR tree-optimization/80591
21935 2017-04-10 Richard Biener <rguenther@suse.de>
21937 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21940 2017-05-02 Richard Biener <rguenther@suse.de>
21942 PR tree-optimization/80549
21943 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
21944 (cleanup_tree_cfg_noloop): Create forwarders to known loop
21945 headers if they do not have a preheader.
21947 2017-05-02 Martin Liska <mliska@suse.cz>
21950 * common.opt: Fix typo.
21951 * doc/invoke.texi: Likewise.
21953 2017-05-01 Jan Beulich <jbeulich@suse.com>
21955 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
21956 swapping, add (x,x,m,x,n) alternative.
21958 2017-05-01 Nathan Sidwell <nathan@acm.org>
21960 * calls.c (combine_pending_stack_adjustment_and_call): Remove
21961 unnecessary unadjusted_alignment check.
21963 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21966 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
21968 * gimplify.c (gimplify_cilk_detach): New function.
21969 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
21970 * tree-core.h: Document EXPR_CILK_SPAWN.
21971 * tree.h (EXPR_CILK_SPAWN): Define.
21973 2017-05-01 David Malcolm <dmalcolm@redhat.com>
21975 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
21976 to use new fixit_hint representation, using the "replace" logic.
21977 (get_line_span_for_fixit_hint): Likewise.
21978 (layout::print_any_fixits): Likewise.
21979 (selftest::test_one_liner_many_fixits): Rename to...
21980 (selftest::test_one_liner_many_fixits_1): ...this, and update
21981 comment and expected output to reflect that the multiple fix-it
21982 hints are now consolidated into one insertion.
21983 (selftest::test_one_liner_many_fixits_2): New test.
21984 (selftest::test_diagnostic_show_locus_one_liner): Update for
21986 (selftest::test_fixit_consolidation): Update for fix-it API
21988 * diagnostic.c (print_parseable_fixits): Likewise.
21989 * edit-context.c (edited_line::m_line_events): Convert from
21990 auto_vec <line_event *> to auto_vec <line_event>.
21991 (class line_event): Convert from abstract base class to a concrete
21992 class, taking over the role of replace_event.
21993 (class insert_event): Delete.
21994 (class replace_event): Rename to class line_event. Convert to
21996 (edit_context::add_fixits): Reimplement.
21997 (edit_context::apply_insert): Delete.
21998 (edit_context::apply_replace): Rename to...
21999 (edit_context::apply_fixit): ...this. Convert to half-open range.
22000 (edited_file::apply_insert): Delete.
22001 (edited_file::apply_replace): Rename to...
22002 (edited_file::apply_fixit): ...this.
22003 (edited_line::~edited_line): Drop deletion of events.
22004 (edited_line::apply_insert): Delete.
22005 (edited_line::apply_replace): Rename to...
22006 (edited_line::apply_fixit): ...this. Convert to half-open range.
22007 Update for change to type of m_line_events.
22008 * edit-context.h (edit_context::apply_insert): Delete.
22009 (edit_context::apply_replace): Rename to...
22010 (edit_context::apply_fixit): ...this.
22012 2017-05-01 Martin Sebor <msebor@redhat.com>
22014 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22017 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22020 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22021 __get_cpuid_max returns 0.
22022 (__get_cpuid_count): Ditto.
22024 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22026 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22027 replacement expression is another instance of one of its arguments.
22029 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22032 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22033 check for stack push/pop autoinc.
22034 * config/i386/i386.c (ix86_agi_dependent): Return false
22035 if the only reason why modified_in_p returned true is that
22036 addr is SP based and set_insn is a push or pop.
22038 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22040 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22043 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22046 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22047 (account_size_time): Use two predicates - exec_pred and
22049 (evaluate_conditions_for_known_args): Compute both clause and
22051 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22052 (inline_summary_t::duplicate): Update.
22053 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22055 (compute_inline_parameters): Likewise.
22056 (estimate_edge_size_and_time): Update caluclation of time.
22057 (estimate_node_size_and_time): Compute both time and nonspecialized
22059 (estimate_ipcp_clone_size_and_time): Update.
22060 (inline_merge_summary): Update.
22061 (do_estimate_edge_time): Update.
22062 (do_estimate_edge_size): Update.
22063 (do_estimate_edge_hints): Update.
22064 (inline_read_section, inline_write_summary): Stream both new predicates.
22065 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22067 (compute_inlined_call_time): Cleanup.
22068 (big_speedup_p): Update.
22069 (edge_badness): Update.
22070 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22071 (size_time_entry): Replace predicate by exec_predicate and
22072 nonconst_predicate.
22073 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22074 (estimate_edge_time): Return also nonspec_time.
22075 (reset_edge_growth_cache): Update.
22077 2017-04-29 Jakub Jelinek <jakub@redhat.com>
22079 PR rtl-optimization/80491
22080 * ifcvt.c (noce_process_if_block): When looking for x setter
22081 with missing else_bb, don't check only the insn right before
22082 cond_earliest, but look for the last insn that x is modified in
22083 within the same bb.
22085 PR rtl-optimization/80491
22086 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
22088 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
22090 PR tree-optimization/80487
22091 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
22093 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22095 PR tree-optimization/79697
22096 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
22097 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
22098 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
22100 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
22101 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
22103 2017-04-28 Martin Sebor <msebor@redhat.com>
22105 PR tree-optimization/80523
22106 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
22107 (init_target_to_host_charmap, target_to_host, target_strtol10): New
22109 (maybe_warn, format_directive, parse_directive): Use new functions.
22110 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
22112 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
22114 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
22116 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
22118 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
22119 target_header_dir): Set correctly.
22120 * configure: Regenerated.
22121 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22122 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22123 instead of SYSTEM_HEADER_DIR.
22125 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
22127 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
22128 (estimate_local_effects): Likewise.
22129 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
22130 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
22131 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
22132 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
22133 do_estimate_edge_time, estimate_edge_time): Likewise.
22134 * ipa-inline-analysis.c (estimate_node_size_and_time,
22135 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
22136 (estimate_time_after_inlining): Remove.
22138 2017-04-28 Martin Liska <mliska@suse.cz>
22140 * doc/gcov.texi: Enhance documentation of gcov.
22142 2017-04-28 Martin Liska <mliska@suse.cz>
22144 * doc/gcov.texi: Sort options in alphabetic order.
22145 * doc/gcov-dump.texi: Likewise.
22146 * doc/gcov-tool.texi: Likewise.
22147 * gcov.c (print_usage): Likewise.
22148 * gcov-dump.c (print_usage): Likewise.
22149 * gcov-tool.c (print_merge_usage_message): Likewise.
22150 (print_rewrite_usage_message): Likewise.
22151 (print_overlap_usage_message): Likewise.
22153 2017-04-28 Martin Liska <mliska@suse.cz>
22155 PR gcov-profile/53915
22156 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
22158 2017-04-28 Martin Liska <mliska@suse.cz>
22160 PR gcov-profile/79891
22161 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
22162 is marked by compiler as living on a line.
22163 (get_cycles_count): Remove usage of the union.
22164 (output_intermediate_file): Likewise.
22165 (find_source): Fix GNU coding style.
22166 (accumulate_line_counts): Remove old non-all block mode.
22167 (output_lines): Remove usage of the union.
22168 * profile.c (output_location): Include all BBs, even if
22169 belonging to a same line (and file) as a previous BB.
22171 2017-04-28 Martin Liska <mliska@suse.cz>
22173 * gcov.c (process_args): Handle new argument 'w'.
22174 (read_graph_file): Assign ID to BBs.
22175 (output_branch_count): Display BB # if verbose flag is set.
22176 (output_lines): Likewise for arcs.
22177 (print_usage): Add '--verbose' option help.
22178 * doc/gcov.texi: Document --verbose (-w) option.
22180 2017-04-28 Martin Liska <mliska@suse.cz>
22182 * gcov.c (struct block_location_info): New struct.
22183 (process_file): Fill up the new structure.
22184 (read_graph_file): Replace usage of encoding by the newly added
22186 (add_line_counts): Likewise.
22187 (accumulate_line_counts): Remove usage of the union.
22188 (function_info::function_info): New function.
22189 (function_info::~function_info): Likewise.
22190 (process_file): Call delete instead of release_function.
22191 (release_function): Release the function.
22192 (release_structures): Call delete instead of release_function.
22193 (solve_flow_graph): Replace usage of num_blocks.
22194 (find_exception_blocks): Likewise.
22195 (output_lines): Fix GNU coding style.
22197 2017-04-28 Martin Liska <mliska@suse.cz>
22200 * coverage.c (coverage_remove_note_file): New function.
22201 * coverage.h: Declare the function.
22202 * toplev.c (finalize): Clean if an error has been seen.
22204 2017-04-28 Martin Liska <mliska@suse.cz>
22206 PR gcov-profile/80031
22207 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
22208 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
22209 * gcov.c (read_graph_file): Read just number of blocks.
22210 * profile.c (branch_prob): Do not stream 0 flags per a basic
22213 2017-04-28 Martin Liska <mliska@suse.cz>
22215 * gcov-dump.c (tag_*): Add new argument to declarations.
22216 (dump_gcov_file): Likewise.
22217 (tag_blocks): Add and use new argument depth.
22218 (tag_arcs): Likewise.
22219 (tag_lines): Likewise.
22220 (tag_counters): Likewise.
22221 (tag_summary): Likewise.
22222 (dump_working_sets): Use depth to do a proper indentation.
22224 2017-04-28 Jakub Jelinek <jakub@redhat.com>
22227 * cgraph.h (symtab_node::debug_symtab): No longer inline.
22228 * symtab.c (symtab_node::debug_symtab): Move definition here.
22230 2017-04-28 Richard Biener <rguenther@suse.de>
22232 * lto-streamer.h (LTO_major_version): Bump to 7.
22234 2017-04-28 Richard Biener <rguenther@suse.de>
22236 * tree-vrp.c (assert_info): New struct.
22237 (add_assert_info): New helper.
22238 (register_edge_assert_for_2): Refactor to add asserts to a vector
22240 (register_edge_assert_for_1): Likewise.
22241 (register_edge_assert_for): Likewise.
22242 (finish_register_edge_assert_for): New helper actually registering
22243 asserts where live on edge.
22244 (find_conditional_asserts): Adjust.
22245 (find_switch_asserts): Likewise.
22246 (evrp_dom_walker::try_find_new_range): Generalize.
22247 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
22249 2017-04-27 Marek Polacek <polacek@redhat.com>
22252 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
22253 arg10 and arg11 to itype.
22255 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
22257 * doc/extend.texi (Object Size Checking): Improve grammar.
22259 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
22262 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
22263 that the logic for permitting reciprocal estimates matches that
22266 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22269 * tree.c (type_cache_hasher::equal): Only compare
22270 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
22271 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
22272 non-aggregate element types.
22273 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
22274 about the flag on ARRAY_TYPEs in the comment, formatting fix.
22276 2017-04-27 Richard Biener <rguenther@suse.de>
22278 PR middle-end/80533
22279 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
22280 stripping ARRAY_REFs from MEM_EXPR make sure we're not
22281 keeping a reference to a trailing array.
22283 2017-04-27 Richard Biener <rguenther@suse.de>
22285 PR middle-end/80539
22286 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
22287 being in loop-closed SSA form conservatively.
22288 (chrec_fold_multiply_poly_poly): Likewise.
22290 2017-04-27 Tamar Christina <tamar.christina@arm.com>
22292 PR middle-end/79665
22293 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
22294 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
22296 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22299 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
22300 (aarch64_function_arg_alignment): Return unsigned int again, but still
22301 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
22302 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22303 Don't emit -Wpsabi note.
22304 (aarch64_function_arg_boundary): Likewise.
22305 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22308 2017-04-26 Nathan Sidwell <nathan@acm.org>
22310 * tree.h (crc32_unsigned_n): Declare.
22311 (crc32_unsigned, crc32_unsigned): Make inline.
22312 * tree.c (crc32_unsigned_bits): Replace with ...
22313 (crc32_unsigned_n): ... this.
22314 (crc32_unsigned, crc32_byte): Remove.
22315 (crc32_string): Remove unnecessary braces.
22317 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22319 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
22320 * ipa-inline-analysis.c (MAX_TIME): Remove.
22321 (account_size_time): Use sreal for time.
22322 (dump_inline_summary): Update.
22323 (estimate_function_body_sizes): Update.
22324 (estimate_edge_size_and_time): Update.
22325 (estimate_calls_size_and_time): Update.
22326 (estimate_node_size_and_time): Update.
22327 (inline_merge_summary): Update.
22328 (inline_update_overall_summary): Update.
22329 (estimate_time_after_inlining): Update.
22330 (inline_read_section): Update.
22331 (inline_write_summary): Update.
22332 * ipa-inline.c (compute_uninlined_call_time): Update.
22333 (compute_inlined_call_time): Update.
22334 (recursive_inlining): Update.
22335 (inline_small_functions): Update.
22336 (dump_overall_stats): Update.
22337 * ipa-inline.h: Include sreal.h.
22338 (size_time_entry): Turn time to sreal.
22339 (inline_summary): Turn self_time nad time to sreal.
22341 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22343 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
22345 (sreal::stream_out, sreal::stream_in): New.
22346 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
22348 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22350 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
22353 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
22356 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
22358 (dimode_scalar_chain::compute_convert_gain): Ditto.
22359 (dimode_scalar_chain::make_vector_copies): Ditto.
22360 (dimode_scalar_chain::convert_reg): Ditto.
22361 (dimode_scalar_chain::convert_insn): Ditto.
22362 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
22363 (VI248_AVX512BW_1): New mode iterator.
22364 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
22365 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
22368 2017-04-25 Martin Sebor <msebor@redhat.com>
22370 PR tree-optimization/80497
22371 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
22372 constants are representable in HOST_WIDE_INT.
22373 (parse_directive): Ditto.
22375 2017-04-25 Martin Sebor <msebor@redhat.com>
22378 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
22379 (new_zero_array): Adjust signature.
22380 (dom_info::dom_init): Used unsigned rather that size_t.
22381 (dom_info::dom_info): Same.
22383 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22384 Jakub Jelinek <jakub@redhat.com>
22387 * config/arm/arm.c: Include gimple.h.
22388 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22389 returns negative, increment ncrn only if it returned positive.
22390 (arm_needs_doubleword_align): Return int instead of bool,
22391 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
22392 members, but if there is any such non-FIELD_DECL
22393 > PARM_BOUNDARY aligned decl, return -1 instead of false.
22394 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22395 returns negative, increment nregs only if it returned positive.
22396 (arm_setup_incoming_varargs): Likewise.
22397 (arm_function_arg_boundary): Emit -Wpsabi note if
22398 arm_needs_doubleword_align returns negative, return
22399 DOUBLEWORD_ALIGNMENT only if it returned positive.
22401 2017-04-25 Marek Polacek <polacek@redhat.com>
22404 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22405 first argument to type.
22407 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
22410 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
22411 type checks to test for compatibility instead of equality.
22413 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22414 Jakub Jelinek <jakub@redhat.com>
22417 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
22419 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
22420 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
22421 the alignment computation, but return their maximum in warn_alignment.
22422 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22423 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
22425 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
22426 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22429 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22431 * config/arc/simdext.md (dmpyh): Fix typo.
22433 2017-04-25 Richard Biener <rguenther@suse.de>
22435 PR tree-optimization/80492
22436 * alias.c (compare_base_decls): Handle registers with asm
22437 specification conservatively.
22438 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22439 compare_base_decls returning dont-know properly.
22441 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22443 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
22444 (legitimate_offset_address_p): New function.
22445 (arc_legitimate_address_p): Use above function.
22447 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22449 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
22451 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22453 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
22454 ACCH registers whenever they are available.
22456 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22458 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
22459 double regs fix when not used.
22461 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22463 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
22465 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
22466 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
22468 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22470 * config/arc/arc.c (arc_output_addsi): Check for h-register class
22471 when emitting short ADD instructions.
22473 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22475 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
22477 (cmpsi_cc_c_insn): Likewise.
22478 (cbranchsi4_scratch): Compute proper instruction length using
22479 compact_hreg_operand.
22480 * config/arc/predicates.md (compact_hreg_operand): New predicate.
22482 2017-04-25 Richard Biener <rguenther@suse.de>
22484 PR middle-end/80509
22485 * passes.c (pass_manager::pass_manager): Initialize
22486 m_name_to_pass_map.
22488 2017-04-25 Richard Biener <rguenther@suse.de>
22490 PR tree-optimization/79201
22491 * tree-ssa-sink.c (statement_sink_location): Handle calls.
22493 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22496 * config/s390/vector.md: Split MEM->GPR vector moves for
22497 non-s_operand addresses.
22499 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22502 * config/s390/predicates.md (reload_const_wide_int_operand): New
22504 * config/s390/s390.md ("movti"): Remove d/P alternative.
22505 ("movti_bigconst"): New pattern definition.
22507 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22510 * s390-protos.h (s390_expand_cs_hqi): Removed.
22511 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
22512 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
22513 modes as well as CCZ1mode and CCZmode.
22514 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
22515 signature of s390_emit_compare_and_swap.
22516 (s390_expand_cs_hqi): Likewise, make static.
22517 (s390_expand_cs_tdsi): Generate an explicit compare before trying
22518 compare-and-swap, in some cases.
22519 (s390_expand_cs): Wrapper function.
22520 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
22522 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
22523 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
22524 patterns for small and large integers. Forbid symref memory operands.
22525 Move expander to s390.c. Require cc register.
22526 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
22527 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
22528 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
22529 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
22530 symref memory operands. Remove CC mode and call s390_match_ccmode
22532 ("atomic_exchange<mode>"): Allow and implement all integer modes.
22534 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22536 * config/s390/s390.md (define_peephole2): New peephole to help
22537 combining the load-and-test pattern with volatile memory.
22539 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22541 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
22542 with CCZmode for TARGET_Z196.
22544 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22546 PR rtl-optimization/80501
22547 * combine.c (make_compound_operation_int): Set subreg_code to SET
22548 even for AND with mask of the sign bit of mode.
22550 PR rtl-optimization/80500
22551 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
22552 sum's initial value.
22554 2017-04-25 Julian Brown <julian@codesourcery.com>
22555 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22557 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
22559 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
22561 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
22563 2017-04-25 Julian Brown <julian@codesourcery.com>
22564 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22566 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
22567 (thunderx2t99_sha): New Reservation.
22569 2017-04-25 Julian Brown <julian@codesourcery.com>
22570 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22572 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
22573 type for 1-element load.
22575 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
22577 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
22579 2017-04-24 Martin Jambor <mjambor@suse.cz>
22581 PR tree-optimization/80293
22582 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
22583 char arrays not totally scalarizable if it is false.
22584 (analyze_all_variable_accesses): Pass correct value in the new
22585 parameter. Add a statistics counter.
22587 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
22589 PR middle-end/79931
22590 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
22592 2017-04-24 Richard Biener <rguenther@suse.de>
22594 PR tree-optimization/80494
22595 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
22596 out for complex types.
22598 2017-04-24 Richard Biener <rguenther@suse.de>
22600 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
22601 * tree-ssa-sccvn.c (print_scc): Print SCC size.
22602 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
22603 (DFS): Adjust and never fail.
22604 (sccvn_dom_walker::fail): Remove.
22605 (sccvn_dom_walker::before_dom_children): Adjust.
22606 (run_scc_vn): Likewise and never fail.
22607 * tree-ssa-pre.c (pass_pre::execute): Adjust.
22608 (pass_fre::execute): Likewise.
22610 2017-04-24 Richard Biener <rguenther@suse.de>
22612 PR tree-optimization/79725
22613 * tree-ssa-sink.c (statement_sink_location): Return whether
22614 failure reason was zero uses. Move that check later.
22615 (sink_code_in_bb): Deal with zero uses by removing the stmt
22618 2017-04-24 Richard Biener <rguenther@suse.de>
22621 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
22622 pointer-based references.
22624 2017-04-24 Richard Biener <rguenther@suse.de>
22627 * pass_manager.h (pass_manager::operator new): Remove.
22628 (pass_manager::operator delete): Likewise.
22629 * passes.c (pass_manager::operator new): Remove.
22630 (pass_manager::operator delete): Likewise.
22631 (pass_manager::pass_manager): Zero individual pass members.
22633 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
22636 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
22637 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
22638 Check "XEXP (src, 1)" operand here.
22639 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
22640 Check "XEXP (src, 1)" operand here.
22641 (dimode_scalar_chain::make_vector_copies): Detect count register
22642 of a shift instruction. Zero extend count register from QImode
22643 to DImode to satisfy vector shift pattern count operand predicate.
22644 Substitute vector shift count operand with a DImode copy.
22645 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
22648 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
22650 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
22651 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22652 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
22653 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22654 (UNSPEC_NOREX_MEM): Remove definition.
22656 2017-04-21 Richard Biener <rguenther@suse.de>
22658 PR tree-optimization/79547
22659 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22660 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
22661 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
22662 without any constraints.
22664 2017-04-21 Richard Biener <rguenther@suse.de>
22666 PR tree-optimization/78847
22667 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
22669 2017-04-21 Richard Biener <rguenther@suse.de>
22671 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
22672 (build_distinct_type_copy): Likewise.
22673 (build_variant_type_copy): Likewise.
22674 * tree.c (build_qualified_type): Pass down mem-stat info.
22675 (build_distinct_type_copy): Likewise.
22676 (build_variant_type_copy): Likewise.
22678 2017-04-21 Richard Biener <rguenther@suse.de>
22680 PR tree-optimization/80237
22681 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
22683 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
22684 for a simplified result.
22686 2016-04-21 Richard Biener <rguenther@suse.de>
22688 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
22689 sth as strict as a simple_iv but a chrec without symbols and an
22690 operand defined in the loop we are peeling (and not some subloop).
22691 (propagate_constants_for_unrolling): Propagate all constants.
22693 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22696 * config/i386/i386.c (print_reg): Remove assert for disalowed
22697 regno values, call output_operand_lossage instead.
22699 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22702 * config/i386/constraints.md (Yc): New register constraint.
22703 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
22704 Use Yc constraint for alternative 2 of operand 0. Remove
22705 preferred_for_speed attribute.
22707 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22709 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
22710 lastprivate clauses in SIMT case.
22712 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
22714 * doc/invoke.texi (-Wextra-semi): Document new warning option.
22716 2017-04-20 Richard Biener <rguenther@suse.de>
22718 PR tree-optimization/57796
22719 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
22720 as N scalar stores.
22721 (vect_model_load_cost): Cost gathers as N scalar loads.
22723 2017-04-20 Richard Biener <rguenther@suse.de>
22725 * ggc-page.c (ggc_allocated_p): Rename to ...
22726 (safe_lookup_page_table_entry): ... this and return the lookup
22728 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
22730 2017-04-20 Richard Biener <rguenther@suse.de>
22732 PR tree-optimization/80453
22733 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
22734 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
22735 from the conditions.
22736 (vn_phi_eq): Pass them down.
22737 (vn_phi_lookup): Record them.
22738 (vn_phi_insert): Likewise.
22740 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
22742 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
22743 uninitialized variable warning to avoid buffer overrun.
22745 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22748 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
22749 is suppressed for '{ 0 }' in C.
22751 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22753 * BASE-VER: Set to 8.0.0.
22755 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
22757 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
22758 priority .init_array and .fini_array section with SECTION_NOTYPE
22761 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22763 PR middle-end/80423
22764 * tree.h (build_array_type): Add typeless_storage default argument.
22765 * tree.c (type_cache_hasher::equal): Also compare
22766 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
22767 (build_array_type): Add typeless_storage argument, set
22768 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
22770 (build_nonshared_array_type): Adjust build_array_type_1 caller.
22771 (build_array_type): Likewise. Add typeless_storage argument.
22773 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22774 Jakub Jelinek <jakub@redhat.com>
22776 PR tree-optimization/80426
22777 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
22778 operation on symbolic operands, also compute the overflow for the
22779 invariant part when the operation degenerates into a negation.
22781 2017-04-19 Jakub Jelinek <jakub@redhat.com>
22784 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
22785 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
22788 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
22790 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
22793 * config/avr/avr.c (tree.h): Include it.
22794 (cgraph.h): Include it.
22795 (avr_encode_section_info): Don't warn for uninitialized progmem
22796 variable if it's just an alias.
22798 2017-04-19 Richard Biener <rguenther@suse.de>
22801 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
22802 when needed by AutoPGO.
22804 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
22807 * doc/lto.texi: Remove an extra 'that'.
22809 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
22811 PR rtl-optimization/80429
22812 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
22813 are only used in debug insns.
22815 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22816 Vladimir Makarov <vmakarov@redhat.com>
22818 * config/sparc/predicates.md (input_operand): Add comment. Return
22819 true for any memory operand when LRA is in progress.
22820 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
22822 2017-04-18 Jeff Law <law@redhat.com>
22825 * mips.md ({return,simple_return}_internal): Do not overwrite
22828 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22830 PR tree-optimization/80443
22831 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
22832 instead of adding 1, subtract -1 and similarly instead of subtracting
22835 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
22837 PR rtl-optimization/80357
22838 * haifa-sched.c (tmp_bitmap): New variable.
22839 (model_recompute): Handle duplicate use records.
22840 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
22841 (free_global_sched_pressure_data): Free it.
22843 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
22846 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22847 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22848 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22849 instead of SYSTEM_HEADER_DIR.
22851 2017-04-18 Jeff Law <law@redhat.com>
22853 PR middle-end/80422
22854 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
22855 predecessors after walking up the insn chain.
22857 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22860 * dwarf2out.c (modified_type_die): Try harder not to emit internal
22861 sizetype type into debug info.
22863 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22866 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
22867 unneeded test for TARGET_UPPER_REGS_SF.
22868 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
22870 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22873 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
22874 instead of gsi_after_labels.
22876 2017-04-18 Jeff Law <law@redhat.com>
22878 * regcprop.c (maybe_mode_change): Avoid creating copies of the
22882 2017-04-13 Jeff Law <law@redhat.com>
22883 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22884 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22886 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
22889 * config/avr/avr.c (intl.h): Include it.
22890 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
22892 2017-04-18 Martin Liska <mliska@suse.cz>
22894 PR gcov-profile/78783
22895 * gcov-tool.c (gcov_output_files): Validate that destination
22896 file is either removed by the tool or by a user.
22898 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22899 Guy Benyei <guybe@mellanox.com>
22901 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
22902 block, and do not negate it, the stored id is already negative.
22904 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22906 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
22908 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22911 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
22912 masks of options that should be turned off if the VSX vector
22913 options are turned off.
22914 (OTHER_P8_VECTOR_MASKS): Likewise.
22915 (OTHER_VSX_VECTOR_MASKS): Likewise.
22916 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
22917 rs6000_disable_incompatible_switches to validate no type switches
22919 (rs6000_incompatible_switch): New function to disallow turning on
22920 other vector options if -mno-vsx, -mno-power8-vector, or
22921 -mno-power9-vector are specified.
22923 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22925 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
22927 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22929 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
22930 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
22931 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
22932 (ARG_POINTER_CFA_OFFSET): Likewise.
22934 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22936 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
22937 conditions to take advantage of various optimizations.
22939 2017-04-13 Jeff Law <law@redhat.com>
22941 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22942 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22943 (zero_extendsidi2_dext): Likewise.
22945 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22948 * fold-const.c (fold_ternary_loc): Revert
22949 use op0 instead of fold_convert_loc (loc, type, arg0) part of
22952 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
22954 PR rtl-optimization/80343
22955 * lra-remat.c (update_scratch_ops): Assign original hard reg to
22956 new scratch pseudo.
22958 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
22961 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
22962 to ubsan_encode_value.
22964 2017-04-13 Jeff Law <law@redhat.com>
22966 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
22967 appearing in DEBUG_INSNs.
22969 2017-04-13 Martin Liska <mliska@suse.cz>
22971 PR gcov-profile/80413
22972 * gcov-io.c (gcov_write_string): Copy to buffer just when
22973 allocated size is greater than zero.
22975 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22978 * dwarf2out.c (decls_for_scope): Ignore declarations of
22979 current_function_decl in BLOCK_NONLOCALIZED_VARS.
22981 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
22984 * ipa-visibility.c (non_local_p): Fix typos.
22985 (localize_node): When localizing symbol in same comdat group,
22986 dissolve the group only when we know external symbols are going
22988 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
22990 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22992 PR tree-optimization/79390
22993 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
22994 order does not result in usable sequence, retry with reversed operand
23000 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23001 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23002 op0 instead of fold_convert_loc (loc, type, arg0).
23004 2017-04-12 Jeff Law <law@redhat.com>
23006 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23007 has a delay slot in the generated code.
23009 * config/cris/cris.md (cris_preferred_reload_class): Return
23010 GENNONACR_REGS rather than GENERAL_REGS.
23012 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23015 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23016 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23017 signedness of the result type.
23019 2017-04-12 Richard Biener <rguenther@suse.de>
23020 Jeff Law <law@redhat.com>
23022 PR tree-optimization/80359
23023 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23024 trim stores to TARGET_MEM_REFs.
23026 2017-04-12 Richard Biener <rguenther@suse.de>
23028 PR tree-optimization/79390
23029 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23030 threading case even more.
23032 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23035 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23036 for quad_address_p for TImode, instead of just not indexed_address.
23038 2017-04-12 Richard Biener <rguenther@suse.de>
23039 Bernd Edlinger <bernd.edlinger@hotmail.de>
23041 PR middle-end/79671
23042 * alias.c (component_uses_parent_alias_set_from): Handle
23043 TYPE_TYPELESS_STORAGE.
23044 (get_alias_set): Likewise.
23045 * tree-core.h (tree_type_common): Add typeless_storage flag.
23046 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23047 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23048 for types containing members with TYPE_TYPELESS_STORAGE.
23049 (place_field): Likewise.
23050 (layout_type): Likewise for ARRAY_TYPE.
23051 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23052 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23053 TYPE_TYPELESS_STORAGE.
23054 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23056 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23059 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23060 first argument to type.
23062 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23066 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23067 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23068 (rs6000_expand_binop_builtin): Likewise.
23069 (rs6000_expand_ternop_builtin): Likewise; also add missing
23070 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23071 vshasigma built-ins.
23072 * doc/extend.texi: Document that vec_xxpermdi's third argument
23073 must be a constant.
23075 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23077 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
23078 Use shift_const cost parameter when calculating gain of STV shifts.
23080 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
23082 PR rtl-optimization/70478
23083 * lra-constraints.c (process_alt_operands): Check memory for
23084 disfavoring memory insn operand.
23086 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23088 PR middle-end/80100
23089 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
23090 left shift in unsigned HOST_WIDE_INT type.
23092 PR rtl-optimization/80385
23093 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
23094 (not (neg X)) into (plus X -1) for complex or non-integral modes.
23097 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
23098 if they have any depend clauses.
23100 2017-04-11 Martin Liska <mliska@suse.cz>
23103 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
23104 * ipa-split.c (split_function): Create a local comdat symbol
23105 if caller is in a comdat group.
23107 2017-04-11 Martin Liska <mliska@suse.cz>
23110 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
23113 2017-04-11 Martin Sebor <msebor@redhat.com>
23115 PR middle-end/80364
23116 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
23117 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
23119 (directive::set_width, directive::set_precision, format_character):
23121 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
23124 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
23127 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
23128 conflict, set target->arch_name instead of target->cpu_name.
23130 2017-04-11 Richard Biener <rguenther@suse.de>
23132 PR tree-optimization/80374
23133 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
23134 build_zero_cst, remove fold_convertible_p check again.
23136 2017-04-11 Martin Liska <mliska@suse.cz>
23139 * ubsan.c (instrument_object_size): Do not instrument register
23142 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23145 * config/i386/i386-builtin-types.def
23146 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
23147 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
23148 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
23149 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
23150 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
23151 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
23152 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
23153 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
23154 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
23155 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
23156 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
23157 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
23158 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
23159 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
23160 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
23161 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
23162 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
23163 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
23164 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
23165 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
23166 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
23167 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
23168 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
23169 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
23170 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
23171 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
23172 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
23173 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
23174 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
23175 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
23176 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
23177 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
23178 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
23179 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
23180 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
23181 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
23182 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
23183 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
23184 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
23185 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
23186 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
23187 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
23188 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
23189 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
23190 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
23191 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
23193 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
23194 flag to second_arg_count, handle 4 argument function type _COUNT
23195 aliases, handle second_arg_count on second argument rather than last.
23197 2017-04-10 Jeff Law <law@redhat.com>
23199 PR tree-optimization/80374
23200 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
23201 record anything if we can not convert integer_zero_node to the
23204 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23207 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23208 Enhance special handling given to the TARGET_P9_MINMAX option in
23209 relation to certain other options.
23211 2017-04-10 Bin Cheng <bin.cheng@arm.com>
23213 PR tree-optimization/80153
23214 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
23215 remove POINTER_PLUS_EXPR's base part directly, rather than through
23218 2017-04-10 Richard Biener <rguenther@suse.de>
23219 Bin Cheng <bin.cheng@arm.com>
23221 PR tree-optimization/80153
23222 * tree-affine.c (aff_combination_to_tree): Get base pointer from
23223 the first element of pointer type aff_tree. Build result expr in
23225 (add_elt_to_tree): Convert to type unconditionally. Remove other
23226 fold_convert calls.
23227 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
23228 (rewrite_use_nonlinear_expr): Check invariant using iv information.
23230 2017-04-10 Richard Biener <rguenther@suse.de>
23232 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23235 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
23237 PR rtl-optimization/70478
23238 * lra-constraints.c (curr_small_class_check): New.
23239 (update_and_check_small_class_inputs): New.
23240 (process_alt_operands): Update curr_small_class_check. Disfavor
23241 alternative insn memory operands. Check available regs for small
23244 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23247 * config/mips/mips.opt (-mvirt): Update description.
23248 * doc/invoke.texi (-mvirt): Likewise.
23250 2017-04-10 Richard Biener <rguenther@suse.de>
23252 PR middle-end/80362
23253 * fold-const.c (fold_binary_loc): Look at unstripped ops when
23254 looking for NEGATE_EXPR in -A / -B to A / B folding.
23256 2017-04-10 Martin Liska <mliska@suse.cz>
23258 PR gcov-profile/80224
23259 * gcov.c (print_usage): Fix usage string.
23260 (get_gcov_intermediate_filename): Remove.
23261 (output_gcov_file): Use both for normal and intermediate format.
23262 (generate_results): Do not initialize special file for
23263 intermediate format.
23265 2017-04-10 Richard Biener <rguenther@suse.de>
23267 PR tree-optimization/80304
23268 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
23271 2017-04-10 Nathan Sidwell <nathan@acm.org>
23274 * config/rs6000/rs6000.c (rs6000_vector_type): New.
23275 (rs6000_init_builtins): Use it.
23277 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23279 * config/arm/arm.md (<mrc>): Add mode to SET source.
23280 (<mrrc>): Likewise.
23282 2017-04-10 Richard Biener <rguenther@suse.de>
23284 PR middle-end/80344
23285 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
23287 2017-04-10 Jakub Jelinek <jakub@redhat.com>
23290 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
23291 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
23292 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
23293 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
23294 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
23295 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
23296 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
23297 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
23298 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
23299 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
23300 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
23301 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
23302 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
23303 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
23304 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
23305 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
23306 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
23307 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
23308 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
23309 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
23310 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
23311 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
23312 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
23314 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
23316 PR rtl-optimization/70478
23317 * lra-constraints.c: Reverse the last patch.
23319 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
23321 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
23322 Add comment for WCHAR_T.
23324 2017-04-08 Martin Liska <mliska@suse.cz>
23327 2017-04-07 Martin Liska <mliska@suse.cz>
23330 * ipa-split.c (split_function): Add function part to a same comdat
23333 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23336 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
23338 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
23340 * rs6000/rs6000.c (vec_load_pendulum): Rename...
23341 (vec_pairing): ...to this.
23342 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
23343 (rs6000_sched_init): Adjust for name change.
23344 (struct rs6000_sched_context): Likewise.
23345 (rs6000_init_sched_context): Likewise.
23346 (rs6000_set_sched_context): Likewise.
23348 2017-04-07 Jakub Jelinek <jakub@redhat.com>
23354 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
23356 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
23357 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
23358 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
23360 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
23362 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
23364 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23366 PR rtl-optimization/70703
23367 * ira-color.c (update_conflict_hard_regno_costs): Use
23368 int64_t instead of HOST_WIDE_INT.
23370 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23372 PR rtl-optimization/70478
23373 * lra-constraints.c (process_alt_operands): Disfavor alternative
23374 insn memory operands.
23376 2017-04-07 Jeff Law <law@redhat.com>
23378 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
23379 CALL and NOTE_INSN_CALL_ARG_LOCATION.
23381 2017-04-07 Martin Liska <mliska@suse.cz>
23384 * config/aarch64/aarch64.c (aarch64_process_target_attr):
23385 Show error message instead of an ICE.
23387 2017-04-07 Martin Liska <mliska@suse.cz>
23390 * ipa-split.c (split_function): Add function part to a same comdat
23393 2017-04-07 Richard Biener <rguenther@suse.de>
23395 PR middle-end/80341
23396 * tree.c (get_unwidened): Also handle ! for_type case for
23398 * convert.c (do_narrow): Split out from ...
23399 (convert_to_integer_1): ... here. Do not pass final truncation
23400 type to get_unwidened for TRUNC_DIV_EXPR.
23402 2017-04-07 Richard Biener <rguenther@suse.de>
23404 * tree-affine.c (wide_int_ext_for_comb): Take type rather
23406 (aff_combination_const): Adjust.
23407 (aff_combination_scale): Likewise.
23408 (aff_combination_add_elt): Likewise.
23409 (aff_combination_add_cst): Likewise.
23410 (aff_combination_convert): Likewise.
23411 (add_elt_to_tree): Likewise. Remove unused argument.
23412 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
23414 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
23416 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
23418 * config/arm/arm.c (arm_default_short_enums): Use
23419 ARM_DEFAULT_SHORT_ENUMS.
23420 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
23422 2017-04-06 Jakub Jelinek <jakub@redhat.com>
23425 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
23426 members with redundant out-of-class redeclaration.
23428 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23431 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
23432 * config/i386/i386.md (*zero_extendsidi2):
23433 Add (?*x,*x) and (?*v,*v) alternatives.
23435 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23438 * config/i386/i386.c (ix86_expand_builtin)
23439 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
23440 mode from insn data. Convert operands to insn operand mode.
23441 Copy operands that don't satisfy insn predicate to a register.
23443 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
23445 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
23448 2017-04-06 Richard Biener <rguenther@suse.de>
23450 PR tree-optimization/80334
23451 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
23452 preserve alignment of accesses.
23454 2017-04-06 Richard Biener <rguenther@suse.de>
23456 PR tree-optimization/80262
23457 * tree-sra.c (build_ref_for_offset): Preserve address-space
23459 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
23460 Drop useless address-space information on MEM_REF offsets.
23462 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
23464 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
23466 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23468 PR rtl-optimization/70703
23469 * ira-color.c (update_conflict_hard_regno_costs): Use
23470 HOST_WIDE_INT instead of long.
23472 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
23475 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
23476 not defined for x86_64 target. Add -mmmx target option when __SSE2__
23478 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
23479 for x86_64 target. Handle -m3dnowa option.
23481 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23483 PR rtl-optimization/70703
23484 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
23485 (update_conflict_hard_regno_costs): Use long instead of unsigned
23486 arithmetic for cost calculation.
23488 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23489 Bernd Edlinger <bernd.edlinger@hotmail.de>
23492 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
23495 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
23498 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
23499 ptr_mode with Pmode throughout.
23500 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
23501 into probe_stack_range and use DImode.
23503 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
23506 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
23507 call_eh_return is true.
23509 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23511 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
23512 Initialize last_match_fntype_index.
23514 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23517 * tree-nvr.c: Include internal-fn.h.
23518 (pass_return_slot::execute): Ignore internal calls without
23521 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23522 Richard Biener <rguenther@suse.de>
23525 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
23526 captures used multiple times, except for the last use.
23527 * generic-match-head.c: Include gimplify.h.
23529 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23531 PR tree-optimization/79390
23532 * target.h (struct noce_if_info): Declare.
23533 * targhooks.h (default_noce_conversion_profitable_p): Declare.
23534 * target.def (noce_conversion_profitable_p): New target hook.
23535 * ifcvt.h (struct noce_if_info): New type, moved from ...
23536 * ifcvt.c (struct noce_if_info): ... here.
23537 (noce_conversion_profitable_p): Renamed to ...
23538 (default_noce_conversion_profitable_p): ... this. No longer
23540 (noce_try_store_flag_constants, noce_try_addcc,
23541 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
23542 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
23543 instead of noce_conversion_profitable_p.
23544 * config/i386/i386.c: Include ifcvt.h.
23545 (ix86_option_override_internal): Don't override
23546 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
23547 (ix86_noce_conversion_profitable_p): New function.
23548 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
23549 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
23550 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
23551 * doc/tm.texi: Regenerated.
23553 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23555 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
23558 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23561 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
23562 instructions for small multiply cores.
23564 2017-04-04 Jeff Law <law@redhat.com>
23566 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
23568 (mips_expand_vec_perm_const): Initialize elements in orig_perm
23569 that are not set by the loop over the elements.
23571 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23574 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
23575 int mode, convert_modes it to mode as unsigned, otherwise use
23576 lowpart_subreg to mode rather than SImode.
23577 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
23578 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
23579 Use DImode instead of SImode for the shift count operand.
23580 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
23583 2017-04-04 Richard Biener <rguenther@suse.de>
23585 PR middle-end/80281
23586 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
23587 arithmetic done for the negate or the plus. Simplify.
23588 (A - (-B) -> A + B): Likewise.
23589 * fold-const.c (split_tree): Make sure to not negate pointers.
23591 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
23593 PR rtl-optimization/60818
23594 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
23595 a compare of comparisons with the thing compared if this results
23596 in a different machine mode.
23598 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
23600 * alias.c (base_alias_check): Fix typo in comment.
23601 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
23602 * cgraphunit.c (symbol_table::compile): Likewise.
23603 * collect2.c (maybe_run_lto_and_relink): Likewise.
23604 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
23605 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
23606 * config/avr/avr.c (avr_map_op_t): Likewise.
23607 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
23608 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
23609 * config/epiphany/epiphany.md (movcc): Likewise.
23610 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
23611 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
23613 * config/mips/mips.c (mips_save_restore_reg): Likewise.
23614 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
23615 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
23616 * config/sh/sh.c (sh_rtx_costs): Likewise.
23617 * fold-const.c (fold_truth_andor): Likewise.
23618 * genautomata.c (collapse_flag): Likewise.
23619 * gengtype.h (struct type::u::s): Likewise.
23620 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
23621 * input.c (FORMAT_AMOUNT): Likewise.
23622 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
23623 (known_aggs_to_agg_replacement_list): Likewise.
23624 * ipa-inline-analysis.c: Likewise.
23625 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
23626 * ipa-polymorphic-call.c
23627 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
23628 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
23629 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
23631 * modulo-sched.c (apply_reg_moves): Likewise.
23632 * omp-expand.c (build_omp_regions_1): Likewise.
23633 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
23634 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
23635 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
23636 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
23637 * value-prof.c: Likewise.
23638 * var-tracking.c (val_reset): Likewise.
23640 2017-04-03 Richard Biener <rguenther@suse.de>
23642 PR tree-optimization/80275
23643 * fold-const.c (split_address_to_core_and_offset): Handle
23646 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
23648 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
23649 descriptors is at least equal to that of functions.
23651 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23653 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
23655 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23658 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
23659 (mov<IMOD4:mode>): New expander.
23660 (*mov<IMOD4:mode>_internal): New insn and split pattern.
23662 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
23664 PR rtl-optimization/79405
23665 * fwprop.c (propagations_left): New variable.
23666 (forward_propagate_into): Decrement it.
23667 (fwprop_init): Initialize it.
23668 (fw_prop): If the variable has reached zero, stop propagating.
23669 (fwprop_addr): Ditto.
23671 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23674 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
23675 a FUNCTION_DECL, pass it as decl instead of origin to
23678 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
23680 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
23683 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
23686 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
23687 TARGET_VSX_SMALL_INTEGER.
23689 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23691 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23692 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
23694 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23696 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
23697 extraction from odd-numbered MSA register.
23699 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23701 PR middle-end/80173
23702 * expmed.c (store_bit_field_1): Don't attempt to create
23703 a word subreg out of hard registers wider than word if they
23704 have HARD_REGNO_NREGS of 1 for their mode.
23706 PR middle-end/80163
23707 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
23708 conversions to integer types wider than word and pointer.
23711 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
23712 (rtx_equal_for_cselib_p): Pass 0 to it.
23713 * cselib.c (cselib_hasher::equal): Likewise.
23714 (rtx_equal_for_cselib_1): Add depth argument. If depth
23715 is 128, don't look up VALUE locs and punt. Increment
23716 depth in recursive calls when walking VALUE locs.
23718 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
23720 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
23721 (make_gcov_file_name): Use the canonical path name for generating
23723 (read_line): Fix handling of files with ascii null bytes.
23725 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
23727 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
23728 to initialise a vector register instead
23729 of using a const_int.
23731 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23733 PR translation/80189
23734 * gimplify.c (omp_default_clause): Use %qs instead of %s in
23735 diagnostic messages.
23737 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
23740 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
23741 (dfp_diex_<mode>): Update mode of operand 1.
23742 * doc/extend.texi (dxex, dxexq): Document change to return type.
23743 (diex, diexq): Document change to argument type.
23745 2017-03-30 Martin Jambor <mjambor@suse.cz>
23748 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
23749 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
23750 it reflects the signature changes performed at the callee side.
23751 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
23752 to cgraph_build_function_type_skip_args.
23753 (build_function_decl_skip_args): Adjust call to the above function.
23755 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23758 * config/i386/sse.md
23759 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
23760 register as dest whenever it is a MEM not rtx_equal_p to the
23761 corresponding dup operand, and when forcing into reg move the
23762 reg into the memory afterwards.
23763 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
23764 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
23765 for the force_reg mode.
23766 (avx512vl_vextractf128<mode>): Use register as dest either
23767 always when a MEM, or when it is a MEM not rtx_equal_p to the
23768 corresponding dup operand, or even not when it is a CONST_VECTOR
23769 depending on the mode and lo vs. hi.
23770 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
23772 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
23773 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
23774 Likewise. Require that operands[2] is even.
23775 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
23776 Remove extraneous parens. Require that operands[2] is a multiple
23778 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
23779 operands[0] is a MEM if <mask_applied>, the predicates/constraints
23780 disallow memory then.
23782 2017-03-30 Richard Biener <rguenther@suse.de>
23784 PR tree-optimization/77498
23785 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
23786 to non-constants over backedges.
23788 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
23790 PR rtl-optimization/80233
23791 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
23792 as last_combined_insn. Do not test for BARRIER_P separately.
23794 2017-03-29 Andreas Schwab <schwab@suse.de>
23797 * calls.c (prepare_call_address): Convert funexp to Pmode before
23798 copying to temp reg.
23800 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23802 PR tree-optimization/80158
23803 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
23804 Handle possible future case of more than one alternate
23806 (replace_rhs_if_not_dup): Likewise.
23807 (replace_one_candidate): Likewise.
23809 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
23811 PR rtl-optimization/80193
23812 * ira.c (ira): Do not check allocation for LRA.
23814 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
23816 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
23817 (nvptx_output_simt_exit): Declare.
23818 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
23819 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
23820 (init_softstack_frame): Move initialization of crtl->is_leaf to...
23821 (nvptx_declare_function_name): ...here. Emit declaration of local
23822 memory space buffer for omp_simt_enter insn.
23823 (nvptx_output_unisimt_switch): New.
23824 (nvptx_output_softstack_switch): New.
23825 (nvptx_output_simt_enter): New.
23826 (nvptx_output_simt_exit): New.
23827 * config/nvptx/nvptx.h (struct machine_function): New fields
23828 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
23829 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
23830 (UNSPECV_SIMT_EXIT): Ditto.
23831 (omp_simt_enter_insn): New insn.
23832 (omp_simt_enter): New expansion.
23833 (omp_simt_exit): New insn.
23834 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
23836 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
23837 (expand_GOMP_SIMT_ENTER_ALLOC): New.
23838 (expand_GOMP_SIMT_EXIT): New.
23839 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
23840 (GOMP_SIMT_ENTER_ALLOC): Ditto.
23841 (GOMP_SIMT_EXIT): Ditto.
23842 * target-insns.def (omp_simt_enter): New insn.
23843 (omp_simt_exit): Ditto.
23844 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
23846 (lower_rec_simd_input_clauses): Implement SIMT privatization.
23847 (lower_rec_input_clauses): Likewise.
23848 (lower_lastprivate_clauses): Handle SIMT privatization.
23850 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
23851 (ompdevlow_adjust_simt_enter): New.
23852 (find_simtpriv_var_op): New.
23853 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
23854 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
23856 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
23857 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
23858 (copy_decl_for_dup_finish): Ditto.
23860 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
23862 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
23865 * config/i386/i386.c (ix86_option_override_internal): Always
23866 allow -mpreferred-stack-boundary=3 for 64-bit targets.
23868 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23870 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
23872 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23874 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
23875 mark new edge's irreducible flag accordign to it.
23876 (vect_do_peeling): Check loop preheader edge's irreducible flag
23877 and pass it to function slpeel_add_loop_guard.
23879 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
23881 PR tree-optimization/80218
23882 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23883 Update block frequencies and counts.
23885 2017-03-28 Richard Biener <rguenther@suse.de>
23887 PR tree-optimization/78644
23888 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
23889 of a simplification result we may not use it at all.
23891 2017-03-28 Richard Biener <rguenther@suse.de>
23894 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
23895 without arguments, generate default definition of a SSA name.
23897 2017-03-28 Richard Biener <rguenther@suse.de>
23899 PR middle-end/80222
23900 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
23901 TYPE_REF_CAN_ALIAS_ALL references.
23902 * fold-const.c (fold_indirect_ref_1): Likewise.
23904 2017-03-28 Martin Liska <mliska@suse.cz>
23907 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
23908 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
23910 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23911 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
23913 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
23914 (EXTRA_SPECS): Define.
23915 (SUBTARGET_EXTRA_SPECS): Likewise.
23916 (SUBTARGET_CPP_SPEC): Likewise.
23917 * config/arc/elf.h (EXTRA_SPECS): Renamed to
23918 SUBTARGET_EXTRA_SPECS.
23919 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
23921 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23923 * config/arc/simdext.md (vst64_insn): Update pattern.
23924 (vld32wh_insn): Likewise.
23925 (vld32wl_insn): Likewise.
23926 (vld64_insn): Likewise.
23927 (vld32_insn): Likewise.
23929 2017-03-28 Marek Polacek <polacek@redhat.com>
23932 * fold-const.c (fold_comparison): Use protected_set_expr_location
23933 instead of SET_EXPR_LOCATION.
23935 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
23937 * tree.c (add_expr): Avoid name lookup warning.
23939 2017-03-27 Jeff Law <law@redhat.com>
23941 PR tree-optimization/80216
23942 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
23943 function name. Limit recursion depth.
23944 (record_temporary_equivalences): Corresponding changes.
23946 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
23948 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
23951 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23954 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
23956 * cfgcleanup.c (reg_note_cfa_p): New array.
23957 (insns_have_identical_cfa_notes): New function.
23958 (old_insns_match_p): Don't cross-jump in between /f
23959 and non-/f instructions. If both i1 and i2 are frame related,
23960 verify all CFA notes, their order and content.
23962 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
23965 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
23966 HImode and SImode with zero extend to DImode to one insn.
23967 (bswap<mode>2_extenddi): Likewise.
23968 (bswapsi2_extenddi): Likewise.
23969 (bswaphi2_extendsi): Likewise.
23970 (bswaphi2): Combine bswap HImode and SImode into one insn.
23971 Separate memory insns from swapping register.
23972 (bswapsi2): Likewise.
23973 (bswap<mode>2): Likewise.
23974 (bswaphi2_internal): Delete, no longer used.
23975 (bswapsi2_internal): Likewise.
23976 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
23977 store, and gpr<-gpr swap insns.
23978 (bswap<mode>2_store): Likewise.
23979 (bswaphi2_reg): Register only splitter, combine with the splitter.
23980 (bswaphi2 splitter): Likewise.
23981 (bswapsi2_reg): Likewise.
23982 (bswapsi2 splitter): Likewise.
23983 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
23984 the insns into load, store, and register/register insns.
23985 (bswapdi2_ldbrx): Likewise.
23986 (bswapdi2_load): Likewise.
23987 (bswapdi2_store): Likewise.
23988 (bswapdi2_reg): Likewise.
23990 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
23992 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
23993 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
23995 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
23998 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24000 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24001 special handling for target option conflicts between dform
24002 options (-mpower9-dform, -mpower9-dform-vector,
24003 -mpower9-dform-scalar) and -mno-direct-move.
24005 2017-03-27 Richard Biener <rguenther@suse.de>
24007 PR tree-optimization/80181
24008 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24010 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24012 * config/arc/predicates.md (move_double_src_operand): Replace the
24013 call to move_double_src_operand with a call to address_operand.
24015 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24017 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24018 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24019 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24021 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24023 * config/arc/predicates.md (long_immediate_loadstore_operand):
24024 Consider scaled addresses cases.
24026 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24028 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24029 restored when in interrupt.
24030 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24031 doesn't have delay slot.
24033 2017-03-27 Richard Biener <rguenther@suse.de>
24036 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24037 inlined thunk clones.
24039 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24042 * asan.c (instrument_derefs): Copy over last operand from
24043 original COMPONENT_REF to the new COMPONENT_REF with
24044 DECL_BIT_FIELD_REPRESENTATIVE.
24045 * ubsan.c (instrument_object_size): Likewise.
24047 2017-03-27 Richard Biener <rguenther@suse.de>
24049 PR tree-optimization/80170
24050 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24051 sure DR/SCEV didnt fold in constants we do not see when looking
24052 at the reference base alignment.
24054 2017-03-27 Richard Biener <rguenther@suse.de>
24056 PR middle-end/80171
24057 * gimple-fold.c (fold_ctor_reference): Properly guard against
24058 NULL return value from canonicalize_constructor_val.
24060 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24063 * config/i386/i386.c (ix86_expand_builtin)
24064 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24065 flags reg setting and flags reg using instructions.
24066 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24067 clobbering instructions to zero extend op2.
24069 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24071 * doc/install.texi (Configuration) <--with-aix-soname>:
24072 Update link to AIX ld.
24074 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24076 PR rtl-optimization/80160
24077 PR rtl-optimization/80159
24078 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
24079 reg_alternate_class into account.
24081 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
24084 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
24085 to consider in curr_insn_transform.
24087 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24089 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
24090 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
24091 and emit_mode_inner.
24093 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24095 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
24096 argument to the overloaded builtin variants. Use the new flag to
24097 deprecate certain builtin variants.
24098 * config/s390/s390-builtin-types.def: Add new builtin types.
24099 * config/s390/s390-builtins.h: Support new flags field for
24100 overloaded builtins.
24101 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
24102 (s390_macro_to_expand): Enable vector float data type.
24103 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24104 builtins by incrementing the __VEC__ version number.
24105 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
24107 (s390_resolve_overloaded_builtin): Emit error messages depending
24108 on the builtin flags.
24109 * config/s390/s390.c (s390_expand_builtin): Support additional
24110 flags argument. Change error message to match the messages
24111 emitted in s390-c.c.
24112 * config/s390/s390.md: New UNSPEC_* constants.
24113 (op_type): Add new instruction types.
24114 * config/s390/vecintrin.h: Add new builtins and test data class
24116 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
24117 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
24118 (VEC_INEXACT, VEC_NOINEXACT): New constants.
24119 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
24120 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
24121 ("vec_mergel<mode>"): V_HW -> VEC_HW.
24123 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
24124 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
24125 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
24126 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
24128 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
24129 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
24130 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
24131 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
24133 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
24134 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
24135 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
24136 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
24137 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
24138 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
24139 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
24141 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
24142 ("vec_scatter_element<V_HW_4:mode>_DI")
24143 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
24144 ("vec_fpint<mode>", "vflls")
24145 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
24146 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
24147 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
24148 ("*vec_cmphe<mode>_cc"): ... these.
24150 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
24151 mode constant instead of magic value.
24153 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24155 * config/s390/s390.c (s390_expand_vec_compare): Support other
24156 vector floating point modes than just V2DF.
24157 (s390_expand_vcond): Likewise.
24158 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
24159 (s390_cannot_change_mode_class): Prevent mode changes between TF
24160 and V1TF in vector registers.
24161 * config/s390/s390.md (DF, SF): New mode attributes.
24162 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
24163 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
24164 SFmode support for VRs.
24165 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
24167 (VFT, VF_HW): New mode iterators.
24168 (vw, sdx): New mode attributes.
24169 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
24170 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
24171 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
24172 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
24173 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
24174 also the new vector floating point modes. Renaming to ...
24176 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
24177 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
24178 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
24179 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
24180 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
24181 ("vec_unordered<mode>"): ... these.
24183 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
24184 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
24185 ("*vec_extendv2df"): New insn definitions.
24187 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24189 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
24190 ("mulditi3_2", "*muldi3_sign"): New patterns.
24191 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
24192 rename the pattern definition.
24194 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24196 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
24198 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
24200 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24202 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
24203 instruction if possible.
24204 * config/s390/vector.md (vec_halfnumelts): New mode
24206 ("*vec_vllezlf<mode>"): New pattern.
24208 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24210 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
24211 ("popcountv4si2", "popcountv2di2"): Rename to ...
24212 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
24213 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
24215 ("popcount<mode>2_vxe"): New pattern.
24217 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24219 * common/config/s390/s390-common.c (processor_flags_table): Add
24221 * config.gcc: Add arch12.
24222 * config/s390/driver-native.c (s390_host_detect_local_cpu):
24223 Default to arch12 for unknown CPU model numbers.
24224 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
24225 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
24226 PROCESSOR_max sanity check.
24227 * config/s390/s390-opts.h (enum processor_type): Add
24229 * config/s390/s390.c (processor_table): Add arch12.
24230 (s390_expand_builtin): Add check for B_VXE flag.
24231 (s390_issue_rate): Add PROCESSOR_ARCH12.
24232 (s390_get_sched_attrmask): Likewise.
24233 (s390_get_unit_mask): Likewise.
24234 (s390_sched_score): Enable z13 scheduling for arch12.
24235 (s390_sched_reorder): Likewise.
24236 (s390_sched_variable_issue): Likewise.
24237 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
24239 (s390_tune_attr): Use z13 scheduling also for arch12.
24240 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
24241 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
24242 (TARGET_VXE_P): New macros.
24243 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
24244 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
24245 * config/s390/s390.opt: Add arch12 as processor_type.
24247 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24249 * config/s390/s390.md
24250 ("fixuns_truncdddi2", "fixuns_trunctddi2")
24251 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
24252 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
24254 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
24255 Rename expanders to ...
24257 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
24258 ("fixuns_truncdddi2_emu"): ... these.
24260 ("fixuns_trunc<mode>si2_emu"): New expander.
24262 ("*fixuns_truncdfdi2_z13"): Rename to ...
24263 ("*fixuns_truncdfdi2_vx"): ... this.
24265 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24267 * config/s390/2964.md: Remove the single element vector compare
24268 instructions which are no longer used.
24269 * config/s390/s390.c (s390_select_ccmode): Remove handling of
24271 (s390_canonicalize_comparison): Remove handling of DFmode
24273 (s390_expand_vec_compare_scalar): Remove function.
24274 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
24275 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
24277 ("*cmp<mode>_ccs"): Add wfcdb instruction.
24279 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24281 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
24283 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
24284 will anyway by matched by mov<mode>_64dfp.
24286 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24288 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
24289 vlef/vstef. Add missing operand to vleif.
24291 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24293 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
24294 pair for all vector types with 64 bit elements.
24295 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
24296 * config/s390/vector.md (V_HW_64): ... here.
24297 (V_128_NOSINGLE): New mode iterator.
24298 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
24299 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
24300 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
24301 ("*vec_load_pairv2di"): Change to ...
24302 ("*vec_load_pair<mode>"): ... this one.
24304 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24306 * config/s390/constraints.md: Add comments.
24307 (jKK): Reject element sizes > 8 bytes.
24308 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
24310 * config/s390/s390.md: Add the s_operand checks formerly in
24311 s390_split_ok_p to various splitters where they are still
24313 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
24314 for 128 bit vectors. Plus two splitters.
24316 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24318 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
24321 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24324 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
24325 error if the boundary argument is not constant.
24327 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24329 PR rtl-optimization/80112
24330 * loop-doloop.c (doloop_condition_get): Don't check condition
24331 if cmp isn't SET with IF_THEN_ELSE src.
24333 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24335 PR tree-optimization/80158
24336 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
24337 replacing a candidate statement, also replace it for the
24338 candidate's alternate interpretation.
24339 (replace_rhs_if_not_dup): Likewise.
24340 (replace_one_candidate): Likewise.
24342 2017-03-24 Richard Biener <rguenther@suse.de>
24344 PR tree-optimization/80167
24345 * graphite-isl-ast-to-gimple.c
24346 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
24348 (translate_isl_ast_to_gimple::get_rename): Likewise.
24350 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24352 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
24353 handling of certain combinations of target options, including the
24354 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
24355 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
24357 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24360 * config/arm/arm.md (*load_multiple): Add reload_completed to
24361 matching condition.
24363 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24364 Richard Biener <rguenth@suse.de>
24366 PR tree-optimization/79908
24367 PR tree-optimization/80136
24368 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24369 been cast away, gimplify_and_add suffices.
24371 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
24373 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
24375 2017-03-23 Richard Biener <rguenther@suse.de>
24377 PR tree-optimization/80032
24378 * gimplify.c (gimple_push_cleanup): Forced unconditional
24379 cleanups still have to go to the conditional_cleanups
24382 2017-03-22 Jakub Jelinek <jakub@redhat.com>
24384 PR tree-optimization/80072
24385 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
24387 (next_operand_entry_id): Change type to unsigned int.
24388 (sort_by_operand_rank): Make sure to return the right return value
24389 even if unsigned fields are bigger than INT_MAX.
24390 (struct oecount): Change cnt and id type to unsigned int.
24391 (oecount_hasher::equal): Formatting fix.
24392 (oecount_cmp): Make sure to return the right return value
24393 even if unsigned fields are bigger than INT_MAX.
24394 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
24397 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
24398 TREE_READONLY on result if writing it more than once.
24401 * doc/invoke.texi (-fsanitize=thread): Document that with
24402 -fnon-call-exceptions atomics are not able to throw
24406 * tsan.c: Include tree-eh.h.
24407 (instrument_builtin_call): Call maybe_clean_eh_stmt or
24408 maybe_clean_or_replace_eh_stmt where needed.
24409 (instrument_memory_accesses): Add cfg_changed argument.
24410 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
24411 if it returned true.
24412 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
24414 PR rtl-optimization/63191
24415 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
24416 wrapper function, moved the whole old content into ...
24417 (ix86_delegitimize_address_1): ... this. New inline function.
24418 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
24419 true as last argument instead of ix86_delegitimize_address.
24421 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24423 * config/aarch64/aarch64.c (generic_branch_cost): Copy
24424 cortexa57_branch_cost.
24426 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24428 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
24430 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24433 * doc/md.texi (Constraints): Document wA constraint.
24434 * config/rs6000/constraints.md (wA): New.
24435 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
24436 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
24437 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
24438 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
24440 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
24443 * gimplify.c (is_oacc_declared): New function.
24444 (oacc_default_clause): Use it to set default flags for acc declared
24445 variables inside parallel regions.
24446 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
24447 declared variables.
24448 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
24449 declare attribute to any decl as necessary.
24451 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24454 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
24455 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
24456 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
24457 (arm_arch_lpae): This.
24458 * config/arm/arm.c (arm_arch7ve): Rename into ...
24459 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
24460 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
24463 2017-03-22 Martin Liska <mliska@suse.cz>
24466 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
24467 error message instead of an ICE.
24469 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24471 * doc/extend.texi (6.11 Additional Floating Types): Revise.
24473 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
24475 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24477 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24480 2017-03-21 Martin Sebor <msebor@redhat.com>
24482 * doc/extend.texi: Use "cannot" instead of "can't."
24483 * doc/hostconfig.texi: Same.
24484 * doc/install.texi: Same.
24485 * doc/invoke.texi: Same.
24486 * doc/loop.texi: Same.
24487 * doc/md.texi: Same.
24488 * doc/objc.texi: Same.
24489 * doc/rtl.texi: Same.
24490 * doc/tm.texi: Same.
24491 * doc/tm.texi.in: Same.
24492 * doc/trouble.texi: Same.
24494 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
24497 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
24498 (collect_checksum_attributes): Set it.
24499 (die_checksum_ordered): Use it.
24501 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24503 PR tree-optimization/79908
24504 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
24505 change: For a VA_ARG whose LHS has been cast away, use
24506 force_gimple_operand to construct the side effects.
24508 2017-03-21 David Malcolm <dmalcolm@redhat.com>
24510 PR translation/80001
24511 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
24512 more amenable to translation.
24513 (oacc_loop_auto_partitions): Likewise.
24515 2017-03-21 Marek Polacek <polacek@redhat.com>
24516 Martin Sebor <msebor@redhat.com>
24518 PR tree-optimization/80109
24519 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
24520 on INTEGRAL_TYPE_P.
24522 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24523 Segher Boessenkool <segher@kernel.crashing.org>
24526 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
24527 check reg_used_between_p between insn and one of succ or succ2
24528 depending on if succ is artificial insn not inserted into insn
24531 2017-03-21 Martin Liska <mliska@suse.cz>
24533 PR gcov-profile/80081
24534 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
24535 * doc/gcc.texi: Include gcov-dump stuff.
24536 * doc/gcov-dump.texi: New file.
24538 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
24540 PR rtl-optimization/79150
24541 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
24542 conditional jump, if the jump is the last insn of the loop.
24544 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24545 Richard Biener <rguenth@suse.de>
24547 PR tree-optimization/79908
24548 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24549 been cast away, use force_gimple_operand to construct the side
24552 2017-03-21 Martin Liska <mliska@suse.cz>
24554 PR libfortran/79956
24555 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
24558 2017-03-21 Brad Spengler <spender@grsecurity.net>
24561 * plugin.c (htab_hash_plugin): New function.
24562 (add_new_plugin): Use it and adjust.
24563 (parse_plugin_arg_opt): Adjust.
24564 (init_one_plugin): Likewise.
24566 2017-03-21 Richard Biener <rguenther@suse.de>
24568 PR tree-optimization/80032
24569 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
24570 if set force the cleanup to happen unconditionally.
24571 (gimplify_target_expr): Push inserted clobbers with force_uncond
24572 to avoid them being removed by control-dependent DCE.
24574 2017-03-21 Richard Biener <rguenther@suse.de>
24576 PR tree-optimization/80122
24577 * tree-inline.c (copy_bb): Do not expans va-arg packs or
24578 va_arg_pack_len when the inlined call stmt requires pack
24580 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
24582 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24585 * tsan.c (instrument_builtin_call): If the memory model argument
24586 is not a constant, assume it is valid.
24589 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
24592 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24594 PR rtl-optimization/79910
24595 * combine.c (can_combine_p): Do not allow combining an I0 or I1
24596 if its dest is used by an insn before I2 (other than the combined
24597 insns themselves, which are properly handled already).
24599 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24602 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24604 * combine.c (record_used_regs): New static function.
24605 (try_combine): Handle situations where there is an additional
24606 instruction between I2 and I3 which needs to have a LOG_LINK
24610 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24612 * combine.c (try_combine): Delete redundant i1 test. Call
24613 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24615 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24618 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
24619 alternatives 13/14.
24621 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24623 PR tree-optimization/80054
24624 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
24625 the optimization if a PHI or any of its arguments is not dominated
24626 by the candidate's basis. Use gphi* rather than gimple* as
24628 (replace_profitable_candidates): Clean up a gimple* variable that
24629 should be a gphi* variable.
24631 2017-03-20 Martin Sebor <msebor@redhat.com>
24634 * doc/extend.texi (attribute constructor): Document present limitation.
24636 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
24639 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
24640 __POWER9_VECTOR__ #ifdef control, change template definition to
24641 use Power9-specific built-in function.
24642 (vec_any_eq): Likewise.
24643 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
24644 to control outcomes from this test.
24645 (vector_ae_<mode>p): For VEC_F modes, likewise.
24647 2017-03-20 Ian Lance Taylor <iant@google.com>
24649 * config/i386/i386.c (ix86_function_regparm): Save an extra
24650 register for -fsplit-stack with DECL_STATIC_CHAIN.
24652 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24655 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
24656 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
24658 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24660 * config/riscv/riscv.c (riscv_print_operand): Use "fence
24662 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
24665 2017-03-20 Marek Polacek <polacek@redhat.com>
24668 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
24670 2017-03-20 Richard Biener <rguenther@suse.de>
24672 PR tree-optimization/80113
24673 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
24674 allocate extra SSA name for PHI def.
24675 (add_close_phis_to_outer_loops): Likewise.
24676 (add_close_phis_to_merge_points): Likewise.
24677 (copy_loop_close_phi_args): Likewise.
24678 (copy_cond_phi_nodes): Likewise.
24680 2017-03-20 Martin Liska <mliska@suse.cz>
24682 PR middle-end/79753
24683 * tree-chkp.c (chkp_build_returned_bound): Do not build
24684 returned bounds for a LHS that's not a BOUNDED_P type.
24686 2017-03-20 Martin Liska <mliska@suse.cz>
24690 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
24691 COMPLEX_CST and VECTOR_CST.
24693 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24696 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
24697 target operand. A new splitter adds the clobber statement in case
24698 the target operand is dead anyway.
24700 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
24702 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
24703 to age-old versions of binutils and glibc.
24705 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
24707 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
24709 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24711 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
24713 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24715 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
24716 requirement for binutils 2.13.
24718 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24720 * combine.c (try_combine): Delete redundant i1 test. Call
24721 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24723 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
24725 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
24726 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
24728 <riscv64-*-elf>: Re-arrange section
24729 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
24730 <riscv32-*-linux>: Likewise.
24731 <riscv64-*-elf>: Likewise
24732 <riscv64-*-linux>: Likewise.
24734 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
24737 * aarch64.opt(verbose-cost-dump): Fix typo.
24739 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
24742 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
24743 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
24745 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24747 * reload.c (find_reloads): When reloading a nonoffsettable address,
24748 use RELOAD_OTHER for it and its address reloads.
24750 PR rtl-optimization/79910
24751 * combine.c (record_used_regs): New static function.
24752 (try_combine): Handle situations where there is an additional
24753 instruction between I2 and I3 which needs to have a LOG_LINK
24756 2017-03-17 Jeff Law <law@redhat.com>
24758 PR tree-optimization/71437
24759 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
24760 conditional in the hash table first.
24761 (vrp_dom_walker::before_dom_children): Extract condition from
24762 ASSERT_EXPR. Record condition, its inverion and any implied
24763 conditions as well.
24765 2017-03-17 Marek Polacek <polacek@redhat.com>
24766 Markus Trippelsdorf <markus@trippelsdorf.de>
24768 PR tree-optimization/80079
24769 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
24772 2017-03-17 Richard Biener <rguenther@suse.de>
24774 PR middle-end/80075
24775 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
24776 Properly verify the LHS before the RHS possibly claims to be
24778 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
24781 2017-03-17 Martin Jambor <mjambor@suse.cz>
24783 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
24784 (List of -O2 options): Likewise.
24785 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
24788 2017-03-17 Tom de Vries <tom@codesourcery.com>
24790 * gcov-dump.c (print_usage): Print bug_report_url.
24792 2017-03-17 Richard Biener <rguenther@suse.de>
24794 PR middle-end/80050
24795 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
24796 (parser::peek): Likewise.
24798 2017-03-17 Richard Biener <rguenther@suse.de>
24800 PR tree-optimization/80048
24801 * sese.c (free_sese_info): Properly release rename_map and
24802 copied_bb_map elements.
24804 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
24806 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
24807 Add linked-list forward and backlinks. Insert on
24808 construction, remove on destruction.
24809 (class pass_store_merging): Add m_stores_head field.
24810 (pass_store_merging::terminate_and_process_all_chains):
24811 Iterate over m_stores_head list.
24812 (pass_store_merging::terminate_all_aliasing_chains):
24814 (pass_store_merging::execute): Check for debug stmts first.
24815 Push new chains onto the m_stores_head stack.
24817 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24820 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
24821 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
24822 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
24824 2017-03-16 Jeff Law <law@redhat.com>
24826 PR tree-optimization/71437
24827 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
24828 member function. Implementation moved into after_dom_children
24829 member function and into the threader's thread_outgoing_edges
24831 (dom_opt_dom_walker::after_dom_children): Simplify by moving
24832 some code into new thread_outgoing_edges.
24833 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
24834 definition. Simplify marker handling (do it here). Assume we always
24835 have the available expression and the const/copies tables.
24836 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
24838 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
24839 * tree-vrp.c (equiv_stack): No longer file scoped.
24840 (vrp_dom_walker): New class.
24841 (vrp_dom_walker::before_dom_children): New member function.
24842 (vrp_dom_walker::after_dom_children): Likewise.
24843 (identify_jump_threads): Setup domwalker. Use it rather than
24844 walking edges in a random order by hand. Simplify setup/finalization.
24845 (finalize_jump_threads): Remove.
24846 (vrp_finalize): Do not call identify_jump_threads here.
24847 (execute_vrp): Do it here instead and call thread_through_all_blocks
24850 PR tree-optimization/71437
24851 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
24853 (simplify_stmt_for_jump_threading): Add basic_block argument. All
24855 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
24856 (dom_opt_dom_walker::thread_across_edge): Remove
24857 handle_dominating_asserts argument. All callers changed.
24858 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
24859 changes. Remove calls to lhs_of_dominating_assert. Other
24860 uses of handle_dominating_asserts turn into unconditional code
24861 (simplify_control_stmt_condition_1): Likewise.
24862 (simplify_control_stmt_condition): Likewise.
24863 (thread_through_normal_block, thread_across_edge): Likewise.
24864 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
24865 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
24866 object if it is not an SSA_NAME.
24867 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
24868 before calling into the VRP specific simplifiers.
24869 (identify_jump_threads): Remove handle_dominating_asserts
24872 2017-03-16 Jakub Jelinek <jakub@redhat.com>
24875 * tree-diagnostic.c (default_tree_printer): No longer static.
24876 * tree-diagnostic.h (default_tree_printer): New prototype.
24878 2017-03-16 Tamar Christina <tamar.christina@arm.com>
24880 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
24881 Change ins into fmov.
24883 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24885 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
24886 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
24887 Use h_con constraint for operand 1.
24888 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
24889 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
24891 2017-03-15 Jeff Law <law@redhat.com>
24893 PR tree-optimization/71437
24894 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
24895 (record_temporary_equivalences): Use it.
24897 PR tree-optimization/71437
24898 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
24899 tree-ssa-scopedtables.
24900 (lookup_avail_expr, build_and_record_new_cond): Likewise.
24901 (record_conditions, record_cond, vuse_eq): Likewise.
24902 (record_edge_info): Adjust to API tweak of record_conditions.
24903 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
24904 (record_temporary_equivalences, optimize_stmt): Likewise.
24905 (eliminate_redundant_computations): Likewise.
24906 (record_equivalences_from_stmt): Likewise.
24907 * tree-ssa-scopedtables.c: Include options.h and params.h.
24908 (vuse_eq): New function, moved from tree-ssa-dom.c
24909 (build_and_record_new_cond): Likewise.
24910 (record_conditions): Likewise. Accept vector of conditions rather
24911 than edge_equivalence structure for first argument.
24912 for the first argument.
24913 (avail_exprs_stack::lookup_avail_expr): New member function, moved
24914 from tree-ssa-dom.c.
24915 (avail_exprs_stack::record_cond): Likewise.
24916 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
24917 from tree-ssa-dom.c.
24918 (avail_exprs_stack): Add new member functions lookup_avail_expr
24920 (record_conditions): Declare.
24922 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
24925 * lra-constraints.c (process_alt_operands): Increase reject for
24926 reloading an input/output operand.
24928 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
24931 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
24932 insns to convert from signed/unsigned char/short to IEEE 128-bit
24934 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
24936 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
24939 * config/i386/i386.c (ix86_vector_duplicate_value): Create
24940 subreg of inner mode for values already in registers.
24942 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
24944 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
24945 iteration reg is used after the loop.
24947 2017-03-14 Martin Sebor <msebor@redhat.com>
24949 PR tree-optimization/79800
24950 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
24951 precision in negative-positive range.
24952 (format_floating): Call non-const overload with adjusted precision.
24954 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24957 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
24960 2017-03-14 Martin Sebor <msebor@redhat.com>
24962 PR middle-end/80020
24963 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
24964 * builtins.def (aligned_alloc): Use it.
24967 * Makefile.in (GTFILES): Add calls.c.
24968 * calls.c: Include "gt-calls.h".
24970 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
24972 PR rtl-optimization/79728
24973 * regs.h (struct target_regs): New field
24974 x_contains_allocatable_regs_of_mode.
24975 (contains_allocatable_regs_of_mode): New macro.
24976 * reginfo.c (init_reg_sets_1): Initialize it, and change
24977 contains_reg_of_mode so it includes global regs as well.
24978 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
24979 rather than contains_regs_of_mode.
24981 2017-03-14 Martin Liska <mliska@suse.cz>
24983 * doc/invoke.texi: Document options that can't be combined with
24984 -fcheck-pointer-bounds.
24986 2017-03-14 Martin Liska <mliska@suse.cz>
24988 PR middle-end/79831
24989 * doc/invoke.texi (-Wchkp): Document the option.
24991 2017-03-14 Martin Liska <mliska@suse.cz>
24993 * Makefile.in: Install gcov-dump.
24995 2017-03-14 Martin Liska <mliska@suse.cz>
24997 * multiple_target.c (expand_target_clones): Bail out for
24998 an invalid attribute.
25000 2017-03-14 Richard Biener <rguenther@suse.de>
25002 * alias.c (struct alias_set_entry): Pack properly.
25003 * cfgloop.h (struct loop): Likewise.
25004 * cse.c (struct set): Likewise.
25005 * ipa-utils.c (struct searchc_env): Likewise.
25006 * loop-invariant.c (struct invariant): Likewise.
25007 * lra-remat.c (struct cand): Likewise.
25008 * recog.c (struct change_t): Likewise.
25009 * rtl.h (struct address_info): Likewise.
25010 * symbol-summary.h (function_summary): Likewise.
25011 * tree-loop-distribution.c (struct partition): Likewise.
25012 * tree-object-size.c (struct object_size_info): Likewise.
25013 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25014 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25015 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25016 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25017 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25018 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25019 (struct _stmt_vec_info): Likewise.
25021 2017-03-14 Martin Liska <mliska@suse.cz>
25024 * multiple_target.c (create_dispatcher_calls): Check that
25025 a target can create a function dispatcher.
25027 2017-03-14 Martin Liska <mliska@suse.cz>
25030 * multiple_target.c (expand_target_clones): Drop local.local
25031 flag for default implementation.
25033 2017-03-14 Richard Biener <rguenther@suse.de>
25035 PR tree-optimization/80030
25036 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25038 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25040 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25041 gcc_fallthrough() instead of __attribute__((fallthrough));
25043 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25045 * doc/gcc.texi: Remove "up" link to (DIR).
25046 * doc/gccint.texi: Ditto.
25048 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25050 * doc/install.texi (Specific) <avr>: Remove reference to
25053 2017-03-13 Jeff Law <law@redhat.com>
25055 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25056 attribute rather than comments.
25058 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25059 match_scratch operand is highest.
25061 2017-03-13 Martin Liska <mliska@suse.cz>
25063 PR middle-end/78339
25064 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25065 is a CHKP clone, use original declaration.
25067 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25069 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25070 (arc_conditional_register_usage): Use a different allocation order
25071 when optimizing for size.
25072 * common/config/arc/arc-common.c (arc_option_optimization_table):
25073 Section anchors default on when optimizing for size.
25075 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25077 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
25079 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25081 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
25082 * config/arc/arc.md (cpu_facility): Add cd variant.
25083 (*movqi_insn): Add code density variant.
25084 (*movhi_insn): Likewise.
25085 (*movqi_insn): Likewise.
25086 (*addsi3_mixed): Likewise.
25087 (subsi3_insn): Likewise.
25089 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25091 * config/arc/arc.md (movsi_cond_exec): Update constraint.
25093 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25095 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
25096 expressions with MINUS and UNARY ops.
25098 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25101 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
25103 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
25104 between vec_select and vector argument.
25105 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
25106 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
25107 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
25108 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
25109 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
25110 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
25112 2017-03-13 Richard Biener <rguenther@suse.de>
25115 * params.def (vect-max-peeling-for-alignment): Fix typo.
25117 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25119 * doc/install.texi (Specific) <mips-*-*>: Remove description of
25120 issue that only occurred with binutils below 2.18.
25122 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25124 * doc/install.texi (Specific) <cris-axis-elf>: No longer
25125 refer to binutils 2.11/2.12 minimum.
25127 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25129 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
25130 ftp.kernel.org and simplify binutils requirement.
25132 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
25134 * doc/invoke.texi (Warning Options): Fix spelling of link-time
25136 (Optimize Options): Ditto. Also remove redundancy.
25138 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25140 PR translation/79848
25141 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
25143 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
25144 to G_ to avoid double translation.
25146 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25148 PR translation/79923
25149 * auto-profile.c (get_combined_location): Convert leading
25150 character of diagnostics to lower case and remove trailing period.
25151 (read_profile): Likewise for various diagnostics.
25152 * config/arm/arm.c (arm_option_override): Remove trailing period
25153 from various diagnostics.
25154 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
25155 (msp430_expand_delay_cycles): Likewise.
25157 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25160 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
25161 full command-line argument, rather than just "str".
25162 (aarch64_validate_march): Likewise.
25163 (aarch64_validate_mtune): Likewise.
25165 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
25167 PR rtl-optimization/78911
25168 * lra-assigns.c (must_not_spill_p): New function.
25169 (spill_for): Use it.
25171 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25173 PR tree-optimization/79981
25174 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
25175 ATOMIC_COMPARE_EXCHANGE ifn result.
25176 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
25177 IFN_ATOMIC_COMPARE_EXCHANGE.
25179 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25182 * opts.c (parse_sanitizer_options): Add missing question mark to
25183 "did you mean" message.
25185 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25187 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
25189 (VMULEUH_UNS): Likewise.
25190 (VMULOUB_UNS): Likewise.
25191 (VMULOUH_UNS): Likewise.
25192 * config/rs6000/rs6000.c (builtin_function_type): Remove
25193 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
25195 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25198 * read-rtl-function.c (function_reader::read_rtx_operand): Update
25199 x with result of extra_parsing_for_operand_code_0.
25200 (function_reader::extra_parsing_for_operand_code_0): Convert
25201 return type from void to rtx, returning x. When reading
25202 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
25203 larger size containing struct block_symbol.
25205 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
25207 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
25208 -mfloat128-hardware without -m64.
25210 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
25213 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
25214 entries to the case statement that marks unsigned arguments to
25215 overloaded functions.
25217 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25219 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
25220 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
25222 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
25225 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
25226 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
25228 2017-03-10 Martin Liska <mliska@suse.cz>
25232 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
25233 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
25236 2017-03-10 Olivier Hainque <hainque@adacore.com>
25238 * tree-switch-conversion (array_value_type): Start by resetting
25239 candidate type to it's main variant.
25241 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25243 PR rtl-optimization/79909
25244 * combine.c (try_combine): Use simplify_replace_rtx on individual
25245 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
25246 of the whole CALL_INSN_FUNCTION_USAGE.
25248 PR tree-optimization/79972
25249 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
25250 get_range_info on SSA_NAMEs. Formatting fixes.
25252 2017-03-10 Richard Biener <rguenther@suse.de>
25253 Jakub Jelinek <jakub@redhat.com>
25255 PR tree-optimization/77975
25256 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
25257 edge to be constant.
25258 (get_val_for): For constant x return it. Formatting fix.
25259 (loop_niter_by_eval): Avoid pointless looping if the next iteration
25260 would use the same bases as the current one.
25262 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25264 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
25265 instead of vec_select for V1TImode.
25266 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
25268 (VSX_LE_128): Add V1TI to this mode iterator.
25269 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
25270 (*vsx_le_perm_store_<mode>): Likewise.
25271 (pre-reload splitter for VSX stores): Likewise.
25272 (post-reload splitter for VSX stores): Likewise.
25273 (*vsx_xxpermdi2_le_<mode>): Likewise.
25274 (*vsx_lxvd2x2_le_<mode>): Likewise.
25275 (*vsx_stxvd2x2_le_<mode>): Likewise.
25277 2017-03-09 Michael Eager <eager@eagercon.com>
25279 Correct failures with --enable-checking=yes,rtl.
25281 * config/microblaze/microblaze.c (microblaze_expand_shift):
25282 Replace GET_CODE test with CONST_INT_P and INTVAL test with
25283 test for const0_rtx.
25284 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
25285 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
25287 2017-03-09 Richard Biener <rguenther@suse.de>
25289 PR tree-optimization/79977
25290 * graphite-scop-detection.c (scop_detection::merge_sese):
25291 Handle the case of extra exits to blocks dominating the entry.
25293 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
25295 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25298 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
25300 PR rtl-optimization/79949
25301 * lra-constraints.c (process_alt_operands): Check memory when
25302 trying to predict a cycle. Print about the overall increase.
25304 2017-03-09 Richard Biener <rguenther@suse.de>
25306 PR middle-end/79971
25307 * gimple-expr.c (useless_type_conversion_p): Preserve
25308 TYPE_SATURATING for fixed-point types.
25310 2017-03-09 Richard Biener <rguenther@suse.de>
25313 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
25314 alignment of BLKmode params.
25316 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25319 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
25320 (VALL_NO_V2Q): Likewise.
25322 * config/aarch64/aarch64-simd.md
25323 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
25325 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
25326 VALL_NO_V2Q mode iterator.
25327 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
25329 2017-03-09 Martin Liska <mliska@suse.cz>
25331 PR tree-optimization/79631
25332 * tree-chkp-opt.c (chkp_is_constant_addr): Call
25333 tree_int_cst_sign_bit just for INTEGER constants.
25335 2017-03-09 Martin Liska <mliska@suse.cz>
25339 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
25342 2017-03-09 Marek Polacek <polacek@redhat.com>
25345 * tree.c (inchash::add_expr): Handle TREE_VEC.
25347 2017-03-09 Martin Liska <mliska@suse.cz>
25350 (chkp_narrow_size_and_offset): New function.
25351 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
25352 (void chkp_parse_bit_field_ref): New function.
25353 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
25354 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
25356 2017-03-09 Martin Liska <mliska@suse.cz>
25359 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
25360 (chkp_find_bounds_1): Remove gcc_unreachable.
25362 2017-03-09 Jakub Jelinek <jakub@redhat.com>
25365 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
25366 BUILT_IN_SYNC*, determine the access type from the size suffix and
25367 always build a MEM_REF with that type. Handle forgotten
25368 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
25371 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
25372 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
25373 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
25374 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
25375 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
25376 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
25377 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
25378 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
25379 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
25380 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
25381 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
25382 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
25383 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
25384 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
25385 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
25386 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
25387 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
25388 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
25389 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
25390 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
25391 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
25392 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
25393 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
25394 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
25395 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
25396 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
25397 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
25398 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
25399 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
25400 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
25401 definitions outside of __OPTIMIZE__ guarded section.
25404 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
25405 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
25406 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
25407 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
25410 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25412 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
25413 ("vfenez<mode>"): Add missing constraints.
25415 2017-03-08 Martin Sebor <msebor@redhat.com>
25418 * config/nds32/nds32.c (nds32_option_override):
25419 Fix misspelled diagnostic.
25421 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25424 * gimplify.c (gimplify_omp_for): Replace index var in outer
25425 taskloop statement with an artificial variable and add
25426 OMP_CLAUSE_PRIVATE clause for it.
25428 2017-03-08 Richard Biener <rguenther@suse.de>
25430 PR tree-optimization/79955
25431 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
25432 for accesses that are completely outside of the variable.
25434 2017-03-08 Andrew Haley <aph@redhat.com>
25436 PR tree-optimization/79943
25437 * tree-ssa-loop-split.c (compute_new_first_bound): When
25438 calculating the new upper bound, (END-BEG) should be added, not
25441 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25443 * config/avr/avr.md (setmemhi): Make sure match_dup
25444 operand number comes before match_scratch.
25446 2017-03-08 Richard Biener <rguenther@suse.de>
25448 PR tree-optimization/79920
25449 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
25450 with ncopies == 1 to ...
25451 (vect_transform_slp_perm_load): ... here. Properly compute
25452 all element loads by iterating VF times over the group. Do
25453 not handle ncopies (computed in a broken way) in
25454 vect_create_mask_and_perm.
25456 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25459 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
25460 is a uniform vector, use uniform_vector_p return value instead of
25461 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
25463 2017-03-07 Marek Polacek <polacek@redhat.com>
25465 PR middle-end/79809
25466 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
25467 (alloca_call_type): Likewise.
25469 2017-03-07 Martin Liska <mliska@suse.cz>
25471 * gcov.c (process_args): Put comment to correct location.
25473 2017-03-07 Martin Liska <mliska@suse.cz>
25475 PR middle-end/68270
25476 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
25477 Use array_at_struct_end_p instead of DECL_CHAIN (field).
25478 (chkp_narrow_bounds_for_field): Likewise.
25479 (chkp_parse_array_and_component_ref): Pass one more argument to
25482 2017-03-07 Richard Biener <rguenther@suse.de>
25484 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
25487 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25489 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
25490 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
25492 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25495 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
25496 to end of description.
25497 (PARAM_MAX_STORES_TO_MERGE): Likewise.
25499 2017-03-07 Jakub Jelinek <jakub@redhat.com>
25501 PR rtl-optimization/79901
25502 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
25504 (*avx512f_<code><mode>3<mask_name>): ... this.
25505 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
25506 iterator instead of VI8_AVX2_AVX512BW.
25508 PR rtl-optimization/79901
25509 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
25510 min/max expander, expand it using expand_vec_cond_expr.
25513 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
25516 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25519 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
25520 to void * for PCH reasons.
25521 * dwarf2out.c (output_loc_operands, output_die): Cast
25522 v.val_vec.array to unsigned char *.
25524 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
25527 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
25530 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
25532 PR rtl-optimization/79571
25533 * lra-constraints.c (process_alt_operands): Calculate static
25534 reject and subtract it from overall when only addresses will be
25537 2017-03-06 Julia Koval <julia.koval@intel.com>
25540 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
25541 incoming stack boundary to 128 for 64-bit targets.
25543 2017-03-06 Richard Biener <rguenther@suse.de>
25545 PR tree-optimization/79894
25546 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
25547 to NULL after folding it.
25549 2017-03-06 Richard Biener <rguenther@suse.de>
25551 PR tree-optimization/79824
25552 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
25553 check disabling peeling for gaps.
25555 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
25557 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25558 attributes): Document gettimeofday.
25560 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25562 * config/s390/s390.c (s390_option_override_internal): Set
25563 PARAM_MIN_VECT_LOOP_BOUND
25565 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25567 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
25568 * config/s390/s390.md: Likewise.
25570 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25573 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
25574 (<avx2_avx512>_perm<mode>): Rename to ...
25575 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
25577 (<avx512>_perm<mode>_mask): Rename to ...
25578 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
25580 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
25581 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
25582 instead of VI8F_256_512.
25583 (avx512f_perm<mode>): New define_expand.
25584 (avx512f_perm<mode>_mask): Likewise.
25585 (avx512f_perm<mode>_1<mask_name>): New define_insn.
25586 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
25588 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25590 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
25591 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
25593 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
25595 2017-03-06 Martin Liska <mliska@suse.cz>
25598 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
25599 when having a SSA NAME w/o VAR_DECL assigned to it.
25601 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25603 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
25604 msa_dpsub_<su>_d): Fix MODE for vec_select.
25606 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25608 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
25610 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
25612 2017-03-06 Richard Biener <rguenther@suse.de>
25614 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
25615 * plugin.c (register_plugin_info): Likewise.
25616 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
25618 2017-03-05 Jakub Jelinek <jakub@redhat.com>
25620 * config/i386/sse.md (sse_storehps, sse_storelps,
25621 avx_<castmode><avxsizesuffix>_<castmode>,
25622 avx512f_<castmode><avxsizesuffix>_<castmode>,
25623 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
25624 in condition that at least one operand is not a MEM.
25626 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25628 PR middle-end/79805
25629 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
25630 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
25632 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
25633 gimple_call_nothrow_p flag based on whether original builtin can throw.
25634 If it can, emit following stmts on the fallthrough edge.
25635 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
25636 don't create new bb if inserting just debug stmts on the edge, try to
25637 insert them on the fallthru bb or just reset debug stmts.
25639 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
25642 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
25643 restore recog_data (including the operand rtxes inside it) around
25644 the call to get_insn_template.
25646 2017-03-03 Martin Sebor <msebor@redhat.com>
25648 PR tree-optimization/79699
25649 * context.c (context::~context): Free MPFR caches to avoid
25650 a memory leak on program exit.
25652 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25654 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
25655 Use wide_int::ulow () instead of .elt (0).
25657 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25659 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
25660 (*pushxf): Limit oF constraint to 32bit targets and add oC
25661 constraint for 64bit targets.
25662 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
25663 (*pushdf): Change rmF constraint to rmC.
25665 2017-03-03 Martin Liska <mliska@suse.cz>
25667 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
25668 Remove unused variable.
25670 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25673 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
25674 is a memory operand, increase num_memory.
25675 (ix86_expand_args_builtin): Likewise.
25677 2017-03-03 Jan Hubicka <jh@suse.cz>
25680 * ipa-devirt.c (maybe_record_node): Properly handle
25681 __cxa_pure_virtual visibility.
25683 2017-03-03 Martin Liska <mliska@suse.cz>
25685 PR tree-optimization/79803
25686 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
25688 (pass_loop_prefetch::execute): Disabled optimization if an
25689 assumption about L1 cache size is not met.
25691 2017-03-03 Martin Liska <mliska@suse.cz>
25693 PR rtl-optimization/79574
25694 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
25695 (hash_scan_set): Likewise.
25696 (dump_hash_table): Likewise.
25697 (hoist_code): Likewise.
25699 2017-03-03 Richard Biener <rguenther@suse.de>
25701 * fixed-value.c (fixed_from_string): Restore use of elt (1)
25702 in place of uhigh ().
25703 (fixed_convert_from_real): Likewise.
25705 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25708 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
25710 2017-03-03 Richard Biener <rguenther@suse.de>
25712 PR middle-end/79818
25713 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
25714 TYPE_OVERFLOW_UNDEFINED check.
25716 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25718 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
25720 (vector_ae_<mode>_p): Likewise.
25721 (vector_nez_<mode>_p): Likewise.
25722 (vector_ne_v2di_p): Likewise.
25723 (vector_ae_v2di_p): Likewise.
25724 (vector_ne_<mode>_p): Likewise.
25725 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
25727 (vsx_tsqrt<mode>2_fe): Likewise.
25729 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
25732 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
25734 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25736 PR rtl-optimization/79780
25737 * cprop.c (one_cprop_pass): When second and further conditional trap
25738 in a single basic block is turned into an unconditional trap, turn it
25739 into a deleted note to avoid RTL verification failures.
25741 2017-03-02 Richard Biener <rguenther@suse.de>
25743 * fold-const.c (const_binop): Use ulow () instead of elt (0).
25745 2017-03-02 Richard Biener <rguenther@suse.de>
25747 PR tree-optimization/79345
25749 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
25750 param and abort the walk, returning -1 if it is hit.
25751 (walk_aliased_vdefs): Take a limit param and pass it on.
25752 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
25753 defaulting to 0 and return a signed int.
25754 * tree-ssa-uninit.c (struct check_defs_data): New struct.
25755 (check_defs): New helper.
25756 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
25757 about uninitialized memory.
25758 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
25759 bogus uninitialized warning.
25760 (fixed_convert_from_real): Likewise.
25762 2017-03-02 Bin Cheng <bin.cheng@arm.com>
25764 PR tree-optimization/66768
25765 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
25766 iv_use if base object can't be determined.
25768 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25770 PR tree-optimization/79345
25771 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
25772 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
25773 (get_pattern_stats): Initialize it.
25774 * genemit.c (gen_expand): Verify match_scratch numbers come after
25775 match_operand/match_dup numbers.
25776 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
25777 match_scratch numbers.
25778 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
25780 * config/s390/s390.md (trunctdsd2): Likewise.
25782 2017-03-02 Richard Biener <rguenther@suse.de>
25784 * wide-int.h (wide_int_storage::operator=): Implement in terms
25787 2017-03-02 Richard Biener <rguenther@suse.de>
25789 PR tree-optimization/79777
25790 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
25791 the to insert expression to sth existing.
25793 2017-03-01 Martin Sebor <msebor@redhat.com>
25795 PR middle-end/79692
25796 * gimple-ssa-sprintf.c
25797 (directive::known_width_and_precision): New function.
25798 (format_integer): Use it.
25799 (get_mpfr_format_length): Consider the full range of precision
25800 when computing %g output with the # flag. Set the likely byte
25801 count to 3 rather than 1 when precision is indeterminate.
25802 (format_floating): Correct the lower bound of precision.
25804 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25806 * doc/invoke.texi: Document default code model for 64-bit Linux.
25808 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25811 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
25812 udiv rather than div since input pattern is unsigned.
25814 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
25816 * config/i386/i386.c (print_reg): Warn for values of
25817 unsupported size in integer register.
25819 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
25822 * config/rs6000/predicates.md (current_file_function_operand): Do
25823 not allow self calls to be local if the function is replaceable.
25825 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
25828 * config/rs6000/altivec.h (vec_ctz and others): Change the
25829 preprocessor macro that controls conditional compilation from
25830 _ARCH_PWR9 to __POWER9_VECTOR__.
25831 (vec_all_ne): Change parameterization of __altivec_scalar_pred
25832 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
25833 control (instead of _ARCH_PWR9 control) so that template
25834 definition uses power9-specific function.
25835 (vec_any_eq): Likewise.
25836 (vec_all_ne): Change macro definition to use a power9-specific
25837 expansion under #ifdef __POWER9_VECTOR__ control (instead of
25838 _ARCH_PWR9 control).
25839 (vec_any_eq) Likewise.
25840 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
25841 expansion for CMPNEF to remove support for xvcmpnesp instruction.
25842 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
25843 support for xvcmpnedp instruction.
25844 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
25845 macro expansion so that Power9 implementation of vec_all_ne does
25846 not use the AltiVec predicate framework.
25847 (VCMPNEH_P): Likewise.
25848 (VCMPNEW_P): Likewise.
25849 (VCMPNED_P): Likewise.
25850 (VCMPNEFP_P): Likewise.
25851 (VCMPNEDP_P): Likewise.
25852 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
25853 implementation of vec_any_eq to not use AltiVec predicate
25855 (VCMPAEH_P): Likewise.
25856 (VCMPAEW_P): Likewise.
25857 (VCMPAED_P): Likewise.
25858 (VCMPAEFP_P): Likewise.
25859 (VCMPAEDP_P): Likewise.
25860 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
25861 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
25862 not use the AltiVec predicate framework.
25863 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
25864 of vec_any_eq to not use AltiVec predicate framework.
25865 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25866 support for predefined __POWER9_VECTOR__ macro to indicate that
25867 Power9 instruction selection is enabled.
25868 (altivec_overloaded_builtins): Remove extraneous
25869 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
25870 function argument types RS6000_BTI_bool_V16QI and
25871 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
25872 entry for overloaded function argument types RS6000_BTI_bool_V4SI
25873 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
25874 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
25875 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
25876 Power9 for implementations of vec_cmpne. Change the signature for
25877 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
25878 (representing vec_all_ne) to remove the previously described first
25879 argument of type RS6000_BTI_INTSI, as this was an artifact of
25880 reliance on the AltiVec predicate framework, which is no longer
25881 used in the implementation of these functions. Add
25882 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
25883 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
25884 since, unlike the AltiVec predicate framework implementation, we
25885 do not share function descriptors between vec_alle and vec_anyeq.
25886 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
25887 set of modes that receive special treatment even when
25888 TARGET_P9_VECTOR is true. The special treatment emits code that
25889 does not depend on Power9 instructions.
25890 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
25891 define_expand to not rely on AltiVec predicate framework.
25892 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25894 (vector_ne_v2di_p): Change this define_expand to not rely on
25895 AltiVec predicate framework.
25896 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
25898 (vector_ne_<mode>_p): Change this define_expand to not rely on
25899 AltiVec predicate framework.
25900 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25902 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
25903 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
25904 define_insn pattern.
25905 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
25906 define_insn pattern because the xvcmpne<VSs>. instruction is not
25908 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
25909 instruction is not supported.
25911 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25913 * config/nvptx/nvptx.c: Include intl.h.
25915 2017-03-01 Martin Jambor <mjambor@suse.cz>
25918 * ipa-prop.h (ipa_bits): Removed field known.
25919 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
25920 to pointers. Adjusted their comments to warn about their sharing.
25921 (ipcp_transformation_summary): Change bits to a vector of pointers.
25922 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
25923 (ipa_get_ipa_bits_for_value): Declare.
25924 * tree-vrp.h (value_range): Mark as GTY((for_user)).
25925 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
25926 (ipa_bits_hash_table): Likewise.
25927 (ipa_vr_ggc_hash_traits): Likewise.
25928 (ipa_vr_hash_table): Likewise.
25929 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
25930 being pointers and vr_known being removed.
25931 (ipa_set_jf_unknown): Likewise.
25932 (ipa_get_ipa_bits_for_value): New function.
25933 (ipa_set_jfunc_bits): Likewise.
25934 (ipa_get_value_range): New overloaded functions.
25935 (ipa_set_jfunc_vr): Likewise.
25936 (ipa_compute_jump_functions_for_edge): Use the above functions to
25937 construct bits and vr parts of jump functions.
25938 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
25939 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25941 (ipcp_grow_transformations_if_necessary): Also allocate
25942 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25944 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
25945 them. Fix too long lines.
25946 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
25947 vr_known being removed.
25948 (ipa_read_jump_function): Use new setter functions to construct bits
25949 and vr parts of jump functions or set them to NULL.
25950 (write_ipcp_transformation_info): Adjust for bits being pointers.
25951 (read_ipcp_transformation_info): Likewise.
25952 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
25954 Include gt-ipa-prop.h.
25955 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
25957 (ipcp_store_bits_results): Likewise.
25958 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
25959 Do not write to existing jump functions but use a temporary instead.
25961 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25964 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
25965 attempt to use its first operand as BIT_FIELD_REF base.
25967 2017-03-01 Richard Biener <rguenther@suse.de>
25969 PR middle-end/79721
25970 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
25971 interpolating formula in wrapping arithmetic.
25972 (chrec_apply): Convert chrec_evaluate return value to wanted type.
25974 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25976 PR tree-optimization/79734
25977 * tree-vect-generic.c (expand_vector_condition): Optimize
25978 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
25979 Handle VEC_COND_EXPR where comparison has different inner width from
25980 type's inner width.
25982 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
25984 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
25985 markup, and similar issues. Remove @opindex entries for things
25986 that aren't options. Add missing -mmpy-option entries.
25988 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25990 PR tree-optimization/79737
25991 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
25992 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
25993 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
25994 instead of byte_size. Formatting fix.
25995 (shift_bytes_in_array_right): Formatting fix.
25997 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26000 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26001 condition on optimize for the leaf function test.
26003 2017-02-28 Martin Liska <mliska@suse.cz>
26006 * read-rtl-function.c (function_reader::handle_unknown_directive):
26007 Bail out when one uses -flto.
26009 2017-02-28 Martin Liska <mliska@suse.cz>
26011 * common.opt: Replace space with tabular for options of <number>
26013 * config/i386/i386.opt: Show <number> value for
26014 -mlarge-data-threshold.
26015 * opts.c (print_filtered_help): Do not display number in hexadecimal
26018 2017-02-28 Martin Liska <mliska@suse.cz>
26020 * common.opt: Fix --help=option -Q for options which are of
26023 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26025 * config/i386/i386.c (print_reg): Error out for values
26026 of 8-bit size in invalid integer register.
26028 2017-02-28 Martin Sebor <msebor@redhat.com>
26030 PR tree-optimization/79691
26031 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26033 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26036 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26037 gcc_unreachable with output_operand_lossage.
26039 2017-02-28 Richard Biener <rguenther@suse.de>
26041 PR tree-optimization/79740
26042 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26044 (visit_nary_op): Insert the nary into the hashtable if we
26045 pattern-matched sth.
26046 * tree-ssa-pre.c (eliminate_insert): Robustify.
26048 2017-02-28 Richard Biener <rguenther@suse.de>
26050 PR middle-end/79731
26051 * fold-const.c (decode_field_reference): Reject out-of-bound
26054 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26056 * config/i386/i386.c: Include intl.h.
26057 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26058 instead of just cond ? "..." : "...".
26059 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26060 * coverage.c (read_counts_file): Likewise.
26061 * omp-offload.c: Include intl.h.
26062 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26063 of just cond ? "..." : "...".
26064 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26065 of just cond ? "..." : "...".
26067 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26070 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26072 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26073 'tune for' CPU name.
26074 * config/arm/arm-cpu-data.h: Regenerated.
26076 2017-02-28 Richard Biener <rguenther@suse.de>
26078 PR tree-optimization/79732
26079 * tree-inline.c (expand_call_inline): Do not shadow var.
26081 2017-02-28 Richard Biener <rguenther@suse.de>
26083 PR tree-optimization/79723
26084 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
26085 address-space properly.
26087 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
26089 * doc/optinfo.texi (Optimization groups): Fix option used for
26091 * doc/invoke.texi (-fopt-info): Document "omp".
26092 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
26093 (OPTGROUP_ALL): Add OPTGROUP_OMP.
26094 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
26095 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
26096 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
26098 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
26100 * dumpfile.c (optgroup_options): Instead of "openmp", associate
26101 OPTGROUP_OMP with "omp".
26103 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
26106 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
26107 for arithmetic shift of unsigned V2DI.
26109 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
26111 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
26112 arc/linux.h headers.
26113 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
26114 (LINK_SPEC): Likewise.
26115 (ARC_TLS_EXTRA_START_SPEC): Likewise.
26116 (EXTRA_SPECS): Likewise.
26117 (STARTFILE_SPEC): Likewise.
26118 (ENDFILE_SPEC): Likewise.
26119 (LIB_SPEC): Likewise.
26120 (TARGET_SDATA_DEFAULT): Likewise.
26121 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
26122 (MULTILIB_DEFAULTS): Likewise.
26123 (DWARF2_UNWIND_INFO): Likewise.
26124 * config/arc/big.h: New file.
26125 * config/arc/elf.h: Likewise.
26126 * config/arc/linux.h: Likewise.
26127 * config/arc/t-uClibc: Remove.
26129 2017-02-27 Bin Cheng <bin.cheng@arm.com>
26131 PR tree-optimization/77536
26132 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
26133 (tree_transform_and_unroll_loop): Use above function to compute the
26134 estimated niter of unrolled loop and use it when scaling profile.
26135 Also use count info rather than frequency if it's non-zero.
26136 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
26137 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
26138 (vect_transform_loop): Call above function.
26140 2017-02-27 Richard Biener <rguenther@suse.de>
26142 PR tree-optimization/45397
26143 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
26144 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
26145 (visit_nary_op): Add pattern matching for CSEing sign-changed
26146 or truncated operations with wider ones.
26148 2017-02-27 Richard Biener <rguenther@suse.de>
26150 PR tree-optimization/79690
26151 * tree-vect-stmts.c (vectorizable_store): Use vector type
26152 built from the DR with address-space.
26154 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
26156 * doc/invoke.texi (Optimize Options): Refine the description
26157 of asan-use-after-return.
26159 2017-02-25 Alan Modra <amodra@gmail.com>
26161 PR rtl-optimization/79584
26162 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
26163 base, not ad->base_term, the reg within base. Remove assertion
26164 that ad->base == ad->base_term. Replace gen_int_mode using
26165 bogus mode with const0_rtx.
26167 2017-02-25 Jakub Jelinek <jakub@redhat.com>
26169 PR middle-end/79396
26170 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
26171 FMA_EXPR like tcc_binary or tcc_unary.
26173 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
26176 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
26178 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
26179 (output_loc_operands): Handle DW_OP_call_ref and
26180 DW_OP_GNU_variable_value.
26181 (struct variable_value_struct): New type.
26182 (struct variable_value_hasher): Likewise.
26183 (variable_value_hash): New variable.
26184 (string_types): Remove.
26185 (copy_loc_descr): New function.
26186 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
26187 (prepend_loc_descr_to_each): New function.
26188 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
26189 instead of add_loc_descr_to_each if the first argument is single
26190 location list and the second has multiple.
26191 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
26192 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
26193 when looking for variable value which doesn't have other location info.
26194 (loc_list_from_tree): Formatting fix.
26195 (gen_array_type_die): Simplify DW_AT_string_length handling.
26196 (adjust_string_types): Remove.
26197 (gen_subprogram_die): Don't call adjust_string_types nor test/set
26198 string_types. Call resolve_variable_values.
26199 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
26200 (resolve_addr_in_expr): Likewise. Add A argument.
26201 (copy_deref_exprloc): Remove deref argument. Adjust for the
26202 original expression being DW_OP_GNU_variable_value with optionally
26203 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
26204 optionally after it.
26205 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
26206 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
26207 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
26208 (variable_value_hasher::hash, variable_value_hasher::equal): New
26210 (resolve_variable_value_in_expr, resolve_variable_value,
26211 resolve_variable_values, note_variable_value_in_expr,
26212 note_variable_value): New functions.
26213 (dwarf2out_early_finish): Call note_variable_value on all toplevel
26216 2017-02-24 Jakub Jelinek <jakub@redhat.com>
26219 * opts.h (handle_generated_option): Add GENERATED_P argument.
26220 * opts-common.c (handle_option): Adjust function comment.
26221 (handle_generated_option): Add GENERATED_P argument, pass it to
26223 (control_warning_option): Pass false to handle_generated_option
26225 * opts.c (maybe_default_option): Pass true to handle_generated_option
26227 * optc-gen.awk: Likewise.
26229 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26231 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
26232 a REG, look at the REG it is a SUBREG of.
26233 (splitter for cmpeqsi_t): Ditto.
26235 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26237 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
26238 the special USEs with the pattern of the insn, not the insn itself.
26240 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
26243 * doc/invoke.texi: Document -mload-store-pairs.
26245 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26246 Sandra Loosemore <sandra@codesourcery.com>
26248 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
26249 argument isn't a CONST_INT.
26250 (nios2_alternate_compare_const): Assert op is a CONST_INT.
26251 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
26252 (nios2_validate_compare): Bypass alternate compare logic if *op2
26253 is not a CONST_INT.
26254 (ldstwm_operation_p): Return false if first_base is not a REG or
26255 if first_offset is not a CONST_INT.
26257 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26259 * config/cris/cris.md: Use correct operand in a define_peephole2.
26261 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26263 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
26265 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26267 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
26268 this_insn if it is an INSN or JUMP_INSN.
26269 (force_offsettable): Look at base, not at addr.
26270 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
26271 on things that aren't necessarily CONST_INTs.
26273 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
26275 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
26276 -mfpmath=sse is the default also for x86-32 targets with SSE2
26277 instruction set when @option{-ffast-math} is enabled
26279 2017-02-24 Jeff Law <law@redhat.com>
26281 PR rtl-optimizatoin/79286
26282 * ira.c (update_equiv_regs): Drop may_trap_p exception to
26285 2017-02-24 Richard Biener <rguenther@suse.de>
26287 PR tree-optimization/79389
26288 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
26291 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
26293 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
26294 function comment to reflect reality.
26295 (loop_exits_before_overflow): Fix typo in function description.
26297 2017-02-24 Richard Biener <rguenther@suse.de>
26299 PR tree-optimization/79389
26300 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
26301 properly that a threading opportunity exists. Detect conditional
26302 copy/constant propagation opportunities.
26304 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
26306 * config/visium/visium.md (type): Add trap.
26307 (b): New mode attribute.
26308 (*btst): Rename into...
26309 (*btst<mode>): ...this and adjust.
26310 (*cbranchsi4_btst_insn): Rename into...
26311 (*cbranch<mode>4_btst_insn): ...this and adjust.
26312 (trap): New define_insn.
26314 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26316 PR tree-optimization/79389
26317 * ifcvt.c (struct noce_if_info): Add rev_cond field.
26318 (noce_reversed_cond_code): New function.
26319 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
26320 reversed_comparison_code. Formatting fix.
26321 (noce_try_store_flag): Test rev_cond != NULL in addition to
26322 reversed_comparison_code.
26323 (noce_try_store_flag_constants): Likewise.
26324 (noce_try_store_flag_mask): Likewise.
26325 (noce_try_addcc): Use rev_cond if non-NULL instead of
26326 reversed_comparison_code.
26327 (noce_try_cmove_arith): Likewise. Formatting fixes.
26328 (noce_try_minmax, noce_try_abs): Clear rev_cond.
26329 (noce_find_if_block): Initialize rev_cond.
26330 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
26331 instead of false as last argument never attempt to reverse it
26334 2017-02-23 Bin Cheng <bin.cheng@arm.com>
26336 PR tree-optimization/79663
26337 * tree-predcom.c (combine_chains): Process refs in reverse order
26338 only for ZERO length chains, and add explaining comment.
26340 2017-02-23 Jeff Law <law@redhat.com>
26342 PR tree-optimization/79578
26343 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
26344 in call to operand_equal_p.
26346 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
26349 * config/i386/cpuid.h: Fix another undefined behavior.
26351 2017-02-23 Richard Biener <rguenther@suse.de>
26353 PR tree-optimization/79683
26354 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
26355 vector types for data-refs.
26357 2017-02-23 Martin Liska <mliska@suse.cz>
26359 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
26361 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26363 PR middle-end/79665
26364 * internal-fn.c (get_range_pos_neg): Moved to ...
26365 * tree.c (get_range_pos_neg): ... here. No longer static.
26366 * tree.h (get_range_pos_neg): New prototype.
26367 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
26368 are known to be in between 0 and signed maximum inclusive, try to
26369 expand both unsigned and signed divmod and use the cheaper one from
26372 2017-02-22 Jeff Law <law@redhat.com>
26374 PR tree-optimization/79578
26375 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
26376 to compare base operands.
26378 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
26381 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
26382 gpc_reg_operand instead of fpr_reg_operand.
26384 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
26386 * config/mips/mips.c (mips_return_in_memory): Force FP
26387 vector types to be returned in memory for o32 ABI.
26389 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26391 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
26392 instead of DW_TAG_member for static data member declarations and don't
26393 set no_linkage_name for static inline data members.
26394 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
26397 2017-02-22 Martin Liska <mliska@suse.cz>
26399 * doc/invoke.texi: Replace inequality signs with square brackets
26402 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26404 PR tree-optimization/68644
26405 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
26407 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
26410 * lra-constraints.c (simplify_operand_subreg): Handle
26411 WORD_REGISTER_OPERATIONS targets.
26413 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26416 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
26417 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
26418 elimination by swapping fld*.
26420 2017-02-22 Richard Biener <rguenther@suse.de>
26422 PR tree-optimization/79673
26423 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
26424 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
26425 irrelevant address-space qualifiers and avoiding a
26426 ADDR_SPACE_CONVERT_EXPR from fold_convert.
26428 2017-02-22 Richard Biener <rguenther@suse.de>
26430 PR tree-optimization/79666
26431 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
26432 to not symbolically negate if that may introduce undefined
26435 2017-02-22 Martin Liska <mliska@suse.cz>
26438 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
26439 * data-streamer-out.c (streamer_write_gcov_count_stream):
26441 * value-prof.c (stream_out_histogram_value): Make assert more
26442 precise based on type of counter.
26444 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
26447 * config/i386/i386.md (standard_x87sse_constant_load splitter):
26448 Use nonimmediate_operand instead of memory_operand for operand 1.
26449 (float-extend standard_x87sse_constant_load splitter): Ditto.
26451 2017-02-21 Jeff Law <law@redhat.com>
26453 PR tree-optimization/79621
26454 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
26455 blocks with edges to themselves.
26457 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26460 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
26461 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
26462 Use gimple_call_builtin_p.
26465 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
26466 on temporarily removed DEBUG_INSNs.
26468 PR tree-optimization/79649
26469 * tree-loop-distribution.c (classify_partition): Give up on
26470 non-generic address space loads/stores.
26472 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
26474 * doc/loop.texi (Loop manipulation): Remove nonexistent
26475 tree_ssa_loop_version from the documentation.
26476 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
26478 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26481 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
26482 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
26483 * config/rs6000/rs6000.c: Include except.h.
26484 (rs6000_expand_split_stack_prologue): Call
26485 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
26487 2017-02-21 Martin Jambor <mjambor@suse.cz>
26490 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
26491 have been analyzed.
26493 2017-02-21 Martin Jambor <mjambor@suse.cz>
26495 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
26496 for backward compatibility only.
26497 * doc/invoke.texi (Option Summary): Remove all references to
26498 -fipa-cp-alignment.
26500 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
26504 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26506 * lra-constraints.c (curr_insn_transform): Handle
26507 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26509 2017-02-21 Martin Liska <mliska@suse.cz>
26511 * config/i386/i386.opt: Replace -masm-dialect with -masm.
26513 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
26515 PR translation/79638
26516 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
26518 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
26521 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
26522 (arm_function_ok_for_sibcall): Return false for an indirect call by
26523 descriptor if all the argument registers are used.
26524 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
26525 alignment of the function.
26527 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26529 PR tree-optimization/61441
26530 * simplify-rtx.c (simplify_const_unary_operation): For
26531 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
26532 the sNaN unmodified.
26534 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26536 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26537 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26538 instead of SYSTEM_HEADER_DIR.
26540 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
26541 Martin Liška <mliska@suse.cz>
26543 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
26544 Fix typos and grammar, use active voice, and clarify.
26546 2017-02-20 Marek Polacek <polacek@redhat.com>
26548 PR middle-end/79537
26549 * gimplify.c (gimplify_expr): Handle unused *&&L;.
26552 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
26554 2017-02-20 Jakub Jelinek <jakub@redhat.com>
26557 * config/i386/i386.c (ix86_expand_builtin): Handle
26558 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
26559 ix86_builtins_isa[fcode].isa as a requirement of those
26560 flags and any other flag in the bitmask.
26561 (ix86_init_mmx_sse_builtins): Use 0 instead of
26562 ~OPTION_MASK_ISA_64BIT as mask.
26563 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
26564 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
26565 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
26566 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
26568 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26571 * lra-constraints.c (split_reg): Check requested split mode
26572 is supported by the register.
26574 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26576 * lra-constraints.c (simplify_operand_subreg): Remove early
26579 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26582 * lra-constraints.c (curr_insn_transform): Tighten condition
26583 for converting SUBREG reloads from OP_OUT to OP_INOUT.
26585 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26588 * lra-constraints.c (curr_insn_transform): Handle
26589 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26591 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
26594 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
26596 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
26598 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
26601 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
26604 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26606 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
26609 2017-02-18 Jakub Jelinek <jakub@redhat.com>
26611 * final.c (last_columnnum, override_columnnum): New variables.
26612 (final_start_function): Set last_columnnum, pass it to begin_prologue
26613 hook and pass 0 to dwarf2out_begin_prologue.
26614 (final_scan_insn): Update override_columnnum. Pass last_columnnum
26615 to source_line debug hook.
26616 (notice_source_line): Compute last_columnnum and for debug_column_info
26617 return true on column changes.
26618 * debug.h (struct gcc_debug_hooks): Add column argument to
26619 source_line and begin_prologue hooks.
26620 (debug_nothing_int_charstar_int_bool): Remove prototype.
26621 (debug_nothing_int_int_charstar,
26622 debug_nothing_int_int_charstar_int_bool): New prototypes.
26623 (dwarf2out_begin_prologue): Add column argument.
26624 * debug.c (do_nothing_debug_hooks): Adjust source_line and
26625 begin_prologue hooks.
26626 (debug_nothing_int_charstar_int_bool): Remove.
26627 (debug_nothing_int_int_charstar,
26628 debug_nothing_int_int_charstar_int_bool): New functions.
26629 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
26630 through to dwarf2out_source_line.
26631 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
26632 (dwarf2out_source_line): Add column argument, emit it if requested.
26633 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
26635 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26636 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26637 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
26638 through to dwarf2out_begin_prologue.
26639 (vmsdbgout_source_line): Add column argument, pass it through to
26640 dwarf2out_source_line.
26641 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
26642 dbxout_source_line caller.
26643 (dbxout_source_line): Add column argument.
26645 * common.opt (gno-column-info, gcolumn-info): New options.
26646 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
26647 (check_die): Also test for multiple DW_AT_decl_column attributes.
26648 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
26649 DW_AT_decl_column if requested.
26650 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
26652 (gen_variable_die): Likewise.
26653 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
26654 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
26657 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
26658 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
26659 (ix86_handle_option): Handle OPT_m3dnowa.
26660 * doc/invoke.texi (-m3dnowa): Document.
26661 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
26662 -m3dnowa instead of -m3dnow -march=athlon.
26665 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
26666 instead of gcc_assert for K, r and R code checks. Formatting fixes.
26668 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26671 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
26672 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
26673 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
26674 generator for vsx_xxpermdi_<mode>_be.
26675 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
26676 force big-endian semantics.
26677 (vsx_xxpermdi_<mode>_be): New define_expand with same
26678 implementation as previous version of vsx_xxpermdi_<mode>.
26680 2017-02-17 Jakub Jelinek <jakub@redhat.com>
26682 PR tree-optimization/79327
26683 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
26684 variable, its initialization and use.
26686 2017-02-17 Julia Koval <julia.koval@intel.com>
26688 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
26689 (OPTION_MASK_ISA_PKU_UNSET): New.
26690 (ix86_handle_option): Handle -mrdpid.
26691 * config/i386/cpuid.h (bit_RDPID): New.
26692 * config/i386/driver-i386.c (host_detect_local_cpu):
26693 Detect RDPID feature.
26694 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
26695 * config/i386/i386-c.c (ix86_target_macros_internal):
26697 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
26698 (ix86_valid_target_attribute_inner_p): Add "rdpid".
26699 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26700 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
26701 * config/i386/i386.md (define_insn "rdpid"): New.
26702 * config/i386/i386.opt Add -mrdpid.
26703 * config/i386/immintrin.h (_rdpid_u32): New.
26705 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
26707 PR rtl-optimization/79541
26708 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
26709 instead of transforming it into USE.
26711 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
26713 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
26714 If HONOR_SNANS (SFmode) force the input to a register.
26715 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
26716 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
26717 an frsp or similar insn.
26719 2017-02-17 Martin Liska <mliska@suse.cz>
26721 PR rtl-optimization/79577
26722 * params.def (selsched-max-sched-times): Increase minimum to 1.
26724 2017-02-17 Martin Liska <mliska@suse.cz>
26726 PR rtl-optimization/79574
26727 * gcse.c (want_to_gcse_p): Prevent integer overflow.
26729 2017-02-17 Martin Liska <mliska@suse.cz>
26731 PR tree-optimization/79529
26732 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
26733 ssa_defined_default_def_p to handle cases which are implicitly
26735 * tree-ssa.c (ssa_defined_default_def_p): New function.
26736 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
26737 which are implicitly defined.
26738 * tree-ssa.h (ssa_defined_default_def_p): Declare.
26740 2017-02-17 Richard Biener <rguenther@suse.de>
26742 PR middle-end/79576
26743 * params.def (max-ssa-name-query-depth): Limit to 10.
26745 2017-02-17 Richard Biener <rguenther@suse.de>
26747 PR tree-optimization/79552
26748 * tree-ssa-structalias.c (visit_loadstore): Properly verify
26751 2017-02-17 Richard Biener <rguenther@suse.de>
26754 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
26756 2017-02-17 Marek Polacek <polacek@redhat.com>
26758 PR middle-end/79536
26759 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
26760 (fold_negate_expr): New wrapper.
26762 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
26764 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
26765 Correct terminology and de-emphasize pre-standard behavior.
26767 2017-02-16 Alan Modra <amodra@gmail.com>
26769 PR rtl-optimization/79286
26770 * ira.c (def_dominates_uses): New function.
26771 (update_equiv_regs): Don't create an equivalence for insns that
26772 may trap where the register def does not dominate the use.
26774 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
26776 PR rtl-optimization/78127
26777 * lra.c (lra): Call lra_eliminate before finish the loop after
26780 2017-02-16 Richard Biener <rguenther@suse.de>
26782 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
26784 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
26785 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
26786 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
26787 (isl_val_int_from_wi): New function.
26788 (extract_affine_gmp): Rename to ...
26789 (extract_affine_wi): ... this, take a widest_int.
26790 (extract_affine_int): Just wrap extract_affine_wi.
26791 (add_param_constraints): Use isl_val_int_from_wi.
26792 (add_loop_constraints): Likewise, and extract_affine_wi.
26794 2017-02-15 Jeff Law <law@redhat.com>
26796 PR middle-end/79521
26797 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
26798 ira_init_register_move_cost_if_necessary.
26800 2017-02-15 Martin Sebor <msebor@redhat.com>
26802 PR middle-end/32003
26803 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
26804 removed in a prior commit.
26806 2017-02-15 Bin Cheng <bin.cheng@arm.com>
26808 PR tree-optimization/79347
26809 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
26810 counters during peeling.
26812 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
26814 * Makefile.in (site.exp): Remove "set ISLVER".
26816 2017-02-15 Jakub Jelinek <jakub@redhat.com>
26819 * real.c (real_from_integer): Call real_convert even for decimal.
26821 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
26824 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
26826 2017-02-14 Andrew Pinski <apinski@cavium.com>
26828 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
26829 cores and change the partno/implementer to be correct.
26830 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
26831 the 'B" as the implementer.
26832 * config/aarch64/aarch64-tune.md: Regenerate.
26834 2017-02-14 Carl Love <cel@us.ibm.com>
26836 * config/rs6000/rs6000.c: Add case statement entry to make the
26837 xvcvuxdsp built-in argument unsigned.
26838 * config/rs6000/vsx.md: Fix the source and return operand types so they
26839 match the instruction definitions from the ISA document. Fix typo
26840 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
26843 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
26846 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
26847 member early_clobber_alts.
26848 * lra-lives.c (reg_early_clobber_p): New.
26849 (process_bb_lives): Use it.
26850 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
26851 (debug_operand_data): Initialize early_clobber_alts.
26852 (setup_operand_alternative): Set up early_clobber_alts.
26853 (collect_non_operand_hard_regs): Ditto. Pass early clobber
26854 alternatives to new_insn_reg.
26855 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
26857 (lra_update_insn_regno_info): Pass the new arg.
26859 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26861 PR middle-end/79505
26862 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
26863 (new_oacc_loop_raw): Don't clear already cleared fields.
26866 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
26867 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
26868 _mm512_prefetch_i64gather_ps): New inline functions and macros.
26870 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
26873 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
26875 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
26878 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
26879 the extra instruction to the right place to store 128-bit constant
26882 2017-02-14 Martin Sebor <msebor@redhat.com>
26884 PR middle-end/79448
26885 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
26886 warning for strings of unknown length.
26888 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26890 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
26892 2017-02-14 Jeff Law <law@redhat.com>
26895 * ira-costs.c (scan_one_insn): Initialize register move costs
26896 for pseudos seen in USE/CLOBBER insns.
26898 PR tree-optimization/79095
26899 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
26900 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
26901 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
26902 if the operands are known to be not equal, then the resulting range
26904 (intersect_ranges): If the new range is ~[0,0] and the old range is
26905 wide, then prefer ~[0,0].
26906 * tree-vrp.c (overflow_comparison_p_1): New function.
26907 (overflow_comparison_p): New function.
26908 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
26909 if NAME is used in an overflow test.
26910 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
26911 overflow check that can be expressed as an equality test, then adjust
26912 ops to be that equality test.
26914 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26916 * config/s390/s390-builtin-types.def: Remove flags argument.
26917 * config/s390/s390.c (s390_init_builtins): Likewise.
26919 2017-02-14 Martin Liska <mliska@suse.cz>
26921 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
26922 vector. Fix trailing white spaces.
26924 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
26926 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
26929 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26931 PR rtl-optimization/68664
26932 * config/arm/arm.c (arm_sched_can_speculate_insn):
26933 New function. Declare prototype.
26934 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26936 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26938 PR rtl-optimization/68664
26939 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
26941 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26943 2017-02-14 Amit Pawar <amit.pawar@amd.com>
26945 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
26946 max skip bytes for function, loop and jump.
26948 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26950 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
26951 ABS_EXPR for gimple dump.
26953 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26956 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
26958 PR tree-optimization/79408
26959 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
26960 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
26961 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
26962 also if rhs1 is INTEGER_CST.
26964 2017-02-14 Richard Biener <rguenther@suse.de>
26966 PR middle-end/79432
26967 * tree-into-ssa.c (insert_phi_nodes): When the function can
26968 have abnormal edges rewrite SSA names with broken use-def
26969 dominance out of SSA and register them for PHI insertion.
26971 2017-02-13 Martin Sebor <msebor@redhat.com>
26973 PR middle-end/79496
26974 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
26975 clearing info.nowrite flag when snprintf size argument is a range.
26977 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26979 * cprop.c (cprop_jump): Add missing space in string literal.
26980 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
26981 (get_constraint_for_component_ref): Likewise.
26982 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
26983 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
26984 * lra-constraints.c (process_alt_operands): Likewise.
26985 * ipa-inline.c (inline_small_functions): Likewise.
26986 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
26987 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26988 * trans-mem.c (diagnose_tm_1_op): Likewise.
26989 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
26990 (grid_parallel_clauses_gridifiable): Likewise.
26992 * config/nvptx/mkoffload.c (process): Add space in between
26995 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
26996 "MOD4_SSE_REGS" and "ALL_REGS".
26998 * spellcheck.c (test_data): Add , in between "foo" and "food".
27000 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27003 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27004 boundary crossing check and subsequent code generation agree.
27006 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27008 * config/aarch64/aarch64.c (has_memory_op): Delete.
27009 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27012 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27014 PR rtl-optimization/79388
27015 PR rtl-optimization/79450
27016 * combine.c (distribute_notes): When removing TEM_INSN for which
27017 corresponding dest has last value recorded, invalidate that last
27020 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27022 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27023 of explicit '@'. Add missing assembly comment marker on branch costs
27026 2017-02-13 Nathan Sidwell <nathan@acm.org>
27028 * gengtype-lex.l (<in_struct>): Add '/'.
27030 2017-02-13 Martin Liska <mliska@suse.cz>
27033 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27035 2017-02-13 Richard Biener <rguenther@suse.de>
27037 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27039 * configure: Re-generate.
27040 * config.in: Likewise.
27041 * graphite-dependences.c: Simplify as if
27042 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27043 * graphite-isl-ast-to-gimple.c: Likewise.
27044 * graphite-optimize-isl.c: Likewise.
27045 * graphite-poly.c: Likewise.
27046 * graphite-sese-to-poly.c: Likewise.
27047 * graphite.h: Likewise.
27048 * toplev.c: Include isl/version.h and use isl_version () for
27049 printing the ISL version.
27050 * doc/install.texi: Update ISL requirement.
27052 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27054 * doc/standards.texi (Standards): Update reference to
27057 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27059 * doc/extend.texi (Named Address Spaces): sourceware.org now
27061 * doc/install.texi (Binaries): Ditto.
27064 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27066 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27067 standard terminology "stringize"/"stringizing" throughout.
27068 * doc/cppinternals.texi: Likewise.
27070 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27072 * doc/extend.texi: Fix some spelling mistakes and typos.
27073 * doc/invoke.texi: Likewise.
27075 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27078 * params.def (inline-min-speedup) Change from 10 to 8.
27080 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27082 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
27085 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27088 * ipa-inline-analysis.c (get_minimal_bb): New function.
27089 (record_modified): Use it.
27090 (remap_edge_change_prob): Handle also ancestor functions.
27092 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
27094 * doc/contrib.texi (Contributors): Remove broken link into
27095 the Mauve CVS repository.
27097 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27099 PR middle-end/79454
27100 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
27101 result computation whenever lhs doesn't have vector mode, not
27102 just when it has BLKmode.
27104 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27106 * doc/makefile.texi (profiledbootstrap): Refer to the
27107 installation instructions only in textual form.
27109 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27112 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
27114 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27116 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
27117 (Specific): Update mingw-w64 reference.
27119 (Specific): Remove broken link to Renesas RX processor.
27121 2017-02-10 Richard Biener <rguenther@suse.de>
27123 * toplev.c (process_options): Do not mention obsolete graphite
27124 options when printing sorry message about missing graphite support.
27125 Mention -floop-nest-optimize.
27127 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
27129 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
27130 (vtst_p16): Likewise.
27131 (vtstq_p8): Likewise.
27132 (vtstq_p16): Likewise.
27134 (vtstq_p64): Likewise.
27135 * config/arm/arm_neon.h (vgetq_lane_p64): New.
27136 (vset_lane_p64): New.
27137 (vsetq_lane_p64): New.
27139 2017-02-10 Jakub Jelinek <jakub@redhat.com>
27141 PR tree-optimization/79411
27142 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
27143 stmt operands are SSA_NAMEs used in abnormal phis.
27144 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
27147 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27150 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
27153 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27155 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
27157 2017-02-09 Jakub Jelinek <jakub@redhat.com>
27159 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
27163 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
27164 not arbitrary TREE_CONSTANT.
27167 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
27168 "omp declare target link" attribute unless is_global_var.
27169 * omp-offload.c (find_link_var_op): Likewise.
27171 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
27172 Chung-Lin Tang <cltang@codesourcery.com>
27174 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
27176 (gimplify_adjust_omp_clauses): Don't delete TILE.
27177 (gimplify_omp_for): Deal with TILE.
27178 * internal-fn.c (expand_GOACC_TILE): New function.
27179 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
27181 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
27182 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
27184 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
27185 avoid DIV for outermost collapse var.
27186 (expand_oacc_for): Insert tile element loop as needed. Adjust.
27187 Remove out of date comments, fix whitespace.
27188 * omp-general.c (omp_extract_for_data): Deal with tiling.
27189 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
27190 adjust OLF_DIM_BASE value.
27191 (struct omp_for_data): Add tiling field.
27192 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
27193 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
27194 for auto loops. Remove default auto determining, moved to
27195 oacc_loop_fixed_partitions.
27196 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
27197 stmts, add e_mask field.
27198 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
27199 (oacc_thread_numbers): Use oacc_dim_call.
27200 (oacc_xform_tile): New.
27201 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
27202 (finish_oacc_loop): Adjust for ifns vector.
27203 (oacc_loop_discover_walk): Append loop abstraction sites to list,
27204 add case for GOACC_TILE fns.
27205 (oacc_loop_xform_loop): Delete.
27206 (oacc_loop_process): Iterate over call list directly, and add
27207 handling for GOACC_TILE fns.
27208 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
27210 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
27211 vector partitioning to outer loops. Assign 2 partitions to loops
27212 when available. Add TILE handling.
27213 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
27214 (execite_oacc_device_lower): Process GOACC_TILE fns,
27215 ignore unknown specs.
27216 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
27217 * tree.c (omp_clause_num_ops): Adjust TILE ops.
27218 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
27220 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
27222 * configure.ac (ACX_BUGURL): Update.
27223 * configure: Regenerate.
27225 2017-02-09 Richard Biener <rguenther@suse.de>
27227 PR tree-optimization/69823
27228 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27229 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
27231 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27233 * config/arc/arc-c.def: Add __NPS400__ definition.
27234 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
27235 (TARGET_NPS400): Define.
27237 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27239 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
27241 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
27242 pointer, arch_info.
27243 (arc_cpu_types): Fill the arch_info field with a pointer into the
27244 arc_arch_types table.
27245 (arc_selected_cpu): Declare.
27246 * config/arc/arc.c (arc_selected_cpu): Make global.
27247 (arc_selected_arch): Delete.
27248 (arc_base_cpu): Delete.
27249 (arc_override_options): Remove references to deleted variables,
27250 update access to arch information.
27251 (ARC_OPT): Update access to arch information.
27252 (ARC_OPTX): Likewise.
27253 * config/arc/arc.h (arc_base_cpu): Remove declaration.
27254 (TARGET_ARC600): Update access to arch information.
27255 (TARGET_ARC601): Likewise.
27256 (TARGET_ARC700): Likewise.
27257 (TARGET_EM): Likewise.
27258 (TARGET_HS): Likewise.
27259 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
27262 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
27265 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
27266 condition/operands for integer GE/LE/GEU/LEU operations.
27268 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
27270 PR translation/79397
27271 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
27274 2017-02-08 Martin Jambor <mjambor@suse.cz>
27277 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
27278 whether allocation happened.
27279 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
27280 nothing was allocated.
27282 2017-02-08 Jakub Jelinek <jakub@redhat.com>
27284 PR tree-optimization/79408
27285 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
27286 constant, but SSA_NAME with a known integer range, use the minimum
27287 of that range instead of op1 to determine if modulo can be replaced
27288 with its first operand.
27290 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27292 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
27294 2017-02-08 Richard Biener <rguenther@suse.de>
27296 PR tree-optimization/71824
27297 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27298 Check all loops contained in the merged region.
27300 2017-02-07 Andrew Pinski <apinski@cavium.com>
27302 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
27304 2017-02-07 Andrew Pinski <apinski@cavium.com>
27306 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
27307 (thunderxt88): Likewise.
27308 (thunderxt81): Disable LSE and change v8.1 to v8.
27309 (thunderxt83): Likewise.
27311 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27312 Richard Biener <rguenther@suse.de>
27314 PR middle-end/79399
27315 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
27316 type from int to size_t.
27317 * ira-costs.c (struct_costs_size): Change type from int to size_t.
27319 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27321 PR rtl-optimization/79386
27322 * cprop.c (bypass_conditional_jumps): Initialize
27323 bypass_last_basic_block already before splitting bbs after
27324 unconditional traps...
27325 (bypass_conditional_jumps): ... rather than here.
27328 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
27329 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
27330 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
27331 fix -masm=intel patterns.
27333 2017-02-07 Richard Biener <rguenther@suse.de>
27335 PR tree-optimization/79256
27336 PR middle-end/79278
27337 * builtins.c (get_object_alignment_2): Use min_align_of_type
27338 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
27339 and ADJUST_FIELD_ALIGN.
27341 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
27343 * doc/tm.texi: Regenerate.
27344 * stor-layout.c (layout_decl): Adjust.
27345 (update_alignment_for_field): Likewise.
27346 (place_field): Likewise.
27347 (min_align_of_type): Likewise.
27348 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
27349 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
27350 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
27351 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
27352 * config/frv/frv.c (frv_adjust_field_align): Likewise.
27353 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
27354 * config/i386/i386.c (x86_field_alignment): Likewise.
27355 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
27356 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
27357 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
27358 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
27359 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
27360 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
27364 2017-01-30 Richard Biener <rguenther@suse.de>
27366 PR tree-optimization/79256
27367 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27368 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27371 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
27373 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
27374 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
27375 builtins to SImode and emit a zero-extend, if necessary.
27377 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27379 * docs/invoke.texi (RISC-V Options): Alphabetize.
27381 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27383 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
27386 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27388 * config/riscv/riscv.c: New file.
27389 * common/config/riscv/riscv-common.c: Likewise.
27390 * config.gcc: Likewise.
27391 * config/riscv/constraints.md: Likewise.
27392 * config/riscv/elf.h: Likewise.
27393 * config/riscv/generic.md: Likewise.
27394 * config/riscv/linux.h: Likewise.
27395 * config/riscv/multilib-generator: Likewise.
27396 * config/riscv/peephole.md: Likewise.
27397 * config/riscv/pic.md: Likewise.
27398 * config/riscv/predicates.md: Likewise.
27399 * config/riscv/riscv-builtins.c: Likewise.
27400 * config/riscv/riscv-c.c: Likewise.
27401 * config/riscv/riscv-ftypes.def: Likewise.
27402 * config/riscv/riscv-modes.def: Likewise.
27403 * config/riscv/riscv-opts.h: Likewise.
27404 * config/riscv/riscv-protos.h: Likewise.
27405 * config/riscv/riscv.h: Likewise.
27406 * config/riscv/riscv.md: Likewise.
27407 * config/riscv/riscv.opt: Likewise.
27408 * config/riscv/sync.md: Likewise.
27409 * config/riscv/t-elf-multilib: Likewise.
27410 * config/riscv/t-linux: Likewise.
27411 * config/riscv/t-linux-multilib: Likewise.
27412 * config/riscv/t-riscv: Likewise.
27413 * configure.ac: Likewise.
27414 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
27415 Waterman as RISC-V maintainers.
27416 * doc/install.texi: Add RISC-V entries.
27417 * doc/invoke.texi: Add RISC-V options section.
27418 * doc/md.texi: Add RISC-V constraints section.
27419 * configure: Regenerated.
27421 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27424 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
27425 false values to be constant vectors with all 0 or all 1 bits set.
27426 (vcondu<mode><mode>): Likewise.
27427 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
27429 (fpmask_comparison_operator): Update comment.
27430 (vecint_comparison_operator): New predicate.
27431 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
27432 vector conditionals when the true and false values are constant
27433 vectors with all 0 bits or all 1 bits set.
27435 2017-02-06 Martin Sebor <msebor@redhat.com>
27437 PR tree-optimization/79376
27438 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
27440 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
27442 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
27443 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
27444 to simplify split condition.
27446 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27448 * omp-expand.c (oxpand_omp_atomic_fetch_op,
27449 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
27452 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
27454 PR rtl-optimization/68664
27455 * target.def (can_speculate_insn): New hook.
27456 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
27457 * doc/tm.texi: Regenerate.
27458 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
27459 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
27460 (rs6000_sched_can_speculate_insn): New function.
27462 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27464 PR tree-optimization/79284
27465 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
27466 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
27467 vectorizable_mask_load_store, vectorizable_operation,
27468 vect_is_simple_cond, get_same_sized_vectype): Use it instead
27469 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
27470 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
27471 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
27472 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
27473 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27474 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
27475 is_gimple_assign (stmt). Replace another such test with
27476 is_gimple_assign (stmt).
27478 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27481 * config/avr/avr.c (rtl-iter.h): Include it.
27482 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
27483 (avr_legitimate_combined_insn): ...and implementation.
27485 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
27487 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
27488 * config/s390/s390.c (s390_const_operand_ok)
27489 (s390_canonicalize_comparison, s390_extract_part)
27490 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
27491 (s390_contiguous_bitmask_p, s390_rtx_costs)
27492 (legitimize_pic_address): Likewise.
27493 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
27494 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
27495 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
27496 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
27497 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
27499 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27501 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
27502 REGNO($0) == REGNO($1).
27504 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27506 * config/s390/linux.h(SIZE_TYPE): Add comment.
27508 2017-02-06 Julian Brown <julian@codesourcery.com>
27509 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27510 Virendra Pathak <virendra.pathak@broadcom.com>
27512 * config/aarch64/aarch64-cores.def: Change the scheduler
27514 * config/aarch64/aarch64.md: Include thunderx2t99.md.
27515 * config/aarch64/thunderx2t99.md: New file.
27517 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27519 * doc/standards.texi (Go Language): Update link to language
27522 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27524 * tree-eh.c (lower_resx): Sanitize profile.
27525 (cleanup_empty_eh_move_lp): Likewise.
27527 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27530 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
27532 * cfgloopmanip.h (loop_version): Update prototype.
27533 * modulo-sched.c (sms_schedule): Update call of loop_version.
27534 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
27535 * tree-parloops.c (gen_parallel_loop): Likewise.
27536 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
27537 * tree-ssa-loop-split.c (split_loop): Likewise.
27538 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
27539 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
27541 2017-02-05 Martin Liska <mliska@suse.cz>
27544 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
27546 (print_operand_address): Initialize a struct to zero.
27548 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27550 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
27551 garbage collector only in textual form.
27553 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27555 * doc/extend.texi (x86 specific memory model extensions for
27556 transactional memory): Simplify a phrase.
27558 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
27561 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
27562 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
27563 (atomic_storedi_1): Likewise.
27565 2017-02-04 Jakub Jelinek <jakub@redhat.com>
27567 PR tree-optimization/79338
27568 * tree-parloops.c (gather_scalar_reductions): Don't call
27569 vect_analyze_loop_form for loop->inner before destroying loop's
27572 2017-02-03 Martin Sebor <msebor@redhat.com>
27574 PR tree-optimization/79327
27575 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
27576 when precision has resulted in leading zeros.
27577 (format_integer): Adjust the likely counter to assume an unknown
27578 argument that may be zero is non-zero.
27580 2017-02-03 Jason Merrill <jason@redhat.com>
27583 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
27584 avoid copying non-taken branch.
27586 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27588 PR tree-optimization/79340
27589 * tree-vect-loop.c (vectorizable_reduction): Release
27590 vec_defs elements after safe_splicing them into other vectors.
27593 PR tree-optimization/79327
27594 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
27595 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
27597 (format_integer): Use wide_int_to_tree instead of build_int_cst
27598 + to_?hwi. If argmin is NULL, just set argmin and argmax to
27599 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
27600 of shortest and longest sequence.
27602 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
27604 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
27605 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
27607 2017-02-03 Walter Lee <walt@tilera.com>
27610 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
27611 after initial stackframe link reg save.
27612 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
27614 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27617 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
27618 wu for stxssp alternative.
27620 2017-02-03 Martin Sebor <msebor@redhat.com>
27622 PR tree-optimization/79352
27623 * gimple-fold.c (get_range_strlen): Add argument.
27624 (get_range_strlen): Change return type to bool.
27625 (get_maxval_strlen): Pass in a dummy argument.
27626 * gimple-fold.h (get_range_strlen): Change return type to bool.
27627 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
27628 * tree.h (array_at_struct_end_p): Add argument.
27629 * tree.c (array_at_struct_end_p): Handle it.
27631 2017-02-03 Martin Liska <mliska@suse.cz>
27634 * multiple_target.c (create_dispatcher_calls): Redirect edge
27635 from a caller of a dispatcher.
27636 (expand_target_clones): Make the clones local.
27637 (ipa_target_clone): Do both target clones and resolvers.
27638 (ipa_dispatcher_calls): Remove the pass.
27639 (pass_dispatcher_calls::gate): Likewise.
27640 (make_pass_dispatcher_calls): Likewise.
27641 * passes.def (pass_target_clone): Put as very first IPA early
27644 2017-02-03 Martin Liska <mliska@suse.cz>
27646 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
27647 in case of a function with ifunc attribute.
27649 2017-02-03 Martin Liska <mliska@suse.cz>
27651 * cgraph.c (cgraph_node::dump): Dump function version info.
27652 * symtab.c (symtab_node::dump_base): Add missing new line.
27654 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27656 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
27657 (ifcombine_ifandif): Use it.
27659 2017-02-03 Martin Liska <mliska@suse.cz>
27661 * doc/invoke.texi: Document default value for
27662 use-after-scope-direct-emission-threshold.
27664 2017-02-03 Martin Liska <mliska@suse.cz>
27666 PR tree-optimization/79339
27667 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
27668 (format_floating): Likewise.
27670 2017-02-03 Martin Liska <mliska@suse.cz>
27673 * ipa-prop.c (ipa_node_params_t::insert): Remove current
27675 (ipa_node_params_t::remove): Likewise.
27676 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
27677 initialization from removed ipa_node_params_t::insert.
27678 (ipa_node_params::~ipa_node_params): Move from removed
27679 ipa_node_params_t::release.
27680 * symbol-summary.h (symbol_summary::m_released): New member.
27681 Do not release a summary twice. Do not allow to call finalizer
27682 for types of a summary that live in GGC memory.
27684 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27686 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
27689 2017-02-02 Martin Sebor <msebor@redhat.com>
27691 PR middle-end/79275
27692 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
27693 (format_string): Tighten up the range of output for non-constant
27694 strings and correct the expected range for wide non-constant strings.
27696 2017-02-02 Martin Sebor <msebor@redhat.com>
27698 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
27700 PR middle-end/32003
27701 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
27703 (-fdump-tree-@var): Add to index and document how to come up
27704 with pass-specific option and dump file names.
27705 (-fdump-passes): Clarify where to look for output.
27707 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27709 PR middle-end/77445
27710 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
27711 statistics of the analyzed path; allow threading for speed when
27712 any of BBs along the path are optimized for speed.
27714 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
27716 PR middle-end/78468
27717 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
27718 settings of the virtual registers.
27721 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
27723 * explow.c (get_dynamic_stack_size): Take known alignment of stack
27724 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
27727 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27729 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
27730 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
27732 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27734 * config/s390/s390.md: Add missing comments with the expanded
27736 * config/s390/vector.md: Likewise.
27737 * config/s390/vx-builtins.md: Likewise.
27739 2017-02-02 Jakub Jelinek <jakub@redhat.com>
27742 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
27743 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
27744 conditions on a single line.
27746 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27748 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27749 __S390_VX__ to __VX__.
27751 2017-02-01 Andrew Pinski <apinski@cavium.com>
27753 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
27754 stmt_info to record_stmt_cost.
27755 (vect_get_known_peeling_cost): Pass stmt_info if known to
27757 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
27758 cpu_vector_cost field into
27759 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
27760 field into vec_int_stmt_cost and vec_fp_stmt_cost.
27761 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
27762 splitting of scalar_stmt_cost and vec_stmt_cost.
27763 (thunderx_vector_cost): Likewise.
27764 (cortexa57_vector_cost): LIkewise.
27765 (exynosm1_vector_cost): Likewise.
27766 (xgene1_vector_cost): Likewise.
27767 (thunderx2t99_vector_cost): Improve after the splitting of the two
27769 (aarch64_builtin_vectorization_cost): Update for the splitting of
27770 scalar_stmt_cost and vec_stmt_cost.
27772 2017-02-01 Torvald Riegel <triegel@redhat.com>
27773 Richard Henderson <rth@redhat.com>
27775 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
27776 conditional on existance of a fast atomic load.
27777 * optabs-query.c (can_atomic_load_p): New function.
27778 * optabs-query.h (can_atomic_load_p): Declare it.
27779 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
27780 no fast atomic load is available for the particular size of access.
27781 (expand_atomic_compare_and_swap): Likewise.
27782 (expand_atomic_load): Likewise.
27783 (expand_atomic_store): Likewise.
27784 (expand_atomic_fetch_op): Likewise.
27785 * testsuite/lib/target-supports.exp
27786 (check_effective_target_sync_int_128): Remove x86 because it provides
27787 no fast atomic load.
27788 (check_effective_target_sync_int_128_runtime): Likewise.
27790 2017-02-01 Richard Biener <rguenther@suse.de>
27792 * graphite.c: Include tree-vectorizer.h for find_loop_location.
27793 (graphite_transform_loops): Provide opt-info for optimized nests.
27794 * tree-parloop.c (parallelize_loops): Provide opt-info for
27795 parallelized loops.
27797 2017-02-01 Richard Biener <rguenther@suse.de>
27799 PR middle-end/79315
27800 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
27801 was not set before.
27803 2017-02-01 Richard Biener <rguenther@suse.de>
27805 PR tree-optimization/71824
27806 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27807 Verify the loops are valid in the merged SESE region.
27808 (scop_detection::can_represent_loop_1): Check analyzing the
27809 evolution of the number of iterations in the region succeeds.
27811 2017-01-31 Ian Lance Taylor <iant@golang.org>
27813 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
27814 REG_ARGS_SIZE note to 32-bit push insns and call insn.
27816 2017-01-31 David Malcolm <dmalcolm@redhat.com>
27818 PR preprocessor/79210
27819 * input.c (get_substring_ranges_for_loc): Replace line_width
27820 assertion with error-handling.
27822 2017-01-31 Richard Biener <rguenther@suse.de>
27824 PR tree-optimization/77318
27825 * graphite-sese-to-poly.c (extract_affine): Fix assert.
27826 (create_pw_aff_from_tree): Take loop parameter.
27827 (add_condition_to_pbb): Pass loop of the condition to
27828 create_pw_aff_from_tree.
27830 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27832 * config/s390/s390.c (s390_asan_shadow_offset): New function.
27833 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
27835 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
27839 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
27841 (convert_int_to_float128): Likewise.
27842 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
27843 (convert_int_to_float128): Likewise.
27844 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
27845 (UNSPEC_IEEE128_CONVERT): Likewise.
27846 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
27847 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
27848 Use local variables for IBM extended format.
27849 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
27850 (fix_trunc<mode>si2_fprs): Likewise.
27851 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
27852 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
27853 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
27854 to know that we can now have integers of all sizes in vector
27856 (fix<uns>_<mode>di2_hw): Likewise.
27857 (float<uns>_<mode>si2_hw): Likewise.
27858 (fix_<mode>si2_hw): Likewise.
27859 (fixuns_<mode>si2_hw): Likewise.
27860 (float<uns>_<mode>di2_hw): Likewise.
27861 (float_<mode>di2_hw): Likewise.
27862 (float_<mode>si2_hw): Likewise.
27863 (floatuns_<mode>di2_hw): Likewise.
27864 (floatuns_<mode>si2_hw): Likewise.
27865 (xscvqp<su>wz_<mode>): Delete, no longer used.
27866 (xscvqp<su>dz_<mode>): Likewise.
27867 (xscv<su>dqp_<mode>): Likewise.
27868 (ieee128_mfvsrd_64bit): Likewise.
27869 (ieee128_mfvsrd_32bit): Likewise.
27870 (ieee128_mfvsrwz): Likewise.
27871 (ieee128_mtvsrw): Likewise.
27872 (ieee128_mtvsrd_64bit): Likewise.
27873 (ieee128_mtvsrd_32bit): Likewise.
27875 2017-01-31 Martin Liska <mliska@suse.cz>
27878 * ipa-prop.c (ipa_free_all_node_params): Call release method
27879 instead of ~sumbol_summary to not to trigger double times
27882 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
27884 PR tree-optimization/71691
27885 * bitmap.h (class auto_bitmap): New.
27886 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
27887 is_maybe_undefined instead of ssa_undefined_value_p.
27889 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27891 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27892 __S390_ARCH_LEVEL__ to __ARCH__.
27894 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27896 PR tree-optimization/79267
27897 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
27898 if should_remove_lhs_p is true.
27900 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
27903 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
27904 (add_alignment_attribute): New.
27905 (base_type_die): Add alignment attribute.
27906 (subrange_type_die): Likewise.
27907 (modified_type_die): Likewise.
27908 (gen_array_type_die): Likewise.
27909 (gen_descr_array_type_die: Likewise.
27910 (gen_enumeration_type_die): Likewise.
27911 (gen_subprogram_die): Likewise.
27912 (gen_variable_die): Likewise.
27913 (gen_field_die): Likewise.
27914 (gen_ptr_to_mbr_type_die): Likewise.
27915 (gen_struct_or_union_type_die): Likewise.
27916 (gen_subroutine_type_die): Likewise.
27917 (gen_typedef_die): Likewise.
27918 (base_type_cmp): Compare alignment attribute.
27920 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27923 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
27924 (setb_unsigned) New pattern for setb with CCUNS.
27925 * config/rs6000/rs6000.c (expand_block_compare): Use a different
27926 subfc./subfe sequence to avoid overflow problems. Generate a
27927 shorter sequence with cmpld/setb for power9.
27928 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
27929 for generating subfc. instruction.
27930 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
27931 now uses this instruction.
27933 2017-01-30 Ian Lance Taylor <iant@google.com>
27936 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
27937 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
27939 2017-01-30 Martin Sebor <msebor@redhat.com>
27941 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
27942 Move constant to the right of a relational operator.
27943 (get_mpfr_format_length, format_character, format_string): Ditto.
27944 (should_warn_p, maybe_warn): Same.
27946 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
27948 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
27951 * asan.c (get_translation_unit_decl): Remove function.
27952 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
27954 2017-01-30 Martin Liska <mliska@suse.cz>
27956 PR gcov-profile/79259
27957 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
27958 -fprofile-generate.
27960 2017-01-30 Martin Liska <mliska@suse.cz>
27963 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
27964 Initialize variables with NULL value.
27966 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
27969 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
27971 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
27973 2017-01-30 Richard Biener <rguenther@suse.de>
27975 PR tree-optimization/79276
27976 * tree-vrp.c (process_assert_insertions): Properly adjust common
27977 when removing a duplicate.
27979 * gcc.dg/torture/pr79276.c: New testcase.
27981 2017-01-30 Richard Biener <rguenther@suse.de>
27983 PR tree-optimization/79256
27984 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27985 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27987 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
27989 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
27992 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
27993 ("*r<noxa>sbg_<mode>_sll_bitmask")
27994 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
27995 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
27996 Use contiguous_bitmask_nowrap_operand.
27998 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28001 * config/rs6000/altivec.h (vec_xl): Revise #define.
28002 (vec_xst): Likewise.
28004 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28006 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28008 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28010 PR rtl-optimization/79194
28011 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28012 traps before call to bypass_conditional_jumps.
28014 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28016 PR tree-optimization/71374
28017 * lra-constraints.c (check_conflict_input_operands): New.
28018 (match_reload): Use it.
28020 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28023 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28024 account to calculate conflict_set.
28026 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28028 PR rtl-optimization/78559
28029 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28030 other_insn in combine.
28032 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28034 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28035 uint16_type_node for BT_UINT16.
28037 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28039 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28040 "RTL Tests" to menu.
28041 (GIMPLE Tests): New node.
28042 (RTL Tests): New node.
28044 2017-01-27 Richard Biener <rguenther@suse.de>
28046 PR tree-optimization/79245
28047 * tree-loop-distribution.c (distribute_loop): Apply cost
28048 modeling also to detected patterns.
28050 2017-01-27 Richard Biener <rguenther@suse.de>
28052 PR tree-optimization/71433
28053 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28054 (compare_assert_loc): New function.
28055 (process_assert_insertions): Sort and optimize assert locations
28056 to remove duplicates and push down identical assertions on
28057 edges to their destination block.
28059 2017-01-27 Richard Biener <rguenther@suse.de>
28061 PR tree-optimization/79244
28062 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28063 out SSA names even if abnormal.
28065 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28067 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28068 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28069 instead of MPFR_RNDN.
28071 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28074 * arm.c (arm_option_override): Don't call build_target_option_node
28075 until after doing all option overrides.
28076 (arm_valid_target_attribute_tree): Likewise.
28078 2017-01-27 Martin Liska <mliska@suse.cz>
28080 * doc/invoke.texi (-fprofile-arcs): Document profiling support
28081 for {cd}tors and C++ {cd}tors.
28083 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
28085 * config/s390/s390.md ("*setmem_long_and")
28086 ("*setmem_long_and_31z"): Use zero_extend instead of and.
28088 2017-01-26 Martin Sebor <msebor@redhat.com>
28090 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
28093 2017-01-26 Martin Sebor <msebor@redhat.com>
28095 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
28096 HAVE_DFmode before using XFmode or DFmode.
28097 (parse_directive): Avoid using the z length modifier to avoid
28098 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
28100 PR middle-end/78703
28101 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
28102 to accept adjustment as an array.
28103 (get_int_range): New function.
28104 (struct directive): Make width and prec arrays.
28105 (directive::set_width, directive::set_precision): Call get_int_range.
28106 (format_integer, format_floating): Handle width and precision ranges.
28107 (format_string, parse_directive): Same.
28109 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28112 * dwarf2out.c (generate_skeleton_bottom_up): For children with
28113 comdat_type_p set, just clone them, but keep the children in the
28117 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
28118 which have direct callers with -fvar-tracking-assignments enabled
28120 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
28121 inside of type units.
28123 2017-01-26 Martin Sebor <msebor@redhat.com>
28125 PR middle-end/78703
28126 * gimple-ssa-sprintf.c (struct result_range): Add likely and
28128 (struct format_result): Replace number_chars, number_chars_min,
28129 and number_chars_max with a single member of struct result_range.
28131 (format_result::operator+=): Adjust.
28132 (struct fmtresult): Remove bounded. Handle likely and unlikely
28134 (fmtresult::adjust_for_width_or_precision): New function.
28135 (fmtresult:type_max_digits): New function.
28136 (bytes_remaining): Handle likely and unlikely counters.
28137 (min_bytes_remaining): Remove.
28138 (format_percent): Simplify.
28139 (format_integer, format_floating): Set likely and unlikely counters.
28140 (get_string_length, format_character, format_string): Same.
28141 (format_plain, should_warn_p): New function.
28142 (maybe_warn): Call should_warn_p. Update diagnostic messages
28143 and handle those for all directives, including plain strings.
28144 (format_directive): Handle likely and unlikely counters.
28145 Remove unnecessary quoting from diagnostics. Add an informational
28147 (add_bytes): Remove.
28148 (pass_sprintf_length::compute_format_length): Simplify.
28149 (try_substitute_return_value): Handle likely and unlikely counters.
28151 2017-01-26 Carl Love <cel@us.ibm.com>
28153 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
28154 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
28156 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
28159 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
28160 endianess for subregs into account.
28161 * lra-constraints.c (lra_constraints): Do risky transformations
28162 always on the first iteration.
28163 * lra-lives.c (check_pseudos_live_through_calls): Add arg
28164 last_call_used_reg_set.
28165 (process_bb_lives): Define and use last_call_used_reg_set.
28166 * lra.c (lra): Always continue after lra_constraints on the first
28169 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
28171 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
28173 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
28175 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28177 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
28178 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
28179 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
28180 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
28181 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
28182 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
28183 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
28184 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
28185 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
28187 2017-01-26 Marek Polacek <polacek@redhat.com>
28190 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
28191 for the third operand.
28193 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28195 PR middle-end/79236
28196 * omp-low.c (struct omp_context): Add simt_stmt field.
28197 (scan_omp_for): Return omp_context *.
28198 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
28199 context to the _simt_ SIMD stmt.
28200 (lower_omp_for): For combined SIMD with sibling _simt_
28201 SIMD, make sure to use the same decls in _looptemp_
28202 clauses as in the sibling.
28204 2017-01-26 David Sherwood <david.sherwood@arm.com>
28206 PR middle-end/79212
28207 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
28210 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28213 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
28214 emit fld b; fld a; if possible.
28216 * brig-builtins.def: Update copyright years.
28217 * config/arm/arm_acle_builtins.def: Update copyright years.
28219 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
28222 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
28223 constraint instead of o for the stxsd instruction.
28225 2017-01-25 Carl Love <cel@us.ibm.com>
28227 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
28228 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
28230 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
28232 * doc/invoke.texi (C++ Dialect Options): Fix typo.
28234 2017-01-25 Richard Biener <rguenther@suse.de>
28236 PR tree-optimization/69264
28237 * target.def (vector_alignment_reachable): Improve documentation.
28238 * doc/tm.texi: Regenerate.
28239 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
28241 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
28242 earlier changes with respect to TYPE_USER_ALIGN.
28243 (vector_alignment_reachable_p): Likewise. Improve dumping.
28245 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28248 * config/arm/arm.md (xordi3): Force constant operand into a register
28251 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28253 * doc/invoke.texi (-fstore-merging): Correct default optimization
28254 levels at which it is enabled.
28255 (-O): Move -fstore-merging from list to...
28258 2017-01-25 Richard Biener <rguenther@suse.de>
28261 * omp-expand.c: Include debug.h.
28262 (expand_omp_taskreg): Make sure to generate early debug before
28263 outlining anything from a function.
28264 (expand_omp_target): Likewise.
28265 (grid_expand_target_grid_body): Likewise.
28267 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
28270 * asan.c (get_translation_unit_decl): New function.
28271 (asan_add_global): Extract modules file name from globals
28272 TRANSLATION_UNIT_DECL name.
28274 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
28277 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
28278 for long calls with APCS frame and VFP.
28280 2017-01-24 David Malcolm <dmalcolm@redhat.com>
28282 * cfg.c (original_copy_tables_initialized_p): New function.
28283 * cfg.h (original_copy_tables_initialized_p): New decl.
28284 * cfgrtl.c (relink_block_chain): Guard the call to
28285 free_original_copy_tables with a call to
28286 original_copy_tables_initialized_p.
28287 * cgraph.h (symtab_node::native_rtl_p): New decl.
28288 * cgraphunit.c (symtab_node::native_rtl_p): New function.
28289 (symtab_node::needed_p): Don't assert for early assembly output
28290 for __RTL functions.
28291 (cgraph_node::finalize_function): Set "force_output" for __RTL
28293 (cgraph_node::analyze): Bail out early for __RTL functions.
28294 (analyze_functions): Update assertion to support __RTL functions.
28295 (cgraph_node::expand): Bail out early for __RTL functions.
28296 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
28298 * function.h (struct function): Update comment for field
28300 * gimple-expr.c: Include "tree-pass.h".
28301 (gimple_has_body_p): Return false for __RTL functions.
28302 * Makefile.in (OBJS): Add run-rtl-passes.o.
28303 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
28305 (gcc::pass_manager::get_clean_slate): New accessor.
28306 * passes.c: Include "insn-addr.h".
28307 (should_skip_pass_p): Add logging. Update logic for running
28308 "expand" to be compatible with both __GIMPLE and __RTL. Guard
28309 property-provider override so it is only done for gimple passes.
28311 (skip_pass): New function.
28312 (execute_one_pass): Call skip_pass when skipping passes.
28313 * read-md.c (md_reader::read_char): Support filtering
28314 the input to a subset of line numbers.
28315 (md_reader::md_reader): Initialize fields
28316 m_first_line and m_last_line.
28317 (md_reader::read_file_fragment): New function.
28318 * read-md.h (md_reader::read_file_fragment): New decl.
28319 (md_reader::m_first_line): New field.
28320 (md_reader::m_last_line): New field.
28321 * read-rtl-function.c (function_reader::create_function): Only
28322 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
28323 curr_properties. Set DECL_INITIAL to a dummy block.
28324 (read_rtl_function_body_from_file_range): New function.
28325 * read-rtl-function.h (read_rtl_function_body_from_file_range):
28327 * run-rtl-passes.c: New file.
28328 * run-rtl-passes.h: New file.
28330 2017-01-24 Jeff Law <law@redhat.com>
28332 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
28335 2017-01-24 Bin Cheng <bin.cheng@arm.com>
28337 PR tree-optimization/79159
28338 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
28339 (record_nonwrapping_iv): Improve boundary using above function if no
28340 value range information.
28342 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
28343 Martin Jambor <mjambor@suse.cz>
28345 * brig-builtins.def: New file.
28346 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
28347 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
28348 (DEF_HSAIL_SAT_BUILTIN): Likewise.
28349 (DEF_HSAIL_INTR_BUILTIN): Likewise.
28350 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
28351 * builtin-types.def (BT_INT8): New.
28352 (BT_INT16): Likewise.
28353 (BT_UINT8): Likewise.
28354 (BT_UINT16): Likewise.
28355 (BT_FN_ULONG): Likewise.
28356 (BT_FN_UINT_INT): Likewise.
28357 (BT_FN_UINT_ULONG): Likewise.
28358 (BT_FN_UINT_LONG): Likewise.
28359 (BT_FN_UINT_PTR): Likewise.
28360 (BT_FN_ULONG_PTR): Likewise.
28361 (BT_FN_INT8_FLOAT): Likewise.
28362 (BT_FN_INT16_FLOAT): Likewise.
28363 (BT_FN_UINT32_FLOAT): Likewise.
28364 (BT_FN_UINT16_FLOAT): Likewise.
28365 (BT_FN_UINT8_FLOAT): Likewise.
28366 (BT_FN_UINT64_FLOAT): Likewise.
28367 (BT_FN_UINT16_UINT32): Likewise.
28368 (BT_FN_UINT32_UINT16): Likewise.
28369 (BT_FN_UINT16_UINT16_UINT16): Likewise.
28370 (BT_FN_INT_PTR_INT): Likewise.
28371 (BT_FN_UINT_PTR_UINT): Likewise.
28372 (BT_FN_LONG_PTR_LONG): Likewise.
28373 (BT_FN_ULONG_PTR_ULONG): Likewise.
28374 (BT_FN_VOID_UINT64_UINT64): Likewise.
28375 (BT_FN_UINT8_UINT8_UINT8): Likewise.
28376 (BT_FN_INT8_INT8_INT8): Likewise.
28377 (BT_FN_INT16_INT16_INT16): Likewise.
28378 (BT_FN_INT_INT_INT): Likewise.
28379 (BT_FN_UINT_FLOAT_UINT): Likewise.
28380 (BT_FN_FLOAT_UINT_UINT): Likewise.
28381 (BT_FN_ULONG_UINT_UINT): Likewise.
28382 (BT_FN_ULONG_UINT_PTR): Likewise.
28383 (BT_FN_ULONG_ULONG_ULONG): Likewise.
28384 (BT_FN_UINT_UINT_UINT): Likewise.
28385 (BT_FN_VOID_UINT_PTR): Likewise.
28386 (BT_FN_UINT_UINT_PTR: Likewise.
28387 (BT_FN_UINT32_UINT64_PTR): Likewise.
28388 (BT_FN_INT_INT_UINT_UINT): Likewise.
28389 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
28390 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
28391 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
28392 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
28393 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
28394 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
28395 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
28396 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
28397 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
28398 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
28399 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
28400 * doc/frontends.texi: List BRIG FE.
28401 * doc/install.texi (Testing): Add BRIG tesring requirements.
28402 * doc/invoke.texi (Overall Options): Mention BRIG.
28403 * doc/standards.texi (Standards): Doucment BRIG HSA version.
28405 2017-01-24 Richard Biener <rguenther@suse.de>
28407 PR translation/79208
28408 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
28410 2017-01-24 Martin Jambor <mjambor@suse.cz>
28413 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
28414 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
28415 and known_contexts.
28417 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
28419 PR middle-end/79123
28420 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
28421 casts from signed to unsigned really don't have a range.
28423 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
28425 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
28426 GMP_RNDx for compatiblity.
28428 2017-01-24 Martin Liska <mliska@suse.cz>
28431 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
28432 that would prevent us to call alloca with -1 as argument.
28434 2017-01-24 Jakub Jelinek <jakub@redhat.com>
28436 * dwarf2out.c (output_compilation_unit_header, output_file_names):
28437 Avoid -Wformat-security warning.
28439 2017-01-23 Andrew Pinski <apinski@cavium.com>
28441 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
28444 2017-01-23 Martin Sebor <msebor@redhat.com>
28446 PR middle-end/78703
28447 * gimple-ssa-sprintf.c (warn_level): New global.
28448 (format_integer): Use it here and throughout the rest of the file.
28449 Use the same switch to compute sign as base.
28450 (maybe_warn): New function.
28451 (format_directive): Factor out warnings into maybe_warn.
28452 Add debugging output. Use warn_level.
28453 (add_bytes): Use warn_level.
28454 (pass_sprintf_length::compute_format_length): Add debugging output.
28455 (try_substitute_return_value): Same.
28456 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
28458 PR middle-end/78703
28459 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
28460 (struct fmtresult, format_integer, format_floating): Adjust.
28461 (fmtresult::fmtresult): Set max correctly in two argument ctor.
28462 (get_string_length, format_string,format_directive): Same.
28463 (pass_sprintf_length::compute_format_length): Same.
28464 (try_substitute_return_value): Simplify slightly.
28466 PR middle-end/78703
28467 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
28468 (fmtresult::operator+=): Outlined.
28469 (struct fmtresult): Add ctors.
28470 (struct conversion_spec): Rename...
28471 (struct directive): ...to this. Add and remove data members.
28472 (directive::set_width, directive::set_precision): New functions.
28473 (format_percent): Use fmtresult ctor.
28474 (get_width_and_precision): Remove.
28475 (format_integer): Make naming changes. Avoid computing width and
28477 (format_floating): Same. Adjust indentation.
28478 (format_character, format_none): New functions.
28479 (format_string): Moved character handling to format_character.
28480 (format_directive): Remove arguments, change return type.
28481 (parse_directive): New function.
28482 (pass_sprintf_length::compute_format_length): Move directive
28483 parsing to parse_directive.
28485 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28487 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
28488 (assign_assembler_name_if_needed): ... this.
28489 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
28490 (assign_assembler_name_if_needed): ... this.
28491 (free_lang_data_in_cgraph): Adjust callers.
28492 * cgraphunit.c (cgraph_node::analyze): Likewise.
28493 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
28496 2017-01-23 Richard Biener <rguenther@suse.de>
28498 PR tree-optimization/79088
28499 PR tree-optimization/79188
28500 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
28501 resetting loop bounds after last path deletion. Reset loop
28502 bounds of the target loop, make code match the comments.
28503 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
28504 Make sure loops need no fixups.
28506 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28508 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
28509 exponent support with double type for first argument.
28510 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
28511 type returned by __builtin_vec_extract_sig,
28512 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
28513 functions from "vector int" to "vector unsigned int" or from
28514 "vector long long int" to "vector unsigned long long int".
28515 Changed type returned by __builtin_vec_extract_exp,
28516 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
28517 functions from "vector int" to "vector unsigned int" or from
28518 "vector long long int" to "vector unsigned long long int".
28519 Changed return type of __builtin_vec_test_data_class,
28520 __builtin_vec_test_data_class_sp, and
28521 __builtin_vec_test_data_class_dp from "vector int" to
28522 "vector bool int" or from "vector long long int" to "vector bool
28523 long long int" and changed second argument type from "unsigned
28524 int" to "int". Added new overloaded function forms "vector float
28525 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
28526 "vector float __builtin_vec_insert_exp_sp (vector float, vector
28527 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
28528 double, vector unsigned long long int)" and "vector double
28529 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
28530 long int)". Changed return type of
28531 __builtin_scalar_test_data_class and
28532 __builtin_scalar_test_data_class_sp and
28533 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
28534 int" and changed second argument from "unsigned int" to "int".
28535 Changed type returned by __builtin_scalar_test_neg,
28536 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
28537 from "int" to "bool int". Added new overloaded function form
28538 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
28539 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
28540 exponent double-precision with floating point first argument.
28541 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
28542 documentation of scalar_test_data_class, scalar_test_neg,
28543 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
28544 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
28545 vec_test_data_class built-in functions to reflect refinements in
28546 their type signatures.
28548 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
28550 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
28552 (aarch64_elf_asm_destructor): Likewise.
28554 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
28556 PR rtl-optimization/78634
28557 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
28558 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
28559 * ifcvt.c (noce_try_cmove): Add missing cost check.
28561 PR rtl-optimization/71724
28562 * combine.c (if_then_else_cond): Look for situations where it is
28563 beneficial to undo the work of one of the recursive calls.
28565 2017-01-23 Bin Cheng <bin.cheng@arm.com>
28567 PR tree-optimization/70754
28568 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
28569 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
28570 combined stmt before it if not NULL.
28571 (combine_chains): Process refs reversely and compute dominance point
28574 2017-01-23 Martin Liska <mliska@suse.cz>
28576 PR tree-optimization/79196
28577 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
28578 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
28580 (strlen_optimize_stmt): Call the renamed function.
28582 2017-01-23 Michael Matz <matz@suse.de>
28584 PR tree-optimization/78384
28585 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
28587 2017-01-23 Richard Biener <rguenther@suse.de>
28589 PR tree-optimization/79186
28590 * tree-vrp.c (register_new_assert_for): Make sure we've seen
28591 both incoming edges before moving an assert.
28593 2017-01-23 Martin Jambor <mjambor@suse.cz>
28595 * ipa-prop.c (load_from_param_1): Removed.
28596 (load_from_unmodified_param): Bits from load_from_param_1 put back
28598 (load_from_param): Removed.
28599 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
28600 with stmt. Reverted back to use of load_from_unmodified_param.
28602 2017-01-23 Martin Jambor <mjambor@suse.cz>
28605 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
28606 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
28607 field a pointer to garbage collected vector, mark lattices and
28608 ipcp_orig_node with GTY((skip)).
28609 (ipa_get_param_count): Adjust to descriptors being a pointer.
28610 (ipa_get_param): Likewise.
28611 (ipa_get_type): Likewise.
28612 (ipa_get_param_move_cost): Likewise.
28613 (ipa_set_param_used): Likewise.
28614 (ipa_get_controlled_uses): Likewise.
28615 (ipa_set_controlled_uses): Likewise.
28616 (ipa_is_param_used): Likewise.
28617 (ipa_node_params_t): Move into garbage collector. New methods insert
28619 (ipa_node_params_sum): Annotate wth GTY(()).
28620 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
28622 (ipa_load_from_parm_agg): Adjust declaration.
28623 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
28624 * ipa-profile.c (ipa_profile): Likewise.
28625 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
28626 (ipa_populate_param_decls): Make descriptors parameter garbage
28628 (ipa_dump_param): Adjust to descriptors being a pointer.
28629 (ipa_alloc_node_params): Likewise.
28630 (ipa_initialize_node_params): Likewise.
28631 (load_from_param_1): Make descriptors parameter garbage collected.
28632 (load_from_unmodified_param): Likewise.
28633 (load_from_param): Likewise.
28634 (ipa_load_from_parm_agg): Likewise.
28635 (ipa_node_params::~ipa_node_params): Removed.
28636 (ipa_free_all_node_params): Remove call to delete operator.
28637 (ipa_node_params_t::insert): New.
28638 (ipa_node_params_t::remove): Likewise.
28639 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
28640 copy known_csts and known_contexts vectors.
28641 (ipa_read_node_info): Adjust to descriptors being a pointer.
28642 (ipcp_modif_dom_walker): Make m_descriptors field garbage
28644 (ipcp_transform_function): Make descriptors variable garbage
28647 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
28649 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
28650 * config/i386/avx512dqintrin.h: Ditto.
28651 * config/i386/avx512fintrin.h: Ditto.
28652 * config/i386/i386.c: Handle new builtins.
28653 * config/i386/i386-builtin.def: Add new builtins.
28654 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
28655 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
28657 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28658 Martin Liska <mliska@suse.cz>
28660 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
28661 * asan.c (asan_expand_poison_ifn): Support stores and use
28662 appropriate ASAN report function.
28663 * internal-fn.c (expand_ASAN_POISON_USE): New function.
28664 * internal-fn.def (ASAN_POISON_USE): Declare.
28665 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
28666 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
28667 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
28668 ASAN_POISON calls w/o LHS.
28669 * tree-ssa.c (execute_update_addresses_taken): Create clobber
28670 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
28671 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
28672 * gimplify.c (asan_poison_variables): Add attribute
28673 use_after_scope_memory to variables that really needs to live
28675 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
28676 having the attribute.
28678 2017-01-23 Martin Liska <mliska@suse.cz>
28680 * asan.c (create_asan_shadow_var): New function.
28681 (asan_expand_poison_ifn): Likewise.
28682 * asan.h (asan_expand_poison_ifn): New declaration.
28683 * internal-fn.c (expand_ASAN_POISON): Likewise.
28684 * internal-fn.def (ASAN_POISON): New builtin.
28685 * sanopt.c (pass_sanopt::execute): Expand
28686 asan_expand_poison_ifn.
28687 * tree-inline.c (copy_decl_for_dup_finish): Make function
28689 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
28690 * tree-ssa.c (is_asan_mark_p): New function.
28691 (execute_update_addresses_taken): Rewrite local variables
28692 (identified just by use-after-scope as addressable) into SSA.
28694 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28696 * doc/install.texi (Specific): opensource.apple.com uses https
28697 now. Remove trailing slash.
28699 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28701 * README.Portability: Remove note on an Irix compatibility issue.
28703 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
28705 * gcov.c (INCLUDE_ALGORITHM): Define.
28706 (INCLUDE_VECTOR): Define.
28707 No longer include <vector> and <algorithm> directly.
28709 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
28711 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
28713 * doc/invoke.texi (Code Gen Options): Ditto.
28715 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
28718 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
28720 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
28722 rtl-optimization/79125
28723 * cprop.c (local_cprop_pass): Handle cases where we make an
28724 unconditional trap.
28726 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
28730 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
28731 read from, for big endian.
28733 2017-01-20 Jiong Wang <jiong.wang@arm.com>
28735 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
28736 register pauth builtins for LP64 only.
28738 2017-01-20 Marek Polacek <polacek@redhat.com>
28741 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
28744 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28746 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
28748 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
28749 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
28750 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
28752 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28755 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
28756 in big-endian mode when they are not a single duplicated value.
28758 2017-01-20 Richard Biener <rguenther@suse.de>
28760 * BASE-VER: Bump to 7.0.1.
28762 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28764 * omp-low.c (omplow_simd_context): New struct. Use it...
28765 (lower_rec_simd_input_clauses): ...here and...
28766 (lower_rec_input_clauses): ...here to hold common data. Adjust all
28767 references to idx, lane, max_vf, is_simt.
28769 2017-01-20 Graham Markall <graham.markall@embecosm.com>
28771 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
28774 2017-01-20 Martin Jambor <mjambor@suse.cz>
28776 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
28777 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
28779 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
28780 (GTFILES): Rename hsa.c to hsa-common.c.
28781 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
28782 * hsa-dump.c: Likewise.
28783 * hsa-gen.c: Likewise.
28784 * hsa-regalloc.c: Likewise.
28785 * ipa-hsa.c: Likewise.
28786 * omp-expand.c: Likewise.
28787 * omp-low.c: Likewise.
28788 * toplev.c: Likewise.
28790 2017-01-20 Marek Polacek <polacek@redhat.com>
28793 * doc/invoke.texi: Document -Wduplicated-branches.
28794 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
28795 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
28796 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
28797 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
28798 return 0 only when not OEP_LEXICOGRAPHIC.
28799 (fold_build_cleanup_point_expr): Use the expression
28800 location when building CLEANUP_POINT_EXPR.
28801 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
28802 * tree.c (add_expr): Handle error_mark_node.
28804 2017-01-20 Martin Liska <mliska@suse.cz>
28807 * tree-profile.c (init_ic_make_global_vars): Do not call
28809 (gimple_init_gcov_profiler): Likewise.
28811 2017-01-20 Martin Liska <mliska@suse.cz>
28814 * cgraph.h (maybe_create_reference): Remove argument and
28816 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
28818 * ipa-cp.c (create_specialized_node): Likewise.
28819 * symtab.c (symtab_node::maybe_create_reference): Handle
28820 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
28822 2017-01-20 Martin Liska <mliska@suse.cz>
28824 * read-rtl-function.c (function_reader::create_function): Use
28825 build_decl instread of build_decl_stat.
28827 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
28829 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
28830 * config/i386/avx512dqintrin.h: Ditto.
28831 * config/i386/avx512fintrin.h: Ditto.
28832 * config/i386/i386-builtin-types.def: Add new types.
28833 * config/i386/i386.c: Handle new types.
28834 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
28835 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
28836 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
28837 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
28838 (__builtin_ia32_kshiftridi): New.
28839 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
28841 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
28845 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
28846 define to rs6000_init_stack_protect_guard.
28847 (rs6000_init_stack_protect_guard): New function.
28849 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28850 Yunqiang Su <yunqiang.su@imgtec.com>
28852 * config.gcc (supported_defaults): Add madd4.
28853 (with_madd4): Add validation.
28854 (all_defaults): Add madd4.
28855 * config/mips/mips.opt (mmadd4): New option.
28856 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28858 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28860 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
28861 (ISA_HAS_FUSED_MADD4): Likewise.
28862 * doc/invoke.texi (-mmadd4): Document the new option.
28863 * doc/install.texi (--with-madd4): Document the new option.
28865 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28867 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
28868 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
28869 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
28870 (aarch64_init_pauth_hint_builtins): New.
28871 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
28872 (aarch64_expand_builtin): Expand new builtins.
28874 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28876 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
28877 * combine-stack-adj.c (no_unhandled_cfa): Handle
28878 REG_CFA_TOGGLE_RA_MANGLE.
28879 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
28880 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
28881 info for return address signing.
28882 (aarch64_expand_epilogue): Likewise.
28884 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28886 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
28887 * config/aarch64/aarch64-protos.h
28888 (aarch64_return_address_signing_enabled): New declaration.
28889 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
28891 (aarch64_expand_prologue): Sign return address before it's pushed onto
28893 (aarch64_expand_epilogue): Authenticate return address fetched from
28895 (aarch64_override_options): Sanity check for ILP32 and ISA level.
28896 (aarch64_attributes): New function attributes for "sign-return-address".
28897 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
28898 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
28899 ("*do_return"): Generate combined instructions according to key index.
28900 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
28901 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
28903 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
28904 * config/aarch64/aarch64.opt (msign-return-address=): New.
28905 * doc/extend.texi (AArch64 Function Attributes): Documents
28906 "sign-return-address=".
28907 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
28909 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28911 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
28912 overall option summary.
28914 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28916 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
28917 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
28918 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
28919 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
28921 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
28923 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
28924 -mpower9-minmax by default for -mcpu=power9.
28925 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
28926 128-bit floating point.
28928 2017-01-20 Alan Modra <amodra@gmail.com>
28930 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
28931 optimizing for size.
28933 2017-01-20 Alan Modra <amodra@gmail.com>
28936 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
28937 for strcmp and strncmp from corresponding builtin decl.
28939 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28941 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
28942 instead of i386/rtems-64.h.
28943 * config/i386/rtems-64.h: Remove.
28945 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28949 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
28951 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
28953 2017-01-19 Tamar Christina <tamar.christina@arm.com>
28955 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
28956 Change int to HOST_WIDE_INT.
28957 * config/aarch64/aarch64-protos.h
28958 (aarch64_simd_gen_const_vector_dup): Likewise.
28959 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
28961 2017-01-19 David Malcolm <dmalcolm@redhat.com>
28963 * langhooks-def.h (lhd_type_for_size): New decl.
28964 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
28965 * langhooks.c (lhd_type_for_size): New function, taken from
28968 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
28970 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
28971 define_bypass for CR latency.
28972 (power9-cracked-alu): Update bypass latency and remove power9-branch.
28973 (power9-alu2): Add define_bypass for CR latency.
28975 (power9-mul): Update insn latency.
28976 (power9-mul-compare): Update insn latency, bypass latency and remove
28979 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28981 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
28983 * config/aarch64/aarch64.md
28984 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
28985 aarch64_nopcrelative_literal_loads.
28986 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
28988 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
28990 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
28991 TARGET_LOONGSON_3A.
28992 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
28994 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
28997 * config.gcc (supported_defaults): Add lxc1-sxc1.
28998 (with_lxc1_sxc1): Add validation.
28999 (all_defaults): Add lxc1-sxc1.
29000 * config/mips/mips.opt (mlxc1-sxc1): New option.
29001 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29003 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29004 __mips_no_lxc1_sxc1.
29005 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29006 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29007 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29009 2017-01-19 Richard Biener <rguenther@suse.de>
29011 PR tree-optimization/72488
29012 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29013 sure to restore SSA info.
29014 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29016 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29018 PR rtl-optimization/79121
29019 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29020 of the inner type when shifting an extended value.
29022 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29025 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29026 interposable aliases.
29028 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29031 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29032 Use the evmergelohi instruction.
29033 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29034 (mov_sitf_e500_subreg8_2_be): Likewise.
29035 (mov_sitf_e500_subreg12_2_le): Likewise.
29036 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29037 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29038 (mov_sitf_e500_subreg8_2_le): Likewise.
29039 (mov_sitf_e500_subreg12_2_be): Likewise.
29041 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29043 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29044 attribute from vecsimple to vecperm.
29045 (altivec_vbpermq2): Likewise.
29047 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29050 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29052 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29053 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29054 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29055 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29056 case where N arg is SIZE_MAX.
29057 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29058 (cmpstrsi): Add pattern.
29060 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29062 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29063 __builtin_vec_revb builtins.
29064 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29065 built-in functions to support generation of the ISA 3.0 XXBR<x>
29066 vector byte reverse instructions.
29067 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29068 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29069 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29070 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29071 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29072 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29073 (P9V_BUILTIN_VEC_REVB): Likewise.
29074 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29075 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29076 (p9_xxbrq_v16qi): Likewise.
29077 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
29078 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
29079 (p9_xxbrh_v8hi): Likewise.
29080 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
29081 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
29082 vec_revb built-in functions.
29084 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
29086 PR rtl-optimization/78952
29087 * config/i386/i386.md (any_extract): New code iterator.
29088 (*insvqi_2): Use any_extract for source operand.
29089 (*insvqi_3): Use any_shiftrt for source operand.
29091 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29093 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
29095 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
29097 2017-01-18 Matthias Klose <doko@ubuntu.com>
29099 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
29101 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29103 * config/rs6000/altivec.h (vec_bperm): Change #define.
29104 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
29105 (altivec_vbpermq2): New define_insn.
29106 (altivec_vbpermd): Likewise.
29107 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
29108 function interface.
29109 (VBPERMD): Likewise.
29110 (VBPERM): New polymorphic function interface.
29111 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
29112 Add entries for P9V_BUILTIN_VEC_VBPERM.
29113 * doc/extend.texi: Add interfaces for vec_bperm.
29115 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29117 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
29118 first letter of error messages.
29119 (s390_resolve_overloaded_builtin): Likewise.
29120 * config/s390/s390.c (s390_expand_builtin): Likewise.
29121 (s390_invalid_arg_for_unprototyped_fn): Likewise.
29122 (s390_valid_target_attribute_inner_p): Likewise.
29123 * config/s390/s390.md ("tabort"): Likewise.
29125 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
29127 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
29128 (ISA_AVOID_DIV_HILO): New macro.
29129 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
29130 (ISA_HAS_DDIV): Likewise.
29132 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29134 * doc/invoke.texi (fabi-version): Correct number of occurrences.
29136 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29138 * doc/invoke.texi (fabi-version): Spelling fix.
29140 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29143 * doc/invoke.texi (fabi-version): Mention mangling fix for
29146 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29149 * doc/invoke.texi (fabi-version): Document discriminator mangling.
29151 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
29154 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
29155 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
29157 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
29159 (stack_protect_test): Ditto.
29160 * config/rs6000/rs6000.opt (mstack-protector-guard=,
29161 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
29163 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
29164 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
29165 -mstack-protector-guard-offset=.
29166 (RS/6000 and PowerPC Options): Ditto.
29168 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29170 * config/i386/i386.h (MASK_CLASS_P): New define.
29171 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
29172 there are no registers from different register sets also when
29173 mask registers are used. Update function comment.
29174 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
29175 to (*k/*r) and (*k/*km) alternatives.
29177 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29179 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
29180 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
29181 (EH_RETURN_HANDLER_RTX): New define.
29182 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
29183 Force frame pointer in EH return functions.
29184 (aarch64_expand_epilogue): Add barrier for eh_return.
29185 (aarch64_final_eh_return_addr): Remove.
29186 (aarch64_eh_return_handler_rtx): New function.
29187 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
29189 (aarch64_eh_return_handler_rtx): New prototype.
29191 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29193 * config/rs6000/altivec.h (vec_rlmi): New #define.
29194 (vec_vrlnm): Likewise.
29195 (vec_rlnm): Likewise.
29196 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
29197 (UNSPEC_VRLNM): Likewise.
29198 (VIlong): New mode iterator.
29199 (altivec_vrl<VI_char>mi): New define_insn.
29200 (altivec_vrl<VI_char>nm): Likewise.
29201 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
29203 (VRLDNM): Likewise.
29204 (RLNM): New polymorphic function entry.
29205 (VRLWMI): New monomorphic function entry.
29206 (VRLDMI): Likewise.
29207 (RLMI): New polymorphic function entry.
29208 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
29209 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
29210 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
29213 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29216 * dwarf2out.c (field_byte_offset): Restore the
29217 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
29218 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
29219 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
29222 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29225 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
29227 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29230 * dwarf2out.c (add_data_member_location_attribute): For constant
29231 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
29232 instead of DW_AT_data_member_location, DW_AT_bit_offset and
29233 DW_AT_byte_size attributes.
29235 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29237 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
29238 after forcing to constant memory when the code model is medium.
29240 2017-01-17 Julia Koval <julia.koval@intel.com>
29243 * config/i386/avx512fintrin.h
29244 (_mm512_i32gather_ps): Change __addr type to void const*.
29245 (_mm512_mask_i32gather_ps): Ditto.
29246 (_mm512_i32gather_pd): Ditto.
29247 (_mm512_mask_i32gather_pd): Ditto.
29248 (_mm512_i64gather_ps): Ditto.
29249 (_mm512_mask_i64gather_ps): Ditto.
29250 (_mm512_i64gather_pd): Ditto.
29251 (_mm512_mask_i64gather_pd): Ditto.
29252 (_mm512_i32gather_epi32): Ditto.
29253 (_mm512_mask_i32gather_epi32): Ditto.
29254 (_mm512_i32gather_epi64): Ditto.
29255 (_mm512_mask_i32gather_epi64): Ditto.
29256 (_mm512_i64gather_epi32): Ditto.
29257 (_mm512_mask_i64gather_epi32): Ditto.
29258 (_mm512_i64gather_epi64): Ditto.
29259 (_mm512_mask_i64gather_epi64): Ditto.
29260 (_mm512_i32scatter_ps): Change __addr type to void*.
29261 (_mm512_mask_i32scatter_ps): Ditto.
29262 (_mm512_i32scatter_pd): Ditto.
29263 (_mm512_mask_i32scatter_pd): Ditto.
29264 (_mm512_i64scatter_ps): Ditto.
29265 (_mm512_mask_i64scatter_ps): Ditto.
29266 (_mm512_i64scatter_pd): Ditto.
29267 (_mm512_mask_i64scatter_pd): Ditto.
29268 (_mm512_i32scatter_epi32): Ditto.
29269 (_mm512_mask_i32scatter_epi32): Ditto.
29270 (_mm512_i32scatter_epi64): Ditto.
29271 (_mm512_mask_i32scatter_epi64): Ditto.
29272 (_mm512_i64scatter_epi32): Ditto.
29273 (_mm512_mask_i64scatter_epi32): Ditto.
29274 (_mm512_i64scatter_epi64): Ditto.
29275 (_mm512_mask_i64scatter_epi64): Ditto.
29276 * config/i386/avx512pfintrin.h
29277 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
29278 (_mm512_mask_prefetch_i32gather_ps): Ditto.
29279 (_mm512_mask_prefetch_i64gather_pd): Ditto.
29280 (_mm512_mask_prefetch_i64gather_ps): Ditto.
29281 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
29282 (_mm512_prefetch_i32scatter_ps): Ditto.
29283 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
29284 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
29285 (_mm512_prefetch_i64scatter_pd): Ditto.
29286 (_mm512_prefetch_i64scatter_ps): Ditto.
29287 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
29288 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
29289 * config/i386/avx512vlintrin.h
29290 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
29291 (_mm_mmask_i32gather_ps): Ditto.
29292 (_mm256_mmask_i32gather_pd): Ditto.
29293 (_mm_mmask_i32gather_pd): Ditto.
29294 (_mm256_mmask_i64gather_ps): Ditto.
29295 (_mm_mmask_i64gather_ps): Ditto.
29296 (_mm256_mmask_i64gather_pd): Ditto.
29297 (_mm_mmask_i64gather_pd): Ditto.
29298 (_mm256_mmask_i32gather_epi32): Ditto.
29299 (_mm_mmask_i32gather_epi32): Ditto.
29300 (_mm256_mmask_i32gather_epi64): Ditto.
29301 (_mm_mmask_i32gather_epi64): Ditto.
29302 (_mm256_mmask_i64gather_epi32): Ditto.
29303 (_mm_mmask_i64gather_epi32): Ditto.
29304 (_mm256_mmask_i64gather_epi64): Ditto.
29305 (_mm_mmask_i64gather_epi64): Ditto.
29306 (_mm256_i32scatter_ps): Change __addr type to void*.
29307 (_mm256_mask_i32scatter_ps): Ditto.
29308 (_mm_i32scatter_ps): Ditto.
29309 (_mm_mask_i32scatter_ps): Ditto.
29310 (_mm256_i32scatter_pd): Ditto.
29311 (_mm256_mask_i32scatter_pd): Ditto.
29312 (_mm_i32scatter_pd): Ditto.
29313 (_mm_mask_i32scatter_pd): Ditto.
29314 (_mm256_i64scatter_ps): Ditto.
29315 (_mm256_mask_i64scatter_ps): Ditto.
29316 (_mm_i64scatter_ps): Ditto.
29317 (_mm_mask_i64scatter_ps): Ditto.
29318 (_mm256_i64scatter_pd): Ditto.
29319 (_mm256_mask_i64scatter_pd): Ditto.
29320 (_mm_i64scatter_pd): Ditto.
29321 (_mm_mask_i64scatter_pd): Ditto.
29322 (_mm256_i32scatter_epi32): Ditto.
29323 (_mm256_mask_i32scatter_epi32): Ditto.
29324 (_mm_i32scatter_epi32): Ditto.
29325 (_mm_mask_i32scatter_epi32): Ditto.
29326 (_mm256_i32scatter_epi64): Ditto.
29327 (_mm256_mask_i32scatter_epi64): Ditto.
29328 (_mm_i32scatter_epi64): Ditto.
29329 (_mm_mask_i32scatter_epi64): Ditto.
29330 (_mm256_i64scatter_epi32): Ditto.
29331 (_mm256_mask_i64scatter_epi32): Ditto.
29332 (_mm_i64scatter_epi32): Ditto.
29333 (_mm_mask_i64scatter_epi32): Ditto.
29334 (_mm256_i64scatter_epi64): Ditto.
29335 (_mm256_mask_i64scatter_epi64): Ditto.
29336 (_mm_i64scatter_epi64): Ditto.
29337 (_mm_mask_i64scatter_epi64): Ditto.
29338 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
29339 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
29340 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
29341 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
29342 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
29343 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
29344 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
29345 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
29346 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
29347 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
29348 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
29349 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
29350 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
29351 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
29352 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
29353 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
29354 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
29355 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
29356 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
29357 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
29358 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
29359 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
29360 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
29361 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
29362 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
29363 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
29364 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
29365 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
29366 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
29367 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
29368 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
29369 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
29370 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
29371 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
29372 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
29373 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
29374 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
29375 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
29376 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
29377 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
29378 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
29379 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
29380 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
29381 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
29382 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
29383 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
29384 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
29385 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
29386 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
29387 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
29388 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
29389 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
29390 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
29391 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
29392 definitions accordingly.
29394 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
29395 Kuan-Lin Chen <kuanlinchentw@gmail.com>
29398 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
29401 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
29404 * ira-conflicts.c (ira_build_conflicts): Update total conflict
29405 hard regs for inner regno.
29407 2017-01-17 Martin Liska <mliska@suse.cz>
29410 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
29411 assumption and add comment.
29413 2017-01-17 Nathan Sidwell <nathan@acm.org>
29415 * ipa-visibility.c (localize_node): New function, broken out of ...
29416 (function_and_variable_visibility): ... here. Call it.
29418 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29420 PR middle-end/77445
29421 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
29422 correctly set frequency of oudgoing edge.
29423 (duplicate_thread_path): Fix profile updating.
29425 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29428 * configure.ac: Add GCC_BASE_VER.
29429 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
29430 version from BASE-VER file.
29431 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
29432 (gcc.o): Depend on $(BASEVER).
29433 * common.opt (dumpfullversion): New option.
29434 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
29435 * doc/invoke.texi: Document -dumpfullversion.
29436 * doc/install.texi: Document --with-gcc-major-version-only.
29437 * configure: Regenerated.
29439 2017-01-17 Richard Biener <rguenther@suse.de>
29441 PR tree-optimization/71433
29442 * tree-vrp.c (register_new_assert_for): Merge same asserts
29443 on all incoming edges.
29444 (process_assert_insertions_for): Handle insertions at the
29447 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
29449 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
29450 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
29452 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
29455 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
29458 2017-01-17 Alan Modra <amodra@gmail.com>
29461 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
29462 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
29463 symbolic stack limit when pic.
29465 2017-01-16 Martin Sebor <msebor@redhat.com>
29467 PR tree-optimization/78608
29468 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
29470 2017-01-16 Jeff Law <law@redhat.com>
29473 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
29474 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
29475 for several include directories that may be relative to sysroot.
29476 * config/i386/x-mingw32 (gplus_includedir): Define.
29477 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
29478 (native_system_includedir): Likewise.
29479 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
29480 override if TARGET_SYSTEM_ROOT is defined.
29481 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
29483 PR tree-optimization/79090
29484 PR tree-optimization/33562
29485 PR tree-optimization/61912
29486 PR tree-optimization/77485
29487 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
29488 and computed trims into the dump file.
29490 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29492 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
29494 2017-01-16 Jakub Jelinek <jakub@redhat.com>
29497 * gimplify.c (gimplify_init_constructor): If want_value and
29498 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
29502 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
29503 sequence. Formatting fixes.
29504 (doloop_optimize): Formatting fixes.
29507 * gcc.c (debug_level_greater_than_spec_func): New function.
29508 (static_spec_functions): Add debug-level-gt spec function.
29509 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
29511 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29512 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29513 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
29514 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
29515 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
29516 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
29518 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
29520 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
29521 QImode fixups to general and mask registers only.
29523 2017-01-16 Carl Love <cel@us.ibm.com>
29525 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
29526 for built-in functions
29527 vector signed char vec_nabs (vector signed char)
29528 vector signed short vec_nabs (vector signed short)
29529 vector signed int vec_nabs (vector signed int)
29530 vector signed long long vec_nabs (vector signed long long)
29531 vector float vec_nabs (vector float)
29532 vector double vec_nabs (vector double)
29533 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
29535 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
29536 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
29537 * doc/extend.texi: Update the documentation file for the new built-in
29540 2017-01-16 Martin Sebor <msebor@redhat.com>
29542 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
29545 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29547 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
29548 UNSPEC_VSX__XXSPLTD to require special splat handling.
29550 2017-01-16 David Malcolm <dmalcolm@redhat.com>
29553 * system.h: Poison strndup.
29555 2017-01-16 Alan Modra <amodra@gmail.com>
29558 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
29561 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
29563 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
29565 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
29567 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
29568 call recog here. Assert that INSN_CODE (insn) is non-negative.
29570 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29573 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
29575 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
29576 in the currently scheduled RTL fragment.
29578 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29580 PR rtl-optimization/78751
29581 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
29584 2017-01-14 Jeff Law <law@redhat.com>
29586 PR tree-optimization/79090
29587 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
29588 variable length stores.
29589 (compute_trims): Delete dead assignment to *trim_tail.
29590 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
29593 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
29595 PR rtl-optimization/78626
29596 PR rtl-optimization/78727
29597 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
29598 of a block, and split such blocks after everything else is finished.
29600 2017-01-14 Alan Modra <amodra@gmail.com>
29603 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
29604 target legitimate_combined_insn.
29605 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
29606 (rs6000_legitimate_combined_insn): New function.
29607 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
29609 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
29610 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
29611 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
29613 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
29615 * doc/frontends.texi (G++ and GCC): Remove references to Java.
29617 2017-01-13 Jeff Law <law@redhat.com>
29619 PR tree-optimization/33562
29620 PR tree-optimization/61912
29621 PR tree-optimization/77485
29622 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
29624 (delete_dead_assignment): Likewise.
29625 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
29626 statement to delete_dead_call and delete_dead_assignment.
29628 2017-01-13 David Malcolm <dmalcolm@redhat.com>
29631 * substring-locations.c (format_warning_va): Strengthen case 1 so
29632 that both endpoints of the substring must be within the format
29633 range for just the substring to be printed.
29635 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
29637 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
29638 * config/i386/i386.c (ix86_target_string): Add missing options
29639 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
29640 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
29641 flags_other and ix86_target_other to flags2_other. Display unknown
29643 (ix86_valid_target_attribute_inner_p): Add missing options and
29644 reorder options by implied ISAs, as in ix86_target_string.
29646 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29648 * hash-table.h (hash_table::too_empty_p): New function.
29649 (hash_table::expand): Use it.
29650 (hash_table::traverse): Likewise.
29651 (hash_table::empty_slot): Use sizeof (value_type) instead of
29652 sizeof (PTR) to convert bytes to elements. Shrink the table
29653 if the current size is excessive for the current number of
29656 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29658 * ira-costs.c (record_reg_classes): Break from the inner loop
29659 early once alt_fail is known to be true. Update outer loop
29660 handling accordingly.
29662 2017-01-13 Jeff Law <law@redhat.com>
29664 * tree-ssa-dse.c (decrement_count): New function.
29665 (increment_start_addr, maybe_trim_memstar_call): Likewise.
29666 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
29667 when we know the partially dead statement is a mem* function.
29669 PR tree-optimization/61912
29670 PR tree-optimization/77485
29671 * tree-ssa-dse.c: Include expr.h.
29672 (maybe_trim_constructor_store): New function.
29673 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
29675 PR tree-optimization/33562
29676 PR tree-optimization/61912
29677 PR tree-optimization/77485
29678 * doc/invoke.texi: Document new dse-max-object-size param.
29679 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
29680 * tree-ssa-dse.c: Include params.h.
29681 (dse_store_status): New enum.
29682 (initialize_ao_ref_for_dse): New, partially extracted from
29684 (valid_ao_ref_for_dse, normalize_ref): New.
29685 (setup_live_bytes_from_ref, compute_trims): Likewise.
29686 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
29687 (maybe_trim_partially_dead_store): Likewise.
29688 (maybe_trim_complex_store): Likewise.
29689 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
29690 Track what bytes live from the original store. Return tri-state
29691 for dead, partially dead or live.
29692 (dse_dom_walker): Add constructor, destructor and new private members.
29693 (delete_dead_call, delete_dead_assignment): New extracted from
29695 (dse_optimize_stmt): Make a member of dse_dom_walker.
29696 Use initialize_ao_ref_for_dse.
29698 PR tree-optimization/33562
29699 PR tree-optimization/61912
29700 PR tree-optimization/77485
29701 * sbitmap.h (bitmap_count_bits): Prototype.
29702 (bitmap_clear_range, bitmap_set_range): Likewise.
29703 * sbitmap.c (bitmap_clear_range): New function.
29704 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
29706 2017-01-13 Martin Liska <mliska@suse.cz>
29709 * function.c (set_cfun): Add new argument force.
29710 * function.h (set_cfun): Likewise.
29711 * ipa-inline-transform.c (inline_call): Use the function when
29712 strict alising from is dropped for function we inline to.
29714 2017-01-13 Richard Biener <rguenther@suse.de>
29716 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
29717 for dumping GIMPLE INTEGER_CSTs.
29719 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29721 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
29722 to 201112L since C++17.
29724 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
29727 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
29728 if -fsanitize=kernel-address is present.
29730 2017-01-13 Richard Biener <rguenther@suse.de>
29732 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
29733 as _Literal ( type ) number in case usual suffixes do not
29734 preserve all information.
29736 2017-01-13 Richard Biener <rguenther@suse.de>
29738 PR tree-optimization/77283
29739 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
29740 and ssa-iterators.h.
29741 (is_feasible_trace): Implement a cost model based on joiner
29744 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
29747 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
29748 char or short to __float128/_Float128 directly.
29750 2017-01-12 Martin Sebor <msebor@redhat.com>
29752 to -Wformat-overflow.
29753 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
29754 (min_bytes_remaining): Same.
29755 (get_string_length): Same.
29756 (format_string): Same.
29757 (format_directive): Same.
29759 (pass_sprintf_length::handle_gimple_call): Same.
29761 2017-01-12 Jakub Jelinek <jakub@redhat.com>
29763 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
29764 info.nowrite calls with no lhs that can't throw. Return bool
29765 whether gsi_remove has been called or not.
29766 (pass_sprintf_length::handle_gimple_call): Return bool whether
29767 try_substitute_return_value called gsi_remove. Formatting fix.
29768 (pass_sprintf_length::execute): Don't use gsi_remove if
29769 handle_gimple_call returned true.
29772 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
29773 be removed due to side-effects, don't remove following barrier nor
29774 turn the successor edge into fallthru edge.
29776 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29779 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
29780 element-reversing loads and stores as not swappable.
29782 2017-01-12 Nathan Sidwell <nathan@acm.org>
29783 Nicolai Stange <nicstange@gmail.com>
29785 * combine.c (try_combine): Don't ignore result of overlap checking
29786 loop. Combine overlap & asm check into single loop.
29788 2017-01-12 Richard Biener <rguenther@suse.de>
29790 * tree-pretty-print.c (dump_generic_node): Provide -gimple
29791 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
29793 2017-01-12 Richard Biener <rguenther@suse.de>
29795 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
29796 and TS_TARGET_OPTION directly derive from TS_BASE.
29797 * tree-core.h (tree_optimization_option): Derive from tree_base.
29798 (tree_target_option): Likewise.
29800 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
29802 * config/i386/i386.c (memory_address_length): Increase len
29803 only when rip_relative_addr_p returns false.
29805 2017-01-11 Julia Koval <julia.koval@intel.com>
29807 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
29808 (OPTION_MASK_ISA_SGX_SET): New.
29809 (ix86_handle_option): Handle OPT_msgx.
29810 * config.gcc: Added sgxintrin.h.
29811 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
29812 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
29813 * config/i386/i386.c (ix86_target_string): Add -msgx.
29815 (ix86_option_override_internal): Handle new options.
29816 (ix86_valid_target_attribute_inner_p): Add sgx.
29817 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
29818 * config/i386/i386.opt: Add msgx.
29819 * config/i386/sgxintrin.h: New file.
29820 * config/i386/x86intrin.h: Add sgxintrin.h.
29822 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29825 * fold-const.c (maybe_nonzero_address): Return 1 for function
29827 (tree_single_nonzero_warnv_p): Don't handle function local objects
29831 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
29834 2017-01-11 David Malcolm <dmalcolm@redhat.com>
29837 * opts.c: Include "spellcheck.h"
29838 (struct string_fragment): New struct.
29839 (struct edit_distance_traits<const string_fragment &>): New
29841 (get_closest_sanitizer_option): New function.
29842 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
29844 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29846 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
29848 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
29849 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
29850 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
29851 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
29852 for initial die_offset if dwarf_split_debug_info.
29853 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
29854 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
29856 (output_skeleton_debug_sections): Formatting fix. Use
29857 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
29858 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
29860 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
29862 * config/arm/cortex-a53.md: Add bypasses for
29863 cortex_a53_r2f_cvt.
29864 (cortex_a53_r2f): Only use for transfers.
29865 (cortex_a53_f2r): Likewise.
29866 (cortex_a53_r2f_cvt): Add reservation for conversions.
29867 (cortex_a53_f2r_cvt): Likewise.
29869 2017-01-11 Tamar Christina <tamar.christina@arm.com>
29871 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
29872 to all inlined functions, change static to extern.
29874 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
29877 * config/arm/arm.c (legitimize_pic_address): Handle reference to
29879 (arm_assemble_integer): Likewise.
29881 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
29883 * config.gcc: Use new awk script to check CPU, FPU and architecture
29884 parameters for --with-... options.
29885 * config/arm/parsecpu.awk: New file
29886 * config/arm/arm-cpus.in: New file.
29887 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
29889 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
29891 * config/arm/t-arm: Update dependency rules.
29892 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
29893 of processing .def files.
29894 * config/arm/genopt.sh: Deleted.
29895 * config/arm/gentune.sh: Deleted.
29896 * config/arm/arm-cores.def: Deleted.
29897 * config/arm/arm-arches.def: Deleted.
29898 * config/arm/arm-fpus.def: Deleted.
29899 * config/arm/arm-tune.md: Regenerated.
29900 * config/arm/arm-tables.opt: Regenerated.
29901 * config/arm/arm-cpu.h: New generated file.
29902 * config/arm/arm-cpu-data.h: New generated file.
29903 * config/arm/arm-cpu-cdata.h: New generated file.
29905 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
29908 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
29910 (input_varpool_node): Unpack dynamically_initialized bit.
29912 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
29914 PR rtl-optimization/79032
29915 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
29916 the alignment of the adjusted memory reference against that of MODE,
29917 instead of the alignment of the original memory reference.
29919 2017-01-11 Martin Jambor <mjambor@suse.cz>
29921 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
29923 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
29924 decorated functions.
29926 2017-01-11 Richard Biener <rguenther@suse.de>
29928 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
29929 set range/nonnull info for PHI results. Do not set it on
29930 stmts marked for removal.
29932 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
29934 * expr.c (store_field): In the bitfield case, fetch the return value
29935 from the registers before applying a single big-endian adjustment.
29936 Always do a final load for a BLKmode value not larger than a word.
29938 2017-01-10 David Malcolm <dmalcolm@redhat.com>
29941 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29942 that we correctly handle column numbers greater than
29943 LINE_MAP_MAX_COLUMN_NUMBER.
29945 2017-01-10 Martin Sebor <msebor@redhat.com>
29947 PR middle-end/78245
29948 * gimple-ssa-sprintf.c (get_destination_size): Call
29949 {init,fini}object_sizes.
29950 * tree-object-size.c (addr_object_size): Adjust.
29951 (pass_through_call): Adjust.
29952 (pass_object_sizes::execute): Adjust.
29953 * tree-object-size.h (fini_object_sizes): Declare.
29955 2017-01-10 Martin Sebor <msebor@redhat.com>
29957 PR tree-optimization/78775
29958 * builtins.c (get_size_range): Move...
29959 * calls.c: ...to here.
29960 (alloc_max_size): Accept zero argument.
29961 (operand_signed_p): Remove.
29962 (maybe_warn_alloc_args_overflow): Call get_size_range.
29963 * calls.h (get_size_range): Declare.
29965 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
29967 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
29968 from TI's devices.csv file as of September 2016.
29969 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
29971 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
29973 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
29974 * doc/invoke.texi: Likewise.
29975 * doc/md.texi: Likewise.
29976 * doc/objc.texi: Likewise.
29978 2017-01-10 Joshua Conner <joshconner@google.com>
29980 * config/arm/fuchsia-elf.h: New file.
29981 * config/fuchsia.h: New file.
29982 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
29983 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
29985 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
29987 2016-01-10 Richard Biener <rguenther@suse.de>
29989 PR tree-optimization/79034
29990 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29991 Propagate out degenerate PHIs in the joiner.
29993 2017-01-10 Martin Liska <mliska@suse.cz>
29995 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
29996 (sort_congruence_classes_by_decl_uid): Likewise.
29997 (sort_congruence_class_groups_by_decl_uid): Likewise.
29998 (sem_item_optimizer::merge_classes): Sort class, groups in these
29999 classes and members in the groups by DECL_UID of declarations.
30000 This would make merge operations stable.
30002 2017-01-10 Martin Liska <mliska@suse.cz>
30004 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30005 usage of m_classes_vec.
30006 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30007 (sem_item_optimizer::get_group_by_hash): Likewise.
30008 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30009 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30010 (sem_item_optimizer::verify_classes): Likewise.
30011 (sem_item_optimizer::process_cong_reduction): Likewise.
30012 (sem_item_optimizer::dump_cong_classes): Likewise.
30013 (sem_item_optimizer::merge_classes): Likewise.
30014 * ipa-icf.h (congruence_class_hash): Rename from
30015 congruence_class_group_hash. Remove declaration of m_classes_vec.
30017 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30019 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30020 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30021 * config.gcc: Add avx512vpopcntdqintrin.h.
30022 * config/i386/avx512vpopcntdqintrin.h: New.
30023 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30024 * config/i386/i386-builtin-types.def: Add new types.
30025 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30026 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30027 __builtin_ia32_vpopcountq_v8di_mask): New.
30028 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30029 __AVX512VPOPCNTDQ__.
30030 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30031 (PTA_AVX512VPOPCNTDQ): Define.
30032 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30033 TARGET_AVX512VPOPCNTDQ_P): Define.
30034 * config/i386/i386.opt: Add mavx512vpopcntdq.
30035 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30036 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30038 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30040 PR middle-end/77484
30041 * predict.def (PRED_CALL): Set to 67.
30043 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30045 * expr.c (store_field): In the bitfield case, if the value comes from
30046 a function call and is of an aggregate type returned in registers, do
30047 not modify the field mode; extract the value in all cases if the mode
30048 is BLKmode and the size is not larger than a word.
30050 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30053 * config/i386/cpuid.h: Fix undefined behavior.
30055 2017-01-04 Jeff Law <law@redhat.com>
30057 PR tree-optimization/79007
30058 PR tree-optimization/67955
30059 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30060 conservative for pt.null when flag_non_call_exceptions is on.
30062 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30064 PR translation/79019
30065 PR translation/79020
30066 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30067 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30068 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30070 * config/avr/avr.opt (maccumulate-args): Likewise.
30071 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30072 * common.opt (freport-bug): Likewise.
30073 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30074 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30075 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30076 translatable string.
30077 * config/i386/i386.c (function_value_32): Likewise.
30078 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
30079 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
30081 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
30082 * common/config/msp430/msp430-common.c (msp430_handle_option):
30084 * symtab.c (symtab_node::verify_base): Likewise.
30085 * opts.c (set_debug_level): Likewise.
30086 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
30087 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
30088 missing whitespace to translatable strings.
30089 * config/avr/avr.md (bswapsi2): Fix typo in comment.
30090 * config/sh/superh.h: Likewise.
30091 * config/i386/xopintrin.h: Likewise.
30092 * config/i386/znver1.md: Likewise.
30093 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
30094 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
30095 * double-int.h (struct double_int): Likewise.
30096 * double-int.c (div_and_round_double): Likewise.
30097 * wide-int.cc: Likewise.
30098 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
30099 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
30100 * cfgcleanup.c (crossjumps_occured): Renamed to ...
30101 (crossjumps_occurred): ... this.
30102 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
30105 PR tree-optimization/78899
30106 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
30107 returning bool return struct loop *, NULL for failure and the new
30109 (versionable_outer_loop_p): Don't version outer loop if it has
30110 dont_vectorized bit set.
30111 (tree_if_conversion): When versioning outer loop, ensure
30112 tree_if_conversion is performed also on the inner loop of the
30113 non-vectorizable outer loop copy.
30114 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
30115 LOOP_VECTORIZED in inner loop of the scalar outer loop and
30116 prevent vectorization of it.
30117 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
30118 the outer loop vectorization of the non-scalar version is attempted
30119 before vectorization of the inner loop in scalar version. If
30120 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
30121 vectorization of its inner loop.
30122 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
30123 has 2 inner loops, rename also on edges from bb whose single pred
30124 is outer_loop->header. Fix typo in function comment.
30126 2017-01-09 Martin Sebor <msebor@redhat.com>
30129 * asan.c (asan_emit_stack_protection): Increase local buffer size
30130 to avoid snprintf truncation warning.
30132 2017-01-09 Andrew Pinski <apinski@cavium.com>
30134 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
30135 to reference thunderx2t99 for the tuning structure
30136 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
30138 (thunderx2t99_extra_costs): This.
30139 * config/aarch64/aarch64-tune.md: Regenerate.
30140 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
30141 (vulcan_addrcost_table): This.
30142 (vulcan_regmove_cost): Rename to ...
30143 (thunderx2t99_regmove_cost): This.
30144 (vulcan_vector_cost): Rename to ...
30145 (thunderx2t99_vector_cost): this.
30146 (vulcan_branch_cost): Rename to ...
30147 (thunderx2t99_branch_cost): This.
30148 (vulcan_tunings): Rename to ...
30149 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
30150 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
30152 2017-01-09 Martin Jambor <mjambor@suse.cz>
30156 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
30157 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
30158 (propagate_vr_accross_jump_function): Use the above function for all
30159 value range computations for pass-through jump functions and type
30160 converasion from explicit value range values.
30161 (ipcp_propagate_stage): Do not attempt to deduce types of formal
30162 parameters from TYPE_ARG_TYPES.
30163 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
30164 (ipa_write_node_info): Stream type of the actual argument.
30165 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
30167 2017-01-09 Martin Liska <mliska@suse.cz>
30170 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
30171 (lookup_compiler): Do not show error message with have_E.
30173 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30175 PR tree-optimization/78938
30176 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
30177 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
30178 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
30179 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
30182 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30184 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
30187 2017-01-09 Richard Biener <rguenther@suse.de>
30189 PR tree-optimization/78997
30190 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
30191 name condition properly.
30193 2017-01-09 Richard Biener <rguenther@suse.de>
30196 * dwarf2out.c (is_cxx): New overload with context.
30197 (is_naming_typedef_decl): Use it.
30199 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30201 * invoke.texi (Option Summary): Correct spacing in option lists
30202 and add line breaks to fix over-long lines.
30204 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30206 PR middle-end/17660
30208 * extend.texi (Common Variable Attributes): Add xref to GCC
30209 Internals manual to explain mode attribute keywords.
30211 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30214 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
30215 and Preprocessor Options.
30216 (Options for Linking): Document -pthread here....
30217 (RS/6000 and PowerPC Options): ...not here.
30218 (Solaris 2 Options): ...or here.
30219 * doc/cppopts.texi: Document -pthread.
30221 2017-01-08 Martin Sebor <msebor@redhat.com>
30223 PR middle-end/77708
30224 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
30225 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
30226 New member functions.
30227 (format_directive): Used them.
30229 (pass_sprintf_length::handle_gimple_call): Same.
30230 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
30231 to avoid truncation for any argument.
30232 (extract_affine_mul): Same.
30233 * tree.c (get_file_function_name): Same.
30235 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30237 PR middle-end/77484
30238 * predict.def (PRED_INDIR_CALL): Set to 86.
30240 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30242 PR preprocessor/54124
30243 * doc/cppopts.texi: Reformat -d subtable to list the full name
30244 of the options. Add cross-reference to the docs for the general
30245 compiler -d options.
30246 * doc/invoke.texi (Developer Options): Add cross-reference to the
30247 preprocessor-specific -d option documentation.
30249 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30251 PR preprocessor/13498
30252 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
30253 redudant material, and reflect new command-line options.
30254 (System Headers): Likewise.
30256 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30258 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
30259 -isystem, and -idirafter. Copy-edit.
30260 * doc/cppopts.texi: Copy-edit. Remove contradiction about
30261 default for -ftrack-macro-expansion. Delete obsolete and
30262 badly-formatted implementation details about -fdebug-cpp output.
30263 * doc/cppwarnopts.texi: Copy-edit.
30265 2017-01-07 David Malcolm <dmalcolm@redhat.com>
30268 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30269 that the transition from a max line width >= 1<<10 to narrower
30270 lines works correctly.
30272 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
30274 * doc/options.texi (PerFunction): New.
30275 * opt-functions.awk (switch_flags): Map both Optimization and
30276 PerFunction to CL_OPTIMIZATION.
30277 * opth-gen.awk: Test for PerFunction flag along with
30279 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
30280 it only when the latter is present. Skip those that don't in
30281 the hash function generator.
30282 * common.opt (fvar-tracking): Mark as PerFunction instead of
30284 (fvar-tracking-assignments): Likewise.
30285 (fvar-tracking-assignments-toggle): Likewise.
30286 (fvar-tracking-uninit): Likewise.
30288 2017-01-07 Jakub Jelinek <jakub@redhat.com>
30290 PR translation/79018
30291 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
30294 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
30297 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
30298 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
30299 TARGET_LONG_JUMP_TABLE_OFFSETS.
30300 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
30301 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
30302 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
30303 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
30304 * config/m68k/m68k.md (tablejump expander): Likewise.
30305 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
30306 TARGET_LONG_JUMP_TABLE_OFFSETS.
30307 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
30308 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
30310 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
30311 David Holsgrove <david.holsgrove@xilinx.com>
30313 * common/config/microblaze/microblaze-common.c
30314 (TARGET_EXCEPT_UNWIND_INFO): Remove.
30315 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
30317 * config/microblaze/microblaze.c (microblaze_must_save_register)
30318 (microblaze_expand_epilogue, microblaze_return_addr): Handle
30320 (microblaze_eh_return): New function.
30321 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
30322 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
30323 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
30324 * config/microblaze/microblaze.md (eh_return): New pattern.
30326 2017-01-06 Jakub Jelinek <jakub@redhat.com>
30328 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
30329 GCC_DIAGNOSTIC_STRINGIFY): Define.
30331 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
30333 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30335 * config/arm/arm.md (<mcrr>): New.
30337 * config/arm/arm.c (arm_arch5te): New.
30338 (arm_option_override): Set arm_arch5te.
30339 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
30341 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
30342 (arm_mcrr_qualifiers): ... this. New.
30343 (MRRC_QUALIFIERS): Define to...
30344 (arm_mrrc_qualifiers): ... this. New.
30345 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
30347 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
30348 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
30349 (MRRCI, mrrc, MRRC): New.
30350 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
30351 VUNSPEC_MRRC2): New.
30353 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30355 * config/arm/arm.md (<mcr>): New.
30357 * config/arm/arm.c (arm_coproc_builtin_available): Add
30358 support for mcr, mrc, mcr2 and mrc2.
30359 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
30360 (arm_mcr_qualifiers): ... this. New.
30361 (MRC_QUALIFIERS): Define to ...
30362 (arm_mrc_qualifiers): ... this. New.
30363 (MCR_QUALIFIERS): Define to ...
30364 (arm_mcr_qualifiers): ... this. New.
30365 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
30367 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
30368 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
30369 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
30370 VUNSPEC_MRC2): New.
30372 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30374 * config/arm/arm.md (*ldc): New.
30378 * config/arm/arm.c (arm_coproc_builtin_available): Add
30379 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
30380 (arm_coproc_ldc_stc_legitimate_address): New.
30381 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
30382 'qualifier_const_pointer'.
30383 (LDC_QUALIFIERS): Define to...
30384 (arm_ldc_qualifiers): ... this. New.
30385 (STC_QUALIFIERS): Define to...
30386 (arm_stc_qualifiers): ... this. New.
30387 * config/arm/arm-protos.h
30388 (arm_coproc_ldc_stc_legitimate_address): New.
30389 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
30390 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
30391 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
30392 stc2, stcl, stc2l): New.
30393 * config/arm/constraints.md (Uz): New.
30394 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
30395 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
30396 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
30397 VUNSPEC_STC2L): New.
30399 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30401 * config/arm/arm.md (<cdp>): New.
30402 * config/arm/arm.c (neon_const_bounds): Rename this ...
30403 (arm_const_bounds): ... this.
30404 (arm_coproc_builtin_available): New.
30405 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
30406 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
30407 (CDP_QUALIFIERS): Define to...
30408 (arm_cdp_qualifiers): ... this. New.
30410 (arm_expand_builtin_args): Add case for 6 arguments.
30411 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
30412 (arm_const_bounds): ... this.
30413 (arm_coproc_builtin_available): New.
30414 * config/arm/arm_acle.h (__arm_cdp): New.
30416 * config/arm/arm_acle_builtins.def (cdp): New.
30418 * config/arm/iterators.md (CDPI,CDP,cdp): New.
30419 * config/arm/neon.md: Rename all 'neon_const_bounds' to
30420 'arm_const_bounds'.
30421 * config/arm/types.md (coproc): New.
30422 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
30423 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
30424 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
30425 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
30427 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30429 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
30430 (UBINOP_QUALIFIERS): New.
30432 (acle_builtin_data): New. Change comment.
30433 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
30434 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
30435 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
30436 arm_acle_builtins.def.
30437 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
30438 (arm_init_acle_builtins): New.
30439 (CRC32_BUILTIN): Remove.
30440 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
30441 crc32cb, crc32ch and crc32cw.
30442 (arm_init_crc32_builtins): Remove.
30443 (arm_init_builtins): Use arm_init_acle_builtins rather
30444 than arm_init_crc32_builtins.
30445 (arm_expand_acle_builtin): New.
30446 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
30447 * config/arm/arm_acle_builtins.def: New.
30449 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30451 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
30452 (arm_builtin_datum): ... this.
30453 (arm_init_neon_builtin): Rename to ...
30454 (arm_init_builtin): ... this. Add a new parameters PREFIX
30455 and USE_SIG_IN_NAME.
30456 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
30457 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
30458 'arm_builtin_datum'.
30459 (arm_init_vfp_builtins): Likewise.
30460 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
30461 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
30462 (arm_expand_neon_args): Rename to ...
30463 (arm_expand_builtin_args): ... this. Rename builtin_arg
30464 enum values and differentiate between ARG_BUILTIN_MEMORY
30465 and ARG_BUILTIN_NEON_MEMORY.
30466 (arm_expand_neon_builtin_1): Rename to ...
30467 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
30468 values, arm_expand_builtin_args and add bool parameter NEON.
30469 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
30470 (arm_expand_vfp_builtin): Likewise.
30471 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
30473 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30475 PR middle-end/77484
30476 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
30477 * predict.c (tree_estimate_probability_bb): Reverse direction of
30478 polymorphic call predictor.
30480 2017-01-06 David Malcolm <dmalcolm@redhat.com>
30482 * passes.c (execute_one_pass): Split out pass-skipping logic into...
30483 (determine_pass_name_match): ...this new function and...
30484 (should_skip_pass_p): ...this new function.
30486 2017-01-06 Nathan Sidwell <nathan@acm.org>
30488 * ipa-visibility.c (function_and_variable_visibility): Reformat
30489 comments and long lines. Remove extrneous if.
30490 * symtab.c (symtab_node::make_decl_local): Fix code format.
30491 (symtab_node::set_section_for_node): Fix comment typo.
30493 2017-01-06 Martin Liska <mliska@suse.cz>
30496 * lra-constraints.c: Rename invariant to lra_invariant.
30497 * predict.c (set_even_probabilities): Initialize e to NULL.
30499 2017-01-05 Martin Sebor <msebor@redhat.com>
30501 PR tree-optimization/78910
30502 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
30503 (format_integer): Correct off-by-one error in the handling
30504 of precision with negative numbers in signed conversions..
30506 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
30508 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
30510 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30512 PR tree-optimization/71016
30513 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
30514 factor_out_conditional_conversion. Formatting fix.
30515 (factor_out_conditional_conversion): Add cond_stmt argument.
30516 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
30517 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
30520 2017-01-05 David Malcolm <dmalcolm@redhat.com>
30522 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
30523 read-rtl-function.o, and selftest-rtl.o.
30524 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
30525 (selftest::aarch64_test_loading_full_dump): New function.
30526 (selftest::aarch64_run_selftests): New function.
30527 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
30528 selftest::aarch64_run_selftests.
30529 * config/i386/i386.c
30530 (selftest::ix86_test_loading_dump_fragment_1): New function.
30531 (selftest::ix86_test_loading_call_insn): New function.
30532 (selftest::ix86_test_loading_full_dump): New function.
30533 (selftest::ix86_test_loading_unspec): New function.
30534 (selftest::ix86_run_selftests): Call the new functions.
30535 * emit-rtl.c (maybe_set_max_label_num): New function.
30536 * emit-rtl.h (maybe_set_max_label_num): New decl.
30537 * function.c (instantiate_decls): Guard call to
30538 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
30539 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
30541 * gensupport.c (gen_reader::gen_reader): Pass "false"
30542 for new "compact" param of rtx_reader.
30543 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
30544 rather than an empty string for NULL strings.
30545 * read-md.c: Potentially include config.h rather than bconfig.h.
30546 Wrap include of errors.h with #ifdef GENERATOR_FILE.
30547 (have_error): New global, copied from errors.c.
30548 (md_reader::read_name): Rename to...
30549 (md_reader::read_name_1): ...this, adding "out_loc" param,
30550 and converting "missing name or number" to returning false, rather
30552 (md_reader::read_name): Reimplement in terms of read_name_1.
30553 (md_reader::read_name_or_nil): New function.
30554 (md_reader::read_string): Handle "(nil)" by returning NULL.
30555 (md_reader::md_reader): Add new param "compact".
30556 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
30557 (md_reader::read_file): New method.
30558 * read-md.h (md_reader::md_reader): Add new param "compact".
30559 (md_reader::read_file): New method.
30560 (md_reader::is_compact): New accessor.
30561 (md_reader::read_name): Convert return type from void to file_location.
30562 (md_reader::read_name_or_nil): New decl.
30563 (md_reader::read_name_1): New decl.
30564 (md_reader::m_compact): New field.
30565 (noop_reader::noop_reader): Pass "false" for new "compact" param
30567 (rtx_reader::rtx_reader): Add new "compact" param.
30568 (rtx_reader::read_rtx_operand): Make virtual and convert return
30569 type from void to rtx.
30570 (rtx_reader::read_until): New decl.
30571 (rtx_reader::handle_any_trailing_information): New virtual function.
30572 (rtx_reader::postprocess): New virtual function.
30573 (rtx_reader::finalize_string): New virtual function.
30574 (rtx_reader::m_in_call_function_usage): New field.
30575 (rtx_reader::m_reuse_rtx_by_id): New field.
30576 * read-rtl-function.c: New file.
30577 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
30578 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
30579 (selftest::verify_three_block_rtl_cfg): New decl.
30580 * read-rtl-function.h: New file.
30581 * read-rtl.c: Potentially include config.h rather than bconfig.h.
30582 For host, include function.h, memmodel.h, and emit-rtl.h.
30583 (one_time_initialization): New function.
30584 (struct compact_insn_name): New struct.
30585 (compact_insn_names): New array.
30586 (find_code): Handle insn codes in compact dumps.
30587 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
30588 (bind_subst_iter_and_attr): Likewise.
30589 (add_condition_to_string): Likewise.
30590 (add_condition_to_rtx): Likewise.
30591 (apply_attribute_uses): Likewise.
30592 (add_current_iterators): Likewise.
30593 (apply_iterators): Likewise.
30594 (initialize_iterators): Guard usage of apply_subst_iterator with
30595 #ifdef GENERATOR_FILE.
30596 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
30597 (md_reader::read_mapping): Likewise.
30598 (add_define_attr_for_define_subst): Likewise.
30599 (add_define_subst_attr): Likewise.
30600 (read_subst_mapping): Likewise.
30601 (check_code_iterator): Likewise.
30602 (rtx_reader::read_rtx): Likewise. Move one-time initialization
30604 (one_time_initialization): New function.
30605 (rtx_reader::read_until): New method.
30606 (read_flags): New function.
30607 (parse_reg_note_name): New function.
30608 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
30609 Handle reuse_rtx ids.
30610 Wrap iterator lookup within #ifdef GENERATOR_FILE.
30611 Add parsing support for RTL dumps, mirroring the special-cases in
30612 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
30613 values, and calling handle_any_trailing_information.
30614 (rtx_reader::read_rtx_operand): Convert return type from void
30615 to rtx, returning return_rtx. Handle case 'e'. Call
30616 finalize_string on XSTR and XTMPL fields.
30617 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
30618 "(nil)" values were omitted. Call the postprocess vfunc on the
30620 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
30621 class ctor. Initialize m_in_call_function_usage. Call
30622 one_time_initialization.
30623 * rtl-tests.c (selftest::test_uncond_jump): Call
30624 set_new_first_and_last_insn.
30625 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
30626 * selftest-rtl.c: New file.
30627 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
30628 (selftest::get_insn_by_uid): New decl.
30629 * selftest-run-tests.c (selftest::run_tests): Call
30630 read_rtl_function_c_tests.
30631 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
30632 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
30635 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
30637 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
30638 operands in a special way. Assert that pos+len <= mode precision.
30640 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30642 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
30643 3 argument Alias with unlimited for the negative form.
30644 (fno-vect-cost-model): Removed.
30646 2017-01-05 Martin Liska <mliska@suse.cz>
30648 * hsa-gen.c (gen_hsa_divmod): New function.
30649 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
30651 2017-01-05 Martin Liska <mliska@suse.cz>
30654 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
30657 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30659 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
30660 small constant length operands.
30662 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30664 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
30665 between loop iterations.
30667 2017-01-05 Martin Liska <mliska@suse.cz>
30670 * gimplify.c (gimplify_decl_expr): Compare to
30671 asan_poisoned_variables instread of checking flags.
30672 (gimplify_target_expr): Likewise.
30673 (gimplify_expr): Likewise.
30674 (gimplify_function_tree): Conditionally initialize
30675 asan_poisoned_variables.
30677 2017-01-04 Jeff Law <law@redhat.com>
30679 PR tree-optimizatin/78812
30680 * rtl.h (contains_mem_rtx_p): Prototype.
30681 * ifcvt.c (containts_mem_rtx_p): Move from here to...
30682 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
30683 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
30684 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
30685 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
30687 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30689 * input.c (assert_char_at_range): Default-initialize actual_range.
30691 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30693 * df-scan.c (df_ref_create_structure): Make regno unsigned,
30694 to match the caller.
30696 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30698 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
30699 insns after final jump in test to emit dummy move.
30701 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30703 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
30704 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
30706 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30708 * multiple_target.c (create_dispatcher_calls): Init e_next.
30709 * tree-ssa-loop-split.c (split_loop): Init border.
30710 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
30713 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
30718 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
30719 (altivec_register_operand): Do not return true if the operand
30720 contains a SUBREG mixing SImode and SFmode.
30721 (vsx_register_operand): Likewise.
30722 (vsx_reg_sfsubreg_ok): New predicate.
30723 (vfloat_operand): Do not return true if the operand contains a
30724 SUBREG mixing SImode and SFmode.
30725 (vint_operand): Likewise.
30726 (vlogical_operand): Likewise.
30727 (gpc_reg_operand): Likewise.
30728 (int_reg_operand): Likewise.
30729 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
30730 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
30731 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
30733 (rs6000_emit_move_si_sf_subreg): New helper function.
30734 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
30735 fixup SUBREGs involving SImode and SFmode.
30736 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
30737 numbers for the new peephole2 optimization.
30738 (peephole2 for SFmode unions): New peephole2 to optimize cases in
30739 the GLIBC math library that do AND/IOR/XOR operations on single
30740 precision floating point.
30741 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
30742 target macros to say whether we need to avoid SUBREGs mixing
30744 (TARGET_ALLOW_SF_SUBREG): Likewise.
30745 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
30746 (UNSPEC_SI_FROM_SF): Likewise.
30747 (iorxor): Change spacing.
30748 (and_ior_xor): New iterator for AND, IOR, and XOR.
30749 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
30750 (movdi_from_sf_zero_ext): Likewise.
30751 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
30752 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
30753 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
30754 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
30755 (fms<mode>4): Likewise.
30756 (fnma<mode>4): Likewise.
30757 (fnms<mode>4): Likewise.
30758 (nfma<mode>4): Likewise.
30759 (nfms<mode>4): Likewise.
30761 2017-01-04 Marek Polacek <polacek@redhat.com>
30764 * doc/invoke.texi: Document -Wpointer-compare.
30766 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30768 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
30771 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
30772 descriptions for -gdwarf-5 and emit them as uleb128 instead of
30775 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
30778 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
30779 documentation of the powerpc_popcntb_ok attribute.
30780 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30781 code to issue warning messages if a requested CPU configuration is
30782 not supported by the binary (assembler and loader) toolchain.
30783 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
30784 made to define a built-in function that has been disabled.
30785 (paired_init_builtins): Add assertion to prevent ICE if attempt is
30786 made to define a built-in function that has been disabled.
30787 (altivec_init_builtins): Add comment explaining why definition
30788 of the DST built-in functions is not preceded by an assertion
30789 check. Add assertions to prevent ICE if attempts are made to
30790 define an altivec predicate or an abs* built-in function that has
30792 (htm_init_builtins): Add comment explaining why definition of the
30793 htm built-in functions is not preceded by an assertion check.
30795 2017-01-04 Jeff Law <law@redhat.com>
30797 PR tree-optimizatin/67955
30798 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
30799 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
30800 the points-to solution does not include pt_null. Use DECL_PT_UID
30803 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
30805 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
30806 Use gen_int_mode instead of gen_lopwart for const_int operands.
30808 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30810 PR tree-optimization/71563
30811 * match.pd: Simplify X << Y into X if Y is known to be 0 or
30812 out of range value - has low bits known to be zero.
30814 2017-01-04 Alan Modra <amodra@gmail.com>
30816 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
30817 * configure: Regenerate.
30818 * config.in: Regenerate.
30820 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30823 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
30824 a substring of the message, but strcmp with the whole message. Ifdef
30825 ENABLE_NLS, translate the message first using dgettext.
30827 2017-01-03 Jeff Law <law@redhat.com>
30829 PR tree-optimizatin/78856
30830 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
30831 (mark_threaded_blocks): Remove code to truncate thread paths that
30832 cross multiple loop headers. Instead invalidate the cached loop
30833 iteration information and handle case of a thread path walking
30834 into an irreducible region.
30836 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
30839 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
30840 assertions. Add support for doing the signbit if the IEEE 128-bit
30841 floating point value is in a GPR.
30842 * config/rs6000/rs6000.md (Fsignbit): Delete.
30843 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
30844 Update the length attribute if the value is in a GPR.
30845 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
30846 the sign or zero extension instruction, since the value is always 0/1.
30847 (signbit<mode>2_dm2): Delete using <Fsignbit>.
30850 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
30851 extracting SImode to a GPR register so that we can generate a
30852 store, limit the vector to be in a traditional Altivec register
30853 for the vextuwrx instruction.
30855 2017-01-03 Ian Lance Taylor <iant@google.com>
30857 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
30859 2017-01-03 Martin Sebor <msebor@redhat.com>
30861 PR tree-optimization/78696
30862 * gimple-ssa-sprintf.c (format_floating): Correct handling of
30863 precision. Use MPFR for %f for greater fidelity. Correct handling
30865 (pass_sprintf_length::compute_format_length): Set width and precision
30866 specified by asrerisk to void_node for vararg functions.
30867 (try_substitute_return_value): Adjust dump output.
30869 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
30871 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
30873 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
30875 * doc/invoke.texi (SPARC options): Document -mlra as the default.
30876 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
30877 -mlra/-mno-lra was passed to the compiler.
30879 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
30881 PR rtl-optimization/65618
30882 * emit-rtl.c (try_split): Move initialization of "before" and
30883 "after" to just before the call to emit_insn_after_setloc.
30885 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
30887 * doc/md.texi (Standard Names): Remove reference to Java frontend.
30889 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
30891 * dwarf2out.c (gen_enumeration_type_die): When
30892 -gno-strict-dwarf, add a DW_AT_encoding attribute.
30894 2017-01-03 Jakub Jelinek <jakub@redhat.com>
30896 PR tree-optimization/78965
30897 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
30898 Change first argument from const call_info & to call_info &. For %n
30899 set info.nowrite to false.
30901 PR middle-end/78901
30902 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
30903 possibly throwing calls.
30905 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
30906 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
30907 and fns handling, rather than in a separate case SSA_NAME.
30909 2017-01-02 Jeff Law <law@redhat.com>
30911 * config/darwin-driver.c (darwin_driver_init): Const-correctness
30912 fixes for first_period and second_period variables.
30914 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
30917 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
30918 (*insvqi_1): New insn pattern.
30919 (*insvqi_1_mem_rex64): Ditto.
30920 (*insvqi_2): Ditto.
30921 (*insvqi_3): Rename from *insvqi.
30923 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
30925 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
30927 * doc/cfg.texi (Edges): Remove reference to Java.
30928 (Maintaining the CFG): Ditto.
30930 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30932 PR middle-end/77674
30933 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
30934 transparent aliases.
30936 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30938 PR middle-end/77484
30939 * predict.def (PRED_CALL): Update hitrate.
30940 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
30941 * predict.c (tree_estimate_probability_bb): Split CALL predictor
30942 into direct/indirect/polymorphic variants.
30944 2017-01-01 Jakub Jelinek <jakub@redhat.com>
30946 Update copyright years.
30948 * gcc.c (process_command): Update copyright notice dates.
30949 * gcov-dump.c (print_version): Ditto.
30950 * gcov.c (print_version): Ditto.
30951 * gcov-tool.c (print_version): Ditto.
30952 * gengtype.c (create_file): Ditto.
30953 * doc/cpp.texi: Bump @copying's copyright year.
30954 * doc/cppinternals.texi: Ditto.
30955 * doc/gcc.texi: Ditto.
30956 * doc/gccint.texi: Ditto.
30957 * doc/gcov.texi: Ditto.
30958 * doc/install.texi: Ditto.
30959 * doc/invoke.texi: Ditto.
30961 Copyright (C) 2017 Free Software Foundation, Inc.
30963 Copying and distribution of this file, with or without modification,
30964 are permitted in any medium without royalty provided the copyright
30965 notice and this notice are preserved.