1 2017-10-20 Richard Biener <rguenther@suse.de>
3 PR tree-optimization/82473
4 * tree-vect-loop.c (vectorizable_reduction): Properly get at
5 the largest input type.
7 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9 * c-attribs.c (handle_nocf_check_attribute): New function.
10 (c_common_attribute_table): Add 'nocf_check' handling.
11 * gimple-parser.c: Add second argument NULL to
12 gimple_build_call_from_tree.
13 * attrib.c (comp_type_attributes): Check nocf_check attribute.
14 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
16 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
17 * common.opt: Add fcf-protection flag.
18 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
19 * flag-types.h: Add enum cf_protection_level.
20 * gimple.c (gimple_build_call_from_tree): Add second parameter.
21 Add 'nocf_check' attribute propagation to gimple call.
22 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
23 (gimple_build_call_from_tree): Update prototype.
24 (gimple_call_nocf_check_p): New function.
25 (gimple_call_set_nocf_check): Likewise.
26 * gimplify.c: Add second argument to gimple_build_call_from_tree.
27 * ipa-icf.c: Add nocf_check attribute in statement hash.
28 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
29 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
30 * toplev.c (process_options): Add flag_cf_protection handling.
32 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
34 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
36 2017-10-20 Richard Biener <rguenther@suse.de>
38 PR tree-optimization/82603
39 * tree-if-conv.c (predicate_mem_writes): Make sure to only
40 remove false predicated stores.
42 2017-10-20 Richard Biener <rguenther@suse.de>
44 * graphite-isl-ast-to-gimple.c
45 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
46 Remove return value and simplify, dump copied stmt after lhs
48 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
49 Reduce dump verbosity.
50 (gsi_insert_earliest): Likewise.
51 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
52 * graphite.c (print_global_statistics): Adjust dumping.
53 (print_graphite_scop_statistics): Likewise.
54 (print_graphite_statistics): Do not dump loops here.
55 (graphite_transform_loops): But here.
57 2017-10-20 Nicolas Roche <roche@adacore.com>
59 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to the include dir.
60 * configure: Regenerate.
62 2017-10-20 Jakub Jelinek <jakub@redhat.com>
65 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
66 functions when optimizing replace GIMPLE_RETURN stmts with
67 calls to __builtin_unreachable ().
70 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
71 for -fsanitize=thread link of executables.
72 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
76 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
78 (<shift_insn><mode>3<mask_name>): Change the last of the 3
79 define_insns for logical vector shifts to use VI248_AVX512BW
80 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
81 condition, useless isa and prefix attributes. Change the first
82 2 of these define_insns to ...
83 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
84 define_insn for avx512vl.
85 (<shift_insn><mode>3): ... and this, new define_insn without
86 masking for non-avx512vl.
89 * config/i386/sse.md (*andnot<mode>3,
90 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
91 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
92 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
93 not applied use empty suffix even for TARGET_AVX512VL.
94 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
95 is applied, supply evex,evex or evex,evex,evex instead of just
98 2017-10-20 Julia Koval <julia.koval@intel.com>
100 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
101 (OPTION_MASK_ISA_GFNI_UNSET): New.
102 (ix86_handle_option): Handle OPT_mgfni.
103 * config/i386/cpuid.h (bit_GFNI): New.
104 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
105 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
106 * config/i386/i386.c (ix86_target_string): Add -mgfni.
107 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
108 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
109 * config/i386/i386.opt: Add mgfni.
111 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
113 * config/msp430/msp430.c (msp430_option_override): Disable
114 -fdelete-null-pointer-checks.
115 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
117 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
119 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
120 of x87 and SSE instructions.
122 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
124 * asan.c (create_cond_insert_point): Do not update edge count.
125 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
126 (afdo_propagate_circuit): Likewise.
127 (afdo_calculate_branch_prob): Likewise.
128 (afdo_annotate_cfg): Likewise.
129 * basic-block.h (struct edge_def): Remove count.
130 (edge_def::count): New accessor.
131 * bb-reorder.c (rotate_loop): Update.
132 (find_traces_1_round): Update.
133 (connect_traces): Update.
134 (sanitize_hot_paths): Update.
135 * cfg.c (unchecked_make_edge): Update.
136 (make_single_succ_edge): Update.
137 (check_bb_profile): Update.
138 (dump_edge_info): Update.
139 (update_bb_profile_for_threading): Update.
140 (scale_bbs_frequencies_int): Update.
141 (scale_bbs_frequencies_gcov_type): Update.
142 (scale_bbs_frequencies_profile_count): Update.
143 (scale_bbs_frequencies): Update.
144 * cfganal.c (connect_infinite_loops_to_exit): Update.
145 * cfgbuild.c (compute_outgoing_frequencies): Update.
146 (find_many_sub_basic_blocks): Update.
147 * cfgcleanup.c (try_forward_edges): Update.
148 (try_crossjump_to_edge): Update
149 * cfgexpand.c (expand_gimple_cond): Update
150 (expand_gimple_tailcall): Update
151 (construct_exit_block): Update
152 * cfghooks.c (verify_flow_info): Update
153 (redirect_edge_succ_nodup): Update
155 (make_forwarder_block): Update
156 (duplicate_block): Update
157 (account_profile_record): Update
158 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
159 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
160 * cfgloopmanip.c (scale_loop_profile): Update.
162 (lv_adjust_loop_entry_edge): Update.
163 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
164 (force_nonfallthru_and_redirect): Update.
165 (purge_dead_edges): Update.
166 (rtl_flow_call_edges_add): Update.
167 * cgraphunit.c (init_lowered_empty_function): Update.
168 (cgraph_node::expand_thunk): Update.
169 * gimple-pretty-print.c (dump_probability): Update.
170 (dump_edge_probability): Update.
171 * gimple-ssa-isolate-paths.c (isolate_path): Update.
172 * haifa-sched.c (sched_create_recovery_edges): Update.
173 * hsa-gen.c (convert_switch_statements): Update.
174 * ifcvt.c (dead_or_predicable): Update.
175 * ipa-inline-transform.c (inline_transform): Update.
176 * ipa-split.c (split_function): Update.
177 * ipa-utils.c (ipa_merge_profiles): Update.
178 * loop-doloop.c (add_test): Update.
179 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
180 * lto-streamer-in.c (input_cfg): Update.
181 (input_function): Update.
182 * lto-streamer-out.c (output_cfg): Update.
183 * modulo-sched.c (sms_schedule): Update.
184 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
185 * predict.c (maybe_hot_edge_p): Update.
186 (unlikely_executed_edge_p): Update.
187 (probably_never_executed_edge_p): Update.
188 (dump_prediction): Update.
189 (drop_profile): Update.
190 (propagate_unlikely_bbs_forward): Update.
191 (determine_unlikely_bbs): Update.
192 (force_edge_cold): Update.
193 * profile.c (compute_branch_probabilities): Update.
194 * reg-stack.c (better_edge): Update.
195 * shrink-wrap.c (handle_simple_exit): Update.
196 * tracer.c (better_p): Update.
197 * trans-mem.c (expand_transaction): Update.
198 (split_bb_make_tm_edge): Update.
199 * tree-call-cdce.c: Update.
200 * tree-cfg.c (gimple_find_sub_bbs): Update.
201 (gimple_split_edge): Update.
202 (gimple_duplicate_sese_region): Update.
203 (gimple_duplicate_sese_tail): Update.
204 (gimple_flow_call_edges_add): Update.
205 (insert_cond_bb): Update.
206 (execute_fixup_cfg): Update.
207 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
208 * tree-complex.c (expand_complex_div_wide): Update.
209 * tree-eh.c (lower_resx): Update.
210 (unsplit_eh): Update.
211 (cleanup_empty_eh_move_lp): Update.
212 * tree-inline.c (copy_edges_for_bb): Update.
213 (freqs_to_counts): Update.
214 (copy_cfg_body): Update.
215 * tree-ssa-dce.c (remove_dead_stmt): Update.
216 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
217 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
218 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
219 (unloop_loops): Update.
220 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
221 * tree-ssa-loop-split.c (connect_loops): Update.
222 (split_loop): Update.
223 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
224 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
225 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
226 * tree-ssa-reassoc.c (branch_fixup): Update.
227 * tree-ssa-tail-merge.c (replace_block_by): Update.
228 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
229 (compute_path_counts): Update.
230 (update_profile): Update.
231 (recompute_probabilities): Update.
232 (update_joiner_offpath_counts): Update.
233 (estimated_freqs_path): Update.
234 (freqs_to_counts_path): Update.
235 (clear_counts_path): Update.
236 (ssa_fix_duplicate_block_edges): Update.
237 (duplicate_thread_path): Update.
238 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
239 (case_bit_test_cmp): Update.
240 (collect_switch_conv_info): Update.
241 (gen_inbound_check): Update.
242 (do_jump_if_equal): Update.
243 (emit_cmp_and_jump_insns): Update.
244 * tree-tailcall.c (decrease_profile): Update.
245 (eliminate_tail_call): Update.
246 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
247 (vect_do_peeling): Update.
248 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
249 * ubsan.c (ubsan_expand_null_ifn): Update.
250 (ubsan_expand_ptr_ifn): Update.
251 * value-prof.c (gimple_divmod_fixed_value): Update.
252 (gimple_mod_pow2): Update.
253 (gimple_mod_subtract): Update.
255 (gimple_stringop_fixed_value): Update.
257 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
260 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
262 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
264 PR rtl-optimization/82395
265 * ira-color.c (allocno_priority_compare_func): Fix comparison step
266 based on non_spilled_static_chain_regno_p.
268 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
270 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
271 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
272 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
274 2017-10-19 Martin Sebor <msebor@redhat.com>
276 PR tree-optimization/82596
277 * tree.c (array_at_struct_end_p): Handle STRING_CST.
279 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
281 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
282 (get_mem_refs_of_builtin_call): Likewise.
283 * builtins.c (expand_builtin_apply): Adjust call to
284 allocate_dynamic_stack_space.
285 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
286 the third argument to allocate_dynamic_stack_space, otherwise -1.
287 (expand_builtin): Deal with all alloca variants.
288 (is_inexpensive_builtin): Likewise.
289 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
290 * calls.c (special_function_p): Deal with all alloca variants.
291 (initialize_argument_information): Adjust call to
292 allocate_dynamic_stack_space.
293 (expand_call): Likewise.
294 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
295 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
296 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
297 use it for the stack usage computation.
298 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
299 * function.c (gimplify_parameters): Call build_alloca_call_expr.
300 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
301 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
302 (in_loop_p): Remove first argument and useless check.
303 (pass_walloca::execute): Remove useless test and adjust call to above.
304 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
305 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
306 (gimplify_call_expr): Deal with all alloca variants.
307 * hsa-gen.c (gen_hsa_alloca): Likewise.
308 (gen_hsa_insns_for_call): Likewise.
309 * ipa-pure-const.c (special_builtin_state): Likewise.
310 * tree-chkp.c (chkp_build_returned_bound): Likewise.
311 * tree-object-size.c (alloc_object_size): Likewise.
312 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
313 (call_may_clobber_ref_p_1): Likewise.
314 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
315 (ccp_fold_stmt): Likewise.
316 (optimize_stack_restore): Likewise.
317 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
318 (mark_all_reaching_defs_necessary_1): Likewise.
319 (propagate_necessity): Likewise.
320 (eliminate_unnecessary_stmts): Likewise.
321 * tree.c (build_common_builtin_nodes): Build
322 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
323 (build_alloca_call_expr): New function.
324 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
325 (CASE_BUILT_IN_ALLOCA): Likewise.
326 (build_alloca_call_expr): Declare.
327 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
329 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
332 * dwarf2out.c (new_die_raw): New static inline function.
333 (new_die): Use it to create the DIE.
334 (add_AT_external_die_ref): Likewise.
335 (clone_die): Likewise.
336 (clone_as_declaration): Likewise.
337 (dwarf2out_vms_debug_main_pointer): Likewise.
338 (base_type_die): Likewise. Remove early return for corner cases.
339 Do not call add_pubtype on the DIE here.
340 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
341 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
342 typedefs for base types with DW_AT_endianity. Make sure a DIE with
343 native order exists for base types, attach the DIE manually and call
344 add_pubtype on it. Do not equate a reverse order DIE to the type.
346 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
348 * config/arm/arm.c (align_ok_ldrd_strd): New function.
349 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
351 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
353 2017-10-19 Jakub Jelinek <jakub@redhat.com>
355 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
356 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
357 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
358 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
359 * opts.c (sanitizer_opts): Add builtin.
360 * ubsan.c (instrument_builtin): New function.
361 (pass_ubsan::execute): Call it.
362 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
363 * doc/invoke.texi: Document -fsanitize=builtin.
365 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
366 builtins, store max (log2 (align), 0) into uchar field instead of
367 align into uptr field.
368 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
369 store uchar 0 field instead of uptr 0 field.
370 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
371 instead of passing one address of struct with 2 locations pass
372 two addresses of structs with 1 location each.
373 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
374 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
375 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
376 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
377 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
378 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
379 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
380 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
382 2017-10-19 Martin Liska <mliska@suse.cz>
385 * file-find.c (remove_prefix): Remove.
386 * file-find.h (remove_prefix): Likewise.
387 * gcc-ar.c: Remove smartness of lookup.
389 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
391 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
392 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
393 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
395 2017-10-19 Jakub Jelinek <jakub@redhat.com>
398 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
399 (setcc + and to xor + setcc): New peephole2.
401 2017-10-19 Tom de Vries <tom@codesourcery.com>
403 * doc/sourcebuild.texi (Test Directives, Variants of
404 dg-require-support): Add dg-require-stack-size.
406 2017-10-19 Martin Liska <mliska@suse.cz>
409 * gimplify.c (gimplify_decl_expr): Do not instrument variables
410 that have a large alignment.
411 (gimplify_target_expr): Likewise.
413 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
415 PR rtl-optimization/82602
416 * ira.c (rtx_moveable_p): Return false for volatile asm.
418 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
421 * config/i386/i386-modes.def (CCGZ): New CC mode.
422 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
423 * config/i386/predicates.md (ix86_comparison_operator):
425 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
426 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
427 with double-word subtraction.
428 (put_condition_code): Handle CCGZmode.
430 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
432 * wide-int.cc (debug (const wide_int &)): New.
433 (debug (const wide_int *)): New.
434 (debug (const widest_int &)): New.
435 (debug (const widest_int *)): New.
437 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
440 * lra-constraints.c (curr_insn_transform): Use non-input operand
441 instead of output one for matched reload.
443 2017-10-18 Bin Cheng <bin.cheng@arm.com>
445 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
446 (tree-ssa-loop-ivopts.h): New header file.
447 (struct builtin_info): New fields.
448 (classify_builtin_1): Compute and record base and offset parts for
449 memset builtin partition by calling strip_offset.
450 (offset_cmp, fuse_memset_builtins): New functions.
451 (finalize_partitions): Fuse adjacent memset partitions by calling
453 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
454 Expose the interface.
455 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
457 2017-10-18 Bin Cheng <bin.cheng@arm.com>
459 PR tree-optimization/82574
460 * tree-loop-distribution.c (find_single_drs): New parameter. Check
461 that data reference must be executed exactly once per iteration
462 against the outermost loop in nest.
463 (classify_partition): Update call to above function.
465 2017-10-18 Richard Biener <rguenther@suse.de>
467 PR tree-optimization/82591
468 * graphite.c (graphite_transform_loops): Move code gen message
470 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
471 Here. Handle scop_to_isl_ast failing.
472 (scop_to_isl_ast): Limit the number of ISL operations.
474 2017-10-18 Richard Biener <rguenther@suse.de>
476 * graphite-isl-ast-to-gimple.c
477 (translate_isl_ast_to_gimple::set_rename): Simplify.
478 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
479 (graphite_copy_stmts_from_block): ... here.
480 (copy_bb_and_scalar_dependences): Simplify.
481 (add_parameters_to_ivs_params): Canonicalize.
482 (generate_entry_out_of_ssa_copies): Simplify.
483 * graphite-sese-to-poly.c (extract_affine_name): Simplify
484 by passing in ISL dimension.
485 (parameter_index_in_region_1): Rename to ...
486 (parameter_index_in_region): ... this.
487 (extract_affine): Adjust assert, pass down parameter index.
488 (add_param_constraints): Use range-info when available.
489 (build_scop_context): Adjust.
490 * sese.c (new_sese_info): Adjust.
491 (free_sese_info): Likewise.
492 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
493 Remove unused typedefs.
494 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
496 2017-10-18 Martin Liska <mliska@suse.cz>
498 * combine.c (simplify_compare_const): Add gcc_fallthrough.
500 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
502 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
503 (s390_sched_init): Do not reset s390_sched_state if we entered the
504 current basic block via a fallthru edge and all others are unlikely.
506 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
508 * config/s390/s390.c (NUM_SIDES): New variable.
509 (LONGRUNNING_THRESHOLD): New variable.
510 (LATENCY_FACTOR): New variable.
511 (s390_sched_score): Decrease score for long-running instructions on
513 (s390_sched_variable_issue): Perform bookkeeping for long-running
516 2017-10-18 Richard Biener <rguenther@suse.de>
518 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
519 Simplify with removal of the parameter rename map.
520 (set_rename): Likewise.
521 (should_copy_to_new_region): Likewise.
522 (graphite_copy_stmts_from_block): Likewise.
523 (copy_bb_and_scalar_dependences): Remove initialization of
524 unused copied_bb_map.
526 (copy_internal_parameters): Likewise.
527 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
528 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
530 (parameter_index_in_region_1): Rename to ...
531 (assign_parameter_index_in_region): ... this. Assert we have
532 a parameter we handle.
533 (scan_tree_for_params): Adjust.
534 * sese.h (parameter_rename_map_t): Remove.
535 (struct sese_info_t): Remove unused parameter_rename_map and
536 copied_bb_map members.
537 * sese.c (new_sese_info): Adjust.
538 (free_sese_info): Likewise.
540 2017-10-18 Martin Liska <mliska@suse.cz>
543 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
546 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
548 * doc/invoke.texi (ffunction-sections and fdata-sections):
551 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
553 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
554 the use statement can throw internally.
556 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
558 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
559 any RTX present on the RHS of a SET.
560 * compare-elim.c (try_eliminate_compare): Restore comment.
562 2017-10-17 Jakub Jelinek <jakub@redhat.com>
564 * langhooks.h (struct lang_hooks): Document that tree_size langhook
565 may be also called on tcc_type nodes.
566 * langhooks.c (lhd_tree_size): Likewise.
568 2017-10-17 David Malcolm <dmalcolm@redhat.com>
570 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
571 format_warning_at_substring.
572 (maybe_warn): Convert source_range * param to a location_t. Pass
573 UNKNOWN_LOCATION rather than NULL to fmtwarn.
574 (format_directive): Remove code to extract source_ranges and
575 source_range * in favor of just a location_t.
576 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
578 * substring-locations.c (format_warning_va): Convert
579 source_range * param to a location_t.
580 (format_warning_at_substring): Likewise.
581 * substring-locations.h (format_warning_va): Likewise.
582 (format_warning_at_substring): Likewise.
584 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
586 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
588 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
589 and scatter/gather ops.
591 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
592 vec_gather_load and vec_scatter_store.
593 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
594 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
595 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
596 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
597 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
598 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
600 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
602 * reg-stack.c (compare_for_stack_reg): Add bool argument.
603 Detect FTST instruction and handle its register pops. Only pop
604 second operand if can_pop_second_op is true.
605 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
606 set can_pop_second_op to false in the compare_for_stack_reg call.
608 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
609 output_fp_compare for stack register operands.
610 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
611 instructions here. Do not emit stack register pops here. Assert
612 that FCOMPP pops next to top stack register. Rewrite function.
614 2017-10-17 Nathan Sidwell <nathan@acm.org>
617 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
618 use DECL_ASSEMBLER_NAME_RAW.
621 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
624 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
625 Wilco Dijkstra <wilco.dijkstra@arm.com>
627 * builtins.c (expand_builtin_update_setjmp_buf): Add a
628 converstion to Pmode from the buf_addr.
630 2017-10-17 Richard Biener <rguenther@suse.de>
632 * graphite-dependences.c (scop_get_reads_and_writes): Change
633 output parameters to references.
635 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
637 PR 71026/tree-optimization
638 * fold-const.c (distribute_real_division): Removed.
639 (fold_binary_loc): Remove calls to distribute_real_divison.
641 2017-10-17 Richard Biener <rguenther@suse.de>
643 * graphite-scop-detection.c
644 (scop_detection::stmt_has_simple_data_refs_p): Always use
645 the full nest as region.
646 (try_generate_gimple_bb): Likewise.
647 * sese.c (scalar_evolution_in_region): Simplify now that
648 SCEV can handle instantiation in regions.
649 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
650 in the non-loop part of a function if requested.
652 2017-10-17 Richard Biener <rguenther@suse.de>
654 PR tree-optimization/82563
655 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
657 (graphite_regenerate_ast_isl): Call it.
658 * graphite-scop-detection.c (build_scops): Remove entry edge split.
660 2017-10-17 Jakub Jelinek <jakub@redhat.com>
662 PR tree-optimization/82549
663 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
664 Formatting fixes. Instead of calling make_bit_field_ref with negative
667 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
669 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
670 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
671 _mm_maskz_reduce_ss): New.
672 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
673 __builtin_ia32_reducess_mask): Ditto..
674 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
675 * config/i386/sse.md (reduces<mode>): Renamed to ...
676 (reduces<mode><mask_scalar_name>): ... this.
677 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
679 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
680 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
682 2017-10-16 David Malcolm <dmalcolm@redhat.com>
684 * Makefile.in (OBJS): Add unique-ptr-tests.o.
685 * selftest-run-tests.c (selftest::run_tests): Call
686 selftest::unique_ptr_tests_cc_tests.
687 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
688 * unique-ptr-tests.cc: New file.
690 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
693 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
695 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
696 (make_hard_regno_born, make_hard_regno_dead): Update
697 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
699 2017-10-16 Jeff Law <law@redhat.com>
701 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
703 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
705 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
707 2017-10-16 Olivier Hainque <hainque@adacore.com>
709 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
710 with_cpu if we were configured for an e500v2 target cpu name.
712 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
714 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
715 * doc/invoke.texi: Document +nodsp as a valid extension for
718 2017-10-16 Martin Liska <mliska@suse.cz>
720 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
721 (test_set_range): Likewise.
722 (test_range_functions): Rename to ...
723 (test_bit_in_range): ... this.
724 (sbitmap_c_tests): Add new test.
726 2017-10-16 Tamar Christina <tamar.christina@arm.com>
728 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
729 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
730 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
732 2017-10-16 Tamar Christina <tamar.christina@arm.com>
734 * config/aarch64/aarch64-builtins.c
735 (aarch64_types_quadopu_lane_qualifiers): New.
736 (TYPES_QUADOPU_LANE): New.
737 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
738 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
739 (aarch64_<sur>dot_laneq<vsi2qi>): New.
740 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
741 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
742 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
743 (Vdottype, DOTPROD): New.
744 (sur): Add SDOT and UDOT.
746 2017-10-16 Tamar Christina <tamar.christina@arm.com>
748 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
749 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
750 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
751 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
752 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
753 (cortex-a75.cortex-a55): Likewise.
754 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
756 2017-10-16 Tamar Christina <tamar.christina@arm.com>
758 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
759 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
760 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
761 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
762 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
763 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
764 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
765 * config/arm/types.md (neon_dot, neon_dot_q): New.
766 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
768 2017-10-16 Tamar Christina <tamar.christina@arm.com>
770 * config/arm/arm.h (TARGET_DOTPROD): New.
771 * config/arm/arm.c (arm_arch_dotprod): New.
772 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
773 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
774 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
775 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
776 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
777 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
778 * doc/invoke.texi (armv8.2-a): Document dotprod
780 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
782 * i386.c (ix86_vec_cost): New function.
783 (ix86_rtx_costs): Handle vector operations better.
784 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
785 * x86-tune-costs.h: Add new costs to all tables.
787 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
789 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
791 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
792 divsd, sqrtss and sqrtsd
793 * x86-tune-costs.h: Add new entries to all costs.
794 (znver1_cost): Fix to match real instruction latencies.
796 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
797 Michael Collison <michael.collison@arm.com>
799 * compare-elim.c: Include emit-rtl.h.
800 (can_merge_compare_into_arith): New function.
801 (try_validate_parallel): Likewise.
802 (try_merge_compare): Likewise.
803 (try_eliminate_compare): Call the above when no previous clobber
805 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
808 2017-10-14 Jakub Jelinek <jakub@redhat.com>
812 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
813 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
817 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
818 to be any operand_equal_p operands. For & (B - 1) require
819 B to be power of 2. Recognize
820 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
822 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
825 * optabs.c (expand_memory_blockage): Fix call of
826 targetm.have_memory_blockage.
828 2017-10-14 Jakub Jelinek <jakub@redhat.com>
831 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
832 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
833 objects to extra_objs instead of overwriting it.
835 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
837 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
838 Use any_fp_register_operand as operand[3] predicate. Simplify
839 equality test for operands[2] and operands[4] memory location.
840 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
841 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
842 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
843 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
844 any_fp_register_operand as operand[1] predicate. Simplify
845 equality test for operands[0] and operands[3] memory location.
846 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
847 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
848 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
850 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
852 * target-insns.def: Add memory_blockage.
853 * optabs.c (expand_memory_blockage): New function.
854 (expand_asm_memory_barrier): Rename ...
855 (expand_asm_memory_blockage): ... to this.
856 (expand_mem_thread_fence): Call expand_memory_blockage
857 instead of expand_asm_memory_barrier.
858 (expand_mem_singnal_fence): Ditto.
859 (expand_atomic_load): Ditto.
860 (expand_atomic_store): Ditto.
861 * doc/md.texi (Standard Pattern Names For Generation):
862 Document memory_blockage instruction pattern.
864 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
866 * config/rl78/rl78.c (rl78_emit_libcall): New function.
867 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
868 * config/rl78/rl78.md: New define_expand "adddi3".
870 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
872 * cfghooks.c (verify_flow_info): Disable check that all probabilities
875 2017-10-13 Jeff Law <law@redhat.com>
877 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
879 2017-10-13 Jakub Jelinek <jakub@redhat.com>
882 * internal-fn.c (expand_mul_overflow): If both operands have
883 the same highpart of -1 or 0 and the topmost bit of lowpart
884 is different, overflow is if res <= 0 rather than res < 0.
886 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
888 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
889 TARGET_P9_VECTOR code for unaligned_load case.
891 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
893 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
895 2017-10-13 Nathan Sidwell <nathan@acm.org>
897 * tree-core.h (tree_contains_struct): Make bool.
898 * tree.c (tree_contains_struct): Likewise.
899 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
900 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
901 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
902 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
903 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
905 2017-10-13 Richard Biener <rguenther@suse.de>
907 * graphite-isl-ast-to-gimple.c
908 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
909 parameters and dominance check.
910 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
911 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
912 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
913 Do not update SSA form here or do intermediate IL verification.
914 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
915 (graphite_initialize): Remove check on the number of loops in
916 the function and inline into graphite_transform_loops.
917 (graphite_finalize): Inline into graphite_transform_loops.
918 (graphite_transform_loops): Perform SSA update and IL verification
920 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
922 2017-10-13 Richard Biener <rguenther@suse.de>
924 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
925 graphite_expression_type_precision): Avoid global constructor
927 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
928 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
929 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
930 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
931 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
932 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
934 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
937 * config/i386/i386.h (ix86_red_zone_size): New.
938 * config/i386/i386.md (push peephole2s): Replace
939 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
941 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
942 Alan Hayward <alan.hayward@arm.com>
943 David Sherwood <david.sherwood@arm.com>
945 * combine.c (can_change_dest_mode): Reject changes in
946 REGMODE_NATURAL_SIZE.
948 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
949 Alan Hayward <alan.hayward@arm.com>
950 David Sherwood <david.sherwood@arm.com>
952 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
953 (expand_debug_source_expr): Likewise.
954 * combine.c (combine_simplify_rtx): Likewise.
955 * cse.c (fold_rtx): Likewise.
956 * fwprop.c (canonicalize_address): Likewise.
957 * targhooks.c (default_shift_truncation_mask): Likewise.
959 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
960 Alan Hayward <alan.hayward@arm.com>
961 David Sherwood <david.sherwood@arm.com>
963 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
964 (widened_mode): Likewise.
965 (expand_unop): Likewise.
966 * ree.c (transform_ifelse): Likewise.
967 (merge_def_and_ext): Likewise.
968 (combine_reaching_defs): Likewise.
969 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
971 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
972 Alan Hayward <alan.hayward@arm.com>
973 David Sherwood <david.sherwood@arm.com>
975 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
976 * combine.c (gen_lowpart_for_combine): Likewise.
977 * dwarf2out.c (rtl_for_decl_location): Likewise.
978 * final.c (alter_subreg): Likewise.
979 * rtlhooks.c (gen_lowpart_general): Likewise.
980 (gen_lowpart_if_possible): Likewise.
982 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
983 Alan Hayward <alan.hayward@arm.com>
984 David Sherwood <david.sherwood@arm.com>
986 * calls.c (expand_call): Use subreg_lowpart_offset.
987 * cse.c (cse_insn): Likewise.
988 * regcprop.c (copy_value): Likewise.
989 (copyprop_hardreg_forward_1): Likewise.
991 2017-10-13 Jakub Jelinek <jakub@redhat.com>
994 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
995 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
996 =Q constraints to +Q and into insn condition add check
997 that operands[0] and operands[1] are equal.
998 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
999 =Q constraints to +Q and into insn condition add check
1000 that operands[0] is equal to either operands[1] or operands[2].
1003 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
1004 instead of handling MINUS_EXPR twice (once for each argument),
1005 canonicalize operand order and handle just once, use rtype where
1006 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
1009 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
1010 any values of __C while still being pattern recognizable as a simple
1013 2017-10-13 Richard Biener <rguenther@suse.de>
1015 PR tree-optimization/82451
1017 2017-10-02 Richard Biener <rguenther@suse.de>
1019 PR tree-optimization/82355
1020 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1021 a mapping for the enclosing loop but avoid generating one for
1023 (copy_bb_and_scalar_dependences): Remove premature codegen
1024 error on PHIs in blocks duplicated into multiple places.
1025 * graphite-scop-detection.c
1026 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1027 in the region use it as loop and nest to analyze the DR in.
1028 (try_generate_gimple_bb): Likewise.
1029 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1030 (add_loop_constraints): For blocks in a loop not in the region
1031 create a dimension with a single iteration.
1032 * sese.h (gbb_loop_at_index): Remove assert.
1034 * cfgloop.c (loop_preheader_edge): For the loop tree root
1035 return the single successor of the entry block.
1036 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1037 Reset the SCEV hashtable and niters.
1038 * graphite-scop-detection.c
1039 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
1040 assert that we only have POLYNOMIAL_CHREC that vary in loops
1041 contained in the region.
1042 (scop_detection::graphite_can_represent_expr): Adjust.
1043 (scop_detection::stmt_has_simple_data_refs_p): For loops
1044 not in the region set loop to NULL. The nest is now the
1045 entry edge to the region.
1046 (try_generate_gimple_bb): Likewise.
1047 * sese.c (scalar_evolution_in_region): Adjust for
1048 instantiate_scev change.
1049 * tree-data-ref.h (graphite_find_data_references_in_stmt):
1050 Make nest parameter the edge into the region.
1051 (create_data_ref): Likewise.
1052 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
1053 entry edge into a region and adjust instantiate_scev calls.
1054 (create_data_ref): Likewise.
1055 (graphite_find_data_references_in_stmt): Likewise.
1056 (find_data_references_in_stmt): Pass the loop preheader edge
1057 from the nest argument.
1058 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
1059 parameter the edge into the region.
1060 (instantiate_parameters): Use the loop preheader edge as entry.
1061 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
1063 (get_instantiated_value_entry): Make instantiate_below parameter
1064 the edge into the region.
1065 (instantiate_scev_name): Likewise. Adjust dominance checks,
1066 when we cannot use loop-based instantiation instantiate by
1067 walking use-def chains.
1068 (instantiate_scev_poly): Adjust.
1069 (instantiate_scev_binary): Likewise.
1070 (instantiate_scev_convert): Likewise.
1071 (instantiate_scev_not): Likewise.
1072 (instantiate_array_ref): Remove.
1073 (instantiate_scev_3): Likewise.
1074 (instantiate_scev_2): Likewise.
1075 (instantiate_scev_1): Likewise.
1076 (instantiate_scev_r): Do not blindly handle N-operand trees.
1077 Do not instantiate array-refs. Handle all constants and invariants.
1078 (instantiate_scev): Make instantiate_below parameter
1079 the edge into the region.
1080 (resolve_mixers): Use the loop preheader edge for the region
1081 parameter to instantiate_scev_r.
1082 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1084 2017-10-13 Richard Biener <rguenther@suse.de>
1086 PR tree-optimization/82525
1087 * graphite-isl-ast-to-gimple.c
1088 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1090 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1091 Fail code generation when we cannot represent the isl integer.
1092 (binary_op_to_tree): Elide modulo operations that are no-ops
1093 in the type we code generate. Remove now superfluous code
1096 2017-10-13 Richard Biener <rguenther@suse.de>
1098 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1099 (scop_detection::harmful_loop_in_region): Remove premature
1100 IV type restriction.
1101 (scop_detection::graphite_can_represent_scev): We can handle
1102 pointer IVs just fine.
1104 2017-10-13 Alan Modra <amodra@gmail.com>
1106 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1107 "Clobbers and Scratch Registers". Add paragraph on
1108 alternative to clobbers for scratch registers and OpenBLAS
1111 2017-10-13 Alan Modra <amodra@gmail.com>
1113 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1114 example of a memory input for a string of known length. Move
1115 commentary out of table. Add a number of new examples
1116 covering array memory inputs.
1118 2017-10-12 Martin Liska <mliska@suse.cz>
1120 PR tree-optimization/82493
1121 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1122 (test_range_functions): New function.
1123 (sbitmap_c_tests): Likewise.
1124 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1125 * selftest.h (sbitmap_c_tests): New function.
1127 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1129 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1131 * config/rs6000/amo.h: Fix spacing issue.
1133 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1136 * config/i386/i386.md (*ashl<mode>3_mask_1,
1137 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1138 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1141 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1143 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
1144 (profile_probability): Set max_probability
1145 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
1147 * profile-count.c (profile_probability::differs_from_p): Do not
1148 rely on max_probaiblity == 10000
1150 2017-10-12 Jeff Law <law@redhat.com>
1152 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
1155 2017-10-12 Martin Sebor <msebor@redhat.com>
1159 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
1160 (handle_alias_pairs): Call it.
1161 * common.opt (-Wattribute-alias): New option.
1162 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
1163 * doc/invoke.texi (-Wattribute-alias): Document.
1165 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
1168 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1170 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1172 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1173 (make_hard_regno_born, make_hard_regno_dead): Update
1174 bb_killed_pseudos and bb_gen_pseudos.
1176 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1178 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
1180 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
1182 * config/alpha/alpha.c (alpha_split_conditional_move):
1183 Use std::swap instead of manually swapping.
1184 (alpha_stdarg_optimize_hook): Ditto.
1185 (alpha_canonicalize_comparison): Ditto.
1187 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1189 * tree-loop-distribution.c (struct builtin_info): New struct.
1190 (struct partition): Refactor fields into struct builtin_info.
1191 (partition_free): Free struct builtin_info.
1192 (build_size_arg_loc, build_addr_arg_loc): Delete.
1193 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
1194 information from struct builtin_info.
1195 (find_single_drs): New function refactored from classify_partition.
1196 Also moved builtin validity checks to this function.
1197 (compute_access_range, alloc_builtin): New functions.
1198 (classify_builtin_st, classify_builtin_ldst): New functions.
1199 (classify_partition): Refactor code into functions find_single_drs,
1200 classify_builtin_st and classify_builtin_ldst.
1201 (distribute_loop): Don't do runtime alias check when distributing
1203 (find_seed_stmts_for_distribution): New function.
1204 (pass_loop_distribution::execute): Refactor code finding seed
1205 stmts into above function. Support distribution for the innermost
1206 two-level loop nest. Adjust dump information.
1208 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1210 * tree-loop-distribution.c: Adjust the general comment.
1211 (NUM_PARTITION_THRESHOLD): New macro.
1212 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
1213 (classify_partition): Skip builtin pattern of loop nest's inner loop.
1214 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
1215 in call to build_partition_graph.
1216 (finalize_partitions): New parameter. Make loop distribution more
1217 conservative by fusing more partitions.
1218 (distribute_loop): Don't do runtime alias check in case of loop nest
1220 (find_seed_stmts_for_distribution): New function.
1221 (prepare_perfect_loop_nest): New function.
1222 (pass_loop_distribution::execute): Refactor code finding seed stmts
1223 and loop nest into above functions. Support loop nest distribution.
1224 Adjust dump information accordingly.
1226 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1228 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
1229 and set PTYPE_SEQUENTIAL for merged partition.
1231 2017-10-12 Richard Biener <rguenther@suse.de>
1233 PR tree-optimization/69728
1235 2017-09-19 Richard Biener <rguenther@suse.de>
1237 PR tree-optimization/69728
1238 * graphite-sese-to-poly.c (schedule_error): New global.
1239 (add_loop_schedule): Handle empty domain by failing the
1241 (build_original_schedule): Handle schedule_error.
1243 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
1244 domain by returning an unchanged schedule.
1246 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1248 * genrecog.c (validate_pattern): For VEC_SELECT verify that
1249 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
1251 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
1253 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
1256 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1259 * expr.c (store_field): Don't optimize away bitsize == 0 store
1260 from CALL_EXPR with addressable return type.
1262 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1264 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
1265 * config/rs6000/rs6000.md (sel): Delete mode attribute.
1266 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
1267 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
1268 TARGET_ISEL instead of TARGET_ISEL<sel>.
1270 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
1272 * config/rs6000/rs6000.c
1273 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
1275 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1277 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
1279 (reg_or_cint_operand): Fix comment.
1280 (reg_or_zero_operand): New predicate.
1281 * config/rs6000/rs6000-protos.h (output_isel): Delete.
1282 * config/rs6000/rs6000.c (output_isel): Delete.
1283 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
1284 instead of reg_or_cint_operand. Output instruction directly (not via
1286 (isel_unsigned_<mode>): Ditto.
1287 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
1288 gpc_reg_operand. Add an instruction alternative for this. Output
1289 instruction directly.
1290 (*isel_reversed_unsigned_<mode>): Ditto.
1292 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1294 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
1295 (TARGET_CANONICALIZE_COMPARISON): Define.
1297 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
1300 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1301 Check whether the dest is REG before adding REG_EQUIV note.
1303 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1306 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1308 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1309 (make_hard_regno_born, make_hard_regno_dead): Update
1310 bb_killed_pseudos and bb_gen_pseudos.
1312 2017-10-11 Nathan Sidwell <nathan@acm.org>
1314 * incpath.h (enum incpath_kind): Name enum, prefix values.
1315 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
1316 * incpath.c (heads, tails): Use INC_MAX.
1317 (add_env_var_paths, add_standard_paths): Use incpath_kind.
1318 (merge_include_chains, split_quote_chain,
1319 register_include_chains): Update incpath_kind names.
1320 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
1321 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
1323 (add_framework_path, darwin_register_objc_includes): Likewise.
1324 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
1326 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1328 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1329 Do not use float_operator operator predicate.
1330 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1331 * config/i386/predicates.md (float_operator): Remove predicate.
1333 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1335 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
1336 (*jccxf_i387): Ditto.
1337 (*jcc<mode>_i387): Ditto.
1338 (*jccu<mode>_i387): Ditto.
1339 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
1340 (*jcc_*_i387 splitters): Remove.
1341 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
1342 * config/i386/i386.c (ix86_split_fp_branch): Remove.
1343 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
1346 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1348 * profile-count.h (slow_safe_scale_64bit): New function.
1349 (safe_scale_64bit): New inline.
1350 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
1351 * profile-count.c: Include wide-int.h
1352 (slow_safe_scale_64bit): New.
1354 2017-10-11 Nathan Sidwell <nathan@acm.org>
1356 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
1357 HAS_DECL_ASSEMBLER_NAME_P.
1358 * gimple-expr.c (gimple_decl_printable_name: Check
1359 HAS_DECL_ASSEMBLER_NAME_P too.
1360 * ipa-utils.h (type_in_anonymous_namespace_p): Check
1361 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
1362 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
1363 * passes.c (rest_of_decl_compilation): Check
1364 HAS_DECL_ASSEMBLER_NAME_P too.
1365 * recog.c (verify_changes): Likewise.
1366 * tree-pretty-print.c (dump_decl_name): Likewise.
1367 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
1369 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
1370 (SET_DECL_ASSEMBLER_NAME): Use it.
1371 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
1372 (COPY_DECL_ASSEMBLER_NAME): Likewise.
1373 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
1375 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1377 * config.gcc (i386, x86_64): Add extra objects.
1378 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
1379 (ix86_min_insn_size): Declare.
1380 (ix86_issue_rate): Declare.
1381 (ix86_adjust_cost): Declare.
1382 (ia32_multipass_dfa_lookahead): Declare.
1383 (ix86_macro_fusion_p): Declare.
1384 (ix86_macro_fusion_pair_p): Declare.
1385 (ix86_bd_has_dispatch): Declare.
1386 (ix86_bd_do_dispatch): Declare.
1387 (ix86_core2i7_init_hooks): Declare.
1388 (ix86_atom_sched_reorder): Declare.
1389 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
1390 (COSTS_N_BYTES): Move to x86-tune-costs.h.
1391 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
1392 (rip_relative_addr_p): Rename to ...
1393 (ix86_rip_relative_addr_p): ... this one; export.
1394 (memory_address_length): Update.
1395 (ix86_issue_rate): Move to x86-tune-sched.c.
1396 (ix86_flags_dependent): Move to x86-tune-sched.c.
1397 (ix86_agi_dependent): Move to x86-tune-sched.c.
1398 (exact_dependency_1): Move to x86-tune-sched.c.
1399 (exact_store_load_dependency): Move to x86-tune-sched.c.
1400 (ix86_adjust_cost): Move to x86-tune-sched.c.
1401 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
1402 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
1403 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
1404 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
1405 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
1406 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
1407 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
1408 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
1409 (min_insn_size): Rename to ...
1410 (ix86_min_insn_size): ... this one; export.
1411 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
1412 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
1413 (core2i7_first_cycle_multipass_backtrack): Move to
1414 x86-tune-sched-core.c.
1415 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
1416 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
1417 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
1418 (ix86_avoid_jump_mispredicts): Update.
1419 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
1420 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
1421 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
1422 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
1423 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
1424 (MAX_INSN): Move to ix86-tune-sched-bd.c.
1425 (MAX_IMM): Move to ix86-tune-sched-bd.c.
1426 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
1427 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
1428 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
1429 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
1430 (MAX_STORE): Move to ix86-tune-sched-bd.c.
1431 (BIG): Move to ix86-tune-sched-bd.c.
1432 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
1433 (enum insn_path): Move to ix86-tune-sched-bd.c.
1434 (get_mem_group): Move to ix86-tune-sched-bd.c.
1435 (is_cmp): Move to ix86-tune-sched-bd.c.
1436 (dispatch_violation): Move to ix86-tune-sched-bd.c.
1437 (is_branch): Move to ix86-tune-sched-bd.c.
1438 (is_prefetch): Move to ix86-tune-sched-bd.c.
1439 (init_window): Move to ix86-tune-sched-bd.c.
1440 (allocate_window): Move to ix86-tune-sched-bd.c.
1441 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
1442 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
1443 (process_end_window): Move to ix86-tune-sched-bd.c.
1444 (allocate_next_window): Move to ix86-tune-sched-bd.c.
1445 (find_constant): Move to ix86-tune-sched-bd.c.
1446 (get_num_immediates): Move to ix86-tune-sched-bd.c.
1447 (has_immediate): Move to ix86-tune-sched-bd.c.
1448 (get_insn_path): Move to ix86-tune-sched-bd.c.
1449 (get_insn_group): Move to ix86-tune-sched-bd.c.
1450 (count_num_restricted): Move to ix86-tune-sched-bd.c.
1451 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
1452 (add_insn_window): Move to ix86-tune-sched-bd.c.
1453 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
1454 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
1455 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
1456 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
1457 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
1458 (do_dispatch): Move to ix86-tune-sched-bd.c.
1459 (has_dispatch): Move to ix86-tune-sched-bd.c.
1460 * config/i386/t-i386: Add new object files.
1461 * config/i386/x86-tune-costs.h: New file.
1462 * config/i386/x86-tune-sched-atom.c: New file.
1463 * config/i386/x86-tune-sched-bd.c: New file.
1464 * config/i386/x86-tune-sched-core.c: New file.
1465 * config/i386/x86-tune-sched.c: New file.
1467 2017-10-11 Liu Hao <lh_mouse@126.com>
1469 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
1470 the generic version below instead.
1471 (should_colorize): Recognize Windows consoles as terminals
1473 * pretty-print.c [__MINGW32__] (write_all): New function.
1474 [__MINGW32__] (find_esc_head): Likewise.
1475 [__MINGW32__] (find_esc_terminator): Likewise.
1476 [__MINGW32__] (eat_esc_sequence): Likewise.
1477 [__MINGW32__] (mingw_ansi_fputs): New function that handles
1479 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
1482 2017-10-11 Richard Biener <rguenther@suse.de>
1484 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
1485 Properly call analyze_scalar_evolution with the loop of the stmt.
1487 2017-10-11 Richard Biener <rguenther@suse.de>
1489 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
1490 * tree-core.h (tree_base): Add chrec_var union member.
1491 * tree.h (CHREC_VAR): Remove.
1492 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
1493 * tree-chrec.h (build_polynomial_chrec): Adjust.
1494 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
1495 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
1497 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
1499 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
1500 * match.pd: ... here.
1501 ((T) X == (T) Y): Relax condition.
1503 2017-10-11 Bin Cheng <bin.cheng@arm.com>
1505 PR tree-optimization/82472
1506 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
1508 (break_alias_scc_partitions): Update postorder number.
1510 2017-10-11 Martin Liska <mliska@suse.cz>
1513 * opts.c (parse_no_sanitize_attribute): Do not use error_value
1515 * opts.h (parse_no_sanitize_attribute): Remove last argument.
1517 2017-10-11 Martin Liska <mliska@suse.cz>
1519 * print-rtl.c (print_insn): Move declaration of idbuf
1520 to same scope as name.
1522 2017-10-11 Martin Liska <mliska@suse.cz>
1527 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1530 2017-10-11 Martin Liska <mliska@suse.cz>
1533 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1536 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
1538 * config.gcc (default_gnu_indirect_function): Default to yes for
1539 arm*-*-linux* with glibc.
1541 2017-10-11 Richard Biener <rguenther@suse.de>
1543 * tree-scalar-evolution.c (get_scalar_evolution): Handle
1544 default-defs and types we do not want to analyze.
1545 (interpret_loop_phi): Replace unreachable code with an assert.
1546 (compute_scalar_evolution_in_loop): Remove and inline ...
1547 (analyze_scalar_evolution_1): ... here, replacing condition with
1548 what makes the intent clearer. Remove handling of cases
1549 get_scalar_evolution now handles.
1551 2017-10-10 Jim Wilson <wilson@tuliptree.org>
1553 PR rtl-optimization/81434
1554 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
1555 comment for main loop. In sched_group_found if, also add checks for
1556 pass and min_cost_group.
1558 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
1560 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
1561 (rs6000_insn_cost): New function.
1562 * config/rs6000/rs6000.md (cost): New attribute.
1564 2017-10-10 Jakub Jelinek <jakub@redhat.com>
1565 H.J. Lu <hongjiu.lu@intel.com>
1569 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
1570 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
1571 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
1572 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
1573 to OPTION_MASK_ISA_AVX512VL - builtins that have both
1574 OPTION_MASK_ISA_MMX and some other bit set require both
1575 mmx and the ISAs without the mmx bit.
1576 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
1577 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
1578 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
1579 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
1580 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
1581 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
1582 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
1583 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
1584 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
1585 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
1586 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
1587 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
1588 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
1589 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
1590 Add OPTION_MASK_ISA_MMX.
1592 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
1594 * config.gcc (armv7*-*-freebsd*): New target.
1595 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
1597 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
1599 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
1600 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
1603 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
1605 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
1606 a template parameter.
1607 (WIDE_INT_REF_FOR): Update accordingly.
1608 * tree.h (wi::int_traits <const_tree>): Delete.
1609 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
1610 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
1611 (wi::tree_to_wide_ref): New typedef.
1612 (wi::to_wide): New function.
1613 * calls.c (get_size_range): Use wi::to_wide when operating on
1615 * cgraph.c (cgraph_node::create_thunk): Likewise.
1616 * config/i386/i386.c (ix86_data_alignment): Likewise.
1617 (ix86_local_alignment): Likewise.
1618 * dbxout.c (stabstr_O): Likewise.
1619 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
1620 * expr.c (const_vector_from_tree): Likewise.
1621 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
1622 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
1623 (fold_negate_expr_1, int_const_binop_1, const_binop)
1624 (fold_convert_const_int_from_real, optimize_bit_field_compare)
1625 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
1626 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
1627 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
1628 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
1629 (fold_not_const, round_up_loc): Likewise.
1630 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
1631 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
1632 (alloca_call_type): Likewise.
1633 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
1634 * godump.c (go_output_typedef): Likewise.
1635 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1636 * internal-fn.c (get_min_precision): Likewise.
1637 * ipa-cp.c (ipcp_store_vr_results): Likewise.
1638 * ipa-polymorphic-call.c
1639 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1640 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
1641 (ipa_modify_call_arguments): Likewise.
1642 * match.pd: Likewise.
1643 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
1644 * print-tree.c (print_node_brief, print_node): Likewise.
1645 * stmt.c (expand_case): Likewise.
1646 * stor-layout.c (layout_type): Likewise.
1647 * tree-affine.c (tree_to_aff_combination): Likewise.
1648 * tree-cfg.c (group_case_labels_stmt): Likewise.
1649 * tree-data-ref.c (dr_analyze_indices): Likewise.
1650 (prune_runtime_alias_test_list): Likewise.
1651 * tree-dump.c (dequeue_and_dump): Likewise.
1652 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
1653 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
1654 * tree-pretty-print.c (dump_generic_node): Likewise.
1655 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
1656 (simple_iv_with_niters): Likewise.
1657 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1658 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
1659 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
1660 * tree-ssa-loop-niter.c (split_to_var_and_offset)
1661 (refine_value_range_using_guard, number_of_iterations_ne_max)
1662 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
1663 (get_cst_init_from_scev, record_nonwrapping_iv)
1664 (scev_var_range_cant_overflow): Likewise.
1665 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
1666 * tree-ssa-pre.c (compute_avail): Likewise.
1667 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1668 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
1669 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1670 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
1671 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
1672 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
1673 (dump_case_nodes, try_switch_expansion): Likewise.
1674 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
1675 (vect_do_peeling): Likewise.
1676 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
1677 * tree-vect-stmts.c (vectorizable_load): Likewise.
1678 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
1679 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
1680 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
1681 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
1682 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
1683 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
1684 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
1685 (evrp_dom_walker::before_dom_children): Likewise.
1686 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
1687 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
1688 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
1689 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
1690 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
1691 (get_range_pos_neg): Likewise.
1692 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
1693 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
1694 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
1695 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
1696 * config/avr/avr.c (avr_fold_builtin): Likewise.
1697 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
1698 * config/msp430/msp430.c (msp430_attr): Likewise.
1699 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
1700 * config/powerpcspe/powerpcspe-c.c
1701 (altivec_resolve_overloaded_builtin): Likewise.
1702 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
1703 (rs6000_expand_ternop_builtin): Likewise.
1704 * config/rs6000/rs6000-c.c
1705 (altivec_resolve_overloaded_builtin): Likewise.
1706 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
1707 (rs6000_expand_ternop_builtin): Likewise.
1708 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
1710 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1712 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
1713 when copying loop nest with only one inner loop.
1715 2017-10-10 Richard Biener <rguenther@suse.de>
1717 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
1718 blocks if SCEV is active.
1719 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
1721 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
1722 (scev_initialize): Assert we are not yet initialized.
1724 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1726 * tree-loop-distribution.c (generate_loops_for_partition): Remove
1727 inner loop's exit stmt by making it always exit the loop, otherwise
1728 we would generate an infinite empty loop.
1730 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1732 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
1733 renaming variables in new preheader if it's deleted.
1735 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1737 * tree-loop-distribution.c (struct partition): Remove unused field
1738 loops of the structure.
1739 (partition_alloc, partition_free): Ditto.
1740 (build_rdg_partition_for_vertex): Ditto.
1742 2017-10-09 Jeff Law <law@redhat.com>
1744 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
1745 return type to match prototype and documentation.
1747 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1749 * config/rs6000/rs6000.c (processor_costs): Move to ...
1750 * config/rs6000/rs6000.h: ... here.
1751 (rs6000_cost): Declare.
1753 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
1755 * except.c (setjmp_fn): New global variable.
1756 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
1757 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
1758 if DONT_USE_BUILTIN_SETJMP is defined.
1760 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1762 * target.def (insn_cost): New hook.
1763 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
1764 * doc/tm.texi: Regenerate.
1765 * rtlanal.c (insn_cost): Use the new hook.
1767 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1769 * combine.c (combine_validate_cost): Compute the new insn_cost,
1770 not just pattern_cost.
1771 (try_combine): Adjust comment.
1773 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1775 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
1777 * combine.c (uid_insn_cost): Adjust comment.
1778 (combine_validate_cost): Adjust comment. Use pattern_cost instead
1780 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
1781 * dse.c (find_shift_sequence): Ditto.
1782 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
1783 (bb_valid_for_noce_process_p): Use pattern_cost.
1784 * rtl.h (insn_rtx_cost): Delete.
1785 (pattern_cost): New prototype.
1786 (insn_cost): New prototype.
1787 * rtlanal.c (insn_rtx_cost): Rename to...
1788 (pattern_cost): ... this.
1791 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
1793 * config/i386/i386.md (*jcc_2): Remove insn pattern.
1794 (*jcc<mode>_0_r_i387): Ditto.
1795 (*jccxf_r_i387): Ditto.
1796 (*jcc<mode>_r_i387): Ditto.
1797 (*jccu<mode>_r_i387): Ditto.
1798 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
1799 (*jcc): Rename from *jcc_1.
1801 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1803 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
1804 deferred rescans after the lvx/stvx recombination pre-pass.
1806 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1808 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
1809 memory operation instruction support.
1810 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
1811 (rs6000-ibm-aix[789]*): Likewise.
1812 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
1813 Document new functions.
1815 2017-10-09 Richard Biener <rguenther@suse.de>
1817 PR tree-optimization/82397
1818 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
1819 equality only for semantically equal trees.
1821 2017-10-09 Richard Biener <rguenther@suse.de>
1823 PR tree-optimization/82449
1824 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
1825 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
1826 allow constant addresses.
1827 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
1830 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1832 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
1835 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1838 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
1841 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1844 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
1846 2017-10-09 Jakub Jelinek <jakub@redhat.com>
1849 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
1850 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
1852 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
1854 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
1855 (WI_BINARY_PREDICATE_RESULT): Likewise.
1856 (wi::binary_traits::operator_result): New type.
1857 (wi::binary_traits::predicate_result): Likewise.
1858 (generic_wide_int::operator~, unary generic_wide_int::operator-)
1859 (generic_wide_int::operator==, generic_wide_int::operator!=)
1860 (generic_wide_int::operator&, generic_wide_int::and_not)
1861 (generic_wide_int::operator|, generic_wide_int::or_not)
1862 (generic_wide_int::operator^, generic_wide_int::operator+
1863 (binary generic_wide_int::operator-, generic_wide_int::operator*):
1865 (operator~, unary operator-, operator==, operator!=, operator&)
1866 (operator|, operator^, operator+, binary operator-, operator*): New
1868 * expr.c (get_inner_reference): Use wi::bit_and_not.
1869 * fold-const.c (fold_binary_loc): Likewise.
1870 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
1871 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
1872 (bit_value_binop): Likewise.
1873 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
1874 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
1875 (extract_range_from_binary_expr_1): Likewise.
1876 (masked_increment): Likewise.
1877 (simplify_bit_ops_using_ranges): Likewise.
1879 2017-10-09 Martin Jambor <mjambor@suse.cz>
1882 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
1883 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
1884 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
1886 (hsa_fixup_mov_insn_type): New function.
1887 (hsa_op_with_type::get_in_type): Use it.
1888 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
1889 immediates in an assert.
1890 (hsa_op_with_type::extend_int_to_32bit): New method.
1891 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
1892 types. Convert to dest type if necessary.
1893 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
1894 (reg_for_gimple_ssa): Pass false as min32int to
1895 hsa_type_for_scalar_tree_type.
1896 (gen_hsa_addr): Fixup type when creating addresable temporary.
1897 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
1898 (gen_hsa_unary_operation): Extend operands and convert to dest type if
1899 necessary. Call hsa_fixup_mov_insn_type.
1900 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
1901 extend operands and convert to dest type if necessary.
1902 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
1903 to dest type if necessary.
1904 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
1905 if conversion nt necessary and size matches.
1906 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
1907 to dest type if necessary.
1908 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
1909 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
1911 (gen_hsa_clrsb): Likewise.
1912 (gen_hsa_ffs): Likewise.
1913 (gen_hsa_divmod): Extend operands and convert to dest type if
1915 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
1917 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1919 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
1920 Remove empty default arguments. Use a brace block as output
1922 (conditional return): Ditto.
1924 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
1925 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
1927 (group_ending_nop): Ditto.
1928 (doloop_end): Ditto.
1929 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
1930 (splitters for those): Ditto.
1932 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1934 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
1935 a conditional jump (and the compare for it) so that pc_rtx is the
1937 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
1938 for the deleted and renamed ctr<mode>_internal[234] patterns.
1939 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
1940 Delete second conditional return pattern.
1941 (ctr<mode>_internal2): Delete this second bdnz pattern.
1942 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
1943 (ctr<mode>_internal4): Delete this second bdz pattern.
1945 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1947 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
1948 (always_initialized_rtx_for_ssa_name_p): New predicate.
1949 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
1950 (finish_out_of_ssa): Free new field of SA.
1951 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
1952 * tree-ssa-coalesce.c: Include tree-ssa.h.
1953 (get_parm_default_def_partitions): Remove extern keyword.
1954 (get_undefined_value_partitions): New function.
1955 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
1956 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
1958 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
1960 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1962 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
1964 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1966 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
1967 for targets that preffer 128bit.
1969 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1971 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
1973 2017-10-08 Olivier Hainque <hainque@adacore.com>
1975 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
1976 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
1978 (thumb_set_return_address): Likewise.
1980 2017-10-08 Olivier Hainque <hainque@adacore.com>
1982 * common/config/arm/arm-common.c (arm_except_unwind_info):
1983 Handle DWARF2_UNWIND_INFO.
1985 2017-10-07 Michael Collison <michael.collison@arm.com>
1987 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
1990 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
1992 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
1993 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
1994 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
1995 defined, force the creation of a new block for a dispatch label.
1997 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
1999 * invoke.texi (Wsuggest-attribute=cold): Document.
2000 * common.opt (Wsuggest-attribute=cold): New
2001 * ipa-pure-const.c (warn_function_cold): New function.
2002 * predict.c (compute_function_frequency): Use it.
2003 * predict.h (warn_function_cold): Declare.
2005 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
2007 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
2010 2017-10-06 Martin Liska <mliska@suse.cz>
2012 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
2013 keyword for member functions.
2014 (struct sanopt_tree_couple): New struct.
2015 (struct sanopt_tree_couple_hash): New function.
2016 (struct sanopt_ctx): Add new hash_map.
2017 (has_dominating_ubsan_ptr_check): New function.
2018 (record_ubsan_ptr_check_stmt): Likewise.
2019 (maybe_optimize_ubsan_ptr_ifn): Likewise.
2020 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
2021 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
2023 2017-10-06 Sudakshina Das <sudi.das@arm.com>
2026 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
2027 aarch64_simd_valid_immediate on CONST_VECTORs.
2028 (aarch64_reg_or_bic_imm): Likewise.
2030 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
2032 PR rtl-optimization/82396
2033 * haifa-sched.c (ready_sort_real): Disable qsort checking.
2035 2017-10-06 Sebastian Pop <sebpop@gmail.com>
2037 * graphite-dependences.c (scop_get_reads): Move code to...
2038 (scop_get_must_writes): Move code to...
2039 (scop_get_may_writes): Move code to...
2040 (scop_get_reads_and_writes): ... here.
2041 (scop_get_dependences): Call scop_get_reads_and_writes.
2043 2017-10-06 Jakub Jelinek <jakub@redhat.com>
2045 PR tree-optimization/82434
2046 * fold-const.h (can_native_encode_type_p,
2047 can_native_encode_string_p): Remove.
2048 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
2049 don't encode anything, just return what would be otherwise returned.
2050 (native_encode_fixed, native_encode_complex, native_encode_vector):
2052 (native_encode_string): Likewise. Inline by hand
2053 can_native_encode_string_p.
2054 (can_native_encode_type_p): Remove.
2055 (can_native_encode_string_p): Remove.
2056 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
2057 STRING_CSTs using can_native_encode_string_p, test all
2058 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
2059 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
2060 argument from native_encode_expr.
2061 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
2062 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
2065 2017-10-06 Richard Biener <rguenther@suse.de>
2067 PR tree-optimization/82397
2068 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2069 operand_equal_p but rely on data_ref_compare_tree for detecting
2071 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2072 to match up with dr_group_sort_cmp.
2074 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2077 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2079 * config/s390/s390-builtin-types.def: Regenerate.
2081 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2084 * config/s390/s390-builtin-types.def: Regenerate.
2085 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2086 Change flag from B_VXE to B_VX.
2087 (s390_vec_min_dbl): Remove B_VXE flag.
2089 2017-10-06 Richard Biener <rguenther@suse.de>
2091 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2092 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2093 translate_isl_ast_to_gimple::is_valid_rename,
2094 translate_isl_ast_to_gimple::get_rename,
2095 translate_isl_ast_to_gimple::get_def_bb_for_const,
2096 translate_isl_ast_to_gimple::get_new_name,
2097 translate_isl_ast_to_gimple::collect_all_ssa_names,
2098 translate_isl_ast_to_gimple::copy_loop_phi_args,
2099 translate_isl_ast_to_gimple::collect_all_ssa_names,
2100 translate_isl_ast_to_gimple::copy_loop_phi_args,
2101 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2102 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2103 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2104 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2105 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2106 translate_isl_ast_to_gimple::copy_cond_phi_args,
2107 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2108 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2109 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2110 translate_isl_ast_to_gimple::rename_uses,
2111 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2112 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2113 (set_rename_for_each_def): Likewise.
2114 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2115 here. Handle rewriting SCEV analyzable uses here.
2116 (copy_bb_and_scalar_dependences): Generate code for PHI
2118 (graphite_regenerate_ast_isl): Adjust.
2119 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2120 (add_write, add_read): New functions.
2121 (build_cross_bb_scalars_def): Use it and simplify.
2122 (build_cross_bb_scalars_use): Likewise.
2123 (graphite_find_cross_bb_scalar_vars): Inline into...
2124 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2125 simulating out-of-SSA. Compute liveout and add dependencies.
2126 (build_scops): Force an empty entry block.
2127 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2129 (sese_build_liveouts): Declare.
2130 (sese_trivially_empty_bb_p): Likewise.
2131 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2132 compute liveout and debug_liveout.
2133 (sese_bad_liveouts_use): Remove.
2134 (sese_reset_debug_liveouts_bb): Likewise.
2135 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2136 (sese_build_liveouts): Build liveout and debug_liveout and store
2138 (new_sese_info): Adjust.
2139 (free_sese_info): Likewise.
2140 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2141 do not build liveout here.
2142 (move_sese_in_condition): Adjust region entry.
2143 (scev_analyzable_p): Match up with chrec_apply requirements.
2144 (sese_trivially_empty_bb_p): New.
2145 * tree-into-ssa.c (get_reaching_def): Properly support generating
2146 default-defs for incremental rewrite of anonymous names.
2148 2017-10-06 Richard Biener <rguenther@suse.de>
2150 * graphite-sese-to-poly.c (extract_affine): For casts increasing
2151 precision do not perform modulo reduction.
2153 2017-10-06 Richard Biener <rguenther@suse.de>
2155 PR tree-optimization/82436
2156 * tree-vect-slp.c (vect_supported_load_permutation_p): More
2157 conservatively choose the vectorization factor when checking
2158 whether we can perform the required load permutation.
2159 (vect_transform_slp_perm_load): Assert when we may not fail.
2161 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2163 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
2164 message for incompatible -msdata=* and -mcall-* options.
2166 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2168 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
2169 rate for post-reload scheduling.
2171 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2173 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
2175 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2177 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
2178 to improve monte carlo in scimark.
2180 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2182 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
2183 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
2184 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
2185 pentium4_cost, nocona_cost): Set reassociation width to 1.
2186 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
2187 width to 2 for fp operations and 1 otherwise.
2188 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
2190 (atom_cost): Set reassociation width to 2.
2191 (slm_cost, generic_cost): Set fp reassociation width
2192 to 2 and 1 otherwise.
2193 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
2194 (core_cost): Set fp reassociation width to 4 and vector to 2.
2195 (ix86_reassociation_width): Rewrite using cost table; special case
2196 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
2197 and TARGET_AVX128_OPTIMAL.
2198 * config/i386/i386.h (processor_costs): Add
2199 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
2200 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
2201 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
2202 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
2203 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
2204 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
2206 2017-10-05 Nathan Sidwell <nathan@acm.org>
2208 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
2210 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2212 * config/arm/arm.c (arm_test_fpu_data): New.
2213 (arm_run_selftests): Call arm_test_fpu_data.
2215 2017-10-04 Nathan Sidwell <nathan@acm.org>
2217 * toplev.c (toplev::main): Remove excess parens on pretty_printer
2219 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
2221 2017-10-04 Sudakshina Das <sudi.das@arm.com>
2223 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
2224 check type for aarch64_simd_valid_immediate.
2225 (aarch64_output_simd_mov_immediate): Update prototype.
2226 (aarch64_simd_valid_immediate): Update prototype.
2227 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
2228 support for ORR-immediate.
2229 (and<mode>3): modified pattern to add support for BIC-immediate.
2230 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
2231 now checks for valid immediate for BIC and ORR based on new enum
2233 (aarch64_output_simd_mov_immediate): Function now used to output
2234 BIC/ORR imm as well based on new enum argument.
2235 * config/aarch64/constraints.md (Do): New vector immediate constraint.
2237 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
2238 (aarch64_reg_or_bic_imm): Likewise.
2240 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2242 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
2243 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
2245 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2249 PR rtl-optimization/82396
2250 * haifa-sched.c (autopref_multipass_init): Simplify
2252 (autopref_rank_data): Simplify sort order.
2253 * sched-int.h (autopref_multipass_data_): Remove
2254 multi_mem_insn_p, min_offset and max_offset.
2256 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2258 * doc/sourcebuild.texi: Document vect_peeling_profitable.
2260 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2262 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
2265 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2267 * doc/sourcebuild.texi: Document vect_long_mult.
2269 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
2271 PR tree-optimization/82413
2272 * fold-const.c (build_range_check): Use widest_int when comparing
2273 the maximum ETYPE value with HIGH.
2275 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2277 PR rtl-optimization/82396
2278 * haifa-sched.c (autopref_multipass_init): Simplify
2280 (autopref_rank_data): Simplify sort order.
2281 * sched-int.h (autopref_multipass_data_): Remove
2282 multi_mem_insn_p, min_offset and max_offset.
2284 2017-10-04 Jakub Jelinek <jakub@redhat.com>
2286 PR tree-optimization/82381
2287 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
2288 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
2291 PR tree-optimization/82374
2292 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
2293 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
2294 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
2295 current_function_decl to the new decl.
2297 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
2299 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
2300 helper macro for IEEE float128 hardware built-in functions.
2301 (SQRTF128_ODD): Add built-in functions with the round-to-odd
2303 (TRUNCF128_ODD): Likewise.
2304 (ADDF128_ODD): Likewise.
2305 (SUBF128_ODD): Likewise.
2306 (MULF128_ODD): Likewise.
2307 (DIVF128_ODD): Likewise.
2308 (FMAF128_ODD): Likewise.
2309 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
2310 UNSPEC_TRUNC_ROUND_TO_ODD.
2311 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
2312 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
2313 floating point round to odd instructions.
2314 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
2315 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
2316 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
2317 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
2318 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
2319 (trunc<mode>sf2_hw): Change the truncate with round to odd
2320 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
2321 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
2322 to odd hardware instructions.
2323 (sub<mode>3_odd): Likewise.
2324 (mul<mode>3_odd): Likewise.
2325 (div<mode>3_odd): Likewise.
2326 (sqrt<mode>2_odd): Likewise.
2327 (fma<mode>4_odd): Likewise.
2328 (fms<mode>4_odd): Likewise.
2329 (nfma<mode>4_odd): Likewise.
2330 (nfms<mode>4_odd): Likewise.
2331 (trunc<mode>df2_odd): Change the truncate with round to odd
2332 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
2334 * doc/extend.texi (PowerPC built-in functions): Update documentation
2335 for existing IEEE float128-bit built-in functions. Add built-in
2336 functions that generate the IEEE 128-bit floating point round to
2339 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
2341 PR rtl-optimization/77729
2342 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
2343 to (X&(C1&~C2))|C2 transformations.
2345 2017-10-03 Martin Jambor <mjambor@suse.cz>
2347 PR tree-optimization/82363
2348 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
2349 mismatch, mark lacc written regardless of racc.
2351 2017-10-03 Jakub Jelinek <jakub@redhat.com>
2353 PR tree-optimization/82381
2354 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
2355 stmt_to_insert nor wheather SSA_NAMEs are default defs.
2356 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
2357 fallthrough into reassoc_stmt_dominates_stmt_p.
2360 * combine.c (combine_instructions): Don't combine in unreachable
2363 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2366 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
2367 function to not use the have_cpu variable. Do not set cpu_index,
2368 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
2370 (rs6000_valid_attribute_p): Remove duplicate initializations of
2371 old_optimize and func_optimize.
2372 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
2373 (rs6000_activate_target_options): Make global.
2374 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
2377 2017-10-02 Jakub Jelinek <jakub@redhat.com>
2379 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
2380 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
2381 Set *poffset to 0 and *psize and *pmax_size to -1 if
2382 *poffset + *psize overflows in HOST_WIDE_INT.
2384 PR tree-optimization/82387
2385 PR tree-optimization/82388
2386 PR tree-optimization/82389
2387 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
2388 instead of live_bytes non-NULL.
2390 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
2393 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
2396 2017-10-02 Richard Biener <rguenther@suse.de>
2398 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
2399 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
2400 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
2402 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2404 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
2405 requested precision matches the type's.
2406 * calls.c (alloc_max_size): Calculate the new candidate size as
2407 a widest_int and use wi::to_widest when comparing it with the
2408 current candidate size.
2409 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
2410 zero rather than integer_zero_node.
2411 * match.pd: Check for a no-op conversion before using wi::add
2412 rather than after. Use tree_to_uhwi when summing small shift
2413 counts into an unsigned int.
2415 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2416 Alan Hayward <alan.hayward@arm.com>
2417 David Sherwood <david.sherwood@arm.com>
2420 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
2421 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
2422 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
2423 POINTER_AND_FP_REGS.
2425 2017-10-02 Richard Biener <rguenther@suse.de>
2427 PR tree-optimization/82355
2428 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2429 a mapping for the enclosing loop but avoid generating one for
2431 (copy_bb_and_scalar_dependences): Remove premature codegen
2432 error on PHIs in blocks duplicated into multiple places.
2433 * graphite-scop-detection.c
2434 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2435 in the region use it as loop and nest to analyze the DR in.
2436 (try_generate_gimple_bb): Likewise.
2437 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2438 (add_loop_constraints): For blocks in a loop not in the region
2439 create a dimension with a single iteration.
2440 * sese.h (gbb_loop_at_index): Remove assert.
2442 2017-10-01 Kevin Buettner <kevinb@redhat.com>
2444 * omp-expand.c (adjust_context_scope): New function.
2445 (expand_parallel_call): Call adjust_context_scope.
2447 2017-10-01 Jeff Law <law@redhat.com>
2449 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
2450 dom_opt_dom_walker class with direct access to private members.
2451 Add comments. Call test_for_singularity.
2452 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
2453 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
2454 m_dummy_cond anymore.
2455 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
2457 (pass_dominator:execute): Build the dummy_cond here and pass it
2458 to the dom_opt_dom_walker ctor.
2459 (test_for_singularity): New function.
2461 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
2462 Maya Rashish <coypu@sdf.org>
2464 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
2465 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
2466 (alpha*-*-netbsd*) Use nbsd_tm_file.
2467 (arm*-*-netbsdelf*) Likewise.
2468 (i[34567]86-*-netbsdelf*) Likewise.
2469 (x86_64-*-netbsd*) Likewise.
2470 (mips*-*-netbsd*) Likewise.
2471 (powerpc-*-netbsd*) Likewise.
2472 (sh*-*-netbsd*) Likewise.
2473 (sparc-*-netbsdelf*) Likewise.
2474 (sparc64-*-netbsd*) Likewise.
2475 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
2477 (vax-*-netbsdelf*) Likewise.
2478 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
2479 (UINT_FAST8_TYPE) Likewise.
2480 (INT_FAST16_TYPE) Check CHAR_FAST16.
2481 (UINT_FAST16_TYPE) Likewise.
2483 2017-09-30 Jakub Jelinek <jakub@redhat.com>
2486 * config/i386/i386.md
2487 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
2488 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
2489 *divmodsi4_zext_2): New define_insn_and_split.
2490 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
2491 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
2492 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
2493 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
2494 New define_insn_and_split.
2495 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
2496 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
2497 operands[1] having DImode when mode is SImode.
2499 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
2500 always SImode for DIV and MOD in REG_EQUAL notes.
2502 2017-09-29 Yury Gribov <tetra2005@gmail.com>
2505 * match.pd: Fix handling of NaNs in pattern.
2507 2017-09-29 Jeff Law <law@redhat.com>
2509 * sbitmap.c (bitmap_bit_in_range_p): New function.
2510 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
2511 * tree-ssa-dse.c (live_bytes_read): New function.
2512 (dse_classify_store): Ignore reads of dead bytes.
2514 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
2515 typos and whitespace errors.
2516 * config/i386/predicates.md (address_no_seg_operand): Likewise.
2517 * config/s390/s390.c (s390_emit_prologue): Likewise.
2519 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2522 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
2523 with a symbol for LRA.
2525 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2527 PR rtl-optimization/82338
2528 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
2530 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
2532 * genmodes.c (calc_wider_mode): Suppress qsort macro.
2533 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
2534 (qsort_chk): Declare.
2535 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
2536 (qsort_chk): New function.
2538 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2540 PR tree-optimization/82337
2541 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
2542 phi definition if the PHI result appears in an abnormal PHI.
2543 (find_basis_for_base_expr): Don't record a basis if the LHS of the
2544 basis appears in an abnormal PHI.
2546 2017-09-29 Richard Biener <rguenther@suse.de>
2548 * graphite-isl-ast-to-gimple.c
2549 (translate_isl_ast_to_gimple::set_codegen_error): New function.
2550 (binary_op_to_tree): Use it.
2551 (get_rename_from_scev): Likewise.
2552 (copy_loop_phi_nodes): Likewise.
2553 (copy_bb_and_scalar_dependences): Likewise.
2554 (translate_pending_phi_nodes): Likewise.
2556 2017-09-29 Jakub Jelinek <jakub@redhat.com>
2559 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
2560 for movabsq $(i32 << shift), r64.
2562 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
2564 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
2565 index when encoding %esp as %rsp to avoid 0x67 prefix.
2567 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
2569 * config/i386/i386.md (*movsf_internal, *movdf_internal):
2570 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
2572 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
2574 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
2575 Extensions with more than 16 double VFP registers.
2576 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
2577 to_clear_mask and all code related to it. Replace the remaining
2578 entry by a sbitmap and adapt code accordingly.
2580 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
2582 * brig-builtins.def: Change pure attributes to const.
2584 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2586 * config.gcc (default_gnu_indirect_function): Default to yes for
2587 sparc*-*-linux* with glibc.
2589 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2591 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
2592 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
2593 when creating .init_array and .fini_array sections with priority
2596 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
2599 * config/aarch64/aarch64.c
2600 (aarch64_builtin_support_vector_misalignment): Always return false
2601 when misalignment is unknown.
2603 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
2605 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
2606 this function to return false if the definition used by the swap
2607 instruction is artificial, or if the memory address from which the
2608 constant value is loaded is not represented by a base address held
2609 in a register or if the base address register is a frame or stack
2610 pointer. Additionally, return false if the base address of the
2611 loaded constant is a SYMBOL_REF but is not considered to be a
2613 (replace_swapped_load_constant): New function.
2614 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
2615 loaded constant vector with a load of a swapped constant vector.
2617 2017-09-27 Carl Love <cel@us.ibm.com>
2619 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
2620 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
2621 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
2624 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
2626 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
2627 first, always call autopref_rank_data otherwise.
2629 2017-09-27 Richard Biener <rguenther@suse.de>
2631 * graphite-scop-detection.c (find_scop_parameters): Move
2632 loop bound handling ...
2633 (gather_bbs::before_dom_children): ... here, avoiding the need
2634 to build scop_info->loop_nest.
2635 (record_loop_in_sese): Remove.
2636 * sese.h (sese_info_t::loop_nest): Remove.
2637 * sese.c (new_sese_info): Do not allocate loop_nest.
2638 (free_sese_info): Do not free loop_nest.
2640 2017-09-27 Jakub Jelinek <jakub@redhat.com>
2643 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
2644 lhs from calls if the lhs has addressable type.
2646 2017-09-27 Richard Biener <rguenther@suse.de>
2648 * graphite.h (scop::max_alias_set): New member.
2649 * graphite-scop-detection.c: Remove references to non-existing
2650 --param in comments.
2651 (build_alias_sets): Record the maximum alias set used for drs.
2652 (build_scops): Support zero as unlimited for
2653 --param graphite-max-arrays-per-scop.
2654 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
2656 (build_poly_sr_1): ... here. Compute alias set based on the
2657 maximum alias set used for drs rather than
2658 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
2660 2017-09-27 Richard Biener <rguenther@suse.de>
2662 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
2663 --param loop-block-tile-size=0 to disable tiling.
2665 2017-09-27 Richard Biener <rguenther@suse.de>
2667 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
2668 (graphite-max-nb-scop-params): Document special value zero.
2669 * domwalk.h (dom_walker::STOP): New symbolical constant.
2670 (dom_walker::dom_walker): Add optional parameter for bb to
2672 (dom_walker::~dom_walker): Declare.
2673 (dom_walker::before_dom_children): Document STOP return value.
2674 (dom_walker::m_user_bb_to_rpo): New member.
2675 (dom_walker::m_bb_to_rpo): Likewise.
2676 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
2677 mapping here if not provided by the user.
2678 (dom_walker::~dom_walker): Free bb to RPO mapping if not
2679 provided by the user.
2680 (dom_walker::STOP): Define.
2681 (dom_walker::walk): Do not compute bb to RPO mapping here.
2682 Support STOP return value from before_dom_children to stop
2684 * graphite-optimize-isl.c (optimize_isl): If the schedule
2685 is the same still generate code if -fgraphite-identity
2686 or -floop-parallelize-all are given.
2687 * graphite-scop-detection.c: Include cfganal.h.
2688 (gather_bbs::gather_bbs): Get and pass through bb to RPO
2690 (gather_bbs::before_dom_children): Return STOP for BBs
2692 (build_scops): Compute bb to RPO mapping and pass it to
2693 the domwalk. Treat --param graphite-max-nb-scop-params=0
2694 as not limiting the number of params.
2695 * graphite.c (graphite_initialize): Remove limit on the
2696 number of basic-blocks in a function.
2697 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
2698 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
2699 default value of 10.
2701 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2703 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
2704 Adjust code to eliminate needing to do the shift right 32-bits
2705 operation after XSCVDPSPN.
2707 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2709 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
2710 ((X / Y) != 0 -> X >= Y): Likewise.
2712 2017-09-26 Carl Love <cel@us.ibm.com>
2714 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
2715 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
2716 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
2717 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
2718 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
2719 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
2720 definitions and overloading.
2721 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
2722 statement for P9V_BUILTIN_XST_LEN_R.
2723 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
2724 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
2725 define_expand and define_insn for the instructions and builtins.
2726 * doc/extend.texi: Update the built-in documentation file for the new
2728 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
2729 define_insn for the instructions
2731 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
2734 * gcc/config/netbsd-protos.h: New file.
2735 * gcc/config/netbsd.c: New file.
2736 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
2737 * gcc/config/t-netbsd: New file.
2738 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
2739 (tmake_file) Add t-netbsd.
2740 (extra_objs) Add netbsd.o.
2742 2017-09-26 Janus Weil <janus@gcc.gnu.org>
2746 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
2748 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2750 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
2751 sign extension from a vector register to a GPR by doing a 32-bit
2752 direct move and then an EXTSW.
2753 (extendsi<mode>2 splitter): Likewise.
2754 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
2755 right or vector extract after doing XSCVDPSPN. Use
2756 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
2758 (movdi_from_sf_zero_ext): Likewise.
2759 (reload_gpr_from_vsxsf): Likewise.
2760 (p8_mfvsrd_4_disf): Delete, no longer used.
2761 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
2762 then needing to move the SFmode to a GPR to use the XSCVDPSP
2763 instruction instead of FRSP and XSCVDPSPN.
2764 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
2765 it is adjacent to the other XSCVSPDP insns.
2766 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
2767 SFmode to be in traditional Altivec registers.
2768 (vsx_xscvdpspn): Eliminate useless alternative constraint.
2769 (vsx_xscvspdpn): Likewise.
2770 (vsx_xscvspdpn_scalar): Likewise.
2772 2017-09-26 Martin Jambor <mjambor@suse.cz>
2774 * tree-sra.c (compare_access_positions): Put integral types first,
2775 stabilize sorting of integral types, remove conditions putting
2776 non-full-precision integers last.
2777 (sort_and_splice_var_accesses): Disable scalarization if a
2778 non-integert would be represented by a non-full-precision integer.
2780 2017-09-26 Joseph Myers <joseph@codesourcery.com>
2782 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
2783 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
2784 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
2785 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
2786 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
2787 conditionals inside the function instead of around it. Call
2788 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
2789 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
2791 2017-09-26 Richard Biener <rguenther@suse.de>
2793 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
2795 (scop_detection::build_scop_breadth): ... this. Removed.
2796 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
2797 (scop_detection::harmful_stmt_in_bb): Likewise.
2798 (scop_detection::graphite_can_represent_stmt): Likewise.
2799 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
2800 (scop_detection::can_represent_loop): Remove recursion, fold in ...
2801 (scop_detection::can_represent_loop_1): ... this. Removed.
2802 (scop_detection::harmful_loop_in_region): Simplify after inlining
2803 the above and remove more quadraticness.
2804 (build_scops): Adjust.
2805 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
2808 2017-09-26 Jakub Jelinek <jakub@redhat.com>
2811 * config/i386/i386.c (ix86_print_operand_address_as): Only test
2812 REGNO (base) == SP_REG if base is a REG.
2815 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
2816 if it is different SSA_NAME.
2817 (optimize_range_tests_cmp_bitwise): New function.
2818 (optimize_range_tests): Call it.
2820 2017-09-26 Richard Biener <rguenther@suse.de>
2822 PR tree-optimization/82321
2823 * graphite.c (canonicalize_loop_closed_ssa): Properly check
2824 for the def being inside the loop.
2826 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2828 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
2830 * config/s390/s390-builtins.def: Fix constraint on op4.
2832 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2834 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
2835 independent expanders.
2836 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
2837 ("vec_ordered", "vec_unordered"): New expanders.
2839 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2841 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
2844 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2846 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
2847 vec_unpacks_lo_v16qi.
2848 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
2850 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2852 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
2853 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
2854 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
2856 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2858 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
2860 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
2862 ("*vec_slb<mode>"): New insn pattern.
2863 ("vec_shr_<mode>"): New expander.
2864 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
2865 and force the shift count operand to V16QImode.
2866 ("vec_srb<mode>"): Set shift count mode to V16QI.
2868 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2870 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
2871 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
2872 ("vec_widen_smult_hi_<mode>"): New expander definitions.
2874 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
2877 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
2879 2017-09-26 Richard Biener <rguenther@suse.de>
2881 PR tree-optimization/82320
2882 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
2885 2017-09-25 Jeff Law <law@redhat.com>
2887 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
2888 prototype for new argument.
2889 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
2890 mostly extracted from rs6000_emit_allocate_stack.
2891 (rs6000_emit_probe_stack_range_stack_clash): New function.
2892 (rs6000_emit_allocate_stack): Call
2893 rs6000_emit_probe_stack_range_stack_clash as needed.
2894 (rs6000_emit_probe_stack_range): Add additional argument
2895 to call to gen_probe_stack_range{si,di}.
2896 (output_probe_stack_range): New.
2897 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
2898 (output_probe_stack_range_stack_clash): New.
2899 (rs6000_emit_prologue): Emit notes into dump file as requested.
2900 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
2901 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
2902 Add additional operand and pass it to output_probe_stack_range.
2904 2017-09-25 Bin Cheng <bin.cheng@arm.com>
2906 PR tree-optimization/82163
2907 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
2908 (checking_verify_loop_closed_ssa): New parameter.
2909 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
2910 (check_loop_closed_ssa_stmt): Delete.
2911 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
2912 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
2913 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
2916 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
2918 * brig-builtins.def: Treat HSAIL barrier builtins as
2919 setjmp/longjump style functions.
2921 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2923 * target.def (constant_alignment): New hook.
2924 * defaults.h (CONSTANT_ALIGNMENT): Delete.
2925 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
2926 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
2927 * doc/tm.texi: Regenerate.
2928 * targhooks.h (default_constant_alignment): Declare.
2929 (constant_alignment_word_strings): Likewise.
2930 * targhooks.c (default_constant_alignment): New function.
2931 (constant_alignment_word_strings): Likewise.
2932 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
2933 instead of CONSTANT_ALIGNMENT.
2934 * varasm.c (align_variable, get_variable_align, build_constant_desc)
2935 (force_const_mem): Likewise.
2936 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
2937 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
2938 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
2939 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2940 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
2942 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
2943 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2944 constant_alignment_word_strings.
2945 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
2946 (CONSTANT_ALIGNMENT): Likewise.
2947 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2948 (arm_constant_alignment): New function.
2949 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
2950 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2951 constant_alignment_word_strings.
2952 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
2953 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2954 constant_alignment_word_strings.
2955 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
2956 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2957 (cris_constant_alignment): New function.
2958 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
2959 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2960 (epiphany_constant_alignment): New function.
2961 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
2962 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2963 constant_alignment_word_strings.
2964 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
2965 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2966 constant_alignment_word_strings.
2967 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
2968 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2969 constant_alignment_word_strings.
2970 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
2971 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
2972 * config/i386/i386.c (ix86_constant_alignment): Make static.
2973 Use the same interface as the target hook.
2974 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2975 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
2976 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2977 constant_alignment_word_strings.
2978 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
2979 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
2980 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2981 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
2982 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2983 constant_alignment_word_strings.
2984 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
2985 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2986 constant_alignment_word_strings.
2987 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
2988 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2989 constant_alignment_word_strings.
2990 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
2991 * config/microblaze/microblaze.c (microblaze_constant_alignment):
2993 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2994 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
2995 * config/mips/mips.c (mips_constant_alignment): New function.
2996 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2997 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
2998 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
2999 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3000 (mmix_constant_alignment): Make static. Use the same interface
3002 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
3003 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3004 constant_alignment_word_strings.
3005 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
3006 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3007 constant_alignment_word_strings.
3008 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
3009 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3010 constant_alignment_word_strings.
3011 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
3012 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3013 (rs6000_constant_alignment): New function.
3014 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
3015 * config/riscv/riscv.c (riscv_constant_alignment): New function.
3016 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3017 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
3018 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3019 (rs6000_constant_alignment): New function.
3020 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
3021 * config/s390/s390.c (s390_constant_alignment): New function.
3022 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3023 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
3024 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3025 constant_alignment_word_strings.
3026 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
3027 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3028 (sparc_constant_alignment): New function.
3029 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
3030 * config/spu/spu.c (spu_constant_alignment): New function.
3031 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3032 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
3033 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3034 constant_alignment_word_strings.
3035 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
3036 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3037 constant_alignment_word_strings.
3038 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
3039 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3040 constant_alignment_word_strings.
3041 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
3042 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3043 (visium_constant_alignment): New function.
3044 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
3045 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3046 (xtensa_constant_alignment): New function.
3047 * system.h (CONSTANT_ALIGNMENT): Poison.
3049 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
3051 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3052 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
3053 (rs6000_builtin_valid_without_lhs): New helper function.
3054 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3055 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
3057 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3059 * target.h (vec_perm_indices): Use unsigned short rather than
3061 (auto_vec_perm_indices): Likewise.
3062 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
3063 Use unsigned int rather than unsigned char.
3064 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3066 2017-09-25 Richard Biener <rguenther@suse.de>
3068 * cfgloop.h (sort_sibling_loops): Declare.
3069 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3070 (sort_sibling_loops): New function sorting the sibling loop list
3072 * graphite.c (graphite_transform_loops): Sort sibling loops.
3074 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3076 * target.def (vec_perm_const_ok): Change sel parameter to
3078 * optabs-query.c (can_vec_perm_p): Update accordingly.
3079 * doc/tm.texi: Regenerate.
3080 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3081 auto_vec_perm_indices and remove separate nelt field.
3082 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3083 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3084 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3085 (aarch64_expand_vec_perm_const): Update accordingly.
3086 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3087 to vec_perm_indices.
3088 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3089 auto_vec_perm_indices and remove separate nelt field.
3090 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3091 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3092 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3094 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3095 to vec_perm_indices.
3096 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3097 sel to vec_perm_indices.
3098 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3099 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3100 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3102 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3105 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3108 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3109 on the FUNCTION_DECL function context if it has a DIE that is a
3112 2017-09-25 Richard Biener <rguenther@suse.de>
3114 PR tree-optimization/82285
3115 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3118 2017-09-25 Tom de Vries <tom@codesourcery.com>
3122 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3125 2017-09-25 Richard Biener <rguenther@suse.de>
3127 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3128 ISL errors other than isl_error_quota happen. Dump if the
3129 schedule is the same.
3130 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3131 errors instead of aborting inside ISL.
3133 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3136 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3137 of libgcc_eh for m64.
3138 * config/i386/darwin64.h: Likewise.
3140 2017-09-25 Richard Biener <rguenther@suse.de>
3143 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
3144 attribute for incomplete types nor twice for complete ones.
3146 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
3149 * config/i386/i386.c (ix86_print_operand_address_as): Encode
3150 %esp as %rsp to avoid 0x67 prefix if there is no index or base
3153 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
3156 * config/i386/i386.opt (mprefer-avx256): Use
3157 ix86_target_flags variable.
3158 * config/i386/i386.c (ix86_target_string): Move
3159 -mprefer-avx256 to flag2_opts.
3161 2017-09-22 Jakub Jelinek <jakub@redhat.com>
3164 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
3165 and x != -1 | y != -1 into (x & y) != -1.
3167 2017-09-22 Steve Ellcey <sellcey@cavium.com>
3169 * config.gcc: Add new case statement to set
3170 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
3171 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
3172 s390x-*-linux* case statements. Added aarch64 to the list of
3173 supported architectures.
3175 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3177 PR tree-optimization/82289
3178 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
3179 STMT_VINFO_RELEVANT_P.
3181 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3182 Alan Hayward <alan.hayward@arm.com>
3183 David Sherwood <david.sherwood@arm.com>
3185 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
3186 for VR_RANGE only; don't allow VR_ANTI_RANGE.
3187 (extract_range_from_binary_expr_1): Don't call
3188 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
3190 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3191 Alan Hayward <alan.hayward@arm.com>
3192 David Sherwood <david.sherwood@arm.com>
3194 * target.def (preferred_vector_alignment): New hook.
3195 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
3197 * doc/tm.texi: Regenerate.
3198 * targhooks.h (default_preferred_vector_alignment): Declare.
3199 * targhooks.c (default_preferred_vector_alignment): New function.
3200 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
3202 (DR_TARGET_ALIGNMENT): New macro.
3203 (aligned_access_p): Update commentary.
3204 (vect_known_alignment_in_bytes): New function.
3205 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
3207 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
3208 Calculate the misalignment based on the target alignment rather than
3210 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
3211 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
3212 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
3213 the target alignment, rather than masking the element misalignment
3214 with the number of elements in a vector. Also use the target
3215 alignment when calculating the maximum number of peels.
3216 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
3217 instead of TYPE_ALIGN_UNIT.
3218 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
3219 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
3220 (vect_create_addr_base_for_vector_ref): Update call accordingly.
3221 (vect_create_data_ref_ptr): Likewise.
3222 (vect_setup_realignment): Realign by ANDing with
3223 -DR_TARGET_MISALIGNMENT.
3224 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
3225 the number of peels based on DR_TARGET_ALIGNMENT.
3226 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
3227 with the guaranteed alignment boundary when deciding whether
3229 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
3230 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
3231 (ensure_base_align): Remove stmt_info parameter. Get the
3232 target base alignment from DR_TARGET_ALIGNMENT.
3233 (vectorizable_store): Update call accordingly. Interpret
3234 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
3236 (vectorizable_load): Likewise.
3238 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3239 Alan Hayward <alan.hayward@arm.com>
3240 David Sherwood <david.sherwood@arm.com>
3242 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
3243 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
3244 (vect_enhance_data_refs_alignment): Likewise.
3246 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3248 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3249 error. Only quit immediately if parsing is complete.
3250 (BEGIN): Initialize fatal_err and parse_done.
3251 (begin fpu, end fpu): Check number of arguments.
3252 (begin arch, end arch): Likewise.
3253 (begin cpu, end cpu): Likewise.
3254 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3255 (optalias): Likewise.
3257 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3259 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3260 * config/arm/arm-isa.h: Delete. Move definitions to ...
3261 * arm-cpus.in: ... here. Use new feature and fgroup values.
3262 * config/arm/arm.c (arm_option_override): Use lower case for feature
3264 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3265 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3266 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3268 (print_isa_bits_for): New function.
3269 (gen_isa): New function.
3270 (gen_comm_data): Use print_isa_bits_for.
3271 (define feature): New keyword.
3272 (define fgroup): New keyword.
3273 * config/arm/t-arm (TM_H): Remove.
3274 (GTM_H): Add arm-isa.h.
3275 (arm-isa.h): Add rule to generate file.
3276 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3277 case for feature bit names.
3279 2017-09-22 Richard Biener <rguenther@suse.de>
3281 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
3283 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
3284 print of no dependency loops ...
3285 * graphite.c (graphite_transform_loops): ... here.
3286 (canonicalize_loop_closed_ssa_form): Work from inner to outer
3288 (same_close_phi_node, remove_duplicate_close_phi,
3289 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
3290 (canonicalize_loop_closed_ssa): ... here and simplify.
3291 * graphite-optimize-isl.c: Include tree-vectorizer.h.
3292 (optimize_isl): Use dump_printf_loc to tell when we stopped
3293 optimizing because of an ISL timeout.
3295 2017-09-22 Richard Biener <rguenther@suse.de>
3297 PR tree-optimization/82291
3298 * tree-if-conv.c (predicate_mem_writes): Make sure to
3299 remove writes in blocks predicated with false.
3301 2017-09-22 Richard Biener <rguenther@suse.de>
3303 * sese.c: Include cfganal.h.
3304 (if_region_set_false_region): Remove.
3305 (create_if_region_on_edge): Likewise.
3306 (move_sese_in_condition): Re-implement without destroying
3309 2017-09-22 Richard Biener <rguenther@suse.de>
3311 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
3312 Verify both BBs contain loop PHI nodes before dispatching to
3314 (graphite_regenerate_ast_isl): Do not recompute dominators,
3315 do not verify three times. Restructure for clarity.
3316 * graphite-scop-detection.c (same_close_phi_node,
3317 remove_duplicate_close_phi, make_close_phi_nodes_unique,
3318 defined_in_loop_p, canonicalize_loop_closed_ssa,
3319 canonicalize_loop_closed_ssa_form): Simplify, remove excess
3320 checking and SSA rewrite, move to ...
3321 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
3322 (graphite_initialize): Do not pass in ctx, do not reset the
3323 SCEV cache, compute only dominators.
3324 (graphite_transform_loops): Allocate ISL ctx after
3325 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
3326 Maintain post-dominators only around build_scops.
3327 * sese.c (if_region_set_false_region): Make static. Free
3328 and recompute dominators.
3329 (move_sese_in_condition): Assert we don't get called with
3330 post-dominators computed.
3331 * sese.h (if_region_set_false_region): Remove.
3333 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
3335 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
3336 mode attribute for TARGET_AVX512VL.
3338 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
3340 * config/i386/i386.opt (mprefer-avx256): New option.
3341 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
3343 (ix86_preferred_simd_mode): Return 256-bit AVX modes
3344 for TARGET_PREFER_AVX256.
3345 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
3347 2017-09-21 Jeff Law <law@redhat.com>
3349 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
3350 Fix dump output if the only stack space is for pushed registers.
3352 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3354 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
3357 2017-09-21 Martin Sebor <msebor@redhat.com>
3360 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
3361 code (in C++) or code that triggers warnings.
3363 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
3365 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
3367 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
3369 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
3370 * sched-rgn.c: Ditto.
3371 * sel-sched-ir.c: Ditto.
3373 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
3375 * toplev.h (set_random_seed): Adjust return type.
3376 * toplev.c (init_local_tick): Move eager initialization of random_seed
3377 to get_random_seed. Adjust comment.
3378 (init_random_seed): Inline to get_random_seed, delete.
3379 (get_random_seed): Initialize random_seed lazily.
3380 (set_random_seed): Do not return previous value.
3381 (print_switch_value): Do not call get_random_seed.
3383 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
3385 * cgraph.c (delete_function_version): New, broken out from...
3386 (cgraph_node::delete_function_version): ...here. Rename to
3387 cgraph_node::delete_function_version_by_decl. Update all uses.
3388 (cgraph_node::remove): Call delete_function_version.
3390 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3393 * tree-inline.c (expand_call_inline): Emit clobber stmts for
3394 VAR_DECLs to which addressable non-volatile parameters are mapped
3395 and for id->retvar after the return value assignment. Clear
3396 id->retval and id->retbnd after inlining.
3398 2017-09-21 Richard Biener <rguenther@suse.de>
3400 PR tree-optimization/82276
3401 PR tree-optimization/82244
3402 * tree-vrp.c (build_assert_expr_for): Set
3403 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
3405 (remove_range_assertions): Revert earlier change.
3407 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
3410 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
3412 2017-09-21 Richard Biener <rguenther@suse.de>
3414 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3415 Restore valid IL after code generation errors.
3416 * graphite.c (graphite_transform_loops): Diagnose code
3417 generation issues as MSG_MISSED_OPTIMIZATION and continue
3418 with processing SCOPs.
3420 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3421 Alan Hayward <alan.hayward@arm.com>
3422 David Sherwood <david.sherwood@arm.com>
3424 * calls.c (compute_argument_addresses): Use simplify_gen_binary
3425 rather than choosing between plus_constant and gen_rtx_<CODE>.
3426 * expr.c (emit_push_insn): Likewise.
3427 (expand_expr_real_2): Likewise.
3429 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3430 Alan Hayward <alan.hayward@arm.com>
3431 David Sherwood <david.sherwood@arm.com>
3433 * loop-unroll.c (split_iv): Call copy_rtx on the step.
3435 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3436 Alan Hayward <alan.hayward@arm.com>
3437 David Sherwood <david.sherwood@arm.com>
3439 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
3440 calling tree_to_uhwi.
3442 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3443 Alan Hayward <alan.hayward@arm.com>
3444 David Sherwood <david.sherwood@arm.com>
3446 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
3447 INTEGER_CST rather than a negative test for ADDR_EXPR.
3449 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3450 Alan Hayward <alan.hayward@arm.com>
3451 David Sherwood <david.sherwood@arm.com>
3453 * tree-vrp.c (extract_range_from_binary_expr_1): Check
3454 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
3456 2017-09-21 Richard Biener <rguenther@suse.de>
3458 PR tree-optimization/71351
3459 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
3460 graphite_create_new_loop_guard): Remove, fold remaining parts
3462 (translate_isl_ast_node_for): ... here and simplify.
3464 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3467 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
3468 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
3469 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
3470 alternative always use QI mode, for -Os imov (=R,R) alternative
3471 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
3474 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3475 Jeff Law <law@redhat.com>
3477 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
3478 (allocate_stack_space): New function, partially extracted from
3480 (s390_emit_prologue): Track offset to most recent stack probe.
3481 Code to allocate space moved into allocate_stack_space.
3482 Dump actions when no stack is allocated.
3483 (s390_prologue_plus_offset): New function.
3484 (s390_emit_stack_probe): Likewise.
3486 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
3488 * common.opt (Wa, Wl, Wp, g, gz=): Add
3490 (gno-column-info): Remove.
3491 (gcolumn-info): Drop RejectNegative.
3493 (gno-record-gcc-switches): Remove.
3494 (grecord-gcc-switches): Drop RejectNegative.
3495 (gno-split-dwarf): Remove.
3496 (gsplit-dwarf): Drop RejectNegative.
3497 (gno-strict-dwarf): Remove.
3498 (gstrict-dwarf): Drop RejectNegative.
3499 * config/darwin.opt (gfull, gused): Add RejectNegative.
3500 * dwarf2out.c (gen_producer_string): Drop
3501 gno-record-gcc-switches handler.
3502 * optc-gen.awk: Add g to prefixes with negative forms.
3503 * opts-common.c (remapping_prefix_p): New.
3504 (find_opt): Check it.
3505 (generate_canonical_option): Test g prefix.
3506 (option_map): Add -gno- mapping.
3507 (add_misspelling_candidates): Check remapping_prefix_p.
3509 2017-09-20 Jeff Law <law@redhat.com>
3511 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
3512 thinko in stack clash protection support.
3514 * explow.c (compute_stack_clash_protection_loop_data): Use
3515 CONST_INT_P instead of explicit test. Verify object is a
3516 CONST_INT_P before looking at INTVAL.
3517 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
3518 instead of explicit test.
3520 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3523 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
3524 address instead of to r1 and r11.
3526 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
3528 * config.gcc: Support "knm".
3529 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
3530 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3532 * config/i386/i386.c (m_KNM): Define.
3533 (processor_target_table): Add "knm".
3535 (ix86_option_override_internal): Add "knm".
3536 (ix86_issue_rate): Add PROCESSOR_KNM.
3537 (ix86_adjust_cost): Ditto.
3538 (ia32_multipass_dfa_lookahead): Ditto.
3539 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
3540 (fold_builtin_cpu): Add M_INTEL_KNM.
3541 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
3542 (processor_type): Add PROCESSOR_KNM.
3543 * config/i386/x86-tune.def: Add m_KNM.
3544 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
3546 2017-09-20 Richard Biener <rguenther@suse.de>
3548 PR tree-optimization/80213
3549 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
3550 are allowed in empty BBs as well.
3551 (canonicalize_loop_closed_ssa): Also look for other complex
3553 (scop_detection::get_sese): Include the loop-closed PHI block
3555 (scop_detection::merge_sese): Remove code adding extra blocks.
3556 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
3557 (build_scops): Assert the final returned scop is invalid.
3559 2017-09-20 Richard Biener <rguenther@suse.de>
3561 PR tree-optimization/82264
3562 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
3564 (vn_phi_lookup): Likewise.
3565 (vn_phi_insert): Likewise.
3567 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3569 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
3570 that fits into uhwi or shwi, add DW_AT_const_value regardless
3571 of early_dwarf without going through RTL, using add_AT_unsigned
3574 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
3575 (DEBUG_LTO_ABBREV_SECTION): Likewise.
3576 (DEBUG_LTO_MACINFO_SECTION): Likewise.
3577 (DEBUG_MACRO_SECTION): Likewise.
3578 (DEBUG_LTO_MACRO_SECTION): Likewise.
3579 (DEBUG_STR_DWO_SECTION): Likewise.
3580 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
3581 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
3582 (DEBUG_LTO_DWO_LINE_SECTION): Define.
3583 (DEBUG_LTO_LINE_STR_SECTION): Define.
3584 (init_sections_and_labels): Initialize debug_line_str_section
3585 variable. Initialize debug_loc_section for -gdwarf-5 to
3586 DEBUG_LOCLISTS_SECTION. Formatting fixes.
3588 2017-09-20 Richard Biener <rguenther@suse.de>
3590 * graphite-sese-to-poly.c (extract_affine): Properly handle
3591 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
3593 2017-09-20 Richard Biener <rguenther@suse.de>
3595 PR tree-optimization/81373
3596 * graphite-scop-detection.c (build_cross_bb_scalars_def):
3597 Force SESE live-out defs to be handled even if they are
3600 2017-09-19 Jeff Law <law@redhat.com>
3602 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
3603 nothing for stack adjustments with REG_STACK_CHECK.
3604 * sched-deps.c (parse_add_or_inc): Reject insns with
3605 REG_STACK_CHECK from dependency breaking.
3606 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
3607 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
3608 * reg-notes.def (STACK_CHECK): New note.
3610 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
3611 (ix86_expand_prologue): Dump stack clash info as needed.
3612 Call ix86_adjust_stack_and_probe_stack_clash as needed.
3614 * function.c (dump_stack_clash_frame_info): New function.
3615 * function.h (dump_stack_clash_frame_info): Prototype.
3616 (enum stack_clash_probes): New enum.
3618 * config/alpha/alpha.c (alpha_expand_prologue): Also check
3619 flag_stack_clash_protection.
3620 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
3621 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
3622 (arm_frame_pointer_required): Likewise.
3623 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
3624 (ia64_expand_prologue): Likewise.
3625 * config/mips/mips.c (mips_expand_prologue): Likewise.
3626 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
3627 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3628 (sparc_flat_expand_prologue): Likewise.
3629 * config/spu/spu.c (spu_expand_prologue): Likewise.
3631 * explow.c: Include "params.h".
3632 (anti_adjust_stack_and_probe_stack_clash): New function.
3633 (get_stack_check_protect): Likewise.
3634 (compute_stack_clash_protection_loop_data): Likewise.
3635 (emit_stack_clash_protection_loop_start): Likewise.
3636 (emit_stack_clash_protection_loop_end): Likewise.
3637 (allocate_dynamic_stack_space): Use get_stack_check_protect.
3638 Use anti_adjust_stack_and_probe_stack_clash.
3639 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
3640 (emit_stack_clash_protection_loop_start): Likewise.
3641 (emit_stack_clash_protection_loop_end): Likewise.
3642 * rtl.h (get_stack_check_protect): Prototype.
3643 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
3644 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
3645 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
3647 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
3649 * doc/tm.texi: Rebuilt.
3650 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3651 get_stack_check_protect.
3652 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
3653 * config/arm/arm.c (arm_expand_prologue): Likewise.
3654 (arm_frame_pointer_required): Likewise.
3655 * config/i386/i386.c (ix86_expand_prologue): Likewise.
3656 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
3657 * config/mips/mips.c (mips_expand_prologue): Likewise.
3658 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
3659 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
3660 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3661 (sparc_flat_expand_prologue): Likewise.
3663 * common.opt (-fstack-clash-protection): New option.
3664 * flag-types.h (enum stack_check_type): Note difference between
3665 -fstack-check= and -fstack-clash-protection.
3666 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
3667 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
3668 * toplev.c (process_options): Issue warnings/errors for cases
3669 not handled with -fstack-clash-protection.
3670 * doc/invoke.texi (-fstack-clash-protection): Document new option.
3671 (-fstack-check): Note additional problem with -fstack-check=generic.
3672 Note that -fstack-check is primarily for Ada and refer users
3673 to -fstack-clash-protection for stack-clash-protection.
3674 Document new params for stack clash protection.
3676 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
3678 * config/i386/i386.c (ix86_split_long_move): Do not handle
3679 address used for LEA in a special way.
3681 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3683 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
3685 2017-09-19 Martin Sebor <msebor@redhat.com>
3688 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
3689 of incompatible types.
3691 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
3693 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3694 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
3695 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3696 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
3698 2017-09-19 Richard Biener <rguenther@suse.de>
3700 PR tree-optimization/82244
3701 * tree-vrp.c (remove_range_assertions): Do not propagate
3702 a constant to abnormals but replace the assert with a copy.
3704 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
3706 PR rtl-optimization/57878
3707 PR rtl-optimization/68988
3708 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
3709 avoidance test involving non_reload_pseudos. Move frequency test
3710 below the general fragmentation avoidance test.
3712 2017-09-19 Richard Biener <rguenther@suse.de>
3714 PR tree-optimization/69728
3715 * graphite-sese-to-poly.c (schedule_error): New global.
3716 (add_loop_schedule): Handle empty domain by failing the
3718 (build_original_schedule): Handle schedule_error.
3720 2017-09-19 Richard Biener <rguenther@suse.de>
3722 * graphite-scop-detection.c (scop_detection::can_represent_loop):
3723 Do not iterate to sibling loops but only to siblings of inner
3726 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
3729 * config/m68k/m68k.md (moveq feeding equality comparison): Check
3730 that the registers are different.
3732 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
3734 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
3735 to processor_model and "amdfam17h" to arch_names_table.
3736 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
3738 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3741 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
3743 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3744 Alan Hayward <alan.hayward@arm.com>
3745 David Sherwood <david.sherwood@arm.com>
3747 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
3749 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
3751 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
3752 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
3753 vect_schedule_slp_instance.
3754 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
3755 Update call to vect_slp_analyze_node_operations. Simplify return
3757 (vect_slp_analyze_bb_1): Update call accordingly.
3758 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
3759 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
3760 (vect_schedule_slp): Update call accordingly.
3762 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3763 Alan Hayward <alan.hayward@arm.com>
3764 David Sherwood <david.sherwood@arm.com>
3766 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
3767 with types that aren't in fact scalar.
3769 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3771 * tree-vect-slp.c (vect_record_max_nunits): New function,
3773 (vect_build_slp_tree_1): ...here.
3774 (vect_build_slp_tree_2): Call it for phis too.
3776 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3778 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
3779 to vect_get_vec_def_for_operand when getting the mask operand.
3781 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3782 Alan Hayward <alan.hayward@arm.com>
3783 David Sherwood <david.sherwood@arm.com>
3785 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
3788 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3789 Alan Hayward <alan.hayward@arm.com>
3790 David Sherwood <david.sherwood@arm.com>
3792 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
3793 calculation for vector booleans.
3795 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3796 Alan Hayward <alan.hayward@arm.com>
3797 David Sherwood <david.sherwood@arm.com>
3799 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
3801 (vect_transform_stmt): ...here.
3802 (vect_analyze_stmt): Use it instead of calling
3803 vectorizable_live_operation directly.
3805 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
3807 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
3808 non-SIMT targets in acc vector loops.
3810 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
3812 * configure.ac: Add arc and check if assembler supports gdwarf2.
3813 * configure: Regenerate.
3815 2017-09-18 Richard Biener <rguenther@suse.de>
3817 PR tree-optimization/82220
3818 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
3819 epilogue niters from the min_profitable_iters compute.
3821 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3824 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
3825 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
3826 (ix86_init_pic_reg): Revert 2017-09-01 changes.
3828 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
3831 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
3832 switching to a new text section.
3834 2017-09-18 Richard Biener <rguenther@suse.de>
3836 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
3838 (build_alias_set): Reject aliases with no access function.
3840 2017-09-18 Richard Biener <rguenther@suse.de>
3842 PR tree-optimization/79622
3843 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
3845 (build_cross_bb_scalars_use): Likewise.
3847 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
3849 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
3851 2017-09-18 Alan Modra <amodra@gmail.com>
3854 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
3855 stack_pointer_rtx for count 0. Update comments. Break up
3856 large rtl expression.
3858 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
3860 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
3861 Increase to 20 bytes.
3862 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
3863 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
3864 or avx version of the stub.
3866 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
3869 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
3870 compute the minimum stack alignment. Also update preferred stack
3871 boundary for leaf functions.
3873 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
3875 PR tree-optimization/82228
3876 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
3879 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3881 * common/config/nds32/nds32-common.c
3882 (nds32_option_optimization_table): Refine formatting.
3883 (nds32_option_optimization_table): Use -fsched-pressure and
3884 -fomit-frame-pointer for specific optimization level.
3886 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3888 * config/nds32/nds32.c: Refine formatting and comments.
3889 * config/nds32/nds32.h: Likewise.
3890 * config/nds32/nds32.md: Likewise.
3891 * config/nds32/nds32-cost.c: Likewise.
3892 * config/nds32/nds32-isr.c: Likewise.
3893 * config/nds32/nds32-md-auxiliary.c: Likewise.
3894 * config/nds32/nds32-multiple.md: Likewise.
3895 * config/nds32/nds32-predicates.c: Likewise.
3897 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
3898 Jakub Jelinek <jakub@redhat.com>
3900 Add support for -std=c++2a.
3901 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
3903 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
3905 2017-09-15 Steve Ellcey <sellcey@cavium.com>
3908 * doc/extend.texi (Common Function Attributes): Add
3909 references to ARM, AArch64, and S/390 specific attributes.
3910 (Function Specific Option Pragmas): Add AArch64 and S/390
3911 to list of back ends that support the target pragma.
3913 2017-09-15 Nathan Sidwell <nathan@acm.org>
3915 * doc/standards.texi: Fix C++17 description. Update URLs for
3918 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3920 * common.opt (Wcast-align=strict): New warning option.
3921 * doc/invoke.texi: Document -Wcast-align=strict.
3923 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
3925 * cgraph.h (cgraph_thunk_info): Add comments.
3926 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
3927 assert for VIRTUAL_* arguments stricter.
3929 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
3931 PR tree-optimization/71026
3932 * match.pd: Move RDIV patterns from fold-const.c
3933 * fold-const.c (distribute_real_division): Removed.
3934 (fold_binary_loc): Remove calls to distribute_real_divison.
3936 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3938 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
3939 c++1z and gnu++1z as deprecated. Change other references to
3940 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
3941 Change -Wc++1z-compat to -Wc++17-compat.
3942 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
3943 * dwarf2out.c (highest_c_language): Handle C++17.
3944 (gen_compile_unit_die): Likewise.
3946 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3948 PR rtl-optimization/82192
3949 * combine.c (make_extraction): Don't look through non-paradoxical
3950 SUBREGs or TRUNCATE if pos + len is or might be bigger than
3953 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3954 Alan Hayward <alan.hayward@arm.com>
3955 David Sherwood <david.sherwood@arm.com>
3957 * target.def (function_arg_offset): New hook.
3958 * targhooks.h (default_function_arg_offset): Declare.
3959 * targhooks.c (default_function_arg_offset): New function.
3960 * function.c (locate_and_pad_parm): Use
3961 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
3962 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
3963 (TARGET_FUNCTION_ARG_OFFSET): ...this.
3964 * doc/tm.texi: Regenerate.
3965 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
3966 * config/spu/spu.c (spu_function_arg_offset): New function.
3967 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
3968 * system.h (FUNCTION_ARG_OFFSET): Poison.
3970 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3971 Alan Hayard <alan.hayward@arm.com>
3972 David Sherwood <david.sherwood@arm.com>
3974 * target.def (truly_noop_truncation): New hook.
3975 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
3976 than TRULY_NOOP_TRUNCATION.
3977 * hooks.h (hook_bool_uint_uint_true): Declare.
3978 * hooks.c (hook_bool_uint_uint_true): New function.
3979 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
3980 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
3981 * doc/tm.texi: Regenerate.
3982 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
3983 rather than TRULY_NOOP_TRUNCATION in comments.
3984 (simplify_comparison): Likewise.
3985 (record_truncated_value): Likewise.
3986 * expmed.c (extract_bit_field_1): Likewise.
3987 (extract_split_bit_field): Likewise.
3988 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
3989 instead of TRULY_NOOP_TRUNCATION.
3990 * function.c (assign_parm_setup_block): Likewise.
3991 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
3992 * rtlhooks.c: Include target.h.
3993 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
3994 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
3995 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
3996 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
3997 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
3998 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
3999 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
4000 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
4001 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
4002 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
4003 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
4004 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
4005 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
4006 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
4007 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
4008 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
4009 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
4010 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
4011 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
4012 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
4013 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
4014 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
4015 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
4016 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
4017 * config/mips/mips.c (mips_truly_noop_truncation): New function.
4018 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4019 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4020 rather than TRULY_NOOP_TRUNCATION in comments.
4021 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
4022 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
4023 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
4024 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
4025 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
4026 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
4027 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
4028 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
4029 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
4030 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
4031 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
4032 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4033 rather than TRULY_NOOP_TRUNCATION in comments.
4034 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
4035 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
4036 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
4037 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
4038 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
4039 TRULY_NOOP_TRUNCATION condition.
4040 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
4041 (TRULY_NOOP_TRUNCATION): Delete.
4042 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
4043 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
4044 * config/spu/spu.c (spu_truly_noop_truncation): New function.
4045 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4046 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
4047 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
4048 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
4049 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4050 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4051 rather than TRULY_NOOP_TRUNCATION in comments.
4052 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
4053 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
4054 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
4055 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
4056 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
4057 * system.h (TRULY_NOOP_TRUNCATION): Poison.
4059 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
4062 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
4063 (*cmp_ior): Likewise.
4064 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
4065 (*ior_scc_scc_cmp): Likewise.
4066 (*and_scc_scc): Likewise.
4067 (*and_scc_scc_cmp): Likewise.
4069 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4070 Alan Hayard <alan.hayward@arm.com>
4071 David Sherwood <david.sherwood@arm.com>
4073 * target.def (can_change_mode_class): New hook.
4074 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4075 (hard_regno_nregs): Likewise.
4076 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4077 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4078 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4079 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4080 (LOAD_EXTEND_OP): Update accordingly.
4081 * doc/tm.texi: Regenerate.
4082 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4083 CANNOT_CHANGE_MODE_CLASS.
4084 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4085 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4086 * combine.c (simplify_set): Update accordingly.
4087 * emit-rtl.c (validate_subreg): Likewise.
4088 * recog.c (general_operand): Likewise.
4089 * regcprop.c (mode_change_ok): Likewise.
4090 * reload1.c (choose_reload_regs): Likewise.
4091 (inherit_piecemeal_p): Likewise.
4092 * rtlanal.c (simplify_subreg_regno): Likewise.
4093 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4094 instead of CANNOT_CHANGE_MODE_CLASS.
4095 (reload_cse_simplify_operands): Likewise.
4096 * reload.c (push_reload): Use targetm.can_change_mode_class
4097 instead of CANNOT_CHANGE_MODE_CLASS.
4098 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4099 REG_CANNOT_CHANGE_MODE_P.
4100 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4101 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4102 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4103 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4104 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4105 (arm_can_change_mode_class): New function.
4106 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4107 than CANNOT_CHANGE_MODE_CLASS in comments.
4108 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4109 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4110 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4111 (ix86_can_change_mode_class): ...this new function, inverting the
4112 sense of the return value.
4113 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4114 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4115 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4116 (ia64_can_change_mode_class): New function.
4117 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4118 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4119 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4120 (m32c_can_change_mode_class): ...this new function, inverting the
4121 sense of the return value.
4122 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4123 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4124 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4125 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4126 (mips_can_change_mode_class): ...this new function, inverting the
4127 sense of the return value.
4128 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4129 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4130 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4131 (msp430_can_change_mode_class): New function.
4132 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4133 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4134 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4135 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4136 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4137 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4138 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4139 (pa_cannot_change_mode_class): Replace with...
4140 (pa_can_change_mode_class): ...this new function, inverting the
4141 sense of the return value.
4142 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4143 than CANNOT_CHANGE_MODE_CLASS in comments.
4144 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4145 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
4146 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4147 (pdp11_cannot_change_mode_class): Replace with...
4148 (pdp11_can_change_mode_class): ...this new function, inverting the
4149 sense of the return value.
4150 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4151 * config/powerpcspe/powerpcspe-protos.h
4152 (rs6000_cannot_change_mode_class_ptr): Delete.
4153 * config/powerpcspe/powerpcspe.c
4154 (rs6000_cannot_change_mode_class_ptr): Delete.
4155 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4156 (rs6000_option_override_internal): Assign to
4157 targetm.can_change_mode_class instead of
4158 rs6000_cannot_change_mode_class_ptr.
4159 (rs6000_cannot_change_mode_class): Replace with...
4160 (rs6000_can_change_mode_class): ...this new function, inverting the
4161 sense of the return value.
4162 (rs6000_debug_cannot_change_mode_class): Replace with...
4163 (rs6000_debug_can_change_mode_class): ...this new function.
4164 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4165 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
4166 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4167 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4168 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
4170 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
4171 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4172 (rs6000_option_override_internal): Assign to
4173 targetm.can_change_mode_class instead of
4174 rs6000_cannot_change_mode_class_ptr.
4175 (rs6000_cannot_change_mode_class): Replace with...
4176 (rs6000_can_change_mode_class): ...this new function, inverting the
4177 sense of the return value.
4178 (rs6000_debug_cannot_change_mode_class): Replace with...
4179 (rs6000_debug_can_change_mode_class): ...this new function.
4180 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4181 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
4182 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
4183 (s390_can_change_mode_class): ...this new function, inverting the
4184 sense of the return value.
4185 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4186 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4187 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
4188 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4189 (sh_cannot_change_mode_class): Replace with...
4190 (sh_can_change_mode_class): ...this new function, inverting the
4191 sense of the return value.
4192 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4193 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4194 (sparc_can_change_mode_class): New function.
4195 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4196 * config/spu/spu.c (spu_can_change_mode_class): New function.
4197 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4198 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4199 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4200 (visium_can_change_mode_class): New function.
4201 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
4203 2017-09-15 Richard Biener <rguenther@suse.de>
4205 PR tree-optimization/82217
4206 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
4207 but not undefined case.
4209 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4212 * postreload.c (reload_cse_simplify_operands): Skip
4213 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
4215 2017-09-15 Richard Biener <rguenther@suse.de>
4217 PR tree-optimization/68823
4218 * graphite-scop-detection.c (build_alias_set): If we have a
4219 possible dependence check whether we can handle them by just
4220 looking at the DRs DR_ACCESS_FNs.
4221 (build_scops): If build_alias_set fails, fail the SCOP.
4223 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4225 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4226 to support float128 built-in functions that require the ISA 3.0
4228 (BU_FLOAT128_3_HW): Likewise.
4229 (SQRTF128): Add support for the IEEE 128-bit square root and fma
4231 (FMAF128): Likewise.
4233 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4234 support for built-in functions that need the ISA 3.0 IEEE 128-bit
4235 floating point instructions.
4236 (rs6000_invalid_builtin): Likewise.
4237 (rs6000_builtin_mask_names): Likewise.
4238 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
4239 (RS6000_BTM_FLOAT128_HW): Likewise.
4240 (RS6000_BTM_COMMON): Likewise.
4241 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
4243 * doc/extend.texi (RS/6000 built-in functions): Document the
4244 IEEE 128-bit floating point square root and fused multiply-add
4247 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
4249 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
4250 reg (r2) isn't in the set of registers defined in the prologue.
4252 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4253 Alan Hayward <alan.hayward@arm.com>
4254 David Sherwood <david.sherwood@arm.com>
4256 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
4257 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
4258 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
4259 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
4260 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
4262 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
4263 max_vectorization_factor.
4264 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
4266 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4267 Alan Hayward <alan.hayward@arm.com>
4268 David Sherwood <david.sherwood@arm.com>
4270 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
4271 (vect_worthwhile_without_simd_p): Declare.
4272 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
4273 (vectorizable_reduction): Use it.
4274 * tree-vect-stmts.c (vectorizable_shift): Likewise.
4275 (vectorizable_operation): Likewise.
4277 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4278 Alan Hayward <alan.hayward@arm.com>
4279 David Sherwood <david.sherwood@arm.com>
4281 * tree-vectorizer.h (vect_get_num_copies): New function.
4282 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
4283 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4284 (vectorizable_induction): Likewise.
4285 (vectorizable_live_operation): Likewise.
4286 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4287 (vectorizable_bswap): Likewise.
4288 (vectorizable_call): Likewise.
4289 (vectorizable_conversion): Likewise.
4290 (vectorizable_assignment): Likewise.
4291 (vectorizable_shift): Likewise.
4292 (vectorizable_operation): Likewise.
4293 (vectorizable_store): Likewise.
4294 (vectorizable_load): Likewise.
4295 (vectorizable_condition): Likewise.
4296 (vectorizable_comparison): Likewise.
4297 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
4299 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4300 Alan Hayward <alan.hayward@arm.com>
4301 David Sherwood <david.sherwood@arm.com>
4303 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
4304 of vect_init_vector.
4306 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4307 Alan Hayward <alan.hayward@arm.com>
4308 David Sherwood <david.sherwood@arm.com>
4310 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
4311 an inline wrapper that provides a location.
4312 (gimple_build_vector): Likewise.
4313 * gimple-fold.c (gimple_build_vector_from_val): New function.
4314 (gimple_build_vector): Likewise.
4315 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
4316 functions to build the initial value. Always return a gimple value.
4317 (get_initial_defs_for_reduction): Likewise. Only compute
4319 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
4320 vect_init_vector on the results from get_initial_def(s)_for_reduction.
4321 (vectorizable_induction): Use gimple_build_vector rather than
4324 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4325 Alan Hayward <alan.hayward@arm.com>
4326 David Sherwood <david.sherwood@arm.com>
4328 * target.h (vec_perm_indices): New typedef.
4329 (auto_vec_perm_indices): Likewise.
4330 * optabs-query.h: Include target.h
4331 (can_vec_perm_p): Take a vec_perm_indices *.
4332 * optabs-query.c (can_vec_perm_p): Likewise.
4333 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
4334 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4335 * tree-vect-generic.c (lower_vec_perm): Likewise.
4336 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
4337 (vect_grouped_load_supported): Likewise.
4338 (vect_shift_permute_load_chain): Likewise.
4339 (vect_permute_store_chain): Use auto_vec_perm_indices.
4340 (vect_permute_load_chain): Likewise.
4341 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
4342 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
4343 Update uses of can_vec_perm_p.
4344 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
4345 mode with a number of elements. Take a vec_perm_indices *.
4346 (vect_create_epilog_for_reduction): Update accordingly.
4347 Use auto_vec_perm_indices.
4348 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
4349 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
4350 (vect_transform_slp_perm_load): Likewise.
4351 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
4352 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
4353 (vect_gen_perm_mask_checked): Likewise.
4354 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
4355 (vect_gen_perm_mask_checked): Likewise.
4356 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
4357 (vectorizable_store): Likewise.
4358 (vectorizable_load): Likewise.
4359 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
4360 (vectorizable_bswap): Likewise.
4362 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4363 Alan Hayward <alan.hayward@arm.com>
4364 David Sherwood <david.sherwood@arm.com>
4366 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
4367 * tree.c (build_vector): Likewise.
4368 (build_vector_from_ctor): Update accordingly.
4369 (build_vector_from_val): Likewise.
4370 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4371 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4372 * tree-vect-generic.c (add_rshift): Likewise.
4373 (expand_vector_divmod): Likewise.
4374 (optimize_vector_constructor): Likewise.
4375 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4376 (vect_transform_slp_perm_load): Likewise.
4377 (vect_schedule_slp_instance): Likewise.
4378 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
4379 (vectorizable_call): Likewise.
4380 (vect_gen_perm_mask_any): Likewise. Add elements in order.
4381 * expmed.c (make_tree): Likewise.
4382 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
4383 a vector passed to build_vector.
4384 (fold_convert_const): Likewise.
4385 (exact_inverse): Likewise.
4386 (fold_ternary_loc): Likewise.
4387 (fold_relational_const): Likewise.
4388 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
4389 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
4390 (const_unop): Likewise. Store the reduction accumulator in a
4391 variable rather than an array.
4392 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
4393 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
4394 the new vector, rather than constructing it after the input arrays.
4395 (native_interpret_vector): Use auto_vec<tree> when building
4396 a vector passed to build_vector. Add elements in order.
4397 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
4398 auto_vec<tree> when building a vector passed to build_vector.
4399 (vect_create_epilog_for_reduction): Likewise.
4400 (vectorizable_induction): Likewise.
4401 (get_initial_def_for_reduction): Likewise. Fix indentation of
4403 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
4405 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
4406 passed to build_vector.
4408 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4409 Alan Hayward <alan.hayward@arm.com>
4410 David Sherwood <david.sherwood@arm.com>
4412 * tree-core.h (tree_base::u): Add an "nelts" field.
4413 (tree_vector): Use VECTOR_CST_NELTS as the length.
4414 * tree.c (tree_size): Likewise.
4415 (make_vector): Initialize VECTOR_CST_NELTS.
4416 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
4417 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
4418 TYPE_VECTOR_SUBPARTS.
4419 * expr.c (const_vector_mask_from_tree): Consistently use "units"
4420 as the number of units, setting it from VECTOR_CST_NELTS.
4421 (const_vector_from_tree): Likewise.
4422 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
4423 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
4424 (fold_negate_expr_1): Likewise.
4425 (fold_convert_const): Likewise.
4426 (const_binop): Likewise. Differentiate the number of output and
4428 (const_unop): Likewise.
4429 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
4430 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
4431 in cases that did the opposite.
4433 2017-09-14 Richard Biener <rguenther@suse.de>
4435 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
4438 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
4440 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
4442 2017-09-14 Jakub Jelinek <jakub@redhat.com>
4445 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
4446 if and where to split a bb, except for splitting before debug insn
4447 sequences followed by non-label real insn. Delete debug insns
4448 in between basic blocks.
4450 * combine.c (make_compound_operation_int): Formatting fixes.
4452 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
4453 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4454 * config/netbsd.h (LINK_EH_SPEC): Likewise.
4455 * config/sol2.h (LINK_EH_SPEC): Likewise.
4456 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4457 * config/s390/linux.h (LINK_SPEC): Likewise.
4458 * config/freebsd.h (LINK_EH_SPEC): Likewise.
4459 * config/openbsd.h (LINK_EH_SPEC): Likewise.
4460 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4461 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
4462 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
4463 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4464 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
4465 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
4467 2017-09-13 Jakub Jelinek <jakub@redhat.com>
4469 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
4471 (ENDFILE_LINUX_SPEC): Likewise.
4472 (LINK_EH_SPEC): Likewise.
4473 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
4474 (LINK_OS_LINUX_SPEC32): Likewise.
4475 (LINK_OS_LINUX_SPEC64): Likewise.
4476 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
4477 (LINK_OS_LINUX_SPEC): Likewise.
4479 2017-09-13 Martin Liska <mliska@suse.cz>
4482 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
4483 CASE_HIGH is NULL_TREE.
4485 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4486 Alan Hayward <alan.hayward@arm.com>
4487 David Sherwood <david.sherwood@arm.com>
4489 * target.def (secondary_memory_needed): New hook.
4490 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
4491 instead of SECONDARY_MEMORY_NEEDED.
4492 (secondary_memory_needed_mode): Likewise.
4493 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
4494 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
4495 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
4496 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
4497 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
4498 * doc/tm.texi: Regenerate.
4499 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
4500 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
4501 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4502 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
4503 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
4504 * config/i386/i386.c (inline_secondary_memory_needed): Put the
4505 mode argument first and change the reg_class arguments to reg_class_t.
4506 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
4507 Make static. Update the call to inline_secondary_memory_needed.
4508 (ix86_register_move_cost): Update the call to
4509 inline_secondary_memory_needed.
4510 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4511 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
4513 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
4514 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4516 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
4517 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
4518 * config/mips/mips.c (mips_secondary_memory_needed): Make static
4519 and match hook interface. Add comment from mips.h.
4520 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4521 * config/mmix/mmix.md (truncdfsf2): Refer to
4522 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4524 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
4525 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
4526 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4527 (pa_secondary_memory_needed): New function.
4528 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
4529 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
4530 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4531 (pdp11_secondary_memory_needed): Make static and match hook interface.
4532 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
4533 * config/powerpcspe/powerpcspe-protos.h
4534 (rs6000_secondary_memory_needed_ptr): Delete.
4535 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
4537 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4538 (rs6000_option_override_internal): Assign to
4539 targetm.secondary_memory_needed rather than
4540 rs6000_secondary_memory_needed_ptr.
4541 (rs6000_secondary_memory_needed): Match hook interface.
4542 (rs6000_debug_secondary_memory_needed): Likewise.
4543 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
4544 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
4545 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
4546 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4547 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
4548 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
4550 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
4551 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4552 (rs6000_option_override_internal): Assign to
4553 targetm.secondary_memory_needed rather than
4554 rs6000_secondary_memory_needed_ptr.
4555 (rs6000_secondary_memory_needed): Match hook interface.
4556 (rs6000_debug_secondary_memory_needed): Likewise.
4557 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
4558 * config/s390/s390.c (s390_secondary_memory_needed): New function.
4559 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4560 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
4561 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4562 (sparc_secondary_memory_needed): New function.
4563 * lra-constraints.c (check_and_process_move): Refer to
4564 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4566 (curr_insn_transform): Likewise.
4567 (process_alt_operands): Use targetm.secondary_memory_needed
4568 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4569 (check_secondary_memory_needed_p): Likewise.
4570 (choose_split_class): Likewise.
4571 * reload.c: Unconditionally include code that was previously
4572 conditional on SECONDARY_MEMORY_NEEDED.
4573 (push_secondary_reload): Use targetm.secondary_memory_needed
4574 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4575 (push_reload): Likewise.
4576 * reload1.c: Unconditionally include code that was previously
4577 conditional on SECONDARY_MEMORY_NEEDED.
4578 (choose_reload_regs): Use targetm.secondary_memory_needed
4579 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4580 (gen_reload): Likewise.
4581 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
4583 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4584 Alan Hayward <alan.hayward@arm.com>
4585 David Sherwood <david.sherwood@arm.com>
4587 * target.def (secondary_memory_needed_mode): New hook:
4588 * targhooks.c (default_secondary_memory_needed_mode): Declare.
4589 * targhooks.h (default_secondary_memory_needed_mode): New function.
4590 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
4591 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
4592 * doc/tm.texi: Regenerate.
4593 * lra-constraints.c (check_and_process_move): Use
4594 targetm.secondary_memory_needed_mode instead of
4595 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
4596 (curr_insn_transform): Likewise.
4597 * reload.c (get_secondary_mem): Likewise.
4598 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4599 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
4601 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4602 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4603 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
4604 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4605 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
4607 * config/powerpcspe/powerpcspe-protos.h
4608 (rs6000_secondary_memory_needed_mode): Delete.
4609 * config/powerpcspe/powerpcspe.c
4610 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4611 (rs6000_secondary_memory_needed_mode): Make static.
4612 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4613 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
4615 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4617 (rs6000_secondary_memory_needed_mode): Make static.
4618 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4619 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
4620 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4621 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4622 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4624 (sparc_secondary_memory_needed_mode): New function.
4625 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
4627 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
4629 * config/aarch64/constraints.md (Umq): New constraint.
4630 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
4632 (mov<mode>): Update condition.
4634 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4636 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
4637 when bitposition is the same.
4639 2017-09-13 Richard Biener <rguenther@suse.de>
4641 * dwarf2out.c (output_die_symbol): Remove.
4642 (output_die): Do not output a DIEs symbol.
4644 2017-09-13 Richard Biener <rguenther@suse.de>
4647 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
4648 default-def to avoid breaking iterator update with the weird
4649 interaction with cgraph_update_edges_for_call_stmt_node.
4651 2017-09-13 Richard Biener <rguenther@suse.de>
4653 * tree-cfg.c (verify_gimple_assign_binary): Add verification
4654 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
4655 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
4656 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
4658 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
4660 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
4661 Disable pc relative literal load irrespective of
4662 TARGET_FIX_ERR_A53_84341 for default.
4664 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
4666 * config/sparc/sparc.c (output_return): Output the source location of
4667 the insn in the delay slot, if any.
4668 (output_sibcall): Likewise.
4670 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
4673 * common.opt (-static-pie): New alias.
4674 (shared): Negate static-pie.
4675 (-no-pie): Update help text.
4677 (static-pie): New option.
4678 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
4679 -static-pie support.
4680 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
4681 (LINK_EH_SPEC): Likewise.
4682 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4683 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4684 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4685 * gcc.c (LINK_COMMAND_SPEC): Likewise.
4686 (init_gcc_specs): Likewise.
4687 (init_spec): Likewise.
4688 (display_help): Update help message for -pie.
4689 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
4692 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
4694 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
4695 (movdi_aarch64): Likewise.
4696 (movti_aarch64): Likewise.
4698 2017-09-12 Simon Wright <simon@pushface.org>
4701 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
4702 calculation of the minor version, always output as 0.
4704 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4707 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4708 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
4709 on it early, rather than manual conversion late. For
4710 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
4711 instead of performing manual conversion.
4713 2017-09-12 Carl Love <cel@us.ibm.com>
4715 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
4716 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
4718 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4719 VMULOSW): Add definitions.
4720 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4721 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4722 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4723 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4724 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4726 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4728 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
4730 (movti_aarch64): Likewise.
4731 (movdf_aarch64): Likewise.
4732 (movtf_aarch64): Likewise.
4733 (load_pairdi): Likewise.
4734 (store_pairdi): Likewise.
4735 (load_pairdf): Likewise.
4736 (store_pairdf): Likewise.
4737 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
4738 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
4739 (ldr_got_small_<mode>): Likewise.
4740 (ldr_got_small_28k_<mode>): Likewise.
4741 (ldr_got_tiny): Likewise.
4742 * config/aarch64/iterators.md (ldst_sz): New.
4743 (ldpstp_sz): Likewise.
4744 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
4746 (thunderx_load): Split load_8 to load_16.
4747 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
4749 (thunderx2t99_storepair_basic): Split store_8 to store_16.
4750 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
4751 (xgene1_store_pair): Split store_8 to store_16.
4752 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
4753 (falkor_st_0_st_sd): Split store_8 to store_16.
4755 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4757 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
4758 and store1/2/3/4 to store_4/8/12/16.
4759 * config/aarch64/aarch64.md: Update for rename.
4760 * config/arm/arm.md: Likewise.: Likewise.
4761 * config/arm/arm.c: Likewise.
4762 * config/arm/thumb1.md: Likewise.
4763 * config/arm/thumb2.md: Likewise.
4764 * config/arm/vfp.md: Likewise.
4765 * config/arm/arm-generic.md: Likewise.
4766 * config/arm/arm1020e.md: Likewise.
4767 * config/arm/arm1026ejs.md: Likewise.
4768 * config/arm/arm1136jfs.md: Likewise.
4769 * config/arm/arm926ejs.md: Likewise.
4770 * config/arm/cortex-a15.md: Likewise.
4771 * config/arm/cortex-a17.md: Likewise.
4772 * config/arm/cortex-a5.md: Likewise.
4773 * config/arm/cortex-a53.md: Likewise.
4774 * config/arm/cortex-a57.md: Likewise.
4775 * config/arm/cortex-a7.md: Likewise.
4776 * config/arm/cortex-a8.md: Likewise.
4777 * config/arm/cortex-a9.md: Likewise.
4778 * config/arm/cortex-m4.md: Likewise.
4779 * config/arm/cortex-m7.md: Likewise.
4780 * config/arm/cortex-r4.md: Likewise.
4781 * config/arm/exynos-m1.md: Likewise.
4782 * config/arm/fa526.md: Likewise.
4783 * config/arm/fa606te.md: Likewise.
4784 * config/arm/fa626te.md: Likewise.
4785 * config/arm/fa726te.md: Likewise.
4786 * config/arm/fmp626.md: Likewise.
4787 * config/arm/iwmmxt.md: Likewise.
4788 * config/arm/ldmstm.md: Likewise.
4789 * config/arm/marvell-pj4.md: Likewise.
4790 * config/arm/xgene1.md: Likewise.
4791 * config/aarch64/thunderx.md: Likewise.
4792 * config/aarch64/thunderx2t99.md: Likewise.
4793 * config/aarch64/falkor.md: Likewise.
4795 2017-09-12 Martin Liska <mliska@suse.cz>
4797 * attribs.c (private_lookup_attribute): New function.
4798 * attribs.h (private_lookup_attribute): Declared here.
4799 (lookup_attribute): Called from this place.
4801 2017-09-12 Richard Biener <rguenther@suse.de>
4803 PR tree-optimization/82157
4804 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
4805 stmts with side-effects.
4807 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4808 Alan Hayward <alan.hayward@arm.com>
4809 David Sherwood <david.sherwood@arm.com>
4811 * target.def (hard_regno_nregs): New hook.
4812 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
4813 * targhooks.h (default_hard_regno_nregs): Declare.
4814 * targhooks.c (default_hard_regno_nregs): New function.
4815 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
4816 (TARGET_HARD_REGNO_NREGS): ...this hook.
4817 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
4818 (CLASS_MAX_NREGS): Likewise.
4819 * doc/tm.texi: Regenerate.
4820 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
4821 instead of HARD_REGNO_NREGS.
4822 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
4823 HARD_REGNO_NREGS in the comment.
4824 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
4825 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
4826 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
4827 Return an unsigned int.
4828 (TARGET_HARD_REGNO_NREGS): Redefine.
4829 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
4830 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
4831 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4832 (arc_hard_regno_nregs): New function.
4833 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
4834 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
4835 (arm_hard_regno_nregs): New function.
4836 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
4837 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
4838 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
4839 (TARGET_HARD_REGNO_NREGS): Redefine.
4840 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
4841 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
4842 (HARD_REGNO_NREGS): Delete.
4843 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
4844 (cr16_hard_regno_nregs): New function.
4845 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
4846 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
4847 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
4848 (cris_hard_regno_nregs): New function.
4849 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
4850 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
4851 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4852 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
4853 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
4854 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
4855 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
4856 (frv_hard_regno_nregs): Make static. Take and return an
4858 (frv_class_max_nregs): Remove outdated copy of documentation.
4859 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
4860 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
4861 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
4862 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
4863 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
4864 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
4865 (TARGET_HARD_REGNO_NREGS): Redefine.
4866 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
4867 (CLASS_MAX_NREGS): Update comment.
4868 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
4869 (ia64_hard_regno_nregs): New function.
4870 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
4871 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
4872 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
4873 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
4874 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
4876 (m32c_hard_regno_nregs): Likewise. Make static.
4877 (TARGET_HARD_REGNO_NREGS): Redefine.
4878 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
4879 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
4880 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
4881 (m68k_hard_regno_nregs): New function.
4882 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
4883 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
4884 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
4885 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
4886 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
4887 Take and return an unsigned int.
4888 (TARGET_HARD_REGNO_NREGS): Redefine.
4889 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
4890 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4891 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
4892 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
4893 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
4894 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
4895 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
4896 (msp430_hard_regno_nregs): Make static. Take and return an
4898 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
4899 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
4900 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
4901 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
4902 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
4903 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
4904 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
4905 (TARGET_HARD_REGNO_NREGS): Redefine.
4906 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
4907 (PA_HARD_REGNO_NREGS): ...this.
4908 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
4909 (PA_HARD_REGNO_NREGS): ...this.
4910 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4911 (pa_hard_regno_nregs): New function.
4912 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
4913 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
4914 (pdp11_hard_regno_nregs): New function.
4915 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
4916 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
4917 (rs6000_hard_regno_nregs_hook): New function.
4918 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
4919 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4920 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
4921 Take and return an unsigned int. Move earlier in file.
4922 (TARGET_HARD_REGNO_NREGS): Redefine.
4923 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
4924 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
4925 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
4926 (rl78_hard_regno_nregs): Make static. Take and return an
4928 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
4929 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
4930 (rs6000_hard_regno_nregs_hook): New function.
4931 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
4932 * config/rx/rx.c (rx_hard_regno_nregs): New function.
4933 (TARGET_HARD_REGNO_NREGS): Redefine.
4934 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
4935 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
4936 instead of HARD_REGNO_NREGS.
4937 (s390_hard_regno_nregs): New function.
4938 (s390_hard_regno_mode_ok): Add comment from s390.h.
4939 (TARGET_HARD_REGNO_NREGS): Redefine.
4940 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
4941 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
4942 (sh_hard_regno_nregs): New function.
4943 (sh_pass_in_reg_p): Use it.
4944 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
4945 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4946 (sparc_hard_regno_nregs): New function.
4947 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
4948 * config/spu/spu.c (spu_hard_regno_nregs): New function.
4949 (spu_function_arg_advance): Use it, supplying a valid register number.
4950 (TARGET_HARD_REGNO_NREGS): Redefine.
4951 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
4952 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
4953 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
4954 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
4955 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
4956 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
4957 (CLASS_MAX_NREGS): Remove copy of old documentation.
4958 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
4959 (visium_hard_regno_nregs): New function.
4960 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
4961 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
4962 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4963 xtensa_hard_regno_nregs): New function.
4964 * system.h (HARD_REGNO_NREGS): Poison.
4966 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4968 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
4969 hard_regno_nregs instead of HARD_REGNO_NREGS.
4970 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
4971 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
4972 (c6x_expand_epilogue): Likewise.
4973 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
4974 (frv_read_iacc_argument): Likewise.
4975 * config/sh/sh.c: Include regs.h.
4976 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4977 (regs_used): Likewise.
4978 (output_stack_adjust): Likewise.
4979 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
4980 * expmed.c: Include regs.h.
4981 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4982 * ree.c: Include regs.h.
4983 (combine_reaching_defs): Use hard_regno_nregs instead of
4985 (add_removable_extension): Likewise.
4987 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4989 * regs.h (hard_regno_nregs): Turn into a function.
4990 (end_hard_regno): Update accordingly.
4991 * caller-save.c (setup_save_areas): Likewise.
4992 (save_call_clobbered_regs): Likewise.
4993 (replace_reg_with_saved_mem): Likewise.
4994 (insert_restore): Likewise.
4995 (insert_save): Likewise.
4996 * combine.c (can_change_dest_mode): Likewise.
4997 (move_deaths): Likewise.
4998 (distribute_notes): Likewise.
4999 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
5000 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
5001 (rs6000_split_multireg_move): Likewise.
5002 (rs6000_register_move_cost): Likewise.
5003 (rs6000_memory_move_cost): Likewise.
5004 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
5005 (rs6000_split_multireg_move): Likewise.
5006 (rs6000_register_move_cost): Likewise.
5007 (rs6000_memory_move_cost): Likewise.
5008 * cselib.c (cselib_reset_table): Likewise.
5009 (cselib_lookup_1): Likewise.
5010 * emit-rtl.c (set_mode_and_regno): Likewise.
5011 * function.c (aggregate_value_p): Likewise.
5012 * ira-color.c (setup_profitable_hard_regs): Likewise.
5013 (check_hard_reg_p): Likewise.
5014 (calculate_saved_nregs): Likewise.
5015 (assign_hard_reg): Likewise.
5016 (improve_allocation): Likewise.
5017 (calculate_spill_cost): Likewise.
5018 * ira-emit.c (modify_move_list): Likewise.
5019 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
5020 (ira_hard_reg_in_set_p): Likewise.
5021 * ira.c (setup_reg_mode_hard_regset): Likewise.
5022 (clarify_prohibited_class_mode_regs): Likewise.
5023 (check_allocation): Likewise.
5024 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5025 (lra_setup_reg_renumber): Likewise.
5026 (setup_try_hard_regno_pseudos): Likewise.
5027 (spill_for): Likewise.
5028 (assign_hard_regno): Likewise.
5029 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
5030 * lra-constraints.c (in_class_p): Likewise.
5031 (lra_constraint_offset): Likewise.
5032 (simplify_operand_subreg): Likewise.
5033 (lra_constraints): Likewise.
5034 (split_reg): Likewise.
5035 (split_if_necessary): Likewise.
5036 (invariant_p): Likewise.
5037 (inherit_in_ebb): Likewise.
5038 * lra-lives.c (process_bb_lives): Likewise.
5039 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
5040 (get_hard_regs): Likewise.
5041 (do_remat): Likewise.
5042 * lra-spills.c (assign_spill_hard_regs): Likewise.
5043 * mode-switching.c (create_pre_exit): Likewise.
5044 * postreload.c (reload_combine_recognize_pattern): Likewise.
5045 * recog.c (peep2_find_free_register): Likewise.
5046 * regcprop.c (kill_value_regno): Likewise.
5047 (set_value_regno): Likewise.
5048 (copy_value): Likewise.
5049 (maybe_mode_change): Likewise.
5050 (find_oldest_value_reg): Likewise.
5051 (copyprop_hardreg_forward_1): Likewise.
5052 * regrename.c (check_new_reg_p): Likewise.
5053 (regrename_do_replace): Likewise.
5054 * reload.c (push_reload): Likewise.
5055 (combine_reloads): Likewise.
5056 (find_dummy_reload): Likewise.
5057 (operands_match_p): Likewise.
5058 (find_reloads): Likewise.
5059 (find_equiv_reg): Likewise.
5060 (reload_adjust_reg_for_mode): Likewise.
5061 * reload1.c (count_pseudo): Likewise.
5062 (count_spilled_pseudo): Likewise.
5063 (find_reg): Likewise.
5064 (clear_reload_reg_in_use): Likewise.
5065 (free_for_value_p): Likewise.
5066 (allocate_reload_reg): Likewise.
5067 (choose_reload_regs): Likewise.
5068 (reload_adjust_reg_for_temp): Likewise.
5069 (emit_reload_insns): Likewise.
5070 (delete_output_reload): Likewise.
5071 * rtlanal.c (subreg_get_info): Likewise.
5072 * sched-deps.c (sched_analyze_reg): Likewise.
5073 * sel-sched.c (init_regs_for_mode): Likewise.
5074 (mark_unavailable_hard_regs): Likewise.
5075 (choose_best_reg_1): Likewise.
5076 (verify_target_availability): Likewise.
5077 * valtrack.c (dead_debug_insert_temp): Likewise.
5078 * var-tracking.c (track_loc_p): Likewise.
5079 (emit_note_insn_var_location): Likewise.
5080 * varasm.c (make_decl_rtl): Likewise.
5081 * reginfo.c (choose_hard_reg_mode): Likewise.
5082 (init_reg_modes_target): Refer directly to
5083 this_target_regs->x_hard_regno_nregs.
5085 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5087 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5088 instead of hard_regno_nregs.
5090 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5092 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5093 end_hard_regno instead of hard_regno_nregs.
5094 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5095 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5096 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5097 * ira-color.c (improve_allocation): Likewise.
5098 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5099 * lra-lives.c (mark_regno_live): Likewise.
5100 (mark_regno_dead): Likewise.
5101 * lra-remat.c (operand_to_remat): Likewise.
5102 * lra.c (collect_non_operand_hard_regs): Likewise.
5103 * postreload.c (reload_combine_note_store): Likewise.
5104 (move2add_valid_value_p): Likewise.
5105 * reload.c (regno_clobbered_p): Likewise.
5107 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5109 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5111 * config/v850/v850.c (v850_reorg): Likewise.
5112 * reload.c (refers_to_regno_for_reload_p): Likewise.
5113 (find_equiv_reg): Likewise.
5114 * reload1.c (reload_reg_reaches_end_p): Likewise.
5116 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5118 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5120 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5121 * config/arm/arm.c (output_move_neon): Likewise.
5122 (arm_attr_length_move_neon): Likewise.
5123 (neon_split_vcombine): Likewise.
5124 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5125 (c6x_mark_reg_written): Likewise.
5126 (c6x_dwarf_register_span): Likewise.
5127 * config/i386/i386.c (ix86_save_reg): Likewise.
5128 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5129 (rws_access_reg): Likewise.
5130 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5131 * mode-switching.c (create_pre_exit): Likewise.
5132 * ree.c (combine_reaching_defs): Likewise.
5133 (add_removable_extension): Likewise.
5134 * regcprop.c (find_oldest_value_reg): Likewise.
5135 (copyprop_hardreg_forward_1): Likewise.
5136 * reload.c (reload_inner_reg_of_subreg): Likewise.
5137 (push_reload): Likewise.
5138 (combine_reloads): Likewise.
5139 (find_dummy_reload): Likewise.
5140 (reload_adjust_reg_for_mode): Likewise.
5141 * reload1.c (find_reload_regs): Likewise.
5142 (forget_old_reloads_1): Likewise.
5143 (reload_reg_free_for_value_p): Likewise.
5144 (reload_adjust_reg_for_temp): Likewise.
5145 (emit_reload_insns): Likewise.
5146 (delete_output_reload): Likewise.
5147 * sel-sched.c (choose_best_reg_1): Likewise.
5148 (choose_best_pseudo_reg): Likewise.
5150 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5151 Alan Hayward <alan.hayward@arm.com>
5152 David Sherwood <david.sherwood@arm.com>
5154 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
5155 * target.def (slow_unaligned_access): New hook.
5156 * targhooks.h (default_slow_unaligned_access): Declare.
5157 * targhooks.c (default_slow_unaligned_access): New function.
5158 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
5159 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
5160 * doc/tm.texi: Regenerate.
5161 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
5162 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
5163 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
5165 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
5166 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
5168 (rs6000_slow_unaligned_access): New function.
5169 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5170 (expand_block_compare): Likewise.
5171 (expand_strn_compare): Likewise.
5172 (rs6000_rtx_costs): Likewise.
5173 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
5174 (riscv_slow_unaligned_access): Likewise.
5175 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
5176 (riscv_slow_unaligned_access_p): ...this and make static.
5177 (riscv_option_override): Update accordingly.
5178 (riscv_slow_unaligned_access): New function.
5179 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5180 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
5181 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5182 (rs6000_slow_unaligned_access): New function.
5183 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5184 (rs6000_rtx_costs): Likewise.
5185 * config/rs6000/rs6000-string.c (expand_block_compare)
5186 (expand_strn_compare): Use targetm.slow_unaligned_access instead
5187 of SLOW_UNALIGNED_ACCESS.
5188 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
5189 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
5190 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
5191 of SLOW_UNALIGNED_ACCESS.
5192 * expmed.c (simple_mem_bitfield_p): Likewise.
5193 * expr.c (alignment_for_piecewise_move): Likewise.
5194 (emit_group_load_1): Likewise.
5195 (emit_group_store): Likewise.
5196 (copy_blkmode_from_reg): Likewise.
5197 (emit_push_insn): Likewise.
5198 (expand_assignment): Likewise.
5199 (store_field): Likewise.
5200 (expand_expr_real_1): Likewise.
5201 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5202 * lra-constraints.c (simplify_operand_subreg): Likewise.
5203 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5204 * gimple-ssa-store-merging.c: Likewise in block comment at start
5206 * tree-ssa-strlen.c: Include target.h.
5207 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
5208 of SLOW_UNALIGNED_ACCESS.
5209 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
5211 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5213 PR rtl-optimization/82185
5214 * expmed.c (emit_store_flag_int): Only test tem if it has been
5217 2017-09-12 Richard Biener <rguenther@suse.de>
5220 * match.pd ((FTYPE) N CMP CST): Fix typo.
5222 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
5224 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
5226 (mips_near_type_p): Add 'short_call' attribute as a synonym
5228 * doc/extend.texi (short_call): Document new function attribute.
5230 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5233 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
5234 assertion check that in the condition.
5235 (get_atomic_generic_size): Likewise. Before testing if parameter
5236 has pointer type, if it has array type, call for C++
5237 default_conversion to perform array-to-pointer conversion.
5239 2017-09-12 Richard Biener <rguenther@suse.de>
5241 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
5242 for operations we cannot scalarize.
5244 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
5246 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
5247 vectors heap vectors. Clean up comments.
5248 Make visited_bbs a reference.
5249 (profitable_jump_thread_path): Make GC
5250 vectors heap vectors. Clean up comments.
5252 (convert_and_register_jump_thread_path): Make GC vectors heap
5254 (check_subpath_and_update_thread_path): Same. Clean up comments.
5255 Make visited_bbs a reference.
5256 (handle_phi): Abstract common code to to
5257 register_jump_thread_path_if_profitable.
5258 Rename VAR_BB to DEF_BB.
5260 Make GC vectors heap vectors.
5261 Make visited_bbs a reference.
5262 (handle_assignment): Same.
5263 (register_jump_thread_path_if_profitable): New.
5264 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
5266 Make GC vectors heap vectors. Clean up comments.
5267 Make visited_bbs a reference.
5268 (find_jump_threads_backwards): Make visited_bbs live in the stack.
5269 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
5272 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
5275 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
5276 words of E_DImode object are reachable by xtensa_uimm8x4 access.
5278 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
5280 Revert r251800 and r251799.
5282 2017-09-11 Martin Jambor <mjambor@suse.cz>
5285 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
5286 arguments in advance.
5287 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
5288 use it to find predecessor edges.
5289 (naive_outof_ssa): Collect vector of predecessors.
5291 2017-09-08 Jason Merrill <jason@redhat.com>
5293 PR c++/70029 - ICE with ref-qualifier and -flto
5294 * langhooks.h (struct lang_hooks_for_types): Add
5295 copy_lang_qualifiers.
5296 * attribs.c (build_type_attribute_qual_variant): Use it.
5297 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
5299 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
5300 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
5302 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
5305 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
5306 (*mulsi3_sp64): New instruction.
5307 (mulsi3): New expander.
5309 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
5311 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
5313 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5315 * sancov.c: Include memmodel.h.
5317 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
5320 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
5323 2017-09-07 Carl Love <cel@us.ibm.com>
5325 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
5326 the sldi instruction.
5328 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
5330 * sancov.c: Include tm_p.h.
5332 2017-09-07 Jakub Jelinek <jakub@redhat.com>
5335 * output.h (switch_to_other_text_partition): New declaration.
5336 * varasm.c (switch_to_other_text_partition): New function.
5337 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
5338 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
5339 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
5340 to the other text partition before emitting LCL label and switch back
5341 after emitting the word after it.
5343 2017-09-07 Richard Biener <rguenther@suse.de>
5345 * passes.def (pass_split_crit_edges): Remove instance before PRE.
5346 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
5347 critical edges here, after loop init.
5348 (pass_data_pre): Remove PROP_no_crit_edges flags.
5349 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
5350 for valueization of call args to avoid leaking VN_TOP.
5351 (visit_use): Assert we do not visit default defs.
5352 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
5353 Use error_mark_node to more easily detect leaking VN_TOP.
5354 All default-defs are varying, not VN_TOP. Mark them visited.
5355 (run_scc_vn): Make code match comment.
5357 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5359 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
5360 OPTION_MASK_FLOAT128_KEYWORD.
5361 (POWERPC_MASKS): Likewise.
5362 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
5363 support for the -mfloat128-type option, and make -mfloat128
5364 default on PowerPC Linux systems. Define or undefine
5365 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
5366 Define __float128 to be __ieee128 if IEEE 128-bit support is
5367 enabled, or undefine it.
5368 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
5369 Delete defining __FLOAT128_TYPE__.
5370 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
5371 -mfloat128-type option and make -mfloat128 default on PowerPC
5373 (TARGET_FLOAT128_TYPE): Likewise.
5374 (-mfloat128-type): Likewise.
5375 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5376 Delete the -mfloat128-type option and make -mfloat128 default on
5377 PowerPC Linux systems. Always use __ieee128 to be the keyword for
5378 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
5379 128-bit floating point is enabled. Change tests from using
5380 -mfloat128-type to -mfloat128.
5381 (rs6000_mangle_type): Use the correct mangling for the __float128
5382 type even if normal long double is restricted to 64-bits.
5383 (floatn_mode): Enable the _Float128 type by default on VSX Linux
5385 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
5386 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
5387 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
5389 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
5390 documentation for -mfloat128.
5392 2017-09-06 Olivier Hainque <hainque@adacore.com>
5394 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
5396 2017-09-06 Wish Wu <wishwu007@gmail.com>
5397 Jakub Jelinek <jakub@redhat.com>
5399 * asan.c (initialize_sanitizer_builtins): Add
5400 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
5401 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
5402 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
5403 BT_FN_VOID_UINT64_PTR variables.
5404 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
5405 (BT_FN_VOID_UINT16_UINT16): Likewise.
5406 (BT_FN_VOID_UINT32_UINT32): Likewise.
5407 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
5408 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
5409 (BT_FN_VOID_UINT64_PTR): Likewise.
5410 * common.opt (flag_sanitize_coverage): New variable.
5411 (fsanitize-coverage=trace-pc): Remove.
5412 (fsanitize-coverage=): Add.
5413 * flag-types.h (enum sanitize_coverage_code): New enum.
5414 * fold-const.c (fold_range_test): Disable non-short-circuit
5415 optimization if flag_sanitize_coverage.
5416 (fold_truth_andor): Likewise.
5417 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
5418 * opts.c (COVERAGE_SANITIZER_OPT): Define.
5419 (coverage_sanitizer_opts): New array.
5420 (get_closest_sanitizer_option): Add OPTS argument, handle also
5421 OPT_fsanitize_coverage_.
5422 (parse_sanitizer_options): Adjusted to also handle
5423 OPT_fsanitize_coverage_.
5424 (common_handle_option): Add OPT_fsanitize_coverage_.
5425 * sancov.c (instrument_comparison, instrument_switch): New function.
5426 (sancov_pass): Add trace-cmp support.
5427 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
5428 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
5429 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
5430 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
5431 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
5432 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
5433 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
5434 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
5435 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
5436 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
5438 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5440 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5441 error. Only quit immediately if parsing is complete.
5442 (BEGIN): Initialize fatal_err and parse_done.
5443 (begin fpu, end fpu): Check number of arguments.
5444 (begin arch, end arch): Likewise.
5445 (begin cpu, end cpu): Likewise.
5446 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5447 (optalias): Likewise.
5449 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5451 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5452 * config/arm/arm-isa.h: Delete. Move definitions to ...
5453 * arm-cpus.in: ... here. Use new feature and fgroup values.
5454 * config/arm/arm.c (arm_option_override): Use lower case for feature
5456 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5457 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5458 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5460 (print_isa_bits_for): New function.
5461 (gen_isa): New function.
5462 (gen_comm_data): Use print_isa_bits_for.
5463 (define feature): New keyword.
5464 (define fgroup): New keyword.
5465 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
5466 (arm-isa.h): Add rule to generate file.
5467 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5468 case for feature bit names.
5470 2017-09-06 Richard Biener <rguenther@suse.de>
5472 * tree-ssa-pre.c (NECESSARY): Remove.
5473 (create_expression_by_pieces): Do not touch pass-local flags.
5474 (insert_into_preds_of_block): Likewise.
5475 (do_pre_regular_insertion): Likewise.
5476 (eliminate_insert): Likewise.
5477 (eliminate_dom_walker::before_dom_children): Likewise.
5478 (fini_eliminate): Do not look at inserted_exprs.
5479 (mark_operand_necessary): Remove.
5480 (remove_dead_inserted_code): Replace with simple work-list
5481 algorithm based on inserted_exprs and SSA uses.
5482 (pass_pre::execute): Re-order fini_eliminate and
5483 remove_dead_inserted_code.
5485 2017-09-06 Olivier Hainque <hainque@adacore.com>
5487 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
5488 for VxWorks 7. Adjust surrounding comments.
5490 2017-09-06 Richard Biener <rguenther@suse.de>
5492 * gimple-ssa-strength-reduction.c
5493 (find_candidates_dom_walker::before_dom_children): Also allow
5496 2017-09-06 Richard Biener <rguenther@suse.de>
5498 PR tree-optimization/82108
5499 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
5500 for gap in the non-permutation SLP case.
5502 2017-09-06 Martin Jambor <mjambor@suse.cz>
5504 PR tree-optimization/82078
5505 * tree-sra.c (sort_and_splice_var_accesses): Move call to
5506 add_access_to_work_queue...
5507 (build_accesses_from_assign): ...here.
5508 (propagate_all_subaccesses): Make sure racc is the group
5509 representative, if there is one.
5511 2017-09-06 Jakub Jelinek <jakub@redhat.com>
5514 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
5517 2017-09-06 Richard Biener <rguenther@suse.de>
5519 * gimple-ssa-strength-reduction.c
5520 (find_candidates_dom_walker::before_doom_children): Use a
5521 type and not a mode check.
5523 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5526 * config/arm/predicates.md (arm_general_adddi_operand): Create new
5528 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
5530 2017-09-05 Jeff Law <law@redhat.com>
5532 PR tree-optimization/64910
5533 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
5534 cases where we have 3 or more operands.
5536 2017-09-05 Jakub Jelinek <jakub@redhat.com>
5539 * omp-low.c (lower_omp_for): Recompute tree invariant if
5540 gimple_omp_for_initial/final is ADDR_EXPR.
5543 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
5544 into gimple val before gimplification fo the COND_EXPR.
5546 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
5548 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
5549 REGION_COPY argument.
5550 (thread_through_all_blocks): Remove unused argument to
5551 duplicate_thread_path.
5553 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5554 Alan Hayward <alan.hayward@arm.com>
5555 David Sherwood <david.sherwood@arm.com>
5557 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
5558 Take a scalar_mode rather than a machine_mode.
5559 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5560 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
5561 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5562 (aarch64_gen_adjusted_ldpstp): Likewise.
5563 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
5565 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5566 Alan Hayward <alan.hayward@arm.com>
5567 David Sherwood <david.sherwood@arm.com>
5569 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
5570 Take a scalar_int_mode instead of a machine_mode.
5571 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5572 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5573 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5574 (aarch64_simd_attr_length_rglist): Delete.
5575 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
5576 a scalar_int_mode instead of a machine_mode.
5577 (aarch64_add_offset): Likewise.
5578 (aarch64_internal_mov_immediate): Likewise
5579 (aarch64_add_constant_internal): Likewise.
5580 (aarch64_add_constant): Likewise.
5581 (aarch64_movw_imm): Likewise.
5582 (aarch64_rtx_arith_op_extract_p): Likewise.
5583 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5584 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5585 Remove assert that the mode isn't a vector.
5586 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5587 (aarch64_expand_mov_immediate): Update calls after above changes.
5588 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
5589 (aarch64_and_bitmask_imm): Check for scalar integer modes.
5590 (aarch64_move_imm): Likewise.
5591 (aarch64_can_const_movi_rtx_p): Likewise.
5592 (aarch64_strip_extend): Likewise.
5593 (aarch64_extr_rtx_p): Likewise.
5594 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
5595 a CONST_INT when the mode parameter is VOIDmode.
5596 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
5598 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5600 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
5601 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5602 (bitwise_type_for_mode): Update accordingly.
5604 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5606 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
5607 * stor-layout.c (mode_for_size_tree): Likewise.
5608 (mode_for_array): Update accordingly.
5609 (layout_decl): Likewise.
5610 (compute_record_mode): Likewise. Only set the mode once.
5612 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5614 * target.def (get_mask_mode): Change return type to opt_mode.
5616 * doc/tm.texi: Regenerate.
5617 * targhooks.h (default_get_mask_mode): Return an opt_mode.
5618 * targhooks.c (default_get_mask_mode): Likewise.
5619 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5620 * optabs-query.c (can_vec_mask_load_store_p): Update use of
5621 targetm.get_mask_mode.
5622 * tree.c (build_truth_vector_type): Likewise.
5624 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5626 * machmode.h (mode_for_vector): Return an opt_mode.
5627 * stor-layout.c (mode_for_vector): Likewise.
5628 (mode_for_int_vector): Update accordingly.
5629 (layout_type): Likewise.
5630 * config/i386/i386.c (emit_memmov): Likewise.
5631 (ix86_expand_set_or_movmem): Likewise.
5632 (ix86_expand_vector_init): Likewise.
5633 (ix86_get_mask_mode): Likewise.
5634 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
5636 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
5637 * expmed.c (extract_bit_field_1): Likewise.
5638 * expr.c (expand_expr_real_2): Likewise.
5639 * optabs-query.c (can_vec_perm_p): Likewise.
5640 (can_vec_mask_load_store_p): Likewise.
5641 * optabs.c (expand_vec_perm): Likewise.
5642 * targhooks.c (default_get_mask_mode): Likewise.
5643 * tree-vect-stmts.c (vectorizable_store): Likewise.
5644 (vectorizable_load): Likewise.
5645 (get_vectype_for_scalar_type_and_size): Likewise.
5647 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5649 * machmode.h (mode_for_int_vector): New function.
5650 * stor-layout.c (mode_for_int_vector): Likewise.
5651 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
5652 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
5653 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5654 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5655 (s390_expand_vcond): Likewise.
5657 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5659 * machmode.h (opt_machine_mode): New type.
5660 (opt_mode<T>): Allow construction from anything that can be
5662 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
5663 (mode_for_size): Return an opt_machine_mode.
5664 * stor-layout.c (mode_for_size): Likewise.
5665 (mode_for_size_tree): Update call accordingly.
5666 (bitwise_mode_for_mode): Likewise.
5667 (make_fract_type): Likewise.
5668 (make_accum_type): Likewise.
5669 * caller-save.c (replace_reg_with_saved_mem): Update call
5671 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5672 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5673 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5674 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5675 * expmed.c (extract_bit_field_1): Likewise.
5676 * reload.c (get_secondary_mem): Likewise.
5677 * varasm.c (assemble_integer): Likewise.
5678 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
5681 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5683 * machmode.h (decimal_float_mode_for_size): New function.
5684 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
5685 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
5686 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
5687 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
5688 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
5689 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
5691 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5693 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
5694 (get_builtin_sync_mode): Likewise.
5695 (expand_ifn_atomic_compare_exchange): Likewise.
5696 (expand_builtin_atomic_clear): Likewise.
5697 (expand_builtin_atomic_test_and_set): Likewise.
5698 (fold_builtin_atomic_always_lock_free): Likewise.
5699 * calls.c (compute_argument_addresses): Likewise.
5700 (emit_library_call_value_1): Likewise.
5701 (store_one_arg): Likewise.
5702 * combine.c (combine_instructions): Likewise.
5703 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
5704 * config/arm/arm.c (arm_function_value): Likewise.
5705 (aapcs_allocate_return_reg): Likewise.
5706 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
5707 * config/i386/i386.c (construct_container): Likewise.
5708 (ix86_gimplify_va_arg): Likewise.
5709 (ix86_expand_sse_cmp): Likewise.
5710 (emit_memmov): Likewise.
5711 (emit_memset): Likewise.
5712 (expand_small_movmem_or_setmem): Likewise.
5713 (ix86_expand_pextr): Likewise.
5714 (ix86_expand_pinsr): Likewise.
5715 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
5716 * config/microblaze/microblaze.c (microblaze_block_move_straight):
5718 * config/mips/mips.c (mips_function_value_1) Likewise.
5719 (mips_block_move_straight): Likewise.
5720 (mips_expand_ins_as_unaligned_store): Likewise.
5721 * config/powerpcspe/powerpcspe.c
5722 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5723 (rs6000_darwin64_record_arg_flush): Likewise.
5724 * config/rs6000/rs6000.c
5725 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5726 (rs6000_darwin64_record_arg_flush): Likewise.
5727 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
5728 (sparc_function_value_1): Likewise.
5729 * config/spu/spu.c (adjust_operand): Likewise.
5730 (spu_emit_branch_or_set): Likewise.
5731 (arith_immediate_p): Likewise.
5732 * emit-rtl.c (gen_lowpart_common): Likewise.
5733 * expr.c (expand_expr_real_1): Likewise.
5734 * function.c (assign_parm_setup_block): Likewise.
5735 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5736 * reload1.c (alter_reg): Likewise.
5737 * stor-layout.c (mode_for_vector): Likewise.
5738 (layout_type): Likewise.
5740 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5742 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
5743 (spu_convert_move): Likewise.
5744 * lower-subreg.c (resolve_simple_move): Likewise.
5746 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5749 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
5750 define_insn to a define_expand.
5751 (altivec_vsum2sws_direct): New define_insn.
5752 (altivec_vsumsws): Convert from a define_insn to a define_expand.
5754 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5756 * config/arm/arm.c (arm_option_params_internal): Improve setting of
5759 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
5764 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
5765 realign stack if stack alignment needed is less than incoming
5768 2017-09-05 Marek Polacek <polacek@redhat.com>
5771 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
5774 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5776 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
5778 2017-09-05 Richard Biener <rguenther@suse.de>
5780 PR tree-optimization/82084
5781 * fold-const.c (can_native_encode_string_p): Handle wide characters.
5783 2017-09-05 Richard Biener <rguenther@suse.de>
5785 PR tree-optimization/82102
5786 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
5788 2017-09-05 Martin Liska <mliska@suse.cz>
5790 PR tree-optimization/82032
5791 * tree-cfg.c (generate_range_test): New function.
5792 * tree-cfg.h (generate_range_test): Declared here.
5793 * tree-cfgcleanup.c (convert_single_case_switch): New function.
5794 (cleanup_control_expr_graph): Use it.
5795 * tree-switch-conversion.c (try_switch_expansion): Remove
5797 (emit_case_nodes): Use generate_range_test.
5799 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5802 * config/i386/i386.md (*<btsc><mode>_mask): Add
5803 TARGET_USE_BT to insn constraint.
5804 (*btr<mode>_mask): Ditto.
5806 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
5808 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
5809 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
5811 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5814 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
5815 TARGET_NEON and TARGET_IWMMXT.
5816 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
5817 TARGET_NEON and TARGET_IWMMXT.
5818 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
5820 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5822 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
5823 (ix86_rewrite_tls_address): Ditto.
5824 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
5825 (ix86_rewrite_tls_address_1): Ditto.
5826 (ix86_rewrite_tls_address): Ditto.
5827 * config/i386/predicates.md (tls_address_pattern): New predicate.
5828 * config/i386/i386.md (TLS address splitter): New splitter.
5830 2017-09-04 Richard Biener <rguenther@suse.de>
5832 PR tree-optimization/82084
5833 * fold-const.h (can_native_encode_string_p): Declare.
5834 * fold-const.c (can_native_encode_string_p): Factor out from ...
5835 (native_encode_string): ... here.
5836 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
5837 vectorizing stores from constants we later cannot handle.
5839 2017-09-04 Marek Polacek <polacek@redhat.com>
5842 * doc/invoke.texi: Update -Wtautological-compare documentation.
5844 2017-09-04 Jeff Law <law@redhat.com>
5846 PR tree-optimization/64910
5847 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
5848 swap the first and last operand if the last is a constant.
5850 2017-09-04 Marek Polacek <polacek@redhat.com>
5853 * convert.c (do_narrow): When sanitizing signed integer overflows,
5854 bail out for signed types.
5855 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
5857 2017-09-04 Richard Biener <rguenther@suse.de>
5859 PR tree-optimization/82060
5860 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5861 Move devirtualization after stmt folding and before EH/AB/noreturn
5862 cleanup to get the stmt refs canonicalized. Use a bool instead
5863 of gimple_modified_p since that doesn't work for NOPs. Schedule
5864 NOPs generated by folding for removal.
5866 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5867 Alan Hayward <alan.hayward@arm.com>
5868 David Sherwood <david.sherwood@arm.com>
5870 * coretypes.h (pad_direction): New enum.
5871 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
5872 (FUNCTION_ARG_PADDING): Likewise.
5873 * target.def (function_arg_padding): New hook.
5874 * targhooks.h (default_function_arg_padding): Declare.
5875 * targhooks.c (default_function_arg_padding): New function.
5876 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
5877 (TARGET_FUNCTION_ARG_PADDING): ...this.
5878 * doc/tm.texi: Regenerate.
5879 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
5880 instead of direction.
5881 (compute_argument_addresses): Likewise.
5882 (load_register_parameters): Likewise.
5883 (emit_library_call_value_1): Likewise.
5884 (store_one_arg): Use targetm.calls.function_arg_padding instead
5885 of FUNCTION_ARG_PADDING.
5886 (must_pass_in_stack_var_size_or_pad): Likewise.
5887 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
5888 (emit_group_store): Likewise.
5889 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
5890 instead of FUNCTION_ARG_PADDING.
5891 (emit_push_insn): Likewise, and propagate enum change throughout
5893 * function.h (direction): Delete.
5894 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
5896 * function.c (assign_parm_find_stack_rtl): Likewise.
5897 (assign_parm_setup_block_p): Likewise.
5898 (assign_parm_setup_block): Likewise.
5899 (gimplify_parameters): Likewise.
5900 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
5901 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
5903 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
5904 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5905 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
5906 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
5907 (aarch64_function_arg_padding): ...this new function.
5908 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
5909 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5910 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
5911 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5912 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
5913 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5914 (arm_pad_arg_upward): Replace with...
5915 (arm_function_arg_padding): ...this new function.
5916 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
5918 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
5919 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
5920 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5921 (ia64_hpux_function_arg_padding): Replace with...
5922 (ia64_function_arg_padding): ...this new function. Use pad_direction
5923 instead of direction. Check for TARGET_HPUX.
5924 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
5925 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5926 (iq2000_function_arg_padding): New function.
5927 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
5928 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
5929 (mips_function_arg_padding): ...this new function.
5930 (mips_pad_reg_upward): Update accordingly.
5931 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5932 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
5933 targetm.calls.function_arg_padding.
5934 (FUNCTION_ARG_PADDING): Delete.
5935 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5936 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
5937 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5938 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
5939 (nios2_block_reg_padding): Return pad_direction instead of direction.
5940 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
5941 instead of direction.
5942 (nios2_function_arg_padding): Likewise. Make static.
5943 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5944 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
5945 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
5946 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
5947 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5948 (pa_function_arg_padding): Make static. Return pad_direction instead
5950 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
5951 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5952 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
5953 instead of direction. Use targetm.calls.function_arg_padding.
5954 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
5955 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5956 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
5957 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
5958 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
5960 (function_arg_padding): Rename to...
5961 (rs6000_function_arg_padding): ...this. Make static. Return
5962 pad_direction instead of direction.
5963 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5964 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
5965 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5966 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
5967 instead of direction. Use targetm.calls.function_arg_padding.
5968 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
5969 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5970 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
5971 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
5972 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5973 (function_arg_padding): Rename to...
5974 (rs6000_function_arg_padding): ...this. Make static. Return
5975 pad_direction instead of direction.
5976 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5977 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
5978 * config/s390/s390.c (s390_function_arg_padding): New function.
5979 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5980 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
5981 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
5982 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5983 (function_arg_padding): Rename to...
5984 (sparc_function_arg_padding): ...this. Make static. Return
5985 pad_direction instead of direction.
5986 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
5987 * config/spu/spu.c (spu_function_arg_padding): New function.
5988 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5989 * system.h (FUNCTION_ARG_PADDING): Poison.
5991 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5992 Alan Hayward <alan.hayward@arm.com>
5993 David Sherwood <david.sherwood@arm.com>
5995 * target.def (modes_tieable_p): New hook.
5996 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
5997 (TARGET_MODES_TIEABLE_P): ...this.
5998 * doc/tm.texi.in: Regenerate.
5999 * hooks.h (hook_bool_mode_mode_true): Declare.
6000 * hooks.c (hook_bool_mode_mode_true): New function.
6001 * combine.c (subst): Use targetm.modes_tieable_p instead of
6003 * dse.c (find_shift_sequence): Likewise.
6004 * expmed.c (extract_low_bits): Likewise.
6005 * lower-subreg.c: Include target.h.
6006 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
6008 * rtlanal.c (rtx_cost): Likewise.
6009 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
6010 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
6011 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
6012 (TARGET_MODES_TIEABLE_P): Redefine.
6013 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
6014 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
6015 (TARGET_MODES_TIEABLE_P): Redefine.
6016 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
6017 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
6018 (arc_modes_tieable_p): New function.
6019 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
6020 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
6021 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
6022 (arm_modes_tieable_p): Make static.
6023 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
6024 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
6025 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
6026 (TARGET_MODES_TIEABLE_P): Redefine.
6027 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
6028 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
6029 (TARGET_MODES_TIEABLE_P): Redefine.
6030 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
6031 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
6032 (cr16_modes_tieable_p): New function.
6033 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
6034 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
6035 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
6036 (TRULY_NOOP_TRUNCATION): Update comment.
6037 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
6038 (TRULY_NOOP_TRUNCATION): Update comment.
6039 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
6040 (frv_modes_tieable_p): New function.
6041 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
6042 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
6043 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
6044 (TARGET_MODES_TIEABLE_P): Redefine.
6045 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
6046 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
6047 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
6048 (TARGET_MODES_TIEABLE_P): Redefine.
6049 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
6050 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
6051 (ia64_modes_tieable_p): New function.
6052 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
6053 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
6054 (iq2000_modes_tieable_p): New function.
6055 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
6056 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
6057 (lm32_modes_tieable_p): New function.
6058 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
6059 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
6060 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
6061 (TARGET_MODES_TIEABLE_P): Redefine.
6062 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
6063 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
6064 (m32r_modes_tieable_p): New function.
6065 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6066 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6067 (m68k_modes_tieable_p): New function.
6068 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6069 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6070 (mcore_modes_tieable_p): New function.
6071 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6072 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6074 (TARGET_MODES_TIEABLE_P): Redefine.
6075 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6076 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6077 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6078 (TARGET_MODES_TIEABLE_P): Redefine.
6079 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6080 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6081 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6082 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6083 (mn10300_modes_tieable_p): ...this and make static.
6084 (TARGET_MODES_TIEABLE_P): Redefine.
6085 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6086 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6087 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6088 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6089 (msp430_modes_tieable_p): Make static.
6090 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6091 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6092 (TARGET_MODES_TIEABLE_P): Redefine.
6093 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6094 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6095 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6096 (TARGET_MODES_TIEABLE_P): Redefine.
6097 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6098 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6099 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6100 (TARGET_MODES_TIEABLE_P): Redefine.
6101 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6102 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6103 (pdp11_modes_tieable_p): New function.
6104 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6105 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6106 (rs6000_modes_tieable_p): New function.
6107 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6108 * config/powerpcspe/powerpcspe.md: Update comment.
6109 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6110 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6111 (TARGET_MODES_TIEABLE_P): Redefine.
6112 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6113 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6114 (rl78_modes_tieable_p): New function.
6115 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6116 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6117 (rs6000_modes_tieable_p): New function.
6118 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6119 * config/rs6000/rs6000.md: Update comment.
6120 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6121 * config/rx/rx.c (rx_modes_tieable_p): New function.
6122 (TARGET_MODES_TIEABLE_P): Redefine.
6123 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6124 * config/s390/s390.c (s390_modes_tieable_p): New function.
6125 (TARGET_MODES_TIEABLE_P): Redefine.
6126 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6127 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6128 (sh_modes_tieable_p): New function.
6129 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6130 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6131 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6132 (sparc_modes_tieable_p): Make static.
6133 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6134 * config/spu/spu.c (spu_modes_tieable_p): New function.
6135 (TARGET_MODES_TIEABLE_P): Redefine.
6136 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6137 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6138 (TARGET_MODES_TIEABLE_P): Redefine.
6139 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6140 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6141 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6142 * config/v850/v850.c (v850_modes_tieable_p): New function.
6143 (TARGET_MODES_TIEABLE_P): Redefine.
6144 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
6145 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
6146 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
6147 (visium_modes_tieable_p): New function.
6148 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
6149 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
6150 (xtensa_modes_tieable_p): New function.
6151 * system.h (MODES_TIEABLE_P): Poison.
6153 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6154 Alan Hayward <alan.hayward@arm.com>
6155 David Sherwood <david.sherwood@arm.com>
6157 * target.def (hard_regno_mode_ok): New hook.
6158 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
6159 (TARGET_HARD_REGNO_MODE_OK): ...this.
6160 * doc/tm.texi.in: Regenerate.
6161 * hooks.h (hook_bool_uint_mode_true): Declare.
6162 * hooks.c (hook_bool_uint_mode_true): New function.
6163 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
6165 * genpreds.c (write_insn_preds_c): Add an include of target.h.
6166 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
6167 instead of HARD_REGNO_MODE_OK.
6168 * caller-save.c: Include target.h.
6169 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
6171 * combine.c (can_combine_p): Likewise.
6172 (combinable_i3pat): Likewise.
6173 (can_change_dest_mode): Likewise.
6174 * expr.c (init_expr_target): Likewise.
6175 (convert_move): Likewise.
6176 (convert_modes): Likewise.
6177 * ira.c (setup_prohibited_class_mode_regs): Likewise.
6178 (setup_prohibited_mode_move_regs): Likewise.
6179 * ira.h (target_ira): Likewise.
6180 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6181 * lra-constraints.c (process_alt_operands): Likewise.
6182 (split_reg): Likewise.
6183 * recog.c (peep2_find_free_register): Likewise.
6184 * ree.c (combine_reaching_defs): Likewise.
6185 * regcprop.c (maybe_mode_change): Likewise.
6186 * reginfo.c (init_reg_sets_1): Likewise.
6187 (choose_hard_reg_mode): Likewise.
6188 (simplifiable_subregs): Likewise.
6189 * regrename.c (check_new_reg_p): Likewise.
6190 * reload.c (find_valid_class): Likewise.
6191 (find_valid_class_1): Likewise.
6192 (reload_inner_reg_of_subreg): Likewise.
6193 (push_reload): Likewise.
6194 (combine_reloads): Likewise.
6195 (find_dummy_reload): Likewise.
6196 (find_reloads): Likewise.
6197 * reload1.c (find_reg): Likewise.
6198 (set_reload_reg): Likewise.
6199 (allocate_reload_reg): Likewise.
6200 (choose_reload_regs): Likewise.
6201 (reload_adjust_reg_for_temp): Likewise.
6202 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
6203 (simplify_subreg_regno): Likewise.
6204 * sel-sched.c (init_regs_for_mode): Likewise.
6205 * varasm.c (make_decl_rtl): Likewise.
6206 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
6207 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
6209 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
6210 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
6211 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6212 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
6213 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
6214 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6215 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
6216 (arc_mode_class): Delete.
6217 (HARD_REGNO_MODE_OK): Delete.
6218 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6219 (arc_hard_regno_mode_ok): Rename old array to...
6220 (arc_hard_regno_mode_ok_modes): ...this.
6221 (arc_conditional_register_usage): Update accordingly.
6222 (arc_mode_class): Make static.
6223 (arc_hard_regno_mode_ok): New function.
6224 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
6225 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
6226 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6227 (arm_hard_regno_mode_ok): Make static.
6228 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
6230 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
6231 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
6232 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
6234 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6235 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
6236 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
6237 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
6238 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
6239 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6240 * config/bfin/predicates.md (valid_reg_operand): Use
6241 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
6242 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
6243 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
6244 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6245 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
6246 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
6247 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6248 (cr16_hard_regno_mode_ok): Make static and return a bool.
6249 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
6250 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6251 (cris_hard_regno_mode_ok): New function.
6252 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
6253 (epiphany_mode_class): Delete.
6254 (HARD_REGNO_MODE_OK): Delete.
6255 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
6256 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6257 (hard_regno_mode_ok): Rename to...
6258 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
6259 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
6260 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
6262 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
6263 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
6264 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6265 (frv_hard_regno_mode_ok): Make static and return a bool.
6266 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
6268 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
6269 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
6270 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
6271 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
6273 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6274 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
6275 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
6276 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
6278 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6279 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
6280 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6281 (ia64_hard_regno_mode_ok): New function.
6282 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
6283 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6284 (iq2000_hard_regno_mode_ok): New function.
6285 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
6286 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6287 (lm32_hard_regno_mode_ok): New function.
6288 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
6289 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
6290 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
6291 instead of HARD_REGNO_MODE_OK.
6292 (m32c_hard_regno_ok): Rename to...
6293 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
6294 (m32c_cannot_change_mode_class): Update accordingly.
6295 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6296 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
6297 (m32r_mode_class): Delete.
6298 (HARD_REGNO_MODE_OK): Delete.
6299 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6300 (m32r_hard_regno_mode_ok): Rename to...
6301 (m32r_hard_regno_modes): ...this.
6302 (m32r_mode_class): Make static.
6303 (m32r_hard_regno_mode_ok): New function.
6304 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
6305 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
6306 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6307 (m68k_hard_regno_mode_ok): Make static.
6308 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
6309 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6310 (mcore_hard_regno_mode_ok): New function.
6311 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
6312 (HARD_REGNO_MODE_OK): Delete.
6313 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
6315 (microblaze_hard_regno_mode_ok_p): ...this and make static.
6316 (microblaze_hard_regno_mode_ok): New function.
6317 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6318 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
6319 (mips_hard_regno_mode_ok): Delete.
6320 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
6321 (mips_hard_regno_mode_ok_p): ...this and make static.
6322 (mips_hard_regno_mode_ok_p): Rename to...
6323 (mips_hard_regno_mode_ok_uncached): ...this.
6324 (mips_hard_regno_mode_ok): New function.
6325 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
6326 of HARD_REGNO_MODE_OK.
6327 (mips_option_override): Update after above name changes.
6328 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6329 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
6330 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
6331 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
6332 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
6333 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6334 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
6335 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
6336 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
6337 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6338 (msp430_hard_regno_mode_ok): Make static and return a bool.
6339 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
6340 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
6341 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
6343 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6344 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
6345 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
6346 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
6347 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
6348 (PA_HARD_REGNO_MODE_OK): ...this
6349 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
6350 (PA_HARD_REGNO_MODE_OK): ...this.
6351 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6352 (pa_hard_regno_mode_ok): New function.
6353 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
6354 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6355 (pdp11_hard_regno_mode_ok): New function.
6356 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
6357 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
6359 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
6361 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6362 (rs6000_hard_regno_mode_ok): Rename to...
6363 (rs6000_hard_regno_mode_ok_uncached): ...this.
6364 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6365 (rs6000_hard_regno_mode_ok): New function.
6366 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
6367 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
6368 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
6369 (riscv_hard_regno_mode_ok): ...this and make static.
6370 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6371 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
6372 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
6373 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6374 (rl78_hard_regno_mode_ok): Make static and return bool.
6375 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
6376 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
6378 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
6379 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6380 (rs6000_hard_regno_mode_ok): Rename to...
6381 (rs6000_hard_regno_mode_ok_uncached): ...this.
6382 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6383 (rs6000_hard_regno_mode_ok): New function.
6384 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
6385 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
6386 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6387 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
6388 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
6389 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
6390 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6391 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
6392 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
6393 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6394 (sh_hard_regno_mode_ok): Make static.
6395 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
6396 instead of HARD_REGNO_MODE_OK.
6397 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
6398 (sparc_mode_class): Delete.
6399 (HARD_REGNO_MODE_OK): Delete.
6400 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6401 (hard_regno_mode_classes): Make static.
6402 (sparc_mode_class): Likewise.
6403 (sparc_hard_regno_mode_ok): New function.
6404 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
6405 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
6406 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
6408 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6409 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
6410 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
6411 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
6412 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
6413 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6414 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
6415 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
6416 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6417 (visium_hard_regno_mode_ok): New function.
6418 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
6419 instead of HARD_REGNO_MODE_OK.
6420 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
6421 (HARD_REGNO_MODE_OK): Delete.
6422 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
6423 (xtensa_hard_regno_mode_ok_p): ...this and make static.
6424 (xtensa_option_override): Update accordingly.
6425 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6426 (xtensa_hard_regno_mode_ok): New function.
6427 * system.h (HARD_REGNO_MODE_OK): Poison.
6429 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6430 Alan Hayward <alan.hayward@arm.com>
6431 David Sherwood <david.sherwood@arm.com>
6433 * target.def (hard_regno_call_part_clobbered): New hook.
6434 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
6435 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
6436 * doc/tm.texi: Regenerate.
6437 * hooks.h (hook_bool_uint_mode_false): Declare.
6438 * hooks.c (hook_bool_uint_mode_false): New function.
6439 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6440 * cselib.c (cselib_process_insn): Use
6441 targetm.hard_regno_call_part_clobbered instead of
6442 HARD_REGNO_CALL_PART_CLOBBERED.
6443 * ira-conflicts.c (ira_build_conflicts): Likewise.
6444 * ira-costs.c (ira_tune_allocno_costs): Likewise.
6445 * lra-constraints.c (need_for_call_save_p): Likewise.
6446 * lra-lives.c: Include target.h.
6447 (check_pseudos_live_through_calls): Use
6448 targetm.hard_regno_call_part_clobbered instead of
6449 HARD_REGNO_CALL_PART_CLOBBERED.
6450 * regcprop.c: Include target.h.
6451 (copyprop_hardreg_forward_1): Use
6452 targetm.hard_regno_call_part_clobbered instead of
6453 HARD_REGNO_CALL_PART_CLOBBERED.
6454 * reginfo.c (choose_hard_reg_mode): Likewise.
6455 * regrename.c (check_new_reg_p): Likewise.
6456 * reload.c (find_equiv_reg): Likewise.
6457 * reload1.c (emit_reload_insns): Likewise.
6458 * sched-deps.c (deps_analyze_insn): Likewise.
6459 * sel-sched.c (init_regs_for_mode): Likewise.
6460 (mark_unavailable_hard_regs): Likewise.
6461 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
6462 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6463 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
6465 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6466 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6467 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
6469 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
6471 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6472 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6473 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
6475 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6476 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6477 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
6479 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6480 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
6482 * config/powerpcspe/powerpcspe.c
6483 (rs6000_hard_regno_call_part_clobbered): New function.
6484 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6485 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6486 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
6488 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6489 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6490 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
6492 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6493 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6494 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
6496 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6497 Alan Hayward <alan.hayward@arm.com>
6498 David Sherwood <david.sherwood@arm.com>
6500 * rtl.h (subreg_memory_offset): Declare.
6501 * emit-rtl.c (subreg_memory_offset): New function.
6502 * expmed.c (store_bit_field_1): Use it.
6503 * expr.c (undefined_operand_subword_p): Likewise.
6504 * simplify-rtx.c (simplify_subreg): Likewise.
6506 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
6508 PR rtl-optimization/57448
6511 * optabs.c (expand_atomic_load): Place compiler memory barriers if
6512 using atomic_load pattern.
6513 (expand_atomic_store): Likewise.
6515 2017-09-04 Jakub Jelinek <jakub@redhat.com>
6518 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
6519 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
6520 handling. Use replace_call_with_value with NULL instead of
6521 gsi_replace, unlink_stmt_vdef and release_defs.
6523 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
6526 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
6528 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6531 * rtl.h (emit_library_call_value_1): Declare.
6532 (emit_library_call): Replace declaration with a series of overloads.
6533 Remove the parameter count argument.
6534 (emit_library_call_value): Likewise.
6535 * calls.c (emit_library_call_value_1): Make global. Replace varargs
6536 with an "rtx_mode_t *".
6537 (emit_library_call_value): Delete.
6538 (emit_library_call): Likewise.
6539 * asan.c (asan_emit_stack_protection): Update calls accordingly.
6540 (asan_emit_allocas_unpoison): Likewise.
6541 * builtins.c (expand_builtin_powi): Likewise.
6542 (expand_asan_emit_allocas_unpoison): Likewise.
6543 * cfgexpand.c (expand_main_function): Likewise.
6544 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
6545 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
6546 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
6547 * config/arm/arm.c (arm_trampoline_init): Likewise.
6548 (arm_call_tls_get_addr): Likewise.
6549 (arm_expand_divmod_libfunc): Likewise.
6550 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
6551 (smulsi3_highpart): Likewise.
6552 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
6553 (c6x_expand_compare): Likewise.
6554 (c6x_expand_movmem): Likewise.
6555 * config/frv/frv.c (frv_trampoline_init): Likewise.
6556 * config/i386/i386.c (ix86_trampoline_init): Likewise.
6557 (ix86_expand_divmod_libfunc): Likewise.
6558 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
6559 (ia64_expand_compare): Likewise.
6560 (ia64_profile_hook): Likewise.
6561 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
6562 (nonlocal_goto): Likewise.
6563 (restore_stack_nonlocal): Likewise.
6564 * config/m32r/m32r.c (block_move_call): Likewise.
6565 (m32r_trampoline_init): Likewise.
6566 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
6567 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
6568 (m68k_call_m68k_read_tp): Likewise.
6569 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
6570 (microblaze_expand_divide): Likewise.
6571 * config/mips/mips.h (mips_args): Likewise.
6572 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
6573 (MIPS_ICACHE_SYNC): Likewise.
6574 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
6575 (nios2_trampoline_init): Likewise.
6576 * config/pa/pa.c (hppa_tls_call): Likewise.
6577 (pa_trampoline_init): Likewise.
6578 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
6579 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
6580 (expand_strn_compare): Likewise.
6581 (rs6000_generate_compare): Likewise.
6582 (rs6000_expand_float128_convert): Likewise.
6583 (output_profile_hook): Likewise.
6584 (rs6000_trampoline_init): Likewise.
6585 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
6586 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
6587 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
6588 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
6589 (rs6000_generate_compare): Likewise.
6590 (rs6000_expand_float128_convert): Likewise.
6591 (output_profile_hook): Likewise.
6592 (rs6000_trampoline_init): Likewise.
6593 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
6594 * config/sh/sh.c (sh_trampoline_init): Likewise.
6595 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
6596 (sparc_emit_float_lib_cmp): Likewise.
6597 (sparc32_initialize_trampoline): Likewise.
6598 (sparc64_initialize_trampoline): Likewise.
6599 (sparc_profile_hook): Likewise.
6600 * config/spu/spu.c (ea_load_store): Likewise.
6601 * config/spu/spu.md (floatunssidf2): Likewise.
6602 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
6603 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
6604 * config/visium/visium.c (expand_block_move_4): Likewise.
6605 (expand_block_move_2): Likewise.
6606 (expand_block_move_1): Likewise.
6607 (expand_block_set_4): Likewise.
6608 (expand_block_set_2): Likewise.
6609 (expand_block_set_1): Likewise.
6610 (visium_trampoline_init): Likewise.
6611 (visium_profile_hook): Likewise.
6612 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
6613 (xtensa_setup_frame_addresses): Likewise.
6614 (xtensa_trampoline_init): Likewise.
6615 * except.c (sjlj_emit_function_enter): Likewise.
6616 (sjlj_emit_function_exit): Likewise.
6617 * explow.c (allocate_dynamic_stack_space): Likewise.
6618 (probe_stack_range): Likewise.
6619 * expr.c (convert_mode_scalar): Likewise.
6620 * optabs.c (expand_binop): Likewise.
6621 (expand_twoval_binop_libfunc): Likewise.
6622 (expand_unop): Likewise.
6623 (prepare_cmp_insn): Likewise.
6624 (prepare_float_lib_cmp): Likewise.
6625 (expand_float): Likewise.
6626 (expand_fix): Likewise.
6627 (expand_fixed_convert): Likewise.
6628 (maybe_emit_sync_lock_test_and_set): Likewise.
6629 (expand_atomic_compare_and_swap): Likewise.
6630 (expand_mem_thread_fence): Likewise.
6631 (expand_atomic_fetch_op): Likewise.
6633 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
6635 * doc/generic.texi (OpenACC): Adjust URL.
6636 * doc/invoke.texi (C Dialect Options): Ditto.
6638 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
6640 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
6641 predicate for operand 1. Add (m,<S>) constraint.
6642 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
6643 Prevent memory operand 1 with register operand 2.
6645 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
6647 PR rtl-optimization/82024
6648 * combine.c (try_combine): If the combination result is a PARALLEL,
6649 and we only need to retain the SET in there that would be placed
6650 at I2, check that we can place that at I3 instead, before doing so.
6652 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6655 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
6657 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
6658 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
6661 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
6662 Jeff Law <law@redhat.com>
6664 * varasm.c (bss_initializer_p): Do not put constants into .bss
6665 (categorize_decl_for_section): Handle bss_initializer_p returning
6666 false when DECL_INITIAL is NULL.
6668 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6671 * config/s390/s390.c (s390_can_inline_p): New function.
6673 2017-09-01 Jeff Law <law@redhat.com>
6675 PR tree-optimization/82052
6676 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
6677 Always initialize the returned slot after a hash table miss
6678 when INSERT is true.
6680 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
6682 * config/s390/s390.md (mem_signal_fence): Remove.
6683 * doc/md.texi (mem_signal_fence): Remove.
6684 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
6686 * target-insns.def (mem_signal_fence): Remove.
6688 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6691 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
6694 * asan.c (create_odr_indicator): Strip name encoding from assembler
6695 name before appending it after __odr_asan_.
6697 2017-09-01 Martin Liska <mliska@suse.cz>
6699 PR tree-optimization/82059
6700 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
6701 frequency only when an edge is redirected.
6703 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6705 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
6706 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
6707 (arc_conditional_register_usage): Remove ARC600 lp_count
6709 (arc_file_start): Emit Tag_ARC_CPU_variation.
6710 (arc_can_use_doloop_p): New conditions to use ZOLs.
6711 (hwloop_fail): New function.
6712 (hwloop_optimize): Likewise.
6713 (hwloop_pattern_reg): Likewise.
6714 (arc_doloop_hooks): New struct, to be used with reorg_loops.
6715 (arc_reorg_loops): New function, calls reorg_loops.
6716 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
6717 (arc600_corereg_hazard): Remove ZOL checking, case handled by
6719 (arc_loop_hazard): Remove function, functionality moved into
6721 (arc_hazard): Remove arc_loop_hazard call.
6722 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
6723 into hwloop_optimize.
6724 (arc_label_align): Remove ZOL handling.
6725 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
6726 * config/arc/arc.md (doloop_begin): Remove pattern.
6727 (doloop_begin_i): Likewise.
6728 (doloop_end_i): Likewise.
6729 (doloop_fallback): Likewise.
6730 (doloop_fallback_m): Likewise.
6731 (doloop_end): Reimplement expand.
6732 (arc_lp): New pattern for LP instruction.
6733 (loop_end): New pattern.
6734 (loop_fail): Likewise.
6735 (decrement_and_branch_until_zero): Likewise.
6736 * config/arc/arc.opt (mlpc-width): New option.
6737 * doc/invoke.texi (mlpc-width): Document option.
6739 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6741 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
6742 (arc_ccfsm_advance): Fix checking for delay slots.
6743 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
6745 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6747 * config/arc/arc.md (movqi_insn): Add stores to save constant long
6749 (movhi_insn): Update store instruction constraint which are saving
6750 6-bit short immediates.
6751 (movsi_insn): Consider also short scaled load operations.
6752 (zero_extendhisi2_i): Use Usd constraint instead of T.
6753 (extendhisi2_i): Add q constraint.
6754 (arc_clzsi2): Add type and length attributes.
6755 (arc_ctzsi2): Likewise.
6756 * config/arc/constraints.md (Usc): Update constraint, the
6757 assembler can parse two relocations for a single instruction.
6759 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6761 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
6762 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
6764 2017-08-31 Olivier Hainque <hainque@adacore.com>
6766 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
6767 match as powerpc-wrs-vxworks*.
6769 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
6771 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
6772 register constraint for by-element operand.
6773 (aarch64_mls_elt_merge<mode>): Likewise.
6775 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6777 * config/arc/arc.c (arc_can_follow_jump): Check for short
6780 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6782 * config.gcc: Use g.opt for arc.
6783 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
6784 functionality moved to ...
6785 (legitimate_scaled_address_p): New function, ...here.
6786 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
6787 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
6788 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
6790 (arc_override_options): Handle G option.
6791 (arc_output_pic_addr_const): Correct function definition.
6792 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
6793 (arc_decl_anon_ns_mem_p): Delete.
6794 (arc_in_small_data_p): Overhaul this function to take into
6795 consideration the value given via G option.
6796 (arc_rewrite_small_data_1): Renamed and corrected old
6797 arc_rewrite_small_data function.
6798 (arc_rewrite_small_data): New function.
6799 (small_data_pattern): Don't use pic_offset_table_rtx.
6800 * config/arc/arc.h (CC1_SPEC): Recognize G option.
6801 * config/arc/simdext.md (movmisalignv2hi): Use
6802 prepare_move_operands function.
6804 (movmisalign*): Likewise.
6805 * doc/invoke.texi (ARC options): Document -G option.
6807 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6809 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
6811 * config/arc/arc.c (arc_print_operand): Output scalled address for
6812 sdata whenever is possible.
6813 (arc_in_small_data_p): Allow sdata for 64bit datum when double
6814 load/stores are available.
6815 (compact_sda_memory_operand): Check for the alignment required by
6816 code density instructions.
6817 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
6819 * config/arc/constraints.md (Usd): Update constraint.
6820 (Us0): New constraint.
6821 (Usc): Update constraint.
6823 2017-08-31 Richard Biener <rguenther@suse.de>
6826 * dwarf2out.c (dwarf2out_early_global_decl): Process each
6829 2017-08-31 Tamar Christina <tamar.christina@arm.com>
6831 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
6832 Resize type_signature.
6834 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6835 Alan Hayward <alan.hayward@arm.com>
6836 David Sherwood <david.sherwood@arm.com>
6838 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
6839 subregs whose inner modes can be stored in GPRs.
6840 (aarch64_classify_index): Likewise.
6842 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6843 Alan Hayward <alan.hayward@arm.com>
6844 David Sherwood <david.sherwood@arm.com>
6846 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6847 (V_INT_EQUIV): ...this.
6848 (v_cmp_result): Rename to...
6849 (v_int_equiv): ...this.
6850 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
6851 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
6852 (copysign<mode>3): Likewise.
6853 (aarch64_simd_bsl<mode>_internal): Likewise.
6854 (aarch64_simd_bsl<mode>): Likewise.
6855 (vec_cmp<mode><mode>): Likewise.
6856 (vcond<mode><mode>): Likewise.
6857 (vcond<v_cmp_mixed><mode>): Likewise.
6858 (vcondu<mode><v_cmp_mixed>): Likewise.
6859 (aarch64_cm<optab><mode>): Likewise.
6860 (aarch64_cmtst<mode>): Likewise.
6861 (aarch64_fac<optab><mode>): Likewise.
6862 (vec_perm_const<mode>): Likewise.
6863 (vcond_mask_<mode><v_cmp_result>): Rename to...
6864 (vcond_mask_<mode><v_int_equiv>): ...this.
6865 (vec_cmp<mode><v_cmp_result>): Rename to...
6866 (vec_cmp<mode><v_int_equiv>): ...this.
6868 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6869 Alan Hayward <alan.hayward@arm.com>
6870 David Sherwood <david.sherwood@arm.com>
6872 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
6874 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
6875 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
6876 (UNSPEC_LD4_DREG): New unspecs.
6877 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
6878 (aarch64_ld2<mode>_dreg_be): Replace with...
6879 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
6881 (aarch64_ld3<mode>_dreg_le)
6882 (aarch64_ld3<mode>_dreg_be): Replace with...
6883 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
6885 (aarch64_ld4<mode>_dreg_le)
6886 (aarch64_ld4<mode>_dreg_be): Replace with...
6887 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
6890 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6892 PR tree-optimization/81987
6893 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
6894 insert an initializer in a location not dominated by the stride
6897 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6899 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
6900 on the entire header of the finally block in the fallthru case.
6902 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6904 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
6906 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
6908 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
6909 rs6000_emit_move_from_cr and call renamed function.
6910 (rs6000_emit_prologue): Call renamed functions.
6911 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
6912 movesi_from_cr, remove volatile CRs.
6914 2017-08-30 Jon Beniston <jon@beniston.com>
6915 Richard Biener <rguenther@suse.de>
6917 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
6918 of VECTOR_MODE_P check.
6919 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
6920 element vector types.
6922 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6924 * df.h (df_read_modify_subreg_p): Remove in favor of...
6925 * rtl.h (read_modify_subreg_p): ...this new function. Take a
6926 const_rtx instead of an rtx.
6927 * cprop.c (local_cprop_find_used_regs): Update accordingly.
6928 * df-problems.c (df_word_lr_mark_ref): Likewise.
6929 * ira-lives.c (mark_pseudo_reg_live): Likewise.
6930 (mark_pseudo_reg_dead): Likewise.
6931 (mark_ref_dead): Likewise.
6932 * reginfo.c (init_subregs_of_mode): Likewise.
6933 * sched-deps.c (sched_analyze_1): Likewise.
6934 * df-scan.c (df_def_record_1): Likewise.
6935 (df_uses_record): Likewise.
6936 (df_read_modify_subreg_p): Remove in favor of...
6937 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
6938 const_rtx instead of an rtx.
6940 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6941 Alan Hayward <alan.hayward@arm.com>
6942 David Sherwood <david.sherwood@arm.com>
6944 * rtl.h (partial_subreg_p): New function.
6945 * caller-save.c (save_call_clobbered_regs): Use it.
6946 * calls.c (expand_call): Likewise.
6947 * combine.c (combinable_i3pat): Likewise.
6948 (simplify_set): Likewise.
6949 (make_extraction): Likewise.
6950 (make_compound_operation_int): Likewise.
6951 (gen_lowpart_or_truncate): Likewise.
6952 (force_to_mode): Likewise.
6953 (make_field_assignment): Likewise.
6954 (reg_truncated_to_mode): Likewise.
6955 (record_truncated_value): Likewise.
6956 (move_deaths): Likewise.
6957 * cse.c (record_jump_cond): Likewise.
6958 (cse_insn): Likewise.
6959 * cselib.c (cselib_lookup_1): Likewise.
6960 * expmed.c (extract_bit_field_using_extv): Likewise.
6961 * function.c (assign_parm_setup_reg): Likewise.
6962 * ifcvt.c (noce_convert_multiple_sets): Likewise.
6963 * ira-build.c (create_insn_allocnos): Likewise.
6964 * lra-coalesce.c (merge_pseudos): Likewise.
6965 * lra-constraints.c (match_reload): Likewise.
6966 (simplify_operand_subreg): Likewise.
6967 (curr_insn_transform): Likewise.
6968 * lra-lives.c (process_bb_lives): Likewise.
6969 * lra.c (new_insn_reg): Likewise.
6970 (lra_substitute_pseudo): Likewise.
6971 * regcprop.c (mode_change_ok): Likewise.
6972 (maybe_mode_change): Likewise.
6973 (copyprop_hardreg_forward_1): Likewise.
6974 * reload.c (push_reload): Likewise.
6975 (find_reloads): Likewise.
6976 (find_reloads_subreg_address): Likewise.
6977 * reload1.c (alter_reg): Likewise.
6978 (eliminate_regs_1): Likewise.
6979 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6981 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
6983 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
6984 back to if statements, including unpack.
6986 2017-08-30 Martin Liska <mliska@suse.cz>
6989 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
6991 (func_checker::compare_asm_inputs_outputs): ... this function.
6992 (func_checker::compare_gimple_asm): Use the function to compare
6993 also ASM constrains.
6994 * ipa-icf-gimple.h: Rename the function.
6996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6997 Alan Hayward <alan.hayward@arm.com>
6998 David Sherwood <david.sherwood@arm.com>
7000 * coretypes.h (complex_mode): New type.
7001 * gdbhooks.py (build_pretty_printer): Handle it.
7002 * machmode.h (complex_mode): New class.
7003 (complex_mode::includes_p): New function.
7004 (is_complex_int_mode): Likewise.
7005 (is_complex_float_mode): Likewise.
7006 * genmodes.c (get_mode_class): Handle complex mode classes.
7007 * function.c (expand_function_end): Use is_complex_int_mode.
7009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7010 Alan Hayward <alan.hayward@arm.com>
7011 David Sherwood <david.sherwood@arm.com>
7013 * coretypes.h (scalar_mode_pod): New typedef.
7014 * gdbhooks.py (build_pretty_printer): Handle it.
7015 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
7016 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
7017 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
7018 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
7021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7022 Alan Hayward <alan.hayward@arm.com>
7023 David Sherwood <david.sherwood@arm.com>
7025 * machmode.h (mode_for_vector): Take a scalar_mode instead
7027 * stor-layout.c (mode_for_vector): Likewise.
7028 * explow.c (promote_mode): Use as_a <scalar_mode>.
7029 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
7031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7032 Alan Hayward <alan.hayward@arm.com>
7033 David Sherwood <david.sherwood@arm.com>
7035 * target.def (preferred_simd_mode): Take a scalar_mode
7036 instead of a machine_mode.
7037 * targhooks.h (default_preferred_simd_mode): Likewise.
7038 * targhooks.c (default_preferred_simd_mode): Likewise.
7039 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
7040 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
7041 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
7042 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
7043 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
7044 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
7045 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
7046 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
7048 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
7049 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
7050 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
7051 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
7052 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
7053 * doc/tm.texi: Regenerate.
7054 * optabs-query.c (can_vec_mask_load_store_p): Return false for
7057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7058 Alan Hayward <alan.hayward@arm.com>
7059 David Sherwood <david.sherwood@arm.com>
7061 * target.def (scalar_mode_supported_p): Take a scalar_mode
7062 instead of a machine_mode.
7063 * targhooks.h (default_scalar_mode_supported_p): Likewise.
7064 * targhooks.c (default_scalar_mode_supported_p): Likewise.
7065 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7066 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7067 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7068 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7069 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7070 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7071 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7072 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7073 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7074 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7075 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7076 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7078 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7079 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7080 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7081 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7082 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7084 * doc/tm.texi: Regenerate.
7086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7087 Alan Hayward <alan.hayward@arm.com>
7088 David Sherwood <david.sherwood@arm.com>
7090 * coretypes.h (opt_scalar_mode): New typedef.
7091 * gdbhooks.py (build_pretty_printers): Handle it.
7092 * machmode.h (mode_iterator::get_2xwider): Add overload for
7094 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7096 * expr.c (convert_mode_scalar): Likewise.
7097 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7098 * optabs.c (expand_float): Likewise.
7099 (expand_fix): Likewise.
7100 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7103 Alan Hayward <alan.hayward@arm.com>
7104 David Sherwood <david.sherwood@arm.com>
7106 * optabs.c (expand_float): Explicitly check for scalars before
7107 using a branching expansion.
7108 (expand_fix): Likewise.
7110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7111 Alan Hayward <alan.hayward@arm.com>
7112 David Sherwood <david.sherwood@arm.com>
7114 * expr.c (convert_mode): Split scalar handling out into...
7115 (convert_mode_scalar): ...this new function. Treat the modes
7118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7119 Alan Hayward <alan.hayward@arm.com>
7120 David Sherwood <david.sherwood@arm.com>
7122 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7124 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7127 Alan Hayward <alan.hayward@arm.com>
7128 David Sherwood <david.sherwood@arm.com>
7130 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7131 rather than a machine_mode.
7132 (fixed_from_string): Likewise.
7133 (fixed_convert): Likewise.
7134 (fixed_convert_from_int): Likewise.
7135 (fixed_convert_from_real): Likewise.
7136 (real_convert_from_fixed): Likewise.
7137 * fixed-value.c (fixed_from_double_int): Likewise.
7138 (fixed_from_string): Likewise.
7139 (fixed_convert): Likewise.
7140 (fixed_convert_from_int): Likewise.
7141 (fixed_convert_from_real): Likewise.
7142 (real_convert_from_fixed): Likewise.
7143 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
7145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7146 Alan Hayward <alan.hayward@arm.com>
7147 David Sherwood <david.sherwood@arm.com>
7149 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
7150 of separate mode class checks. Do not allow vector modes here.
7151 (immed_wide_int_const): Use as_a <scalar_mode>.
7152 * explow.c (trunc_int_for_mode): Likewise.
7153 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
7154 (wi::shwi): Likewise.
7155 (wi::min_value): Likewise.
7156 (wi::max_value): Likewise.
7157 * dwarf2out.c (loc_descriptor): Likewise.
7158 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
7161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7162 Alan Hayward <alan.hayward@arm.com>
7163 David Sherwood <david.sherwood@arm.com>
7165 * tree.h (SCALAR_TYPE_MODE): New macro.
7166 * expr.c (expand_expr_addr_expr_1): Use it.
7167 (expand_expr_real_2): Likewise.
7168 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
7169 (fold_convert_const_fixed_from_int): Likewise.
7170 (fold_convert_const_fixed_from_real): Likewise.
7171 (native_encode_fixed): Likewise
7172 (native_encode_complex): Likewise
7173 (native_encode_vector): Likewise.
7174 (native_interpret_fixed): Likewise.
7175 (native_interpret_real): Likewise.
7176 (native_interpret_complex): Likewise.
7177 (native_interpret_vector): Likewise.
7178 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
7179 (simd_clone_adjust_argument_types): Likewise.
7180 (simd_clone_init_simd_arrays): Likewise.
7181 (simd_clone_adjust): Likewise.
7182 * stor-layout.c (layout_type): Likewise.
7183 * tree.c (build_minus_one_cst): Likewise.
7184 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7185 * tree-inline.c (estimate_move_cost): Likewise.
7186 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
7187 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
7188 (vectorizable_reduction): Likewise.
7189 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
7190 (vect_recog_mixed_size_cond_pattern): Likewise.
7191 (check_bool_pattern): Likewise.
7192 (adjust_bool_pattern): Likewise.
7193 (search_type_for_mask_1): Likewise.
7194 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7195 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7196 (vectorizable_load): Likewise.
7197 (vectorizable_store): Likewise.
7198 * ubsan.c (ubsan_encode_value): Likewise.
7199 * varasm.c (output_constant): Likewise.
7201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7202 Alan Hayward <alan.hayward@arm.com>
7203 David Sherwood <david.sherwood@arm.com>
7205 * coretypes.h (scalar_mode): New class.
7206 * machmode.h (scalar_mode): Likewise.
7207 (scalar_mode::includes_p): New function.
7208 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
7209 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
7210 * genmodes.c (get_mode_class): Handle remaining scalar modes.
7211 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
7212 * expmed.c (store_bit_field_1): Likewise.
7213 (extract_bit_field_1): Likewise.
7214 * expr.c (write_complex_part): Likewise.
7215 (read_complex_part): Likewise.
7216 (emit_move_complex_push): Likewise.
7217 (expand_expr_real_2): Likewise.
7218 * function.c (assign_parm_setup_reg): Likewise.
7219 (assign_parms_unsplit_complex): Likewise.
7220 * optabs.c (expand_binop): Likewise.
7221 * rtlanal.c (subreg_get_info): Likewise.
7222 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7223 * varasm.c (output_constant_pool_2): Likewise.
7225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7226 Alan Hayward <alan.hayward@arm.com>
7227 David Sherwood <david.sherwood@arm.com>
7229 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7231 (expmed_mult_highpart_optab): Likewise.
7232 (expmed_mult_highpart): Likewise.
7234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7235 Alan Hayward <alan.hayward@arm.com>
7236 David Sherwood <david.sherwood@arm.com>
7238 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
7239 instead of a machine_mode.
7240 (builtin_memset_read_str): Likewise.
7241 * builtins.c (c_readstr): Likewise.
7242 (builtin_memcpy_read_str): Likewise.
7243 (builtin_strncpy_read_str): Likewise.
7244 (builtin_memset_read_str): Likewise.
7245 (builtin_memset_gen_str): Likewise.
7246 (expand_builtin_signbit): Use scalar_int_mode for local variables.
7247 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
7248 instead of a machine_mode.
7249 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
7251 (make_extraction): Likewise.
7252 (try_widen_shift_mode): Take and return scalar_int_modes instead
7254 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
7255 a scalar_int_mode instead of a machine_mode.
7256 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
7257 (avr_addr_space_pointer_mode): Likewise.
7258 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
7259 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
7260 (msp430_unwind_word_mode): Likewise.
7261 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
7262 (spu_addr_space_pointer_mode): Likewise.
7263 (spu_addr_space_address_mode): Likewise.
7264 (spu_libgcc_cmp_return_mode): Likewise.
7265 (spu_libgcc_shift_count_mode): Likewise.
7266 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
7267 (rl78_addr_space_pointer_mode): Likewise.
7268 (fl78_unwind_word_mode): Likewise.
7269 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
7271 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
7272 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
7273 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
7274 (mips_valid_pointer_mode): Likewise.
7275 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
7276 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
7277 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
7279 (ft32_addr_space_address_mode): Likewise.
7280 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
7281 scalar_int_mode instead of a machine_mode.
7282 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
7284 (m32c_addr_space_address_mode): Likewise.
7285 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
7286 (rs6000_eh_return_filter_mode): Likewise.
7287 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
7288 (rs6000_eh_return_filter_mode): Likewise.
7289 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
7290 (s390_libgcc_shift_count_mode): Likewise.
7291 (s390_unwind_word_mode): Likewise.
7292 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
7294 * target.def (mode_rep_extended): Likewise.
7295 (valid_pointer_mode): Likewise.
7296 (addr_space.valid_pointer_mode): Likewise.
7297 (eh_return_filter_mode): Return a scalar_int_mode rather than
7299 (libgcc_cmp_return_mode): Likewise.
7300 (libgcc_shift_count_mode): Likewise.
7301 (unwind_word_mode): Likewise.
7302 (addr_space.pointer_mode): Likewise.
7303 (addr_space.address_mode): Likewise.
7304 * doc/tm.texi: Regenerate.
7305 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
7307 (do_jump): Use scalar_int_mode for local variables.
7308 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
7309 rather than a machine_mode.
7310 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
7311 (scompare_loc_descriptor_wide): Likewise.
7312 (scompare_loc_descriptor_narrow): Likewise.
7313 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
7315 * except.c (sjlj_emit_dispatch_table): Likewise.
7316 (expand_builtin_eh_copy_values): Likewise.
7317 * explow.c (convert_memory_address_addr_space_1): Likewise.
7318 Take a scalar_int_mode rather than a machine_mode.
7319 (convert_memory_address_addr_space): Take a scalar_int_mode rather
7320 than a machine_mode.
7321 (memory_address_addr_space): Use scalar_int_mode for local variables.
7322 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
7323 rather than a machine_mode.
7324 * expmed.c (mask_rtx): Likewise.
7325 (init_expmed_one_conv): Likewise.
7326 (expand_mult_highpart_adjust): Likewise.
7327 (extract_high_half): Likewise.
7328 (expmed_mult_highpart_optab): Likewise.
7329 (expmed_mult_highpart): Likewise.
7330 (expand_smod_pow2): Likewise.
7331 (expand_sdiv_pow2): Likewise.
7332 (emit_store_flag_int): Likewise.
7333 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
7335 (extract_low_bits): Likewise.
7336 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
7338 * expr.c (pieces_addr::adjust): Likewise.
7339 (can_store_by_pieces): Likewise.
7340 (store_by_pieces): Likewise.
7341 (clear_by_pieces_1): Likewise.
7342 (expand_expr_addr_expr_1): Likewise.
7343 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
7344 (expand_expr_real_1): Likewise.
7345 (try_casesi): Likewise.
7346 * final.c (shorten_branches): Likewise.
7347 * fold-const.c (fold_convert_const_int_from_fixed): Change the
7348 type of "mode" to machine_mode.
7349 * internal-fn.c (expand_arith_overflow_result_store): Take a
7350 scalar_int_mode rather than a machine_mode.
7351 (expand_mul_overflow): Use scalar_int_mode for local variables.
7352 * loop-doloop.c (doloop_modify): Likewise.
7353 (doloop_optimize): Likewise.
7354 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
7355 than a machine_mode.
7356 (expand_doubleword_shift_condmove): Likewise.
7357 (expand_doubleword_shift): Likewise.
7358 (expand_doubleword_clz): Likewise.
7359 (expand_doubleword_popcount): Likewise.
7360 (expand_doubleword_parity): Likewise.
7361 (expand_absneg_bit): Use scalar_int_mode for local variables.
7362 (prepare_float_lib_cmp): Likewise.
7363 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
7364 rather than a machine_mode.
7365 (convert_memory_address_addr_space): Likewise.
7366 (get_mode_bounds): Likewise.
7367 (get_address_mode): Return a scalar_int_mode rather than a
7369 * rtlanal.c (get_address_mode): Likewise.
7370 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
7371 than a machine_mode.
7372 * targhooks.c (default_mode_rep_extended): Likewise.
7373 (default_valid_pointer_mode): Likewise.
7374 (default_addr_space_valid_pointer_mode): Likewise.
7375 (default_eh_return_filter_mode): Return a scalar_int_mode rather
7376 than a machine_mode.
7377 (default_libgcc_cmp_return_mode): Likewise.
7378 (default_libgcc_shift_count_mode): Likewise.
7379 (default_unwind_word_mode): Likewise.
7380 (default_addr_space_pointer_mode): Likewise.
7381 (default_addr_space_address_mode): Likewise.
7382 * targhooks.h (default_eh_return_filter_mode): Likewise.
7383 (default_libgcc_cmp_return_mode): Likewise.
7384 (default_libgcc_shift_count_mode): Likewise.
7385 (default_unwind_word_mode): Likewise.
7386 (default_addr_space_pointer_mode): Likewise.
7387 (default_addr_space_address_mode): Likewise.
7388 (default_mode_rep_extended): Take a scalar_int_mode rather than
7390 (default_valid_pointer_mode): Likewise.
7391 (default_addr_space_valid_pointer_mode): Likewise.
7392 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
7394 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
7395 rather than a machine_mode.
7396 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
7397 for local variables.
7398 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7399 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
7400 than a machine_mode.
7402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7403 Alan Hayward <alan.hayward@arm.com>
7404 David Sherwood <david.sherwood@arm.com>
7406 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
7407 the mode argument to scalar_int_mode.
7408 (do_jump_by_parts_zero_rtx): Likewise.
7409 (do_jump_by_parts_equality_rtx): Likewise.
7410 (do_jump_by_parts_greater): Take a mode argument.
7411 (do_jump_by_parts_equality): Likewise.
7412 (do_jump_1): Update calls accordingly.
7414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7415 Alan Hayward <alan.hayward@arm.com>
7416 David Sherwood <david.sherwood@arm.com>
7418 * is-a.h (safe_dyn_cast): New function.
7419 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
7420 (jump_table_for_label): Likewise.
7421 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
7422 instead of an rtx_insn *.
7423 (shorten_branches): Use dyn_cast instead of LABEL_P and
7424 JUMP_TABLE_DATA_P. Use jump_table_for_label and
7425 rtx_jump_table_data::get_data_mode.
7426 (final_scan_insn): Likewise.
7428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7429 Alan Hayward <alan.hayward@arm.com>
7430 David Sherwood <david.sherwood@arm.com>
7432 * combine.c (try_combine): Use is_a <scalar_int_mode> when
7433 trying to combine a full-register integer set with a subreg
7436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7437 Alan Hayward <alan.hayward@arm.com>
7438 David Sherwood <david.sherwood@arm.com>
7440 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
7441 that is always either address_mode or pointer_mode.
7443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7444 Alan Hayward <alan.hayward@arm.com>
7445 David Sherwood <david.sherwood@arm.com>
7447 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
7448 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
7455 is_a <scalar_int_mode> instead of checking for a nonzero
7456 precision. Forcibly convert op_mode to a scalar_int_mode
7457 in that case. More clearly differentiate the operand and
7458 result modes and use the former when deciding what the value
7459 of a count-bits operation should be. Use is_int_mode instead
7460 of checking for a MODE_INT. Remove redundant check for whether
7461 this mode has a zero precision.
7463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7464 Alan Hayward <alan.hayward@arm.com>
7465 David Sherwood <david.sherwood@arm.com>
7467 * optabs.c (widen_leading): Change the type of the mode argument
7468 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
7469 (widen_bswap): Likewise.
7470 (expand_parity): Likewise.
7471 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
7472 (expand_ffs): Likewise.
7473 (epand_unop): Check for scalar integer modes before calling the
7476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7477 Alan Hayward <alan.hayward@arm.com>
7478 David Sherwood <david.sherwood@arm.com>
7480 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
7482 (expand_expr_real_1): Update call accordingly.
7484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7485 Alan Hayward <alan.hayward@arm.com>
7486 David Sherwood <david.sherwood@arm.com>
7488 * expmed.c (store_bit_field_using_insv): Add op0_mode and
7489 value_mode arguments. Use scalar_int_mode internally.
7490 (store_bit_field_1): Rename the new integer mode from imode
7491 to op0_mode and use it instead of GET_MODE (op0). Update calls
7492 to store_split_bit_field, store_bit_field_using_insv and
7493 store_fixed_bit_field.
7494 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
7495 Use scalar_int_mode internally. Use a bit count rather than a mode
7496 when calculating the largest bit size for get_best_mode.
7497 Update calls to store_split_bit_field and store_fixed_bit_field_1.
7498 (store_fixed_bit_field_1): Add mode and value_mode arguments.
7499 Remove assertion that OP0 has a scalar integer mode.
7500 (store_split_bit_field): Add op0_mode and value_mode arguments.
7501 Update calls to extract_fixed_bit_field.
7502 (extract_bit_field_using_extv): Add an op0_mode argument.
7503 Use scalar_int_mode internally.
7504 (extract_bit_field_1): Rename the new integer mode from imode to
7505 op0_mode and use it instead of GET_MODE (op0). Update calls to
7506 extract_split_bit_field, extract_bit_field_using_extv and
7507 extract_fixed_bit_field.
7508 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
7509 to extract_split_bit_field and extract_fixed_bit_field_1.
7510 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
7511 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
7513 (extract_split_bit_field): Add an op0_mode argument. Update call
7514 to extract_fixed_bit_field.
7516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7517 Alan Hayward <alan.hayward@arm.com>
7518 David Sherwood <david.sherwood@arm.com>
7520 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
7521 * explow.c (hard_function_value): Likewise.
7522 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
7523 convert_to_mode call outside the loop.
7524 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
7525 for the mode iterator. Require the mode specified by max_pieces
7527 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
7529 (copy_blkmode_to_reg): Likewise.
7530 (set_storage_via_setmem): Likewise.
7531 * optabs.c (prepare_cmp_insn): Likewise.
7532 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7533 * stor-layout.c (finish_bitfield_representative): Likewise.
7535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7536 Alan Hayward <alan.hayward@arm.com>
7537 David Sherwood <david.sherwood@arm.com>
7539 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
7540 * expr.c (convert_move): Use them.
7541 (convert_modes): Likewise.
7542 (store_expr_with_bounds): Likewise.
7544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7545 Alan Hayward <alan.hayward@arm.com>
7546 David Sherwood <david.sherwood@arm.com>
7548 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
7549 parameter for the mode of "x". Remove the "known_x", "known_mode"
7550 and "known_ret" arguments. Change the type of the mode argument
7552 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
7553 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
7554 (reg_num_sign_bit_copies_for_combine): Likewise.
7555 * rtlanal.c (nonzero_bits1): Likewise.
7556 (num_sign_bit_copies1): Likewise.
7557 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
7558 (reg_num_sign_bit_copies_general): Likewise.
7559 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
7560 (reg_nonzero_bits_general): Likewise.
7562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7563 Alan Hayward <alan.hayward@arm.com>
7564 David Sherwood <david.sherwood@arm.com>
7566 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
7567 than in subroutines. Return 1 for non-integer modes.
7568 (cached_num_sign_bit_copies): Change the type of the mode parameter
7570 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
7571 classes. Handle CONST_INT_P first and then check whether X also
7572 has a scalar integer mode. Check the same thing for inner registers
7573 of a SUBREG and for values that are being extended or truncated.
7575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7576 Alan Hayward <alan.hayward@arm.com>
7577 David Sherwood <david.sherwood@arm.com>
7579 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
7580 in subroutines. Return the mode mask for non-integer modes.
7581 (cached_nonzero_bits): Change the type of the mode parameter
7583 (nonzero_bits1): Likewise. Remove early exit for other mode
7584 classes. Handle CONST_INT_P first and then check whether X
7585 also has a scalar integer mode.
7587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7588 Alan Hayward <alan.hayward@arm.com>
7589 David Sherwood <david.sherwood@arm.com>
7591 * expr.c (widest_int_mode_for_size): Make the comment match the code.
7592 Return a scalar_int_mode and assert that the size is greater than
7594 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
7595 (op_by_pieces_d::op_by_pieces_d): Likewise.
7596 (op_by_pieces_d::run): Likewise.
7597 (can_store_by_pieces): Likewise.
7599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7600 Alan Hayward <alan.hayward@arm.com>
7601 David Sherwood <david.sherwood@arm.com>
7603 * combine.c (extract_left_shift): Add a mode argument and update
7605 (make_compound_operation_int): Change the type of the mode parameter
7606 to scalar_int_mode and update the call to extract_left_shift.
7608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7609 Alan Hayward <alan.hayward@arm.com>
7610 David Sherwood <david.sherwood@arm.com>
7612 * combine.c (simplify_and_const_int): Change the type of the mode
7613 parameter to scalar_int_mode.
7614 (simplify_and_const_int_1): Likewise. Update recursive call.
7616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7617 Alan Hayward <alan.hayward@arm.com>
7618 David Sherwood <david.sherwood@arm.com>
7620 * combine.c (simplify_compare_const): Check that the mode is a
7621 scalar_int_mode (rather than VOIDmode) before testing its
7623 (simplify_comparison): Move COMPARISON_P handling out of the
7624 loop and restrict the latter part of the loop to scalar_int_modes.
7625 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
7626 and when considering SUBREG_REGs. Use is_int_mode instead of
7627 checking GET_MODE_CLASS against MODE_INT.
7629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7630 Alan Hayward <alan.hayward@arm.com>
7631 David Sherwood <david.sherwood@arm.com>
7633 * combine.c (try_widen_shift_mode): Move check for equal modes to...
7634 (simplify_shift_const_1): ...here. Use scalar_int_mode for
7635 shift_unit_mode and for modes involved in scalar shifts.
7637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7638 Alan Hayward <alan.hayward@arm.com>
7639 David Sherwood <david.sherwood@arm.com>
7641 * combine.c (force_int_to_mode): New function, split out from...
7642 (force_to_mode): ...here. Keep xmode up-to-date and use it
7643 instead of GET_MODE (x).
7645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7646 Alan Hayward <alan.hayward@arm.com>
7647 David Sherwood <david.sherwood@arm.com>
7649 * optabs-query.h (extraction_insn::struct_mode): Change type to
7650 opt_scalar_int_mode and update comment.
7651 (extraction_insn::field_mode): Change type to scalar_int_mode.
7652 (extraction_insn::pos_mode): Likewise.
7653 * combine.c (make_extraction): Update accordingly.
7654 * optabs-query.c (get_traditional_extraction_insn): Likewise.
7655 (get_optab_extraction_insn): Likewise.
7656 * recog.c (simplify_while_replacing): Likewise.
7657 * expmed.c (narrow_bit_field_mem): Change the type of the mode
7658 parameter to opt_scalar_int_mode.
7660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7661 Alan Hayward <alan.hayward@arm.com>
7662 David Sherwood <david.sherwood@arm.com>
7664 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
7665 to a scalar_int_mode instead of a machine_mode.
7666 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
7667 (get_best_mode): Return a boolean and use a pointer argument to store
7668 the selected mode. Replace the limit mode parameter with a bit limit.
7669 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
7670 for the values returned by bit_field_mode_iterator::next_mode.
7671 (store_bit_field): Update call to get_best_mode.
7672 (store_fixed_bit_field): Likewise.
7673 (extract_fixed_bit_field): Likewise.
7674 * expr.c (optimize_bitfield_assignment_op): Likewise.
7675 * fold-const.c (optimize_bit_field_compare): Likewise.
7676 (fold_truth_andor_1): Likewise.
7677 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
7678 Update for new type of m_mode.
7679 (get_best_mode): As above.
7681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7682 Alan Hayward <alan.hayward@arm.com>
7683 David Sherwood <david.sherwood@arm.com>
7685 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
7686 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
7687 (store_bit_field): Check is_a <scalar_int_mode> before calling
7688 strict_volatile_bitfield_p.
7689 (extract_bit_field): Likewise.
7691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7692 Alan Hayward <alan.hayward@arm.com>
7693 David Sherwood <david.sherwood@arm.com>
7695 * target.def (cstore_mode): Return a scalar_int_mode.
7696 * doc/tm.texi: Regenerate.
7697 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
7698 * targhooks.h (default_cstore_mode): Likewise.
7699 * targhooks.c (default_cstore_mode): Likewise, using a forced
7701 * expmed.c (emit_cstore): Expect the target of the cstore to be
7704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7705 Alan Hayward <alan.hayward@arm.com>
7706 David Sherwood <david.sherwood@arm.com>
7708 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
7710 (niter_desc): Likewise mode.
7711 (iv_analyze): Add a mode parameter.
7713 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
7714 and change its type to scalar_int_mode.
7715 * loop-iv.c: Update commentary at head of file.
7716 (iv_constant): Pass the mode paraeter before the rtx it describes
7717 and change its type to scalar_int_mode. Remove VOIDmode handling.
7718 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
7719 (iv_extend): Likewise.
7720 (shorten_into_mode): Likewise.
7721 (iv_add): Use scalar_int_mode.
7722 (iv_mult): Likewise.
7723 (iv_shift): Likewise.
7724 (canonicalize_iv_subregs): Likewise.
7725 (get_biv_step_1): Pass the outer_mode parameter before the rtx
7726 it describes and change its mode to scalar_int_mode. Also change
7727 the type of the returned inner_mode to scalar_int_mode.
7728 (get_biv_step): Likewise, turning outer_mode from a pointer
7729 into a direct parameter. Update call to get_biv_step_1.
7730 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
7731 iv_constant and get_biv_step.
7732 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
7733 and change its type to scalar_int_mode. Don't initialise iv->mode
7734 to VOIDmode and remove later checks for its still being VOIDmode.
7735 Update calls to iv_analyze_op and iv_analyze_expr. Check
7736 is_a <scalar_int_mode> when changing the mode under consideration.
7737 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
7738 Update call to iv_analyze_expr.
7739 (iv_analyze_op): Add a mode parameter. Reject subregs whose
7740 inner register is not also a scalar_int_mode. Update call to
7742 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
7743 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
7744 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
7745 separate mode class checks. Update calls to iv_analyze. Remove
7746 fix-up of VOIDmodes after iv_analyze_biv.
7747 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
7748 don't have a scalar_int_mode. Update call to biv_p.
7750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7751 Alan Hayward <alan.hayward@arm.com>
7752 David Sherwood <david.sherwood@arm.com>
7754 * cfgexpand.c (convert_debug_memory_address): Use
7755 as_a <scalar_int_mode>.
7756 * combine.c (expand_compound_operation): Likewise.
7757 (make_extraction): Likewise.
7758 (change_zero_ext): Likewise.
7759 (simplify_comparison): Likewise.
7760 * cse.c (cse_insn): Likewise.
7761 * dwarf2out.c (minmax_loc_descriptor): Likewise.
7762 (mem_loc_descriptor): Likewise.
7763 (loc_descriptor): Likewise.
7764 * expmed.c (init_expmed_one_mode): Likewise.
7765 (synth_mult): Likewise.
7766 (emit_store_flag_1): Likewise.
7767 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
7768 of a comparison with size.
7769 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
7770 (reduce_to_bit_field_precision): Likewise.
7771 * function.c (expand_function_end): Likewise.
7772 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
7773 * loop-doloop.c (doloop_modify): Likewise.
7774 * optabs.c (expand_binop): Likewise.
7775 (expand_unop): Likewise.
7776 (expand_copysign_absneg): Likewise.
7777 (prepare_cmp_insn): Likewise.
7778 (maybe_legitimize_operand): Likewise.
7779 * recog.c (const_scalar_int_operand): Likewise.
7780 * rtlanal.c (get_address_mode): Likewise.
7781 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7782 (simplify_cond_clz_ctz): Likewise.
7783 * tree-nested.c (get_nl_goto_field): Likewise.
7784 * tree.c (build_vector_type_for_mode): Likewise.
7785 * var-tracking.c (use_narrower_mode): Likewise.
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 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
7792 * builtins.c (expand_builtin_signbit): Use it.
7793 * cfgexpand.c (expand_debug_expr): Likewise.
7794 * dojump.c (do_jump): Likewise.
7795 (do_compare_and_jump): Likewise.
7796 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
7797 * expmed.c (make_tree): Likewise.
7798 * expr.c (expand_expr_real_2): Likewise.
7799 (expand_expr_real_1): Likewise.
7800 (try_casesi): Likewise.
7801 * fold-const-call.c (fold_const_call_ss): Likewise.
7802 * fold-const.c (unextend): Likewise.
7803 (extract_muldiv_1): Likewise.
7804 (fold_single_bit_test): Likewise.
7805 (native_encode_int): Likewise.
7806 (native_encode_string): Likewise.
7807 (native_interpret_int): Likewise.
7808 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
7809 * internal-fn.c (expand_addsub_overflow): Likewise.
7810 (expand_neg_overflow): Likewise.
7811 (expand_mul_overflow): Likewise.
7812 (expand_arith_overflow): Likewise.
7813 * match.pd: Likewise.
7814 * stor-layout.c (layout_type): Likewise.
7815 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7816 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
7817 * tree-ssanames.c (get_range_info): Likewise.
7818 * tree-switch-conversion.c (array_value_type) Likewise.
7819 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
7820 (vect_recog_divmod_pattern): Likewise.
7821 (vect_recog_mixed_size_cond_pattern): Likewise.
7822 * tree-vrp.c (extract_range_basic): Likewise.
7823 (simplify_float_conversion_using_ranges): Likewise.
7824 * tree.c (int_fits_type_p): Likewise.
7825 * ubsan.c (instrument_bool_enum_load): Likewise.
7826 * varasm.c (mergeable_string_section): Likewise.
7827 (narrowing_initializer_constant_valid_p): Likewise.
7828 (output_constant): Likewise.
7830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7831 Alan Hayward <alan.hayward@arm.com>
7832 David Sherwood <david.sherwood@arm.com>
7834 * machmode.h (NARROWEST_INT_MODE): New macro.
7835 * expr.c (alignment_for_piecewise_move): Use it instead of
7836 GET_CLASS_NARROWEST_MODE (MODE_INT).
7837 (push_block): Likewise.
7838 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
7840 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7843 Alan Hayward <alan.hayward@arm.com>
7844 David Sherwood <david.sherwood@arm.com>
7846 * postreload.c (move2add_valid_value_p): Change the type of the
7847 mode parameter to scalar_int_mode.
7848 (move2add_use_add2_insn): Add a mode parameter and use it instead
7850 (move2add_use_add3_insn): Likewise.
7851 (reload_cse_move2add): Update accordingly.
7853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7854 Alan Hayward <alan.hayward@arm.com>
7855 David Sherwood <david.sherwood@arm.com>
7857 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
7859 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
7860 * optabs.c (expand_unop): Likewise.
7862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7863 Alan Hayward <alan.hayward@arm.com>
7864 David Sherwood <david.sherwood@arm.com>
7866 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
7867 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
7868 (popcount_loc_descriptor): Likewise.
7869 (bswap_loc_descriptor): Likewise.
7870 (rotate_loc_descriptor): Likewise.
7871 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
7872 calling the functions above.
7874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7875 Alan Hayward <alan.hayward@arm.com>
7876 David Sherwood <david.sherwood@arm.com>
7878 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
7880 (try_combine): Likewise.
7881 (simplify_if_then_else): Likewise.
7882 * cse.c (cse_insn): Likewise.
7883 * dwarf2out.c (mem_loc_descriptor): Likewise.
7884 * emit-rtl.c (gen_lowpart_common): Likewise.
7885 * simplify-rtx.c (simplify_truncation): Likewise.
7886 (simplify_binary_operation_1): Likewise.
7887 (simplify_const_relational_operation): Likewise.
7888 (simplify_ternary_operation): Likewise.
7889 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
7896 * reload.c (push_reload): Likewise.
7897 (find_reloads): Likewise.
7899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7900 Alan Hayward <alan.hayward@arm.com>
7901 David Sherwood <david.sherwood@arm.com>
7903 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
7904 (make_compound_operation_int): Likewise.
7905 (change_zero_ext): Likewise.
7906 * expr.c (convert_move): Likewise.
7907 (convert_modes): Likewise.
7908 * fwprop.c (forward_propagate_subreg): Likewise.
7909 * loop-iv.c (get_biv_step_1): Likewise.
7910 * optabs.c (widen_operand): Likewise.
7911 * postreload.c (move2add_valid_value_p): Likewise.
7912 * recog.c (simplify_while_replacing): Likewise.
7913 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7914 (simplify_binary_operation_1): Likewise. Remove redundant
7915 mode equality check.
7917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7918 Alan Hayward <alan.hayward@arm.com>
7919 David Sherwood <david.sherwood@arm.com>
7921 * combine.c (combine_simplify_rtx): Add checks for
7922 is_a <scalar_int_mode>.
7923 (simplify_if_then_else): Likewise.
7924 (make_field_assignment): Likewise.
7925 (simplify_comparison): Likewise.
7926 * ifcvt.c (noce_try_bitop): Likewise.
7927 * loop-invariant.c (canonicalize_address_mult): Likewise.
7928 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7931 Alan Hayward <alan.hayward@arm.com>
7932 David Sherwood <david.sherwood@arm.com>
7934 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
7935 is_a <scalar_int_mode> instead of != BLKmode.
7937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7938 Alan Hayward <alan.hayward@arm.com>
7939 David Sherwood <david.sherwood@arm.com>
7941 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
7942 instead of != VOIDmode.
7943 * combine.c (if_then_else_cond): Likewise.
7944 (change_zero_ext): Likewise.
7945 * dwarf2out.c (mem_loc_descriptor): Likewise.
7946 (loc_descriptor): Likewise.
7947 * rtlanal.c (canonicalize_condition): Likewise.
7948 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
7950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7951 Alan Hayward <alan.hayward@arm.com>
7952 David Sherwood <david.sherwood@arm.com>
7954 * simplify-rtx.c (simplify_binary_operation_1): Use
7955 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
7957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7958 Alan Hayward <alan.hayward@arm.com>
7959 David Sherwood <david.sherwood@arm.com>
7961 * wide-int.h (int_traits<unsigned char>) New class.
7962 (int_traits<unsigned short>) Likewise.
7963 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
7964 Use GET_MODE_UNIT_PRECISION and remove redundant test for
7966 * combine.c (set_nonzero_bits_and_sign_copies): Use
7967 is_a <scalar_int_mode>.
7968 (find_split_point): Likewise.
7969 (combine_simplify_rtx): Likewise.
7970 (simplify_logical): Likewise.
7971 (expand_compound_operation): Likewise.
7972 (expand_field_assignment): Likewise.
7973 (make_compound_operation): Likewise.
7974 (extended_count): Likewise.
7975 (change_zero_ext): Likewise.
7976 (simplify_comparison): Likewise.
7977 * dwarf2out.c (scompare_loc_descriptor): Likewise.
7978 (ucompare_loc_descriptor): Likewise.
7979 (minmax_loc_descriptor): Likewise.
7980 (mem_loc_descriptor): Likewise.
7981 (loc_descriptor): Likewise.
7982 * expmed.c (init_expmed_one_mode): Likewise.
7983 * lra-constraints.c (lra_constraint_offset): Likewise.
7984 * optabs.c (prepare_libcall_arg): Likewise.
7985 * postreload.c (move2add_note_store): Likewise.
7986 * reload.c (operands_match_p): Likewise.
7987 * rtl.h (load_extend_op): Likewise.
7988 * rtlhooks.c (gen_lowpart_general): Likewise.
7989 * simplify-rtx.c (simplify_truncation): Likewise.
7990 (simplify_unary_operation_1): Likewise.
7991 (simplify_binary_operation_1): Likewise.
7992 (simplify_const_binary_operation): Likewise.
7993 (simplify_const_relational_operation): Likewise.
7994 (simplify_subreg): Likewise.
7995 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7996 * var-tracking.c (adjust_mems): Likewise.
7997 (prepare_call_arguments): Likewise.
7999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8000 Alan Hayward <alan.hayward@arm.com>
8001 David Sherwood <david.sherwood@arm.com>
8003 * machmode.h (is_int_mode): New fuction.
8004 * combine.c (find_split_point): Use it.
8005 (combine_simplify_rtx): Likewise.
8006 (simplify_if_then_else): Likewise.
8007 (simplify_set): Likewise.
8008 (simplify_shift_const_1): Likewise.
8009 (simplify_comparison): Likewise.
8010 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
8011 * cse.c (notreg_cost): Likewise.
8012 (cse_insn): Likewise.
8013 * cselib.c (cselib_lookup_1): Likewise.
8014 * dojump.c (do_jump_1): Likewise.
8015 (do_compare_rtx_and_jump): Likewise.
8016 * dse.c (get_call_args): Likewise.
8017 * dwarf2out.c (rtl_for_decl_init): Likewise.
8018 (native_encode_initializer): Likewise.
8019 * expmed.c (emit_store_flag_1): Likewise.
8020 (emit_store_flag): Likewise.
8021 * expr.c (convert_modes): Likewise.
8022 (store_field): Likewise.
8023 (expand_expr_real_1): Likewise.
8024 * fold-const.c (fold_read_from_constant_string): Likewise.
8025 * gimple-ssa-sprintf.c (get_format_string): Likewise.
8026 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
8027 * optabs.c (expand_binop): Likewise.
8028 (expand_unop): Likewise.
8029 (expand_abs_nojump): Likewise.
8030 (expand_one_cmpl_abs_nojump): Likewise.
8031 * simplify-rtx.c (mode_signbit_p): Likewise.
8032 (val_signbit_p): Likewise.
8033 (val_signbit_known_set_p): Likewise.
8034 (val_signbit_known_clear_p): Likewise.
8035 (simplify_relational_operation_1): Likewise.
8036 * tree.c (vector_type_mode): Likewise.
8038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8039 Alan Hayward <alan.hayward@arm.com>
8040 David Sherwood <david.sherwood@arm.com>
8042 * machmode.h (smallest_mode_for_size): Fix formatting.
8043 (smallest_int_mode_for_size): New function.
8044 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
8045 instead of smallest_mode_for_size.
8046 * combine.c (make_extraction): Likewise.
8047 * config/arc/arc.c (arc_expand_movmem): Likewise.
8048 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
8049 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8050 * config/s390/s390.c (s390_expand_insv): Likewise.
8051 * config/sparc/sparc.c (assign_int_registers): Likewise.
8052 * config/spu/spu.c (spu_function_value): Likewise.
8053 (spu_function_arg): Likewise.
8054 * coverage.c (get_gcov_type): Likewise.
8055 (get_gcov_unsigned_t): Likewise.
8056 * dse.c (find_shift_sequence): Likewise.
8057 * expmed.c (store_bit_field_1): Likewise.
8058 * expr.c (convert_move): Likewise.
8059 (store_field): Likewise.
8060 * internal-fn.c (expand_arith_overflow): Likewise.
8061 * optabs-query.c (get_best_extraction_insn): Likewise.
8062 * optabs.c (expand_twoval_binop_libfunc): Likewise.
8063 * stor-layout.c (layout_type): Likewise.
8064 (initialize_sizetypes): Likewise.
8065 * targhooks.c (default_get_mask_mode): Likewise.
8066 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8069 Alan Hayward <alan.hayward@arm.com>
8070 David Sherwood <david.sherwood@arm.com>
8072 * machmode.h (opt_mode::else_blk): New function.
8073 (int_mode_for_mode): Declare.
8074 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8075 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8077 * cfgexpand.c (expand_debug_expr): Likewise.
8078 * combine.c (gen_lowpart_or_truncate): Likewise.
8079 (gen_lowpart_for_combine): Likewise.
8080 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8081 * config/avr/avr.c (avr_to_int_mode): Likewise.
8082 (avr_out_plus_1): Likewise.
8083 (avr_out_plus): Likewise.
8084 (avr_out_round): Likewise.
8085 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8086 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8087 (s390_expand_vcond): Likewise.
8088 * config/spu/spu.c (spu_split_immediate): Likewise.
8089 (spu_expand_mov): Likewise.
8090 * dse.c (get_stored_val): Likewise.
8091 * expmed.c (store_bit_field_1): Likewise.
8092 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8094 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8095 (extract_low_bits): Likewise.
8096 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8097 handling rather than repeating the check.
8098 (emit_group_store): Likewise.
8099 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8100 * optabs.c (expand_absneg_bit): Likewise.
8101 (expand_copysign_absneg): Likewise.
8102 (expand_copysign_bit): Likewise.
8103 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8104 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8105 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8106 * var-tracking.c (prepare_call_arguments): Likewise.
8107 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8108 int_mode_for_mode instead of mode_for_size.
8109 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8112 Alan Hayward <alan.hayward@arm.com>
8113 David Sherwood <david.sherwood@arm.com>
8115 * machmode.h (int_mode_for_size): New function.
8116 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8117 instead of mode_for_size.
8118 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8120 * combine.c (expand_field_assignment): Use int_mode_for_size
8121 instead of mode_for_size.
8122 (make_extraction): Likewise.
8123 (simplify_shift_const_1): Likewise.
8124 (simplify_comparison): Likewise.
8125 * dojump.c (do_jump): Likewise.
8126 * dwarf2out.c (mem_loc_descriptor): Likewise.
8127 * emit-rtl.c (init_derived_machine_modes): Likewise.
8128 * expmed.c (flip_storage_order): Likewise.
8129 (convert_extracted_bit_field): Likewise.
8130 * expr.c (copy_blkmode_from_reg): Likewise.
8131 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8132 * internal-fn.c (expand_mul_overflow): Likewise.
8133 * lower-subreg.c (simple_move): Likewise.
8134 * optabs-libfuncs.c (init_optabs): Likewise.
8135 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8136 * tree.c (vector_type_mode): Likewise.
8137 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8138 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8139 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8140 * tree-vect-stmts.c (vectorizable_load): Likewise.
8141 (vectorizable_store): Likewise.
8143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8144 Alan Hayward <alan.hayward@arm.com>
8145 David Sherwood <david.sherwood@arm.com>
8147 * coretypes.h (pod_mode): New type.
8148 (scalar_int_mode_pod): New typedef.
8149 * machmode.h (pod_mode): New class.
8150 (int_n_data_t::m): Change type to scalar_int_mode_pod.
8151 * genmodes.c (emit_mode_int_n): Update accordingly.
8152 * lower-subreg.h (target_lower_subreg): Change type to
8153 scalar_int_mode_pod.
8154 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
8155 scalar_int_mode_pod.
8157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8158 Alan Hayward <alan.hayward@arm.com>
8159 David Sherwood <david.sherwood@arm.com>
8161 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
8162 machine_mode to scalar_int_mode.
8163 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
8164 (rs6000_option_override_internal): Remove cast to int.
8165 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
8166 machine_mode to scalar_int_mode.
8167 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
8168 (rs6000_option_override_internal): Remove cast to int.
8169 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
8170 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
8172 * config/s390/s390.c (s390_expand_builtin): Likewise.
8173 * coretypes.h (scalar_int_mode): New type.
8174 (opt_scalar_int_mode): New typedef.
8175 * machmode.h (scalar_int_mode): New class.
8176 (scalar_int_mode::includes_p): New function.
8177 (byte_mode): Change type to scalar_int_mode.
8178 (word_mode): Likewise.
8179 (ptr_mode): Likewise.
8180 * emit-rtl.c (byte_mode): Likewise.
8181 (word_mode): Likewise.
8182 (ptr_mode): Likewise.
8183 (init_derived_machine_modes): Update accordingly.
8184 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
8185 and MODE_PARTIAL_INT.
8186 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
8187 opt_scalar_int_mode.
8189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8190 Alan Hayward <alan.hayward@arm.com>
8191 David Sherwood <david.sherwood@arm.com>
8193 * target.def (libgcc_floating_mode_supported_p): Take a
8195 * doc/tm.texi: Regenerate.
8196 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
8198 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8199 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
8202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8203 Alan Hayward <alan.hayward@arm.com>
8204 David Sherwood <david.sherwood@arm.com>
8206 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
8207 * doc/tm.texi: Regenerate.
8208 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
8209 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
8210 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
8211 * targhooks.h (default_floatn_mode): Likewise.
8212 * targhooks.c (default_floatn_mode): Likewise.
8213 * tree.c (build_common_tree_nodes): Update accordingly.
8215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8216 Alan Hayward <alan.hayward@arm.com>
8217 David Sherwood <david.sherwood@arm.com>
8219 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
8220 (mode_iterator::iterate_p): Likewise.
8221 (mode_iterator::get_wider): Likewise.
8222 * expr.c (init_expr_target): Use opt_scalar_float_mode.
8224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8225 Alan Hayward <alan.hayward@arm.com>
8226 David Sherwood <david.sherwood@arm.com>
8228 * coretypes.h (opt_scalar_float_mode): New typedef.
8229 * machmode.h (float_mode_for_size): New function.
8230 * emit-rtl.c (double_mode): Delete.
8231 (init_emit_once): Use float_mode_for_size.
8232 * stor-layout.c (layout_type): Likewise.
8233 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
8235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8236 Alan Hayward <alan.hayward@arm.com>
8237 David Sherwood <david.sherwood@arm.com>
8239 * output.h (assemble_real): Take a scalar_float_mode.
8240 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
8241 * config/arm/arm.md (consttable_4): Likewise.
8242 (consttable_8): Likewise.
8243 (consttable_16): Likewise.
8244 * config/mips/mips.md (consttable_float): Likewise.
8245 * config/s390/s390.c (s390_output_pool_entry): Likewise.
8246 * varasm.c (assemble_real): Take a scalar_float_mode.
8247 (output_constant_pool_2): Update accordingly.
8248 (output_constant): Likewise.
8250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8251 Alan Hayward <alan.hayward@arm.com>
8252 David Sherwood <david.sherwood@arm.com>
8254 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
8255 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
8256 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
8257 (native_encode_real): Likewise.
8258 (native_interpret_real): Likewise.
8259 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
8260 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8263 Alan Hayward <alan.hayward@arm.com>
8264 David Sherwood <david.sherwood@arm.com>
8266 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
8267 <scalar_float_mode>. Simplify.
8268 (gen_extend_conv_libfunc): Likewise.
8270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8271 Alan Hayward <alan.hayward@arm.com>
8272 David Sherwood <david.sherwood@arm.com>
8274 * coretypes.h (scalar_float_mode): New type.
8275 * machmode.h (mode_traits::from_int): Use machine_mode if
8276 USE_ENUM_MODES is defined.
8277 (is_a): New function.
8279 (dyn_cast): Likewise.
8280 (scalar_float_mode): New class.
8281 (scalar_float_mode::includes_p): New function.
8282 (is_float_mode): Likewise.
8283 * gdbhooks.py (MachineModePrinter): New class.
8284 (build_pretty_printer): Use it for scalar_float_mode.
8285 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
8286 (format_helper::format_helper): Turn into a template.
8287 * genmodes.c (get_mode_class): New function.
8288 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
8289 or machine_mode if none.
8290 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
8291 as_a <scalar_float_mode>.
8292 * dwarf2out.c (mem_loc_descriptor): Likewise.
8293 (insert_float): Likewise.
8294 (add_const_value_attribute): Likewise.
8295 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8296 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
8297 (expand_unop): Update accordingly.
8298 (expand_abs_nojump): Likewise.
8299 (expand_copysign_absneg): Take a scalar_float_mode.
8300 (expand_copysign_bit): Likewise.
8301 (expand_copysign): Update accordingly.
8303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8304 Alan Hayward <alan.hayward@arm.com>
8305 David Sherwood <david.sherwood@arm.com>
8307 * coretypes.h (opt_mode): New class.
8308 * machmode.h (opt_mode): Likewise.
8309 (opt_mode::else_void): New function.
8310 (opt_mode::require): Likewise.
8311 (opt_mode::exists): Likewise.
8312 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
8313 (GET_MODE_2XWIDER_MODE): Likewise.
8314 (mode_iterator::get_wider): Update accordingly.
8315 (mode_iterator::get_2xwider): Likewise.
8316 (mode_iterator::get_known_wider): Likewise, turning into a template.
8317 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
8318 forcing a wider mode to exist.
8319 * config/cr16/cr16.h (LONG_REG_P): Likewise.
8320 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8321 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
8322 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
8323 * lower-subreg.c (init_lower_subreg): Likewise.
8324 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
8325 on the final iteration.
8326 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
8327 a wider mode exists before asking for a move pattern.
8328 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
8329 forcing a wider mode to exist.
8330 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
8331 returning false if no such mode exists.
8332 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
8333 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
8334 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
8335 Avoid checking for a MODE_INT if we already know the mode is not a
8337 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
8338 forcing a wider mode to exist.
8339 (expmed_mult_highpart_optab): Likewise.
8340 (expmed_mult_highpart): Likewise.
8341 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
8343 * lto-streamer-in.c (lto_input_mode_table): Likewise.
8344 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
8345 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8346 * internal-fn.c (expand_mul_overflow): Update use of
8347 GET_MODE_2XWIDER_MODE.
8348 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8349 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
8350 GET_MODE_WIDER_MODE.
8351 (convert_plusminus_to_widen): Likewise.
8352 * tree-switch-conversion.c (array_value_type): Likewise.
8353 * var-tracking.c (emit_note_insn_var_location): Likewise.
8354 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8355 Return false inside rather than outside the loop if no wider mode
8357 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
8358 and GET_MODE_2XWIDER_MODE
8359 (can_compare_p): Use else_void.
8360 * gdbhooks.py (OptMachineModePrinter): New class.
8361 (build_pretty_printer): Use it for opt_mode.
8363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8364 Alan Hayward <alan.hayward@arm.com>
8365 David Sherwood <david.sherwood@arm.com>
8367 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
8368 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
8370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8371 Alan Hayward <alan.hayward@arm.com>
8372 David Sherwood <david.sherwood@arm.com>
8374 * machmode.h (mode_traits): New structure.
8375 (get_narrowest_mode): New function.
8376 (mode_iterator::start): Likewise.
8377 (mode_iterator::iterate_p): Likewise.
8378 (mode_iterator::get_wider): Likewise.
8379 (mode_iterator::get_known_wider): Likewise.
8380 (mode_iterator::get_2xwider): Likewise.
8381 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
8382 (FOR_EACH_MODE): Likewise.
8383 (FOR_EACH_MODE_FROM): Likewise.
8384 (FOR_EACH_MODE_UNTIL): Likewise.
8385 (FOR_EACH_WIDER_MODE): Likewise.
8386 (FOR_EACH_2XWIDER_MODE): Likewise.
8387 * builtins.c (expand_builtin_strlen): Use new mode iterators.
8388 * combine.c (simplify_comparison): Likewise
8389 * config/i386/i386.c (type_natural_mode): Likewise.
8390 * cse.c (cse_insn): Likewise.
8391 * dse.c (find_shift_sequence): Likewise.
8392 * emit-rtl.c (init_derived_machine_modes): Likewise.
8393 (init_emit_once): Likewise.
8394 * explow.c (hard_function_value): Likewise.
8395 * expmed.c (extract_fixed_bit_field_1): Likewise.
8396 (extract_bit_field_1): Likewise.
8397 (expand_divmod): Likewise.
8398 (emit_store_flag_1): Likewise.
8399 * expr.c (init_expr_target): Likewise.
8400 (convert_move): Likewise.
8401 (alignment_for_piecewise_move): Likewise.
8402 (widest_int_mode_for_size): Likewise.
8403 (emit_block_move_via_movmem): Likewise.
8404 (copy_blkmode_to_reg): Likewise.
8405 (set_storage_via_setmem): Likewise.
8406 (compress_float_constant): Likewise.
8407 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8408 * optabs-query.c (get_best_extraction_insn): Likewise.
8409 * optabs.c (expand_binop): Likewise.
8410 (expand_twoval_unop): Likewise.
8411 (expand_twoval_binop): Likewise.
8412 (widen_leading): Likewise.
8413 (widen_bswap): Likewise.
8414 (expand_parity): Likewise.
8415 (expand_unop): Likewise.
8416 (prepare_cmp_insn): Likewise.
8417 (prepare_float_lib_cmp): Likewise.
8418 (expand_float): Likewise.
8419 (expand_fix): Likewise.
8420 (expand_sfix_optab): Likewise.
8421 * postreload.c (move2add_use_add2_insn): Likewise.
8422 * reg-stack.c (reg_to_stack): Likewise.
8423 * reginfo.c (choose_hard_reg_mode): Likewise.
8424 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8425 * stor-layout.c (mode_for_size): Likewise.
8426 (smallest_mode_for_size): Likewise.
8427 (mode_for_vector): Likewise.
8428 (finish_bitfield_representative): Likewise.
8429 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
8430 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
8431 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8432 * var-tracking.c (prepare_call_arguments): Likewise.
8434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8435 Alan Hayward <alan.hayward@arm.com>
8436 David Sherwood <david.sherwood@arm.com>
8438 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
8439 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
8440 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
8441 * machmode.h (mode_size): Move earlier in file.
8442 (mode_precision): Likewise.
8443 (mode_inner): Likewise.
8444 (mode_nunits): Likewise.
8445 (mode_unit_size): Likewise.
8446 (unit_unit_precision): Likewise.
8447 (mode_wider): Likewise.
8448 (mode_2xwider): Likewise.
8449 (machine_mode): New class.
8450 (mode_to_bytes): New function.
8451 (mode_to_bits): Likewise.
8452 (mode_to_precision): Likewise.
8453 (mode_to_inner): Likewise.
8454 (mode_to_unit_size): Likewise.
8455 (mode_to_unit_precision): Likewise.
8456 (mode_to_nunits): Likewise.
8457 (GET_MODE_SIZE): Use mode_to_bytes.
8458 (GET_MODE_BITSIZE): Use mode_to_bits.
8459 (GET_MODE_PRECISION): Use mode_to_precision.
8460 (GET_MODE_INNER): Use mode_to_inner.
8461 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
8462 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
8463 (GET_MODE_NUNITS): Use mode_to_nunits.
8464 * system.h (ALWAYS_INLINE): New macro.
8465 * config/powerpcspe/powerpcspe-c.c
8466 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
8467 int for arg1_mode and arg2_mode.
8469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8470 Alan Hayward <alan.hayward@arm.com>
8471 David Sherwood <david.sherwood@arm.com>
8473 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8474 Prefix mode names with E_ in case statements.
8475 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8476 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
8477 (aarch64_split_simd_move): Likewise.
8478 (aarch64_gen_storewb_pair): Likewise.
8479 (aarch64_gen_loadwb_pair): Likewise.
8480 (aarch64_gen_store_pair): Likewise.
8481 (aarch64_gen_load_pair): Likewise.
8482 (aarch64_get_condition_code_1): Likewise.
8483 (aarch64_constant_pool_reload_icode): Likewise.
8484 (get_rsqrte_type): Likewise.
8485 (get_rsqrts_type): Likewise.
8486 (get_recpe_type): Likewise.
8487 (get_recps_type): Likewise.
8488 (aarch64_gimplify_va_arg_expr): Likewise.
8489 (aarch64_simd_container_mode): Likewise.
8490 (aarch64_emit_load_exclusive): Likewise.
8491 (aarch64_emit_store_exclusive): Likewise.
8492 (aarch64_expand_compare_and_swap): Likewise.
8493 (aarch64_gen_atomic_cas): Likewise.
8494 (aarch64_emit_bic): Likewise.
8495 (aarch64_emit_atomic_swap): Likewise.
8496 (aarch64_emit_atomic_load_op): Likewise.
8497 (aarch64_evpc_trn): Likewise.
8498 (aarch64_evpc_uzp): Likewise.
8499 (aarch64_evpc_zip): Likewise.
8500 (aarch64_evpc_ext): Likewise.
8501 (aarch64_evpc_rev): Likewise.
8502 (aarch64_evpc_dup): Likewise.
8503 (aarch64_gen_ccmp_first): Likewise.
8504 (aarch64_gen_ccmp_next): Likewise.
8505 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8506 (alpha_emit_xfloating_libcall): Likewise.
8507 (emit_insxl): Likewise.
8508 (alpha_arg_type): Likewise.
8509 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
8510 (arc_preferred_simd_mode): Likewise.
8511 (arc_secondary_reload): Likewise.
8512 (get_arc_condition_code): Likewise.
8513 (arc_print_operand): Likewise.
8514 (arc_legitimate_constant_p): Likewise.
8515 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8516 * config/arc/arc.md (casesi_load): Likewise.
8517 (casesi_compact_jump): Likewise.
8518 * config/arc/predicates.md (proper_comparison_operator): Likewise.
8519 (cc_use_register): Likewise.
8520 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8521 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8522 (arm_init_iwmmxt_builtins): Likewise.
8523 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
8524 (neon_expand_vector_init): Likewise.
8525 (arm_attr_length_move_neon): Likewise.
8526 (maybe_get_arm_condition_code): Likewise.
8527 (arm_emit_vector_const): Likewise.
8528 (arm_preferred_simd_mode): Likewise.
8529 (arm_output_iwmmxt_tinsr): Likewise.
8530 (thumb1_output_casesi): Likewise.
8531 (thumb2_output_casesi): Likewise.
8532 (arm_emit_load_exclusive): Likewise.
8533 (arm_emit_store_exclusive): Likewise.
8534 (arm_expand_compare_and_swap): Likewise.
8535 (arm_evpc_neon_vuzp): Likewise.
8536 (arm_evpc_neon_vzip): Likewise.
8537 (arm_evpc_neon_vrev): Likewise.
8538 (arm_evpc_neon_vtrn): Likewise.
8539 (arm_evpc_neon_vext): Likewise.
8540 (arm_validize_comparison): Likewise.
8541 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
8542 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
8543 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
8544 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
8545 (c6x_preferred_simd_mode): Likewise.
8546 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
8547 (epiphany_rtx_costs): Likewise.
8548 * config/epiphany/predicates.md (proper_comparison_operator):
8550 * config/frv/frv.c (condexec_memory_operand): Likewise.
8551 (frv_emit_move): Likewise.
8552 (output_move_single): Likewise.
8553 (output_condmove_single): Likewise.
8554 (frv_hard_regno_mode_ok): Likewise.
8555 (frv_matching_accg_mode): Likewise.
8556 * config/h8300/h8300.c (split_adds_subs): Likewise.
8557 (h8300_rtx_costs): Likewise.
8558 (h8300_print_operand): Likewise.
8559 (compute_mov_length): Likewise.
8560 (output_logical_op): Likewise.
8561 (compute_logical_op_length): Likewise.
8562 (compute_logical_op_cc): Likewise.
8563 (h8300_shift_needs_scratch_p): Likewise.
8564 (output_a_shift): Likewise.
8565 (compute_a_shift_length): Likewise.
8566 (compute_a_shift_cc): Likewise.
8567 (expand_a_rotate): Likewise.
8568 (output_a_rotate): Likewise.
8569 * config/i386/i386.c (classify_argument): Likewise.
8570 (function_arg_advance_32): Likewise.
8571 (function_arg_32): Likewise.
8572 (function_arg_64): Likewise.
8573 (function_value_64): Likewise.
8574 (ix86_gimplify_va_arg): Likewise.
8575 (ix86_legitimate_constant_p): Likewise.
8576 (put_condition_code): Likewise.
8577 (split_double_mode): Likewise.
8578 (ix86_avx256_split_vector_move_misalign): Likewise.
8579 (ix86_expand_vector_logical_operator): Likewise.
8580 (ix86_split_idivmod): Likewise.
8581 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
8582 (ix86_build_const_vector): Likewise.
8583 (ix86_build_signbit_mask): Likewise.
8584 (ix86_match_ccmode): Likewise.
8585 (ix86_cc_modes_compatible): Likewise.
8586 (ix86_expand_branch): Likewise.
8587 (ix86_expand_sse_cmp): Likewise.
8588 (ix86_expand_sse_movcc): Likewise.
8589 (ix86_expand_int_sse_cmp): Likewise.
8590 (ix86_expand_vec_perm_vpermi2): Likewise.
8591 (ix86_expand_vec_perm): Likewise.
8592 (ix86_expand_sse_unpack): Likewise.
8593 (ix86_expand_int_addcc): Likewise.
8594 (ix86_split_to_parts): Likewise.
8595 (ix86_vectorize_builtin_gather): Likewise.
8596 (ix86_vectorize_builtin_scatter): Likewise.
8597 (avx_vpermilp_parallel): Likewise.
8598 (inline_memory_move_cost): Likewise.
8599 (ix86_tieable_integer_mode_p): Likewise.
8600 (x86_maybe_negate_const_int): Likewise.
8601 (ix86_expand_vector_init_duplicate): Likewise.
8602 (ix86_expand_vector_init_one_nonzero): Likewise.
8603 (ix86_expand_vector_init_one_var): Likewise.
8604 (ix86_expand_vector_init_concat): Likewise.
8605 (ix86_expand_vector_init_interleave): Likewise.
8606 (ix86_expand_vector_init_general): Likewise.
8607 (ix86_expand_vector_set): Likewise.
8608 (ix86_expand_vector_extract): Likewise.
8609 (emit_reduc_half): Likewise.
8610 (ix86_emit_i387_round): Likewise.
8611 (ix86_mangle_type): Likewise.
8612 (ix86_expand_round_sse4): Likewise.
8613 (expand_vec_perm_blend): Likewise.
8614 (canonicalize_vector_int_perm): Likewise.
8615 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
8616 (expand_vec_perm_1): Likewise.
8617 (expand_vec_perm_interleave3): Likewise.
8618 (expand_vec_perm_even_odd_pack): Likewise.
8619 (expand_vec_perm_even_odd_1): Likewise.
8620 (expand_vec_perm_broadcast_1): Likewise.
8621 (ix86_vectorize_vec_perm_const_ok): Likewise.
8622 (ix86_expand_vecop_qihi): Likewise.
8623 (ix86_expand_mul_widen_hilo): Likewise.
8624 (ix86_expand_sse2_abs): Likewise.
8625 (ix86_expand_pextr): Likewise.
8626 (ix86_expand_pinsr): Likewise.
8627 (ix86_preferred_simd_mode): Likewise.
8628 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
8629 * config/i386/sse.md (*andnot<mode>3): Likewise.
8630 (<mask_codefor><code><mode>3<mask_name>): Likewise.
8631 (*<code><mode>3): Likewise.
8632 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
8633 (ia64_expand_atomic_op): Likewise.
8634 (ia64_arg_type): Likewise.
8635 (ia64_mode_to_int): Likewise.
8636 (ia64_scalar_mode_supported_p): Likewise.
8637 (ia64_vector_mode_supported_p): Likewise.
8638 (expand_vec_perm_broadcast): Likewise.
8639 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
8640 (iq2000_function_arg_advance): Likewise.
8641 (iq2000_function_arg): Likewise.
8642 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
8643 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
8644 (m68k_libcall_value): Likewise.
8645 (m68k_function_value): Likewise.
8646 (sched_attr_op_type): Likewise.
8647 * config/mcore/mcore.c (mcore_output_move): Likewise.
8648 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
8650 (microblaze_function_arg): Likewise.
8651 * config/mips/mips.c (mips16_build_call_stub): Likewise.
8652 (mips_print_operand): Likewise.
8653 (mips_mode_ok_for_mov_fmt_p): Likewise.
8654 (mips_vector_mode_supported_p): Likewise.
8655 (mips_preferred_simd_mode): Likewise.
8656 (mips_expand_vpc_loongson_even_odd): Likewise.
8657 (mips_expand_vec_unpack): Likewise.
8658 (mips_expand_vi_broadcast): Likewise.
8659 (mips_expand_vector_init): Likewise.
8660 (mips_expand_vec_reduc): Likewise.
8661 (mips_expand_msa_cmp): Likewise.
8662 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
8663 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
8664 (cc_flags_for_mode): Likewise.
8665 * config/msp430/msp430.c (msp430_print_operand): Likewise.
8666 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
8667 (nds32_output_casesi_pc_relative): Likewise.
8668 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8669 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
8670 (nvptx_gen_unpack): Likewise.
8671 (nvptx_gen_pack): Likewise.
8672 (nvptx_gen_shuffle): Likewise.
8673 (nvptx_gen_wcast): Likewise.
8674 (nvptx_preferred_simd_mode): Likewise.
8675 * config/pa/pa.c (pa_secondary_reload): Likewise.
8676 * config/pa/predicates.md (base14_operand): Likewise.
8677 * config/powerpcspe/powerpcspe-c.c
8678 (altivec_resolve_overloaded_builtin): Likewise.
8679 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
8681 (rs6000_preferred_simd_mode): Likewise.
8682 (output_vec_const_move): Likewise.
8683 (rs6000_expand_vector_extract): Likewise.
8684 (rs6000_split_vec_extract_var): Likewise.
8685 (reg_offset_addressing_ok_p): Likewise.
8686 (rs6000_legitimate_offset_address_p): Likewise.
8687 (rs6000_legitimize_address): Likewise.
8688 (rs6000_emit_set_const): Likewise.
8689 (rs6000_const_vec): Likewise.
8690 (rs6000_emit_move): Likewise.
8691 (spe_build_register_parallel): Likewise.
8692 (rs6000_darwin64_record_arg_recurse): Likewise.
8693 (swap_selector_for_mode): Likewise.
8694 (spe_init_builtins): Likewise.
8695 (paired_init_builtins): Likewise.
8696 (altivec_init_builtins): Likewise.
8697 (do_load_for_compare): Likewise.
8698 (rs6000_generate_compare): Likewise.
8699 (rs6000_expand_float128_convert): Likewise.
8700 (emit_load_locked): Likewise.
8701 (emit_store_conditional): Likewise.
8702 (rs6000_output_function_epilogue): Likewise.
8703 (rs6000_handle_altivec_attribute): Likewise.
8704 (rs6000_function_value): Likewise.
8705 (emit_fusion_gpr_load): Likewise.
8706 (emit_fusion_p9_load): Likewise.
8707 (emit_fusion_p9_store): Likewise.
8708 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
8709 (fusion_gpr_mem_load): Likewise.
8710 (fusion_addis_mem_combo_load): Likewise.
8711 (fusion_addis_mem_combo_store): Likewise.
8712 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
8713 (fusion_gpr_mem_load): Likewise.
8714 (fusion_addis_mem_combo_load): Likewise.
8715 (fusion_addis_mem_combo_store): Likewise.
8716 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8718 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
8719 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8720 (rs6000_preferred_simd_mode): Likewise.
8721 (output_vec_const_move): Likewise.
8722 (rs6000_expand_vector_extract): Likewise.
8723 (rs6000_split_vec_extract_var): Likewise.
8724 (reg_offset_addressing_ok_p): Likewise.
8725 (rs6000_legitimate_offset_address_p): Likewise.
8726 (rs6000_legitimize_address): Likewise.
8727 (rs6000_emit_set_const): Likewise.
8728 (rs6000_const_vec): Likewise.
8729 (rs6000_emit_move): Likewise.
8730 (rs6000_darwin64_record_arg_recurse): Likewise.
8731 (swap_selector_for_mode): Likewise.
8732 (paired_init_builtins): Likewise.
8733 (altivec_init_builtins): Likewise.
8734 (rs6000_expand_float128_convert): Likewise.
8735 (emit_load_locked): Likewise.
8736 (emit_store_conditional): Likewise.
8737 (rs6000_output_function_epilogue): Likewise.
8738 (rs6000_handle_altivec_attribute): Likewise.
8739 (rs6000_function_value): Likewise.
8740 (emit_fusion_gpr_load): Likewise.
8741 (emit_fusion_p9_load): Likewise.
8742 (emit_fusion_p9_store): Likewise.
8743 * config/rx/rx.c (rx_gen_move_template): Likewise.
8744 (flags_from_mode): Likewise.
8745 * config/s390/predicates.md (s390_alc_comparison): Likewise.
8746 (s390_slb_comparison): Likewise.
8747 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
8748 (s390_vector_mode_supported_p): Likewise.
8749 (s390_cc_modes_compatible): Likewise.
8750 (s390_match_ccmode_set): Likewise.
8751 (s390_canonicalize_comparison): Likewise.
8752 (s390_emit_compare_and_swap): Likewise.
8753 (s390_branch_condition_mask): Likewise.
8754 (s390_rtx_costs): Likewise.
8755 (s390_secondary_reload): Likewise.
8756 (__SECONDARY_RELOAD_CASE): Likewise.
8757 (s390_expand_cs): Likewise.
8758 (s390_preferred_simd_mode): Likewise.
8759 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
8760 * config/sh/sh.c (sh_print_operand): Likewise.
8761 (dump_table): Likewise.
8762 (sh_secondary_reload): Likewise.
8763 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8764 * config/sh/sh.md (casesi_worker_1): Likewise.
8765 (casesi_worker_2): Likewise.
8766 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
8767 (fcc_comparison_operator): Likewise.
8768 * config/sparc/sparc.c (sparc_expand_move): Likewise.
8769 (emit_soft_tfmode_cvt): Likewise.
8770 (sparc_preferred_simd_mode): Likewise.
8771 (output_cbranch): Likewise.
8772 (sparc_print_operand): Likewise.
8773 (sparc_expand_vec_perm_bmask): Likewise.
8774 (vector_init_bshuffle): Likewise.
8775 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8776 (spu_vector_mode_supported_p): Likewise.
8777 (spu_expand_insv): Likewise.
8778 (spu_emit_branch_or_set): Likewise.
8779 (spu_handle_vector_attribute): Likewise.
8780 (spu_builtin_splats): Likewise.
8781 (spu_builtin_extract): Likewise.
8782 (spu_builtin_promote): Likewise.
8783 (spu_expand_sign_extend): Likewise.
8784 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8785 (tilegx_simd_int): Likewise.
8786 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
8787 (tilepro_simd_int): Likewise.
8788 * config/v850/v850.c (const_double_split): Likewise.
8789 (v850_print_operand): Likewise.
8790 (ep_memory_offset): Likewise.
8791 * config/vax/vax.c (vax_rtx_costs): Likewise.
8792 (vax_output_int_move): Likewise.
8793 (vax_output_int_add): Likewise.
8794 (vax_output_int_subtract): Likewise.
8795 * config/visium/predicates.md (visium_branch_operator): Likewise.
8796 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
8797 (visium_print_operand_address): Likewise.
8798 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8799 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
8800 (xtensa_expand_conditional_branch): Likewise.
8801 (xtensa_copy_incoming_a7): Likewise.
8802 (xtensa_output_literal): Likewise.
8803 * dfp.c (decimal_real_maxval): Likewise.
8804 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8807 Alan Hayward <alan.hayward@arm.com>
8808 David Sherwood <david.sherwood@arm.com>
8810 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
8811 (mode_nunits_inline): Likewise.
8812 (mode_inner_inline): Likewise.
8813 (mode_unit_size_inline): Likewise.
8814 (mode_unit_precision_inline): Likewise.
8815 (emit_insn_modes_h): Likewise. Also emit a #define of the
8817 (emit_mode_wider): Add an E_ prefix to mode names.
8818 (emit_mode_complex): Likewise.
8819 (emit_mode_inner): Likewise.
8820 (emit_mode_adjustments): Likewise.
8821 (emit_mode_int_n): Likewise.
8822 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
8823 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
8824 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
8825 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
8826 (CRC32_BUILTIN, ENTRY): Likewise.
8827 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
8828 (aarch64_pop_regs): Likewise.
8829 (aarch64_process_components): Likewise.
8830 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
8831 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
8832 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
8833 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
8834 * config/arm/arm.c (arm_init_libfuncs): Likewise.
8835 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
8837 * config/i386/i386-builtin.def (pcmpestr): Likewise.
8838 (pcmpistr): Likewise.
8839 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
8840 * config/mmix/mmix.c (mmix_output_condition): Likewise.
8841 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8843 * config/rl78/rl78.c (mduc_regs): Likewise.
8844 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8845 (htm_expand_builtin): Likewise.
8846 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
8847 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
8848 * config/xtensa/xtensa.c (print_operand): Likewise.
8849 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
8850 (NUM_MODE_VECTOR_INT): Likewise.
8851 * genoutput.c (null_operand): Likewise.
8852 (output_operand_data): Likewise.
8853 * genrecog.c (print_parameter_value): Likewise.
8854 * lra.c (debug_operand_data): Likewise.
8856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8858 * dwarf2out.c (scompare_loc_descriptor_wide)
8859 (scompare_loc_descriptor_narrow): New functions, split out from...
8860 (scompare_loc_descriptor): ...here.
8861 * expmed.c (emit_store_flag_int): New function, split out from...
8862 (emit_store_flag): ...here.
8864 2017-08-30 Richard Biener <rguenther@suse.de>
8866 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
8867 (dwarf2out_early_finish): Move setting of AT_pubnames from
8868 early debug output to early finish.
8870 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8872 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
8873 and -mdata-region to the assembler.
8875 2017-08-30 Richard Biener <rguenther@suse.de>
8877 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
8879 (gen_subprogram_die): Add DW_AT_object_pointer only early.
8880 (dwarf2out_early_global_decl): Only generate a DIE for the
8881 abstract origin if it doesn't already exist or is a declaration DIE.
8882 (resolve_addr): Do not add the linkage name twice when
8883 generating a stub DIE for the DW_TAG_GNU_call_site target.
8885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8887 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8888 Use machine_mode rather than int for arg1_mode.
8890 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8893 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
8894 that the second argument of the built-in functions to unpack
8895 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
8896 switch statement instead a lot of if statements.
8897 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
8898 Allow 64-bit values to be in Altivec registers as well as
8899 traditional floating point registers.
8900 (pack<mode>, FMOVE128_VSX iterator): Likewise.
8902 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
8904 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
8905 MAX_REGS_PER_ADDRESS == 1.
8907 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
8909 * config/i386/i386.opt (flag_fentry): Do not init to -1.
8910 * config/i386/i386.c (ix86_option_override_internal): Simplify
8911 setting of opts->x_flag_entry.
8913 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8914 Jakub Jelinek <jakub@redhat.com>
8915 Richard Biener <rguenther@suse.de>
8917 PR tree-optimization/81503
8918 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
8919 folded constant fits in the target type; reorder tests for clarity.
8921 2017-08-29 Martin Liska <mliska@suse.cz>
8923 * passes.def: Include pass_lower_switch.
8924 * stmt.c (dump_case_nodes): Remove and move to
8925 tree-switch-conversion.
8926 (case_values_threshold): Likewise.
8927 (expand_switch_as_decision_tree_p): Likewise.
8928 (emit_case_decision_tree): Likewise.
8929 (expand_case): Likewise.
8930 (balance_case_nodes): Likewise.
8931 (node_has_low_bound): Likewise.
8932 (node_has_high_bound): Likewise.
8933 (node_is_bounded): Likewise.
8934 (emit_case_nodes): Likewise.
8935 (struct simple_case_node): New struct.
8936 (add_case_node): Remove.
8937 (emit_case_dispatch_table): Use vector instead of case_list.
8938 (reset_out_edges_aux): Remove.
8939 (compute_cases_per_edge): Likewise.
8940 (expand_case): Build list of simple_case_node.
8941 (expand_sjlj_dispatch_table): Use it.
8942 * tree-switch-conversion.c (struct case_node): Moved from
8943 stmt.c and adjusted.
8944 (emit_case_nodes): Likewise.
8945 (node_has_low_bound): Likewise.
8946 (node_has_high_bound): Likewise.
8947 (node_is_bounded): Likewise.
8948 (case_values_threshold): Likewise.
8949 (reset_out_edges_aux): Likewise.
8950 (compute_cases_per_edge): Likewise.
8951 (add_case_node): Likewise.
8952 (dump_case_nodes): Likewise.
8953 (balance_case_nodes): Likewise.
8954 (expand_switch_as_decision_tree_p): Likewise.
8955 (emit_jump): Likewise.
8956 (emit_case_decision_tree): Likewise.
8957 (try_switch_expansion): Likewise.
8958 (do_jump_if_equal): Likewise.
8959 (emit_cmp_and_jump_insns): Likewise.
8960 (fix_phi_operands_for_edge): New function.
8961 (record_phi_operand_mapping): Likewise.
8962 (class pass_lower_switch): New pass.
8963 (pass_lower_switch::execute): New function.
8964 (make_pass_lower_switch): Likewise.
8965 (conditional_probability):
8966 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
8967 * tree-pass.h: Add make_pass_lower_switch.
8969 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8972 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
8975 2017-08-29 Richard Biener <rguenther@suse.de>
8977 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
8978 we do not add a DW_AT_inline attribute twice.
8979 (gen_subprogram_die): Remove code setting DW_AT_inline on
8980 DECL_ABSTRACT_P nodes.
8982 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
8984 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
8985 calls to internal functions.
8986 (gimplify_modify_expr): Likewise.
8987 * tree-call-cdce.c (use_internal_fn): Likewise.
8988 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
8989 (convert_to_divmod): Set the nothrow flag.
8990 * tree-if-conv.c (predicate_mem_writes): Likewise.
8991 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8992 (vectorizable_call): Likewise.
8993 (vectorizable_store): Likewise.
8994 (vectorizable_load): Likewise.
8995 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
8996 (vect_recog_mask_conversion_pattern): Likewise.
8998 2017-08-29 Martin Liska <mliska@suse.cz>
9001 * gcc.c (driver_handle_option): Add new argument.
9002 * opts-common.c (handle_option): Pass
9003 target_option_override_hook.
9004 * opts-global.c (lang_handle_option): Add new option.
9005 (set_default_handlers): Add new argument.
9006 (decode_options): Likewise.
9007 * opts.c (target_handle_option): Likewise.
9008 (common_handle_option): Call target_option_override_hook.
9009 * opts.h (struct cl_option_handler_func): Add hook for
9010 target option override.
9011 (struct cl_option_handlers): Likewise.
9012 (set_default_handlers): Add new argument.
9013 (decode_options): Likewise.
9014 (common_handle_option): Likewise.
9015 (target_handle_option): Likewise.
9016 * toplev.c (toplev::main): Pass targetm.target_option.override
9019 2017-08-29 Richard Biener <rguenther@suse.de>
9020 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9022 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
9023 life to the active subtree.
9025 2017-08-28 Jeff Law <law@redhat.com>
9027 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
9028 derive_equivalences.
9029 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
9031 (edge_info::derive_equivalences): New private member function
9033 * tree-ssa-dom.c (class edge_info): Changed from a struct
9034 to a class. Add ctor/dtor, methods and data members.
9035 (edge_info::edge_info): Renamed from allocate_edge_info.
9036 Initialize additional members.
9037 (edge_info::~edge_info): New.
9038 (free_dom_edge_info): Delete the edge info.
9039 (record_edge_info): Use new class & associated member functions.
9040 Tighten forms for testing for edge equivalences.
9041 (record_temporary_equivalences): Iterate over the simple
9042 equivalences rather than assuming there's only one per edge.
9043 (cprop_into_successor_phis): Iterate over the simple
9044 equivalences rather than assuming there's only one per edge.
9045 (optimize_stmt): Use operand_equal_p rather than pointer
9046 equality for mini-DSE code.
9048 2017-08-28 Nathan Sidwell <nathan@acm.org>
9050 * gcc.c (execute): Fold SIGPIPE handling into switch
9051 statement. Adjust internal error message.
9053 2017-08-28 Richard Biener <rguenther@suse.de>
9056 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
9057 Do nothing for removed DIEs.
9059 2017-08-28 Richard Biener <rguenther@suse.de>
9061 PR tree-optimization/81977
9062 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
9065 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9068 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9069 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9070 using targetm.gen_mem_thread_fence.
9072 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9075 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9076 predicate to register_operand. Reorder operands.
9077 (*btr<mode>): Ditto.
9078 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9079 (*btr<mode>_mask): Ditto.
9081 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9083 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9084 * config/rs6000/xmmintrin.h: New file.
9085 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9087 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9090 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9091 parameter and_insn and return it.
9092 (recombine_lvx_pattern): Insert a copy to ensure availability of
9093 the base register of the copied masking operation at the point of
9094 the instruction replacement.
9095 (recombine_stvx_pattern): Likewise.
9097 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9099 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9100 undocumented switches.
9101 (-mpower9-dform-vector): Likewise.
9102 (-mpower9-dform): Likewise.
9103 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9104 comments to delete references to -mpower9-dform* switches.
9105 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9106 Delete reference to -mpower9-dform* switches, test for
9107 -mpower9-vector instead.
9108 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9109 (OTHER_P9_VECTOR_MASKS): Likewise.
9110 (POWERPC_MASKS): Likewise.
9111 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9112 tests against -mpower9-dform* to -mpower9-vector. Delete code
9113 that checked for -mpower9-dform* consistancy with other options.
9114 Add test for -mpower9-misc to enable other power9 switches.
9115 (rs6000_init_hard_regno_mode_ok): Likewise.
9116 (rs6000_option_override_internal): Likewise.
9117 (rs6000_emit_prologue): Likewise.
9118 (rs6000_emit_epilogue): Likewise.
9119 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9120 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9121 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9123 (emit_fusion_p9_store): Likewise.
9124 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9125 resetting these macros if the assembler does not support ISA 3.0
9127 (TARGET_P9_DFORM_VECTOR): Likewise.
9128 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9129 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9131 2017-08-25 Alan Modra <amodra@gmail.com>
9133 PR rtl-optimization/81747
9134 * cse.c (cse_extended_basic_block): Don't attempt to record
9135 equivalences for degenerate conditional branches that branch
9136 to their fall-through.
9138 2017-08-24 Martin Sebor <msebor@redhat.com>
9141 * gimple-fold.c (size_must_be_zero_p): New function.
9142 (gimple_fold_builtin_memory_op): Call it.
9144 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
9146 * config/rs6000/mm_malloc.h: New file.
9148 2017-08-24 Bin Cheng <bin.cheng@arm.com>
9150 PR tree-optimization/81913
9151 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
9152 analysis when either IVs in condition can wrap.
9154 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
9156 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
9157 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
9159 2017-08-24 Richard Biener <rguenther@suse.de>
9162 * targhooks.c (default_target_can_inline_p): Properly
9163 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
9164 is present and always compare.
9165 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
9166 infer -mfpmath=sse from TARGET_SSE_P.
9167 (ix86_can_inline_p): Properly use target_option_default_node when
9168 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
9170 2017-08-24 Richard Biener <rguenther@suse.de>
9173 * dwarf2out.c (output_die): Handle flag_generate_offload like
9175 (output_comp_unit): Likewise.
9176 (gen_array_type_die): Likewise.
9177 (dwarf2out_early_finish): Likewise.
9178 (note_variable_value_in_expr): Likewise.
9179 (dwarf2out_finish): Likewise. Adjust assert.
9180 * cgraphunit.c (symbol_table::compile): Move setting of
9181 flag_generate_offload earlier ...
9182 (symbol_table::finalize_compilation_unit): ... here, before
9183 early debug finalization.
9185 2017-08-24 Richard Biener <rguenther@suse.de>
9187 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
9188 and ipa-fnsummary.h.
9189 (ix86_can_inline_p): When ix86_fpmath flags do not match
9190 check whether the callee uses FP math at all.
9192 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
9195 * tree-ssanames.c (get_nonzero_bits): Use element_precision
9196 instead of TYPE_PRECISION.
9198 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
9199 Alan Hayward <alan.hayward@arm.com>
9200 David Sherwood <david.sherwood@arm.com>
9202 * combine.c (make_extraction): Use subreg_offset_from_lsb.
9204 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
9206 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
9208 (ix86_frame::stack_realign_allocate): New field.
9209 (struct machine_frame_state): Modify comments.
9210 (machine_frame_state::sp_realigned_fp_end): New field.
9211 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
9213 (sp_valid_at): Add assertion to assure no attempt to access invalid
9214 offset of a realigned stack.
9215 (fp_valid_at): Likewise.
9216 (choose_baseaddr): Modify comments.
9217 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
9218 ix86_expand_prologue.
9219 (ix86_expand_prologue): Modify stack realignment and allocation.
9220 (ix86_expand_epilogue): Modify comments.
9221 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
9222 avx2_runtime, avx512f, and avx512f_runtime.
9224 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
9226 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
9227 (mstackrealign): Do not init to -1.
9228 * config/i386/i386.c (ix86_option_override_internal):
9229 Check opts_set, not opts when setting default value of
9230 opts->x_ix86_force_align_arg_pointer.
9232 2017-08-23 Richard Biener <rguenther@suse.de>
9234 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
9235 lang_hooks.decl_printable_name.
9236 * print-rtl-function.c (print_rtx_function): Likewise.
9237 * tree-pretty-print.c (dump_function_header): Likewise.
9239 2017-08-23 Richard Biener <rguenther@suse.de>
9242 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
9243 -g0 at compile-time.
9245 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9248 * doc/sourcebuild.texi (Other hardware attributes):
9251 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9254 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
9255 Fix single-use check.
9257 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
9259 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
9261 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9263 * doc/install.texi: Modify to add more details on running selected
9266 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9268 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
9269 is combined with -mabi=ms.
9270 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
9273 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9275 PR tree-optimization/81488
9276 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
9277 and cached_basis fields.
9278 (MAX_SPREAD): New constant.
9279 (alloc_cand_and_find_basis): Initialize new fields.
9280 (clear_visited): New function.
9281 (create_phi_basis_1): Rename from create_phi_basis, set visited
9282 and cached_basis fields.
9283 (create_phi_basis): New wrapper function.
9284 (phi_add_costs_1): Rename from phi_add_costs, add spread
9285 parameter, set visited field, short-circuit when limits reached.
9286 (phi_add_costs): New wrapper function.
9287 (record_phi_increments_1): Rename from record_phi_increments, set
9289 (record_phi_increments): New wrapper function.
9290 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
9291 (phi_incr_cost): New wrapper function.
9292 (all_phi_incrs_profitable_1): Rename from
9293 all_phi_incrs_profitable, set visited field.
9294 (all_phi_incrs_profitable): New wrapper function.
9296 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9297 Alan Hayward <alan.hayward@arm.com>
9298 David Sherwood <david.sherwood@arm.com>
9300 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
9301 that takes the outer and inner modes.
9302 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
9303 comparison as the canonical test for a paradoxical subreg.
9304 * combine.c (simplify_set): Use paradoxical_subreg_p.
9305 (make_extraction): Likewise.
9306 (force_to_mode): Likewise.
9307 (rtx_equal_for_field_assignment_p): Likewise.
9308 (gen_lowpart_for_combine): Likewise.
9309 (simplify_comparison): Likewise.
9310 * cse.c (equiv_constant): Likewise.
9311 * expmed.c (store_bit_field_1): Likewise.
9312 * final.c (alter_subreg): Likewise.
9313 * fwprop.c (propagate_rtx): Likewise.
9314 (forward_propagate_subreg): Likewise.
9315 * ira-conflicts.c (ira_build_conflicts): Likewise.
9316 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
9317 * lra-constraints.c (curr_insn_transform): Likewise.
9318 (split_reg): Likewise.
9319 * lra-eliminations.c (move_plus_up): Likewise.
9320 (lra_eliminate_regs_1): Likewise.
9321 * recog.c (general_operand): Likewise.
9322 * ree.c (combine_reaching_defs): Likewise.
9323 * reload.c (push_reload): Likewise.
9324 (find_reloads): Likewise.
9325 * reload1.c (elimination_effects): Likewise.
9326 (compute_reload_subreg_offset): Likewise.
9327 (choose_reload_regs): Likewise.
9328 * rtlanal.c (subreg_lsb_1): Likewise.
9329 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9330 (simplify_subreg): Likewise.
9331 * var-tracking.c (track_loc_p): Likewise.
9332 * emit-rtl.c (byte_lowpart_offset): Likewise.
9333 (paradoxical_subreg_p): Delete out-of-line definition.
9335 2017-08-22 Jeff Law <law@redhat.com>
9337 PR tree-optimization/81741
9338 PR tree-optimization/71947
9339 * tree-ssa-dom.c: Include tree-inline.h.
9340 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
9341 equivalences if one is more expensive to compute than the other.
9342 * tree-ssa-scopedtables.h (class const_or_copies): Make
9343 record_const_or_copy_raw method private.
9344 (class avail_exprs_stack): New method simplify_binary_operation.
9345 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
9346 avail_exprs_stack::simplify_binary_operation as needed.
9347 (avail_exprs_stack::simplify_binary_operation): New function.
9349 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
9351 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
9352 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
9353 (DOT_SYMBOLS): Likewise.
9354 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
9355 (RELOCATABLE_NEEDS_FIXUP): Likewise.
9356 (RS6000_ABI_NAME): Likewise.
9357 (TARGET_CMODEL): Likewise.
9358 (TOC_SECTION_ASM_OP): Likewise.
9359 (SET_CMODEL): New macro.
9360 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
9362 2017-08-22 Richard Biener <rguenther@suse.de>
9364 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
9365 to allow for free-lang-data replacements similar to verify_type_variant.
9367 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9368 Alan Hayward <alan.hayward@arm.com>
9369 David Sherwood <david.sherwood@arm.com>
9371 * config/aarch64/aarch64.md (casesi): Use DImode rather than
9372 VOIDmode for the LABEL_REF.
9374 2017-08-22 Richard Biener <rguenther@suse.de>
9376 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
9378 2017-08-22 Richard Biener <rguenther@suse.de>
9380 * common.opt (feliminate-dwarf2-dups): Ignore.
9381 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
9382 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
9383 same_die_p_wrap, compute_section_prefix,
9384 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
9385 (comdat_symbol_id, comdat_symbol_number): Likewise.
9386 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
9388 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
9389 (output_die): Mark unreachable path unreachable.
9390 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
9391 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
9392 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
9393 (dwarf2out_early_finish): Likewise.
9395 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
9397 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
9399 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
9402 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
9403 not VAR_P. Filter attribute warnings with OPT_Wattributes.
9404 (avr_attribute_table) <io, io_low, address>: Initialize
9405 .decl_required with true.
9407 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
9409 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
9410 undocumented debugging options.
9411 (-mvsx-scalar-double): Likewise.
9412 (-mallow-df-permute): Likewise.
9413 (-mvectorize-builtins): Likewise.
9414 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9415 (rs6000_builtin_vectorized_function): Likewise.
9416 (rs6000_builtin_md_vectorized_function): Likewise.
9417 (rs6000_opt_vars): Likewise.
9419 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
9422 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
9423 (*btrq_imm): Rename from *btrq.
9424 (*btcq_imm): Rename from *btcq.
9425 (btsc): New code attribute.
9426 (*<btsc><mode>): New insn pattern.
9427 (*btr<mode>): Ditto.
9428 (*<btsc><mode>_mask): New insn_and_split pattern.
9429 (*btr<mode>_mask): Ditto.
9431 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9432 Alan Hayward <alan.hayward@arm.com>
9433 David Sherwood <david.sherwood@arm.com>
9435 * function.c (pad_below): Simplify padding calculation.
9437 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9438 Alan Hayward <alan.hayward@arm.com>
9439 David Sherwood <david.sherwood@arm.com>
9441 * target.def (function_prologue): Remove frame size argument.
9442 (function_epilogue): Likewise.
9443 * doc/tm.texi: Regenerate.
9444 * final.c (final_start_function): Update call to function_prologue.
9445 (final_end_function): Update call to function_epilogue.
9446 (default_function_pro_epilogue): Remove frame size argument.
9447 * output.h (default_function_pro_epilogue): Likewise.
9448 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
9449 (arm_output_function_prologue): Likewise.
9450 * config/frv/frv.c (frv_function_prologue): Likewise.
9451 (frv_function_epilogue): Likewise.
9452 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
9453 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
9454 (ia64_output_function_epilogue): Likewise.
9455 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
9456 (m32r_output_function_epilogue): Likewise.
9457 * config/microblaze/microblaze.c (microblaze_function_prologue)
9458 (microblaze_function_epilogue): Likewise.
9459 * config/mips/mips.c (mips_output_function_prologue): Likewise.
9460 (mips_output_function_epilogue): Likewise.
9461 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
9462 (mmix_target_asm_function_epilogue): Likewise.
9463 * config/msp430/msp430.c (msp430_start_function): Likewise.
9464 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
9465 (nds32_asm_function_epilogue): Likewise.
9466 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
9467 * config/pa/pa.c (pa_output_function_prologue): Likewise.
9468 (pa_output_function_epilogue): Likewise.
9469 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
9470 (rs6000_output_function_epilogue): Likewise.
9471 * config/rl78/rl78.c (rl78_start_function): Likewise.
9472 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
9473 (rs6000_output_function_epilogue): Likewise.
9474 * config/rx/rx.c (rx_output_function_prologue): Likewise.
9475 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
9476 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
9477 (sparc_asm_function_epilogue): Likewise.
9479 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9481 * tree.h (type_has_mode_precision_p): New function.
9482 * convert.c (convert_to_integer_1): Use it.
9483 * expr.c (expand_expr_real_2): Likewise.
9484 (expand_expr_real_1): Likewise.
9485 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
9486 * match.pd: Likewise.
9487 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
9488 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
9489 * tree-tailcall.c (process_assignment): Likewise.
9490 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9491 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
9492 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
9493 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9494 (vectorizable_assignment): Likewise.
9495 (vectorizable_shift): Likewise.
9496 (vectorizable_operation): Likewise.
9497 * tree-vrp.c (register_edge_assert_for_2): Likewise.
9499 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
9501 * match.pd: Add pow (C, x) simplification.
9503 2017-08-21 Richard Biener <rguenther@suse.de>
9505 PR tree-optimization/81900
9506 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
9507 for blocks with abnormal predecessors.
9508 (compute_antic): Do not set visited flag prematurely.
9510 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
9513 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
9515 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9517 * stor-layout.h (vector_type_mode): Move to...
9518 * tree.h (vector_type_mode): ...here.
9519 * stor-layout.c (vector_type_mode): Move to...
9520 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
9522 2017-08-21 Richard Biener <rguenther@suse.de>
9524 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
9525 register_external_die hooks.
9526 (debug_false_tree_charstarstar_uhwistar): Declare.
9527 (debug_nothing_tree_charstar_uhwi): Likewise.
9528 * debug.c (do_nothing_debug_hooks): Adjust.
9529 (debug_false_tree_charstarstar_uhwistar): New do nothing.
9530 (debug_nothing_tree_charstar_uhwi): Likewise.
9531 * dbxout.c (dbx_debug_hooks): Adjust.
9532 (xcoff_debug_hooks): Likewise.
9533 * sdbout.c (sdb_debug_hooks): Likewise.
9534 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9535 * dwarf2out.c (macinfo_label_base): New global.
9536 (dwarf2out_register_external_die): New function for the
9537 register_external_die hook.
9538 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
9539 (dwarf2_debug_hooks): Use them.
9540 (dwarf2_lineno_debug_hooks): Adjust.
9541 (struct die_struct): Add with_offset flag.
9542 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
9543 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
9544 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
9545 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
9546 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
9547 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
9548 defining section names for the early LTO debug variants.
9549 (reset_indirect_string): New helper.
9550 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
9551 (print_dw_val): Add support for offsetted symbol references.
9552 (get_ultimate_context): Split out from is_cxx.
9553 (is_cxx): Use get_ultimate_context.
9554 (is_fortran): Add decl overload.
9555 (compute_comp_unit_symbol): Split out worker from
9556 compute_section_prefix.
9557 (compute_section_prefix): Call compute_comp_unit_symbol and
9558 set comdat_type_p here.
9559 (output_die): Skip DIE symbol output for the LTO added one.
9560 Handle DIE symbol references with offset.
9561 (output_comp_unit): Guard section name mangling properly.
9562 For LTO debug sections emit a symbol at the section beginning
9563 which we use to refer to its DIEs.
9564 (add_abstract_origin_attribute): For DIEs registered via
9565 dwarf2out_register_external_die directly refer to the early
9566 DIE rather than indirectly through the shadow one we created.
9567 Remove obsolete call to dwarf2out_abstract_function for
9568 non-function/block origins.
9569 (gen_array_type_die): When generating early LTO debug do
9570 not emit DW_AT_string_length.
9571 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
9572 late when in LTO. As suggested place a gcc_unreachable for
9573 the DECL_ABSTRACT_P case.
9574 (gen_subprogram_die): Avoid another specification DIE
9575 for early built declarations/definitions for the late LTO case.
9576 (gen_variable_die): Add type references for late duplicated VLA dies
9578 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
9579 we have the abstract instance already.
9580 (process_scope_var): Adjust decl DIE contexts in LTO which
9581 first puts them in limbo.
9582 (gen_decl_die): Do not generate type DIEs late apart from
9583 types for VLAs or for decls we do not yet have a DIE. Do not
9584 call dwarf2out_abstract_function late.
9585 (dwarf2out_early_global_decl): Make sure to create DIEs
9586 for abstract instances of a decl first.
9587 (dwarf2out_late_global_decl): Adjust comment.
9588 (output_macinfo_op): With multiple macro sections use
9589 macinfo_label_base to distinguish labels.
9590 (output_macinfo): Likewise. Update macinfo_label_base.
9591 Pass in the line info label.
9592 (note_variable_value_in_expr): When generating LTO resolve
9593 all variable values here by generating DIEs as needed.
9594 (init_sections_and_labels): Add early LTO debug flag parameter
9595 and generate different sections and names if set. Add generation
9596 counter for the labels so we can have multiple of them.
9597 (reset_dies): Helper to allow DIEs to be output multiple times.
9598 (dwarf2out_finish): When outputting DIEs to the fat part of an
9599 LTO object first reset DIEs.
9600 (dwarf2out_early_finish): Output early DIEs when generating LTO.
9601 (modified_type_die): Check for decl_ultimate_origin being self
9603 (gen_type_die_with_usage): Likewise.
9604 (gen_typedef_die): Allow decl_ultimate_origin being self.
9605 (set_decl_abstract_flags): Remove.
9606 (set_block_abstract_flags): Likewise.
9607 (dwarf2out_abstract_function): Treat the early generated DIEs
9608 as the abstract copy and only add DW_AT_inline and
9609 DW_AT_artificial here and call set_decl_origin_self.
9610 If the DIE has an abstract origin don't do anything.
9611 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
9612 if we have none yet (Go fails to build one, PR78628).
9613 (variably_modified_type_p): Prevent endless recursion for Ada
9614 cyclic pointer types.
9615 * lto-streamer-in.c: Include debug.h.
9616 (dref_queue): New global.
9617 (lto_read_tree_1): Stream in DIE references.
9618 (lto_input_tree): Register DIE references.
9619 (input_function): Stream DECL_DEBUG_ARGS.
9620 * lto-streamer-out.c: Include debug.h.
9621 (lto_write_tree_1): Output DIE references.
9622 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
9623 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
9624 (output_function): Stream DECL_DEBUG_ARGS.
9625 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
9626 Stream DECL_ABSTRACT_ORIGIN.
9627 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
9628 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
9629 DECL_CONTEXT for file-scope decls.
9630 * lto-streamer.h (struct dref_entry): Declare.
9631 (dref_queue): Likewise.
9632 * cfgexpand.c (pass_expand::execute): Do not call the
9633 outlining_inline_function hook here.
9634 * lto-wrapper.c (debug_obj): New global.
9635 (tool_cleanup): Unlink it if required.
9636 (debug_objcopy): New function.
9637 (run_gcc): Handle early debug sections in the IL files by
9638 extracting them to separate files, partially linkin them and
9639 feeding the result back as result to the linker.
9640 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
9641 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
9642 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
9643 sections into a separate segment.
9644 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
9646 (darwin_asm_dwarf_section): Likewise.
9647 (darwin_asm_output_dwarf_offset): Likewise.
9648 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
9650 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9651 Alan Hayward <alan.hayward@arm.com>
9652 David Sherwood <david.sherwood@arm.com>
9654 * read-md.h (md_reader::record_potential_iterator_use): Replace
9655 pointer argument with an rtx and an index.
9656 * read-rtl.c (iterator_group::apply_iterator): Likewise.
9657 (apply_mode_iterator): Likewise.
9658 (apply_code_iterator): Likewise.
9659 (apply_int_iterator): Likewise.
9660 (apply_subst_iterator): Likewise.
9661 (record_iterator_use): Likewise.
9662 (record_attribute_use): Likewise.
9663 (md_reader::record_potential_iterator_use): Likewise. Update calls
9664 to record_iterator_use and apply_iterator.
9665 (iterator_use): Replace ptr with x and index.
9666 (attribute_use): Likewise.
9667 (apply_attribute_uses): Update calls to apply_iterator.
9668 (apply_iterators): Likewise. Update initialization of iterator_use.
9669 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
9670 and record_potential_iterator_use.
9671 (rtx_reader::read_rtx_operand): Likewise.
9673 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9674 Alan Hayward <alan.hayward@arm.com>
9675 David Sherwood <david.sherwood@arm.com>
9677 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
9680 2017-08-21 Richard Biener <rguenther@suse.de>
9683 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
9684 at struct end conservatively when comparing common bases.
9686 2017-08-21 Richard Biener <rguenther@suse.de>
9688 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
9689 (mem_ref_in_stmt): Remove.
9690 (determine_max_movement): Use ref index to get at the reference.
9691 (invariantness_dom_walker::before_dom_children): Deal with
9692 lim data already initialized.
9693 (gather_mem_refs_stmt): Initialize lim data and record ref index.
9695 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9697 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
9698 (TARGET_ISA_ROUND): Ditto.
9699 (TARGET_ROUND): Ditto.
9700 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
9701 * config/i386/i386.md: Ditto.
9702 * config/i386/sse.md: Ditto.
9703 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
9704 with OPTION_MASK_ISA_SSE4_1.
9706 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9709 * doc/extend.texi (x86 Built-in Functions): Correct the name of
9710 __builtin_ia32_lzcnt_u16.
9712 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9715 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
9716 (rs6000_set_current_function): Rewrite function to use it.
9718 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
9721 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
9722 and TYPE_WARN_IF_NOT_ALIGN.
9723 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
9724 (handle_warn_if_not_align): New.
9725 (place_union_field): Call handle_warn_if_not_align.
9726 (place_field): Call handle_warn_if_not_align.
9727 Copy TYPE_WARN_IF_NOT_ALIGN.
9728 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
9729 (layout_type): Likewise.
9730 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
9732 (tree_decl_common): Add warn_if_not_align.
9733 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
9734 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
9735 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
9736 (DECL_WARN_IF_NOT_ALIGN): Likewise.
9737 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
9738 * doc/extend.texi: Document warn_if_not_aligned attribute.
9739 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
9741 2017-08-17 Martin Liska <mliska@suse.cz>
9744 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
9745 (get_data_dependence): Use it as pointer type.
9746 (distribute_loop): Likewise.
9748 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9750 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
9751 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
9752 (p8_vmrgow_<mode>_direct): New define_insn.
9753 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
9754 handle endianness for vmrgew and vmrgow permute patterns.
9756 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9758 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
9759 * config/rs6000/rs6000-cpus.def: Remove comment.
9760 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
9761 (POWERPC_MASKS): Likewise.
9762 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
9763 use of TARGET_VSX_TIMODE.
9764 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
9765 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
9766 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
9767 (rs6000_option_override_internal): Remove dead code.
9768 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
9769 (rs6000_legitimize_reload_address): Likewise.
9770 (rs6000_legitimate_address_p): Likewise.
9771 (rs6000_opt_masks): Delete "vsx-timode".
9772 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
9773 from function comment.
9774 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
9775 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
9776 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
9778 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
9779 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
9780 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
9783 (define_peephole2): Likewise.
9785 2017-08-17 Martin Sebor <msebor@redhat.com>
9788 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
9789 past the end of an array.
9790 (test_pp_format): Add test cases.
9792 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
9794 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
9795 missing ECF_NOTHROW flags.
9797 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9800 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
9801 operands residing in integer registers.
9802 (*vsx_le_perm_load_<mode>): Likewise.
9803 (*vsx_le_perm_store_<mode>): Likewise.
9804 (define_peephole2): Add peepholes to optimize the above.
9806 2017-08-17 Marek Polacek <polacek@redhat.com>
9809 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
9810 to mimic what shorten_compare did. Change the return type to bool.
9811 (fold_cond_expr_with_comparison): Update call to
9812 operand_equal_for_comparison_p.
9813 (fold_ternary_loc): Likewise.
9815 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
9817 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
9819 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
9821 2017-08-17 Richard Biener <rguenther@suse.de>
9823 * tree-ssa-structalias.c (solve_graph): When propagating
9824 to successors update the graphs succ edges and avoid duplicate work.
9826 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
9829 * config/i386/i386.c (ix86_option_override_internal): Save target
9830 specific options after ix86_stack_protector_guard_reg was changed.
9832 2017-08-17 Richard Biener <rguenther@suse.de>
9834 PR tree-optimization/81827
9835 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
9836 (new_var_info): Initialize it conservatively.
9837 (get_call_vi): Mark register vars.
9838 (new_scalar_tmp_constraint_exp): Likewise.
9839 (handle_rhs_call): Likewise.
9840 (handle_const_call): Likewise.
9841 (create_function_info_for): Likewise.
9842 (solve_constraints): Sort varinfos to separate register from
9843 non-register vars to pack points-to solution bitmaps during
9846 2017-08-17 Marek Polacek <polacek@redhat.com>
9848 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
9850 2017-08-17 Richard Biener <rguenther@suse.de>
9852 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
9853 to true when overflow is undefined and we saturated the result.
9855 2017-08-17 Alan Modra <amodra@gmail.com>
9858 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
9859 Don't use store multiple if only one reg needs saving.
9860 (interesting_frame_related_regno): New function.
9861 (rs6000_frame_related): Don't emit frame info for regs that
9863 (rs6000_emit_epilogue): Likewise.
9865 2017-08-16 Nathan Sidwell <nathan@acm.org>
9867 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
9868 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
9869 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
9870 * tree.c (free_lang_data_in_type): Use else-if chain. Always
9871 clear TYPE_LANG_1. Remove obsolete member-function stripping.
9872 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
9873 (verify_type): Adjust for TYPE_BINFO move.
9874 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
9875 process TYPE_BINFO directly.
9876 (hash_tree): Likewise.
9877 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9879 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9882 2017-08-16 David Malcolm <dmalcolm@redhat.com>
9884 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
9886 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
9889 * config/i386/i386.md (*anddi_1_btr): Change predicates of
9890 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
9891 Add ix86_binary_operator_ok to insn constraint.
9892 (*iordi_1_bts): Ditto.
9893 (*xordi_1_btc): Ditto.
9894 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
9895 Update corresponding peephole2 pattern.
9899 2017-08-16 Bin Cheng <bin.cheng@arm.com>
9901 PR tree-optimization/81832
9902 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
9903 copy loop header which has IFN_LOOP_DIST_ALIAS call.
9905 2017-08-16 Marek Polacek <polacek@redhat.com>
9908 * fold-const.c (fold_indirect_ref_1): Restore original behavior
9909 regarding size_zero_node.
9911 2017-08-16 Martin Liska <mliska@suse.cz>
9914 * config.gcc: Respect previously set extra_objs in case
9917 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
9919 PR tree-optimization/81835
9920 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
9921 the phi SSA_NAME. Check that the condition in a COND_EXPR does
9922 not depend on the phi.
9924 2017-08-16 Alan Modra <amodra@gmail.com>
9926 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
9929 2017-08-16 Alan Modra <amodra@gmail.com>
9931 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
9932 (save_reg_p): ..into this. Update all callers.
9933 (first_reg_to_save): Simplify.
9935 2017-08-16 Alan Modra <amodra@gmail.com>
9937 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
9940 2017-08-15 Joseph Myers <joseph@codesourcery.com>
9944 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
9945 constant count if that count is less than 32.
9947 2017-08-15 Nathan Sidwell <nathan@acm.org>
9949 * gcc.c (execute): Emit friendlier message if inferior is killed
9950 by an external cause.
9952 2017-08-15 Richard Biener <rguenther@suse.de>
9954 PR tree-optimization/81790
9955 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
9956 CONSTRUCTORs from simplifying and VN.
9958 2017-08-14 Martin Sebor <msebor@redhat.com>
9960 * builtin-attrs.def: Add comments.
9962 2017-08-14 Martin Sebor <msebor@redhat.com>
9965 * doc/extend.texi (attribute nonstring): Document new attribute.
9967 2017-08-14 Martin Sebor <msebor@redhat.com>
9970 * tree-diagnostic.c (default_tree_printer): Handle %G.
9971 * gimple-pretty-print.h (percent_G_format): Declare new function.
9972 * gimple-pretty-print.c (percent_G_format): Define.
9973 * tree-pretty-print.c (percent_K_format): Add argument.
9975 2017-08-14 Martin Sebor <msebor@redhat.com>
9977 PR translation/79998
9978 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
9979 Remove a stray space.
9981 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
9984 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
9985 (*iordi_1_bts): Ditto.
9986 (*xordi_1_btc): Ditto.
9988 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9991 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
9992 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9994 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
9995 quoted strings, and make more translator-friendly.
9996 (darwin_rs6000_override_options): Likewise.
9997 (rs6000_option_override_internal): Likewise.
9998 (rs6000_return_in_memory): Fix overlong line.
9999 (init_cmulative_args): Use quoted strings, and make more
10000 translator-friendly.
10001 (rs6000_pass_by_reference): Fix overlong line.
10002 (def_builtin): Use quoted strings.
10003 (altivec_expand_predicate_builtin): Use quoted strings, and make
10004 more translator-friendly.
10005 (htm_expand_builtin): Use quoted strings.
10006 (cpu_expand_builtin): Use quoted strings, and make more
10007 translator-friendly.
10008 (altivec_expand_builtin): Likewise.
10009 (paired_expand_predicate_builtin): Likewise.
10010 (rs6000_invalid_builtin): Likewise.
10011 (builtin_function_type): Use quoted strings.
10012 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
10013 more translator-friendly.
10014 (rs6000_trampoline_init): Likewise.
10015 (rs6000_handle_altivec_attribute): Likewise.
10016 (rs6000_inner_target_options): Use quoted strings.
10017 (rs6000_disable_incompatible_switches): Likewise.
10018 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
10019 strings, and make more translator-friendly.
10020 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
10022 2017-08-14 Bin Cheng <bin.cheng@arm.com>
10024 PR tree-optimization/81799
10025 * tree-loop-distribution.c (version_loop_by_alias_check): Force
10026 cond_expr to simple gimple operand.
10028 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
10030 PR middle-end/46932
10031 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
10033 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
10037 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
10038 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
10039 TARGET_GASISR_PROLOGUES.
10040 * config/avr/avr.c (avr_option_override): Same.
10041 (avr_pass_pre_proep::execute): Same.
10043 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
10046 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
10047 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
10049 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
10051 * config/i386/i386.md (*load_tp_<mode>): Redefine as
10052 define_insn_and_split. Split to a memory load from 0 in
10053 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
10054 using PTR mode iterator.
10055 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
10056 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
10057 (*add_tp_<mode>): Redefine as define_insn_and_split.
10058 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
10059 address space. Merge with *add_tp_x32 using PTR mode iterator.
10060 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
10061 Split to an add with a memory load from 0 in
10062 DEFAULT_TLS_SEG_REG address space.
10064 2017-08-12 Andrew Pinski <apinski@cavium.com>
10066 * config/aarch64/aarch64-option-extensions.def (rdma):
10067 Fix feature string to what Linux prints out in /proc/cpuinfo.
10069 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10072 * dwarf2out.c (modified_type_die): For C typedef types that have
10073 an ultimate origin, process the ultimate origin instead of the
10075 (gen_typedef_die): Assert that input DECLs have no ultimate
10077 (gen_type_die_with_usage): For typedef variants that have an
10078 ultimate origin, just call gen_decl_die on the original DECL.
10079 (process_scope_var): Avoid creating DIEs for local typedefs and
10080 concrete static variables.
10082 2017-08-12 Alan Modra <amodra@gmail.com>
10086 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10087 match gnu-user.h startfile.
10088 (ENDFILE_LINUX_SPEC): Similarly.
10090 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10093 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10095 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10097 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10098 * config/aarch64/aarch64.md (mov<mode>): Change.
10099 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10100 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10101 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10103 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10105 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10106 for storage order barriers.
10108 2017-08-11 Martin Liska <mliska@suse.cz>
10111 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10112 variables of void type.
10114 2017-08-11 Martin Liska <mliska@suse.cz>
10116 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10117 TARGET_SUPPORTS_ALIASES.
10118 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10119 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10120 (optimize_weakref): Likewise.
10121 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10122 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10123 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10125 2017-08-11 Martin Liska <mliska@suse.cz>
10128 * config/i386/i386.c (make_resolver_func): Do complete
10129 refactoring of the function.
10131 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10134 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10135 * config/i386/i386.c (ix86_stack_protect_guard): Use
10136 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10137 * doc/invoke.texi (x86 Options): Document
10138 -mstack-protector-guard-symbol= option.
10140 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10142 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10143 * config/i386/i386.c (ix86_split_stack_guard): New function.
10144 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
10145 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
10146 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
10147 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
10148 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
10149 (split_stack_space_check): Call ix86_split_stack_guard.
10151 2017-08-10 Martin Sebor <msebor@redhat.com>
10153 * print-tree.c (print_node): Print location using the established
10155 Replace spaces with colons.
10156 (debug_raw, debug): Ditto.
10158 2017-08-10 Martin Sebor <msebor@redhat.com>
10161 * pretty-print.c (pp_format): Correct the handling of %s precision.
10163 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
10166 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10168 (ix86_finalize_stack_frame_flags): This. Also clear
10169 frame_pointer_needed if -fno-omit-frame-pointer is used without
10171 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10172 with ix86_finalize_stack_frame_flags.
10173 (ix86_expand_epilogue): Likewise.
10174 (ix86_expand_split_stack_prologue): Likewise.
10175 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
10177 2017-08-10 Martin Liska <mliska@suse.cz>
10180 * c-attribs.c (handle_target_attribute):
10181 Report warning for an empty string argument of target attribute.
10183 2017-08-09 Jakub Jelinek <jakub@redhat.com>
10186 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
10188 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
10189 or DECL_NONLOCAL labels.
10190 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
10191 or DECL_NONLOCAL labels here.
10193 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
10195 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
10196 to indicate when early gimple folding has been disabled.
10197 (rs6000_gimple_fold_builtin): Add debug content.
10198 (rs6000_invalid_builtin): Fix whitespace.
10199 (rs6000_expand_builtin): Fix whitespace.
10200 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
10202 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
10205 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
10206 SAVE_MULTIPLE if not all the registers that saves, should be saved.
10208 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
10210 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
10211 (qdf24xx): Likewise.
10212 * config/aarch64/aarch64.md: Include falkor.md.
10213 * config/aarch64/falkor.md: New.
10215 2017-08-09 Marek Polacek <polacek@redhat.com>
10218 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
10219 * diagnostic.c (emit_diagnostic): Add a comment.
10220 (emit_diagnostic_valist): New function.
10222 2017-08-09 Marek Polacek <polacek@redhat.com>
10225 * input.c (make_location): New overload.
10226 * input.h (make_location): Declare.
10228 2017-08-08 Alan Modra <amodra@gmail.com>
10229 H.J. Lu <hongjiu.lu@intel.com>
10232 * gcc.c (NO_PIE_SPEC): Delete.
10233 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
10235 (LINK_PIE_SPEC): ..to here.
10236 (LINK_COMMAND_SPEC): Support -no-pie.
10237 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
10238 chain of crtbegin*.o selection, update for PIE_SPEC changes and
10240 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
10241 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
10242 (ENDFILE_CRTEND_SPEC): Similarly.
10244 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
10247 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
10248 (mstack-protector-guard-offset=): Ditto.
10249 * config/i386/i386.c (ix86_option_override): Handle
10250 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
10252 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
10253 ix86_stack_protect_guard_offset variables.
10254 (TARGET_STACK_PROTECT_GUARD): Always define.
10255 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
10256 and -mstack-protector-guard-offset= options.
10258 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10260 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
10261 boundary case for the last candidate.
10263 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10265 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
10266 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
10268 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10270 PR middle-end/19706
10271 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
10272 * config/aarch64/aarch64-builtins.c
10273 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
10274 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
10275 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
10277 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10278 Andrew Pinski <pinskia@gmail.com>
10280 PR middle-end/19706
10281 * internal-fn.def (XORSIGN): New.
10282 * optabs.def (xorsign_optab): New.
10283 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
10284 (convert_expand_mult_copysign): New.
10285 (pass_optimize_widening_mul::execute): Call
10286 convert_expand_mult_copysign.
10288 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10290 PR tree-optimization/81354
10291 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
10292 Insert on edges rather than explicitly creating landing pads.
10293 (analyze_candidates_and_replace): Commit edge inserts.
10295 2017-08-08 Richard Biener <rguenther@suse.de>
10297 PR middle-end/81719
10298 * tree-ssa-loop-niter.c: Include tree-dfa.h.
10299 (expand_simple_operations): Also look through ADDR_EXPRs with
10300 MEM_REF bases treating them as POINTER_PLUS_EXPR.
10302 2017-08-08 Richard Biener <rguenther@suse.de>
10304 PR tree-optimization/81723
10305 * tree-vect-slp.c (struct bst_traits): New hash traits.
10306 (bst_fail): New global.
10307 (vect_build_slp_tree_2): New worker, split out from ...
10308 (vect_build_slp_tree): ... this now wrapping it with using
10309 bst_fail set to cache SLP tree build fails. Properly handle
10311 (vect_analyze_slp_instance): Allocate and free bst_fail.
10313 2017-08-08 Martin Liska <mliska@suse.cz>
10316 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
10317 LABEL_DECLs that can be from a different function.
10319 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10321 PR tree-optimization/81744
10322 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
10323 loop's number of iterations.
10325 2017-08-08 Martin Liska <mliska@suse.cz>
10327 * asan.c: Include header files.
10328 * attribs.c (build_decl_attribute_variant): New function moved
10330 (build_type_attribute_qual_variant): Likewise.
10331 (cmp_attrib_identifiers): Likewise.
10332 (simple_cst_list_equal): Likewise.
10333 (omp_declare_simd_clauses_equal): Likewise.
10334 (attribute_value_equal): Likewise.
10335 (comp_type_attributes): Likewise.
10336 (build_type_attribute_variant): Likewise.
10337 (lookup_ident_attribute): Likewise.
10338 (remove_attribute): Likewise.
10339 (merge_attributes): Likewise.
10340 (merge_type_attributes): Likewise.
10341 (merge_decl_attributes): Likewise.
10342 (merge_dllimport_decl_attributes): Likewise.
10343 (handle_dll_attribute): Likewise.
10344 (attribute_list_equal): Likewise.
10345 (attribute_list_contained): Likewise.
10346 * attribs.h (lookup_attribute): New function moved from tree.[ch].
10347 (lookup_attribute_by_prefix): Likewise.
10348 * bb-reorder.c: Include header files.
10349 * builtins.c: Likewise.
10350 * calls.c: Likewise.
10351 * cfgexpand.c: Likewise.
10352 * cgraph.c: Likewise.
10353 * cgraphunit.c: Likewise.
10354 * convert.c: Likewise.
10355 * dwarf2out.c: Likewise.
10356 * final.c: Likewise.
10357 * fold-const.c: Likewise.
10358 * function.c: Likewise.
10359 * gimple-expr.c: Likewise.
10360 * gimple-fold.c: Likewise.
10361 * gimple-pretty-print.c: Likewise.
10362 * gimple.c: Likewise.
10363 * gimplify.c: Likewise.
10364 * hsa-common.c: Likewise.
10365 * hsa-gen.c: Likewise.
10366 * internal-fn.c: Likewise.
10367 * ipa-chkp.c: Likewise.
10368 * ipa-cp.c: Likewise.
10369 * ipa-devirt.c: Likewise.
10370 * ipa-fnsummary.c: Likewise.
10371 * ipa-inline.c: Likewise.
10372 * ipa-visibility.c: Likewise.
10374 * lto-cgraph.c: Likewise.
10375 * omp-expand.c: Likewise.
10376 * omp-general.c: Likewise.
10377 * omp-low.c: Likewise.
10378 * omp-offload.c: Likewise.
10379 * omp-simd-clone.c: Likewise.
10380 * opts-global.c: Likewise.
10381 * passes.c: Likewise.
10382 * predict.c: Likewise.
10383 * sancov.c: Likewise.
10384 * sanopt.c: Likewise.
10385 * symtab.c: Likewise.
10386 * toplev.c: Likewise.
10387 * trans-mem.c: Likewise.
10388 * tree-chkp.c: Likewise.
10389 * tree-eh.c: Likewise.
10390 * tree-into-ssa.c: Likewise.
10391 * tree-object-size.c: Likewise.
10392 * tree-parloops.c: Likewise.
10393 * tree-profile.c: Likewise.
10394 * tree-ssa-ccp.c: Likewise.
10395 * tree-ssa-live.c: Likewise.
10396 * tree-ssa-loop.c: Likewise.
10397 * tree-ssa-sccvn.c: Likewise.
10398 * tree-ssa-structalias.c: Likewise.
10399 * tree-ssa.c: Likewise.
10400 * tree-streamer-in.c: Likewise.
10401 * tree-vectorizer.c: Likewise.
10402 * tree-vrp.c: Likewise.
10403 * tsan.c: Likewise.
10404 * ubsan.c: Likewise.
10405 * varasm.c: Likewise.
10406 * varpool.c: Likewise.
10407 * tree.c: Remove functions moved to attribs.[ch].
10408 * tree.h: Likewise.
10409 * config/aarch64/aarch64.c: Add attrs.h header file.
10410 * config/alpha/alpha.c: Likewise.
10411 * config/arc/arc.c: Likewise.
10412 * config/arm/arm.c: Likewise.
10413 * config/avr/avr.c: Likewise.
10414 * config/bfin/bfin.c: Likewise.
10415 * config/c6x/c6x.c: Likewise.
10416 * config/cr16/cr16.c: Likewise.
10417 * config/cris/cris.c: Likewise.
10418 * config/darwin.c: Likewise.
10419 * config/epiphany/epiphany.c: Likewise.
10420 * config/fr30/fr30.c: Likewise.
10421 * config/frv/frv.c: Likewise.
10422 * config/ft32/ft32.c: Likewise.
10423 * config/h8300/h8300.c: Likewise.
10424 * config/i386/winnt.c: Likewise.
10425 * config/ia64/ia64.c: Likewise.
10426 * config/iq2000/iq2000.c: Likewise.
10427 * config/lm32/lm32.c: Likewise.
10428 * config/m32c/m32c.c: Likewise.
10429 * config/m32r/m32r.c: Likewise.
10430 * config/m68k/m68k.c: Likewise.
10431 * config/mcore/mcore.c: Likewise.
10432 * config/microblaze/microblaze.c: Likewise.
10433 * config/mips/mips.c: Likewise.
10434 * config/mmix/mmix.c: Likewise.
10435 * config/mn10300/mn10300.c: Likewise.
10436 * config/moxie/moxie.c: Likewise.
10437 * config/msp430/msp430.c: Likewise.
10438 * config/nds32/nds32-isr.c: Likewise.
10439 * config/nds32/nds32.c: Likewise.
10440 * config/nios2/nios2.c: Likewise.
10441 * config/nvptx/nvptx.c: Likewise.
10442 * config/pa/pa.c: Likewise.
10443 * config/pdp11/pdp11.c: Likewise.
10444 * config/powerpcspe/powerpcspe.c: Likewise.
10445 * config/riscv/riscv.c: Likewise.
10446 * config/rl78/rl78.c: Likewise.
10447 * config/rx/rx.c: Likewise.
10448 * config/s390/s390.c: Likewise.
10449 * config/sh/sh.c: Likewise.
10450 * config/sol2.c: Likewise.
10451 * config/sparc/sparc.c: Likewise.
10452 * config/spu/spu.c: Likewise.
10453 * config/stormy16/stormy16.c: Likewise.
10454 * config/tilegx/tilegx.c: Likewise.
10455 * config/tilepro/tilepro.c: Likewise.
10456 * config/v850/v850.c: Likewise.
10457 * config/vax/vax.c: Likewise.
10458 * config/visium/visium.c: Likewise.
10459 * config/xtensa/xtensa.c: Likewise.
10461 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10464 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
10465 constraints since the -mupper-regs-* switches have been
10467 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
10468 into a vector from a double word element that was extracted from
10469 another vector, and eliminate extra XXPERMDI instructions.
10470 (vsx_concat_<mode>_2): Likewise.
10471 (vsx_concat_<mode>_3): Likewise.
10472 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
10473 concat to allow optimizing inserts from previous extracts.
10475 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
10477 * config/i386/i386.c (ix86_stack_protect_guard): Generate
10478 memory reference to a SSP offset in TLS address space.
10479 (ix86_print_operand) <case '@'>: Remove.
10480 (ix86_print_operand_punct_valid_p): Remove '@' code.
10481 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
10482 UNSPEC_SP_TLS_TEST.
10483 (stack_tls_protect_set_<mode>): Remove.
10484 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
10485 (stack_tls_protect_test_<mode>): Remove.
10486 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
10488 2017-08-07 Olivier Hainque <hainque@adacore.com>
10491 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
10493 2017-08-07 Douglas Rupp <rupp@adacore.com>
10495 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
10496 variable was referenced as multidir in command.
10498 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10501 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
10504 2017-08-07 Martin Liska <mliska@suse.cz>
10506 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
10507 * config/rl78/rl78.c: Add include of attribs.h.
10508 * config/sh/sh.c: Likewise.
10509 * config/v850/v850.c: Likewise.
10511 2017-08-07 Tom de Vries <tom@codesourcery.com>
10513 PR middle-end/78266
10514 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
10516 2017-08-07 Martin Liska <mliska@suse.cz>
10518 * config/mips/mips.c: Include attribs.h.
10520 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
10523 * doc/invoke.texi: Document change in behvaior for -Ofast for
10526 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
10528 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
10530 (aarch64_pop_regs): Likewise.
10531 (aarch64_gen_load_pair): Likewise.
10532 (aarch64_save_callee_saves): Likewise.
10533 (aarch64_restore_callee_saves): Likewise.
10535 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10537 * config/i386/i386.c: Revert the last change.
10539 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10542 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10544 (ix86_finalize_stack_frame_flags): This. Also clear
10545 frame_pointer_needed if -fno-omit-frame-pointer is used without
10547 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10548 with ix86_finalize_stack_frame_flags.
10549 (ix86_expand_epilogue): Likewise.
10550 (ix86_expand_split_stack_prologue): Likewise.
10552 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10555 * config/i386/i386.c (get_builtin_code_for_version): Set priority
10556 to P_AES for Westmere.
10558 2017-08-07 Jonathan Yong <10walls@gmail.com>
10560 * config/i386/mingw.opt (fset-stack-executable): Removed.
10561 * config/i386/cygming.opt (fset-stack-executable): Moved
10563 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
10565 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
10567 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
10569 2017-08-07 Marek Polacek <polacek@redhat.com>
10571 PR middle-end/81737
10572 * fold-const.c (fold_indirect_ref_1): Check type_domain.
10574 2017-08-07 Martin Liska <mliska@suse.cz>
10576 * attribs.h (canonicalize_attr_name): New function.
10577 (cmp_attribs): Move from c-format.c and adjusted.
10578 (is_attribute_p): Moved from tree.h.
10579 * tree-inline.c: Add new includes.
10580 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
10581 (private_is_attribute_p): Remove.
10582 (private_lookup_attribute): Likewise.
10583 (private_lookup_attribute_by_prefix): Simplify.
10584 (remove_attribute): Use is_attribute_p.
10585 * tree.h: Remove removed declarations.
10587 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10589 PR middle-end/81698
10590 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
10591 instead of computing it in the function. Formatting fix.
10592 (expand_case): Don't rely on default_edge being the first edge,
10593 clear it if removing it, pass default_edge to
10594 emit_case_dispatch_table.
10595 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
10598 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
10600 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
10601 insn in the function, emit NOP after the insn.
10603 2017-08-06 Tom de Vries <tom@codesourcery.com>
10605 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
10608 2017-08-06 Tom de Vries <tom@codesourcery.com>
10610 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
10613 2017-08-04 Yury Gribov <tetra2005@gmail.com>
10615 PR tree-optimization/57371
10616 * match.pd: New pattern.
10618 2017-08-04 Marek Polacek <polacek@redhat.com>
10620 PR middle-end/81695
10621 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
10622 perform the computation in offset_int.
10624 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10626 PR tree-optimization/81136
10627 * tree-vectorizer.h: Include tree-hash-traits.h.
10628 (vec_base_alignments): New typedef.
10629 (vec_info): Add a base_alignments field.
10630 (vect_record_base_alignments): Declare.
10631 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
10633 (DR_IS_CONDITIONAL_IN_STMT): New macro.
10634 (create_data_ref): Add an is_conditional_in_stmt argument.
10635 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
10636 the is_conditional_in_stmt field.
10637 (data_ref_loc): Add an is_conditional_in_stmt field.
10638 (get_references_in_stmt): Set the is_conditional_in_stmt field.
10639 (find_data_references_in_stmt): Update call to create_data_ref.
10640 (graphite_find_data_references_in_stmt): Likewise.
10641 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
10642 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
10643 (vect_record_base_alignment): New function.
10644 (vect_record_base_alignments): Likewise.
10645 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
10646 for nested statements even if we fail to compute a misalignment.
10647 Use pooled base alignments for unconditional references.
10648 (vect_find_same_alignment_drs): Compare base addresses instead
10650 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
10651 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
10653 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10655 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
10656 Add an explicit name for the enum. Use auto_vec for slp_instances
10657 and grouped_stores.
10658 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
10660 (_bb_vec_info): Add a constructor and destructor.
10661 (vinfo_for_stmt): Return NULL for uids of -1 as well.
10662 (destroy_loop_vec_info): Delete.
10663 (vect_destroy_datarefs): Likewise.
10664 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
10665 (vec_info::vec_info): New function.
10666 (vec_info::~vec_info): Likewise.
10667 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
10668 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
10669 destroy_loop_vec_info.
10670 * tree-vect-loop.c (new_loop_vec_info): Replace with...
10671 (_loop_vec_info::_loop_vec_info): ...this.
10672 (destroy_loop_vec_info): Replace with...
10673 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
10674 the stmt_vec_infos. Leave handling of vec_info information to its
10675 destructor. Remove explicit vector releases.
10676 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
10677 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
10678 * tree-vect-slp.c (new_bb_vec_info): Replace with...
10679 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
10680 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
10681 (destroy_bb_vec_info): Replace with...
10682 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
10683 information to its destructor.
10684 (vect_slp_analyze_bb_1): Use new and delete instead of
10685 new_bb_vec_info and destroy_bb_vec_info.
10686 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
10689 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10691 * tree-data-ref.h (subscript): Add access_fn field.
10692 (data_dependence_relation): Add could_be_independent_p.
10693 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
10694 (same_access_functions): Move to tree-data-ref.c.
10695 * tree-data-ref.c (ref_contains_union_access_p): New function.
10696 (access_fn_component_p): Likewise.
10697 (access_fn_components_comparable_p): Likewise.
10698 (dr_analyze_indices): Add a reference to access_fn_component_p.
10699 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
10701 (constant_access_functions): Likewise.
10702 (add_other_self_distances): Likewise.
10703 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
10704 (initialize_data_dependence_relation): Use XCNEW and remove
10705 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
10706 of access functions that have the same type. Allow the
10707 subsequence to end with different bases in some circumstances.
10708 Record the chosen access functions in SUB_ACCESS_FN.
10709 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
10710 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
10711 (subscript_dependence_tester_1): Likewise dra and drb.
10712 (build_classic_dist_vector): Update calls accordingly.
10713 (subscript_dependence_tester): Likewise.
10714 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
10715 DDR_COULD_BE_INDEPENDENT_P.
10716 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
10717 comp_alias_ddrs instead of may_alias_ddrs.
10718 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
10720 (vect_analyze_data_ref_dependence): Use it if
10721 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
10722 distance vectors if that fails.
10723 (dependence_distance_ge_vf): New function.
10724 (vect_prune_runtime_alias_test_list): Use it. Don't clear
10725 LOOP_VINFO_MAY_ALIAS_DDRS.
10727 2017-08-04 Richard Biener <rguenther@suse.de>
10729 PR middle-end/81705
10730 * fold-const.c (fold_binary_loc): Properly restrict
10731 minus_var0 && minus_var1 case when associating undefined overflow
10734 2017-08-04 Tom de Vries <tom@codesourcery.com>
10736 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
10738 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10740 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10741 Don't start diagnostic messages with a capital letter.
10742 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10744 (rs6000_invalid_builtin): Likewise.
10745 (rs6000_trampoline_init): Likewise.
10747 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10750 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
10751 after setting changeable df flags.
10753 2017-08-03 Richard Biener <rguenther@suse.de>
10755 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
10756 up if the use is in USE - X.
10758 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
10760 * toplev.c (dumpfile.h): New include.
10761 (internal_error_reentered): New static function. Use it...
10762 (internal_error_function): ...here to handle reentered internal_error.
10764 2017-08-03 Richard Biener <rguenther@suse.de>
10766 PR middle-end/81148
10767 * fold-const.c (split_tree): Add minus_var and minus_con
10768 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
10769 here but always use minus_*.
10770 (associate_trees): Assert we never associate with MINUS_EXPR
10771 and NULL first operand. Do not recurse for PLUS_EXPR operands
10772 when associating as MINUS_EXPR either.
10773 (fold_binary_loc): Track minus_var and minus_con.
10775 2017-08-03 Tom de Vries <tom@codesourcery.com>
10778 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
10779 ACCEL_COMPILER, apply finish_options on
10780 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
10782 2017-08-03 Tom de Vries <tom@codesourcery.com>
10785 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
10786 function_entry_patch_area_size > 0.
10788 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10791 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
10792 instead of 10??LU, perform unit multiplication in wide_int,
10793 don't change alloc_object_size_limit if the limit is larger
10796 PR tree-optimization/81655
10797 PR tree-optimization/81588
10798 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
10799 the case when ranges[i].low and high are 1 for unsigned type with
10802 PR middle-end/81052
10803 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
10804 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
10806 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10808 * tree-vrp.h: Add include guard.
10810 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
10813 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
10814 (ud2): New insn pattern.
10815 * config/i386/i386.c (ix86_expand_epilogue):
10816 For naked functions, generate ud2 instead of trap insn.
10818 2017-08-02 Marek Polacek <polacek@redhat.com>
10821 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
10823 2017-08-02 Tom de Vries <tom@codesourcery.com>
10824 Cesar Philippidis <cesar@codesourcery.com>
10826 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
10827 Add missing edge probabilities.
10829 2017-08-02 Tamar Christina <tamar.christina@arm.com>
10831 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
10832 Correct endianness.
10834 2017-08-02 Jakub Jelinek <jakub@redhat.com>
10836 PR middle-end/79499
10837 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
10838 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
10839 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
10840 prologue_seq sequences - if any.
10842 2017-08-02 Richard Biener <rguenther@suse.de>
10844 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
10845 via vectors if supported, integer extracts via punning if supported
10846 or otherwise vector extracts.
10848 2017-08-02 Richard Biener <rguenther@suse.de>
10850 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
10852 (bitmap_insert_into_set): ... this.
10854 2017-08-02 Richard Biener <rguenther@suse.de>
10856 PR tree-optimization/81633
10858 2015-08-17 Alan Hayward <alan.hayward@arm.com>
10860 PR tree-optimization/71752
10861 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
10863 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
10865 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
10866 (machine_function::call_ms2sysv_pad_out): Remove field.
10867 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
10868 (ix86_compute_frame_layout): Likewise.
10870 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
10873 * config/i386/i386.c (ix86_set_func_type): Disallow naked
10874 attribute with interrupt attribute.
10876 2017-08-01 Andrew Pinski <apinski@cavium.com>
10878 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
10879 BIT_INSERT_EXPR's operand 1
10880 to see if the types precision matches.
10882 2017-08-01 Martin Liska <mliska@suse.cz>
10884 PR middle-end/70140
10885 * builtins.c (expand_builtin_memcpy_args): Remove.
10886 (expand_builtin_memcpy): Call newly added function
10887 expand_builtin_memory_copy_args.
10888 (expand_builtin_memcpy_with_bounds): Likewise.
10889 (expand_builtin_mempcpy): Remove last argument.
10890 (expand_builtin_mempcpy_with_bounds): Likewise.
10891 (expand_builtin_memory_copy_args): New function created from
10892 expand_builtin_mempcpy_args with small modifications.
10893 (expand_builtin_mempcpy_args): Remove.
10894 (expand_builtin_stpcpy): Remove unused argument.
10895 (expand_builtin): Likewise.
10896 (expand_builtin_with_bounds): Likewise.
10898 2017-08-01 Martin Liska <mliska@suse.cz>
10901 Make mempcpy more optimal (PR middle-end/70140).
10903 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10906 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10907 __builtin_vec_cmpne verify both arguments are compatible vectors
10908 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
10909 verify arg1_type is a pointer or array type. For __builtin_vec_st,
10910 move computation of aligned to after checking the argument types.
10914 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10915 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
10918 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
10919 Doug Rupp <rupp@adacore.com>
10920 Olivier Hainque <hainque@adacore.com>
10922 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
10923 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
10925 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
10926 for TARGET_OS_CPP_BUILTIN.
10927 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
10928 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
10930 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
10931 passing required abi options to the assembler for EABI configurations.
10932 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
10933 of .text.hot and .text.unlikely sections for kernel modules when
10934 using ARM style exceptions.
10935 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
10936 options. Add EXTRA_CC1_SPEC.
10937 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
10939 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
10941 (ARM_TARGET2_DWARF_FORMAT): Define.
10942 * config/arm/t-vxworks: Adjust multilib control to removal of the
10943 Diab command line options.
10945 2017-08-01 Martin Liska <mliska@suse.cz>
10947 PR gcov-profile/81561
10948 * gcov.c (unblock): Make unblocking safe as we need to preserve
10949 index correspondence of blocks and block_lists.
10951 2017-08-01 Richard Biener <rguenther@suse.de>
10953 PR tree-optimization/81181
10954 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10955 (compute_antic): ... end of iteration here.
10957 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
10959 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
10960 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
10961 (ftree-slp-vectorize): Likewise.
10962 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
10963 can no longer be set independent of flag_tree_loop_vectorize.
10964 * omp-general.c (emp_max_vf): Likewise.
10965 * opts.c (enable_fdo_optimizations): Remove references to
10966 flag_tree_vectorize, these are now implicit.
10967 (common_handle_option): Remove handling for OPT_ftree_vectorize,
10968 and leave it for the options machinery.
10970 2017-08-01 Martin Liska <mliska@suse.cz>
10972 PR middle-end/70140
10973 * builtins.c (expand_builtin_memcpy_args): Remove.
10974 (expand_builtin_memcpy): Call newly added function
10975 expand_builtin_memory_copy_args.
10976 (expand_builtin_memcpy_with_bounds): Likewise.
10977 (expand_builtin_mempcpy): Remove last argument.
10978 (expand_builtin_mempcpy_with_bounds): Likewise.
10979 (expand_builtin_memory_copy_args): New function created from
10980 expand_builtin_mempcpy_args with small modifications.
10981 (expand_builtin_mempcpy_args): Remove.
10982 (expand_builtin_stpcpy): Remove unused argument.
10983 (expand_builtin): Likewise.
10984 (expand_builtin_with_bounds): Likewise.
10986 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10989 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
10990 print "ds:" only for immediates in generic address space.
10992 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10995 * config/i386/i386.c (ix86_funciton_naked): New prototype.
10996 (ix86_function_ok_for_sibcall): Return false for naked functions.
10998 2017-08-01 Richard Biener <rguenther@suse.de>
11000 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
11001 (compute_antic): Seed worklist with exit block predecessors.
11002 * cfganal.c (dfs_find_deadend): For a cycle return the source
11003 of the edge closing it.
11005 2017-08-01 Tamar Christina <tamar.christina@arm.com>
11007 * config/aarch64/aarch64.c
11008 (aarch64_can_const_movi_rtx_p): Move 0 check.
11010 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11012 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
11013 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
11015 * match.pd: Ditto in address comparison pattern.
11017 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11019 PR tree-optimization/81627
11020 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
11021 closed ssa form for store-store chain.
11023 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11025 PR tree-optimization/81620
11026 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
11027 for store-store chain.
11029 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11031 PR tree-optimization/81588
11032 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
11033 ranges[i].in_p, invert comparison code ccode. For >/>=,
11034 swap rhs1 and rhs2 and comparison code unconditionally,
11035 for </<= don't do that. Don't swap rhs1/rhs2 again if
11036 ranges[i].in_p, instead invert comparison code ccode if
11037 opcode or oe->rank is BIT_IOR_EXPR.
11040 * optabs.def (vec_extract_optab, vec_init_optab): Change from
11041 a direct optab to conversion optab.
11042 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
11043 with GET_MODE_INNER as last argument instead of optab_handler.
11044 * expmed.c (extract_bit_field_1): Likewise. Use vector from
11045 vector extraction if possible and optab is available.
11046 * expr.c (store_constructor): Use convert_optab_handler instead
11047 of optab_handler. Use vector initialization from smaller
11048 vectors if possible and optab is available.
11049 * tree-vect-stmts.c (vectorizable_load): Likewise.
11050 * doc/md.texi (vec_extract, vec_init): Document that the optabs
11051 now have two modes.
11052 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
11053 of vec_init from half-sized vectors with the same element mode.
11054 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
11055 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
11056 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
11057 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
11058 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
11059 after mode in gen_vec_extract* calls.
11060 (vec_extract<mode>): Renamed to ...
11061 (vec_extract<mode><ssescalarmodelower>): ... this.
11062 (vec_extract<mode><ssehalfvecmodelower>): New expander.
11063 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
11064 element mode after mode in gen_vec_init* calls.
11065 (VEC_INIT_HALF_MODE): New mode iterator.
11066 (vec_init<mode>): Renamed to ...
11067 (vec_init<mode><ssescalarmodelower>): ... this.
11068 (vec_init<mode><ssehalfvecmodelower>): New expander.
11069 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11070 (vec_extractv2sfsf): ... this.
11071 (vec_initv2sf): Renamed to ...
11072 (vec_initv2sfsf): ... this.
11073 (vec_extractv2si): Renamed to ...
11074 (vec_extractv2sisi): ... this.
11075 (vec_initv2si): Renamed to ...
11076 (vec_initv2sisi): ... this.
11077 (vec_extractv4hi): Renamed to ...
11078 (vec_extractv4hihi): ... this.
11079 (vec_initv4hi): Renamed to ...
11080 (vec_initv4hihi): ... this.
11081 (vec_extractv8qi): Renamed to ...
11082 (vec_extractv8qiqi): ... this.
11083 (vec_initv8qi): Renamed to ...
11084 (vec_initv8qiqi): ... this.
11085 * config/rs6000/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/rs6000/paired.md (vec_initv2sf): Renamed to ...
11091 (vec_initv2sfsf): ... this.
11092 * config/rs6000/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, altivec_vreve<mode>2): Add
11095 element mode after mode in gen_vec_init* calls.
11096 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11097 (vec_init<mode><Vel>): ... this.
11098 (vec_extract<mode>): Renamed to ...
11099 (vec_extract<mode><Vel>): ... this.
11100 * config/aarch64/iterators.md (Vel): New mode attribute.
11101 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11102 Add element mode after mode in gen_vec_extract* calls.
11103 * config/s390/vector.md (non_vec_l): New mode attribute.
11104 (vec_extract<mode>): Renamed to ...
11105 (vec_extract<mode><non_vec_l>): ... this.
11106 (vec_init<mode>): Renamed to ...
11107 (vec_init<mode><non_vec_l>): ... this.
11108 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11109 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11111 * config/arm/iterators.md (V_elem_l): New mode attribute.
11112 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11113 (vec_extract<mode><V_elem_l>): ... this.
11114 (vec_extractv2di): Renamed to ...
11115 (vec_extractv2didi): ... this.
11116 (vec_init<mode>): Renamed to ...
11117 (vec_init<mode><V_elem_l>): ... this.
11118 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11119 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11120 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11121 Add element mode after gen_vec_extract* calls.
11122 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11123 (vec_init<mode><unitmode>): ... this.
11124 (vec_extract<mode>): Renamed to ...
11125 (vec_extract<mode><unitmode>): ... this.
11126 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11127 (vec_init<mode><unitmode>): ... this.
11128 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11129 (vec_initv2sfsf): ... this.
11130 (vec_extractv2sf): Renamed to ...
11131 (vec_extractv2sfsf): ... this.
11132 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11133 Add element mode after gen_vec_extract* calls.
11134 * config/mips/mips.md (unitmode): New mode iterator.
11135 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11136 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11137 * config/spu/spu.md (inner_l): New mode attribute.
11138 (vec_init<mode>): Renamed to ...
11139 (vec_init<mode><inner_l>): ... this.
11140 (vec_extract<mode>): Renamed to ...
11141 (vec_extract<mode><inner_l>): ... this.
11142 * config/sparc/sparc.md (veltmode): New mode iterator.
11143 (vec_init<VMALL:mode>): Renamed to ...
11144 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
11145 * config/ia64/vect.md (vec_initv2si): Renamed to ...
11146 (vec_initv2sisi): ... this.
11147 (vec_initv2sf): Renamed to ...
11148 (vec_initv2sfsf): ... this.
11149 (vec_extractv2sf): Renamed to ...
11150 (vec_extractv2sfsf): ... this.
11151 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
11152 (vec_init<mode>): Renamed to ...
11153 (vec_init<mode><VEC_base_l>): ... this.
11154 (vec_extract<mode>): Renamed to ...
11155 (vec_extract<mode><VEC_base_l>): ... this.
11156 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
11157 (vec_initv2sfsf): ... this.
11158 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
11159 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11160 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
11161 gen_vec_init* calls.
11163 2017-08-01 Richard Biener <rguenther@suse.de>
11165 PR tree-optimization/81297
11166 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
11167 TREE_OVERFLOW from INTEGER_CSTs.
11169 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
11171 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
11173 2017-07-31 Carl Love <cel@us.ibm.com>
11175 * config/rs6000/rs6000-c: Add support for built-in functions
11176 vector signed char vec_xl_be (signed long long, signed char *);
11177 vector unsigned char vec_xl_be (signed long long, unsigned char *);
11178 vector signed int vec_xl_be (signed long long, signed int *);
11179 vector unsigned int vec_xl_be (signed long long, unsigned int *);
11180 vector signed long long vec_xl_be (signed long long, signed long long *);
11181 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
11182 vector signed short vec_xl_be (signed long long, signed short *);
11183 vector unsigned short vec_xl_be (signed long long, unsigned short *);
11184 vector double vec_xl_be (signed long long, double *);
11185 vector float vec_xl_be (signed long long, float *);
11186 * config/rs6000/altivec.h (vec_xl_be): Add #define.
11187 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
11188 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
11190 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
11191 (altivec_expand_builtin): Add switch statement to call
11192 altivec_expand_xl_be for each builtin.
11193 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
11194 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
11195 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
11196 __builtin_vsx_le_be_v16qi.
11197 * doc/extend.texi: Update the built-in documentation file for the
11198 new built-in functions.
11200 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11203 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
11205 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11207 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11209 * config.gcc: Add z14.
11210 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
11211 CPU model numbers for z13s and z14.
11212 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
11214 * config/s390/s390-opts.h (enum processor_type): Rename
11215 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11216 * config/s390/s390.c (processor_table): Add field for CPU name to
11217 be passed to Binutils.
11218 (s390_asm_output_machine_for_arch): Use the new field in
11219 processor_table for Binutils.
11220 (s390_expand_builtin): Replace arch12 with z14.
11221 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11222 (s390_get_sched_attrmask): Likewise.
11223 (s390_get_unit_mask): Likewise.
11224 * config/s390/s390.opt: Add z14 to processor_type enum.
11226 2017-07-31 Martin Jambor <mjambor@suse.cz>
11229 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
11230 regardless of optimization level.
11232 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
11233 Martin Liska <mliska@suse.cz>
11235 * predict.def: Remove old comment and adjust probability.
11236 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
11237 PREDICT statements.
11239 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11242 * config/i386/i386.c (ix86_function_naked): New function.
11243 (ix86_can_use_return_insn_p): Return false for naked functions.
11244 (ix86_expand_prologue): Skip prologue for naked functions.
11245 (ix86_expand_epilogue): Skip epilogue for naked functions
11246 and emit trap instruction.
11247 (ix86_warn_func_return): New function.
11248 (ix86_attribute_table): Add "naked" attribute specification.
11249 (TARGET_WARN_FUNC_RETURN): Define.
11250 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
11252 2017-07-31 Martin Liska <mliska@suse.cz>
11254 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
11255 (dump_gimple_bb_header): Always dump BB info.
11256 (pp_cfg_jump): Do not append info about BB when dumping a jump.
11258 2017-07-31 Martin Liska <mliska@suse.cz>
11261 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
11262 also with current_function_decl non-null equality.
11264 2017-07-31 Jakub Jelinek <jakub@redhat.com>
11267 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
11268 change type to the element type, instead add eltype variable and
11269 use it where we are interested in the element type.
11271 PR tree-optimization/81603
11272 * ipa-polymorphic-call.c
11273 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
11274 offset arithmetic in offset_int, bail out if the resulting bit offset
11275 doesn't fit into shwi.
11277 2017-07-31 Martin Liska <mliska@suse.cz>
11279 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
11280 (gimplify_save_expr): Fix comment.
11282 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
11285 * config/i386/i386.c (ix86_function_arg): Update arguments for
11287 (ix86_compute_frame_layout): Set the initial stack offset to
11288 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
11289 INCOMING_FRAME_SP_OFFSET.
11290 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
11291 stack before exception handler returns.
11292 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
11293 the 'ERROR_CODE' for exception handler.
11295 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
11297 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
11298 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
11299 (ASM_OUTPUT_REG_POP): Ditto.
11300 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
11301 instead of asm_fprintf to output pure string.
11303 2017-07-29 Jakub Jelinek <jakub@redhat.com>
11305 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
11306 to imported_module_or_decl hook.
11307 (debug_nothing_tree_tree_tree_bool): Remove.
11308 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
11309 * debug.c (do_nothing_debug_hooks): Use
11310 debug_nothing_tree_tree_tree_bool_bool instead of
11311 debug_nothing_tree_tree_tree_bool.
11312 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11313 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
11314 * sdbout.c (sdb_debug_hooks): Likewise.
11315 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
11316 (gen_namespace_die): Add DW_AT_export_symbols attribute if
11318 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
11319 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
11320 attribute, don't add anything.
11322 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11324 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
11325 (fold_build2_stat_loc): Likewise.
11326 (fold_build3_stat_loc): Likewise.
11327 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
11328 (fold_build1_loc): Remove macro.
11329 (fold_build2_loc): Likewise.
11330 (fold_build3_loc): Likewise.
11332 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11334 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
11335 (gimple_build_debug_bind_source_stat): Likewise.
11336 * gimple.h (gimple_build_debug_bind): Remove macro.
11337 (gimple_build_debug_bind_source): Likewise.
11339 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11341 * bitmap.c (bitmap_alloc): Adjust.
11342 (bitmap_gc_alloc): Likewise.
11343 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
11345 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11347 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
11348 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
11349 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
11350 (bitmap_gc_alloc_stat): Likewise.
11351 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
11353 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11355 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
11356 * rtl.h (shallow_copy_rtx): Remove macro.
11358 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11360 * emit-rtl.c (gen_raw_REG): Adjust.
11361 * gengenrtl.c (gendef): Likewise.
11362 * rtl.c (rtx_alloc_stat): Remove _stat from name.
11363 * rtl.h (rtx_alloc): Remove macro.
11365 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11367 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
11368 (build_tree_list_stat): Likewise.
11369 * tree.h (build_tree_list): Remove macro.
11370 (build_tree_list_vec): Likewise.
11372 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11374 * tree.c (make_vector_stat): Remove _stat from name.
11375 (build_vector_stat): Likewise.
11376 * tree.h (make_vector_stat): Remove macro.
11377 (build_vector_stat): Likewise.
11379 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11381 * tree.h (build_var_debug_value): Remove prototype.
11383 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11385 * tree.c (tree_cons_stat): Remove _stat from name.
11386 * tree.h (tree_cons): Remove macro.
11388 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11390 * tree.c (build_vl_exp_stat): Remove _stat from name.
11391 * tree.h (build_vl_exp): Remove macro.
11393 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11395 * tree.c (build_decl_stat): Remove _stat from name.
11396 * tree.h (build_decl): Remove macro.
11398 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11400 * gimple.c (gimple_build_with_ops_stat): Adjust.
11401 (gimple_alloc_stat): Remove _stat from name.
11402 * gimple.h (gimple_alloc): Remove macro.
11404 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11406 * tree.c (make_tree_vec_stat): Remove _stat from name.
11407 (grow_tree_vec_stat): Likewise.
11408 * tree.h (make_tree_vec_stat): Adjust prototype.
11409 (grow_tree_vec_stat): Likewise.
11410 (make_tree_vec): Remove macro.
11411 (grow_tree_vec): Likewise.
11413 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11415 * fold-const.c (fold_build1_stat_loc): Adjust.
11416 (fold_build2_stat_loc): Likewise.
11417 (fold_build3_stat_loc): Likewise.
11418 * tree.c (build0_stat): Remove _stat from name.
11419 (build1_stat): Likewise.
11420 (build2_stat): Likewise.
11421 (build3_stat): Likewise.
11422 (build4_stat): Likewise.
11423 (build5_stat): Likewise.
11424 * tree.h (build1_loc): Remove macro, and rename _stat function
11426 (build2_loc): Likewise.
11427 (build3_loc): Likewise.
11428 (build4_loc): Likewise.
11429 (build5_loc): Likewise.
11431 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11433 * tree.c (make_int_cst_stat): Remove _stat from name.
11434 * tree.h (make_int_cst_stat): Adjust prototype.
11435 (make_int_cst): Remove macro.
11437 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11439 * tree.c (make_tre_binfo_stat): Remove _stat from name.
11440 * tree.h (make_tree_binfo_stat): Adjust prototype.
11441 (make_tree_binfo): Remove.
11443 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11445 * tree.c (copy_node_stat): Rename to copy_node.
11446 (build_distinct_type_copy): Adjust.
11447 * tree.h (copy_node_stat): Adjust prototype.
11448 (copy_node): Remove macro.
11450 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11452 * tree.c (make_node_stat): rename to make_node.
11453 (build_tree_list_stat): Adjust.
11454 (build0_stat): Likewise.
11455 (build2_stat): Likewise.
11456 (build3_stat): Likewise.
11457 (build4_stat): Likewise.
11458 (build5_stat): Likewise.
11459 (build_decl_stat): Likewise.
11460 * tree.h (make_node_stat): Adjust prototype.
11461 (make_node): remove macro.
11463 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
11465 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
11466 (PPC_FEATURE2_SCV): Likewise.
11467 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
11469 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11471 * config/aarch64/aarch64.c
11472 (aarch64_internal_mov_immediate): Add new special pattern.
11473 * config/aarch64/aarch64.md (*movdi_aarch64):
11474 Add reg/32bit const mov case.
11476 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11477 Richard Sandiford <richard.sandiford@linaro.org>
11479 * config/aarch64/aarch64.md (mov<mode>): Generalize.
11480 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11481 Add integer and movi cases.
11482 (movi-split-hf-df-sf split, fp16): New.
11483 (enabled): Added TARGET_FP_F16INST.
11484 * config/aarch64/iterators.md (GPF_HF): New.
11485 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
11487 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11489 * config/aarch64/aarch64.c
11490 (aarch64_simd_container_mode): Add prototype.
11491 (aarch64_expand_mov_immediate): Add HI support.
11492 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
11493 (aarch64_can_const_movi_rtx_p): New.
11494 (aarch64_preferred_reload_class):
11495 Remove restrictions of using FP registers for certain SIMD operations.
11496 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
11497 (aarch64_valid_floating_const): Add integer move validation.
11498 (aarch64_simd_imm_scalar_p): Remove.
11499 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
11500 (aarch64_legitimate_constant_p): Expand list of supported cases.
11501 * config/aarch64/aarch64-protos.h
11502 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
11503 (aarch64_reinterpret_float_as_int): New.
11504 (aarch64_simd_imm_scalar_p): Remove.
11505 * config/aarch64/constraints.md (Uvi): New.
11506 (Dd): Split into Ds and new Dd.
11507 * config/aarch64/aarch64.md (*movsi_aarch64):
11509 (*movdi_aarch64): Add SIMD mov case.
11511 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11513 * tree-predcom.c: (struct chain): Handle store-store chain in which
11514 stores for elimination only store loop invariant values.
11515 (execute_pred_commoning_chain): Ditto.
11516 (prepare_initializers_chain_store_elim): Ditto.
11517 (prepare_finalizers): Ditto.
11518 (is_inv_store_elimination_chain): New function.
11519 (initialize_root_vars_store_elim_1): New function.
11521 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11523 * tree-predcom.c: Revise general description of the pass.
11524 (enum chain_type): New enum type for store elimination.
11525 (struct chain): New field supporting store elimination.
11526 (struct component): Ditto.
11527 (dump_chain): Dump store-stores chain.
11528 (release_chain): Release resources.
11529 (split_data_refs_to_components): Compute and create component
11530 contains only stores for elimination.
11531 (get_chain_last_ref_at): New function.
11532 (make_invariant_chain): Initialization.
11533 (make_rooted_chain): Specify chain type in parameter and record it.
11534 (add_looparound_copies): Skip for store-stores chain.
11535 (determine_roots_comp): Compute type of chain and pass it to
11537 (initialize_root_vars_store_elim_2): New function.
11538 (finalize_eliminated_stores): New function.
11539 (remove_stmt): Handle store for elimination.
11540 (execute_pred_commoning_chain): Execute predictive commoning on
11541 store-store chains.
11542 (determine_unroll_factor): Skip unroll for store-stores chain.
11543 (prepare_initializers_chain_store_elim): New function.
11544 (prepare_initializers_chain): Hanlde store-store chain.
11545 (prepare_finalizers_chain, prepare_finalizers): New function.
11546 (tree_predictive_commoning_loop): Return integer value indicating
11547 if loop is unrolled or lcssa form is corrupted.
11548 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
11550 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11552 * tree-predcom.c (initialize_root): Delete.
11553 (execute_pred_commoning_chain): Initialize root vars and replace
11554 reference of non-combined chain directly, rather than call above
11557 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11559 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
11560 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
11562 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11564 * tree-predcom.c (struct chain): New field init_seq.
11565 (release_chain): Release init_seq.
11566 (prepare_initializers_chain): Record intialization stmts in above
11568 (insert_init_seqs): New function.
11569 (tree_predictive_commoning_loop): Call insert_init_seqs.
11571 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11573 * tree-predcom.c (determine_roots_comp): Skip trivial components.
11575 2017-07-28 Richard Biener <rguenther@suse.de>
11577 * match.pd: Remove superfluous :c.
11578 * genmatch.c (simplify::id): Add member.
11579 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
11581 (current_id): New global.
11582 (dt_node::parent): Move from ...
11583 (dt_operand::parent): ... here. Add for_id member.
11584 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
11585 (decision_tree::find_node): Relax order requirement when
11586 merging DT_TRUE nodes to ones inbetween the current simplify
11587 and the one we try to merge with. Add diagnostic whenever
11588 we need to enforce pattern order by not merging.
11589 (decision_tree::insert): Set current_id.
11590 (decision_tree::print_node): Dump parent node and for_id.
11591 (parser::last_id): Add member.
11592 (parser::push_simplify): Assign unique id.
11593 (parser::parser): Initialize last_id.
11595 2017-07-28 Martin Liska <mliska@suse.cz>
11598 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
11599 gimple_build_debug_bind.
11601 2017-07-28 Richard Biener <rguenther@suse.de>
11603 PR tree-optimization/81502
11604 * match.pd: Add pattern combining BIT_INSERT_EXPR with
11606 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
11608 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
11609 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
11610 for BIT_FIELD_REF args.
11611 * fold-const.c (make_bit_field_ref): Likewise.
11612 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
11614 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11617 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
11618 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
11619 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
11620 Or it into SANITIZER_UNDEFINED.
11621 * ubsan.c: Include gimple-fold.h and varasm.h.
11622 (ubsan_expand_ptr_ifn): New function.
11623 (instrument_pointer_overflow): New function.
11624 (maybe_instrument_pointer_overflow): New function.
11625 (instrument_object_size): Formatting fix.
11626 (pass_ubsan::execute): Call instrument_pointer_overflow
11627 and maybe_instrument_pointer_overflow.
11628 * internal-fn.c (expand_UBSAN_PTR): New function.
11629 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
11630 * sanitizer.def (__ubsan_handle_pointer_overflow,
11631 __ubsan_handle_pointer_overflow_abort): New builtins.
11632 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
11633 * internal-fn.def (UBSAN_PTR): New internal function.
11634 * opts.c (sanitizer_opts): Add pointer-overflow.
11635 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
11636 * fold-const.c (build_range_check): Compute pointer range check in
11637 integral type if pointer arithmetics would be needed. Formatting
11640 2017-07-28 Martin Liska <mliska@suse.cz>
11643 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
11644 parameters that are of a variable-length.
11646 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
11648 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
11650 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
11651 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
11652 (CRT_CALL_STATIC_FUNCTION): Likewise.
11653 (ASM_DEFAULT_SPEC): New define.
11654 (ASM_SPEC32): Likewise.
11655 (ASM_SPEC64): Likewise.
11656 (ASM_SPEC_COMMON): Likewise.
11657 (ASM_SPEC): Likewise.
11658 (INVALID_64BIT): Likewise.
11659 (LINK_OS_DEFAULT_SPEC): Likewise.
11660 (LINK_OS_SPEC32): Likewise.
11661 (LINK_OS_SPEC64): Likewise.
11662 (POWERPC_LINUX): Likewise.
11663 (PTRDIFF_TYPE): Likewise.
11664 (RESTORE_FP_PREFIX): Likewise.
11665 (RESTORE_FP_SUFFIX): Likewise.
11666 (SAVE_FP_PREFIX): Likewise.
11667 (SAVE_FP_SUFFIX): Likewise.
11668 (SIZE_TYPE): Likewise.
11669 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
11670 (TARGET_64BIT): Likewise.
11671 (TARGET_64BIT): Likewise.
11672 (TARGET_AIX): Likewise.
11673 (WCHAR_TYPE_SIZE): Likewise.
11674 (WCHAR_TYPE): Undefine.
11675 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
11676 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
11677 (CPP_OS_RTEMS_SPEC): Delete.
11678 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
11679 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
11681 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
11683 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11685 PR tree-optimization/81578
11686 * tree-parloops.c (build_new_reduction): Bail out if
11687 reduction_code isn't one of the standard OpenMP reductions.
11688 Move the details printing after that decision.
11690 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11692 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
11693 related to reload_in_progress.
11694 (splat_input_operand): Likewise.
11695 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
11697 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
11699 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
11700 (TARGET_INSTANTIATE_DECLS): Likewise.
11701 (legitimate_indexed_address_p): Delete reload_in_progress code.
11702 (rs6000_debug_legitimate_address_p): Likewise.
11703 (rs6000_eliminate_indexed_memrefs): Likewise.
11704 (rs6000_emit_le_vsx_store): Likewise.
11705 (rs6000_emit_move_si_sf_subreg): Likewise.
11706 (rs6000_emit_move): Likewise.
11707 (register_to_reg_type): Likewise.
11708 (rs6000_pre_atomic_barrier): Likewise.
11709 (rs6000_machopic_legitimize_pic_address): Likewise.
11710 (rs6000_allocate_stack_temp): Likewise.
11711 (rs6000_address_for_fpconvert): Likewise.
11712 (rs6000_address_for_altivec): Likewise.
11713 (rs6000_secondary_memory_needed_rtx): Delete function.
11714 (rs6000_check_sdmode): Likewise.
11715 (rs6000_alloc_sdmode_stack_slot): Likewise.
11716 (rs6000_instantiate_decls): Likewise.
11717 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
11718 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
11719 Delete reload_in_progress.
11720 (*vec_reload_and_plus_<mptrsize>): Likewise.
11721 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
11722 (vsx_div_v2di): Likewise.
11723 (vsx_udiv_v2di): Likewise.
11725 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11727 * config/rs6000/rs6000.opt (mlra): Replace with stub.
11728 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
11729 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
11730 (rs6000_debug_reg_global): Delete print of LRA status.
11731 (rs6000_option_override_internal): Delete dead LRA related code.
11732 (rs6000_lra_p): Delete function.
11733 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
11735 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
11737 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
11738 * config/riscv/rtems.h: New file.
11740 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11741 Sudakshina Das <sudi.das@arm.com>
11743 * config/aarch64/aarch64.md
11744 (define_split for and<mode>3nr_compare): Move
11745 non aarch64_logical_operand to a register.
11746 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
11747 register immediate operand to a register.
11748 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
11750 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11752 PR middle-end/81564
11753 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
11755 2017-07-27 Richard Biener <rguenther@suse.de>
11757 PR tree-optimization/81573
11758 PR tree-optimization/81494
11759 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
11760 multi defuse cycle case.
11762 2017-07-27 Richard Biener <rguenther@suse.de>
11764 PR tree-optimization/81571
11765 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
11768 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
11770 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
11771 earlier and only if MASK_FPU is set. Adjust formatting.
11773 2017-07-27 Martin Liska <mliska@suse.cz>
11775 * opt-functions.awk: Add validation of value of Init.
11776 * optc-gen.awk: Pass new argument.
11778 2017-07-27 Martin Liska <mliska@suse.cz>
11780 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11781 Fix wrong condition.
11783 2017-07-27 Martin Liska <mliska@suse.cz>
11785 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
11786 BBs and edges seen by autoFDO.
11788 2017-07-27 Richard Biener <rguenther@suse.de>
11790 PR tree-optimization/81502
11791 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
11792 with incompatible but same sized type.
11793 (execute_update_addresses_taken): Likewise.
11795 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
11797 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
11798 flag_tree_loop_vectorize rather than flag_tree_vectorize.
11800 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11803 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
11804 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
11805 Change s_operand to memory_operand.
11807 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
11809 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11810 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
11811 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
11812 Emit instructions rather than returning an expression. Handle TFmode
11813 and KFmode by casting to TImode.
11814 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
11815 (rs6000_emit_le_vsx_store): Likewise.
11816 * config/rs6000/vsx.md (VSX_TI): New iterator.
11817 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
11818 (*vsx_le_undo_permute_<mode>): Likewise.
11819 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
11820 emit the split sequence.
11821 (*vsx_le_perm_store_<mode>): Likewise.
11823 2017-07-27 Jakub Jelinek <jakub@redhat.com>
11825 PR tree-optimization/81555
11826 PR tree-optimization/81556
11827 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
11828 if true, force CHANGED for the recursive invocation.
11829 (reassociate_bb): Remember original length of ops array, pass
11830 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
11832 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
11833 attributes for noipa attribute. For naked attribute use
11834 lookup_attribute first before lookup_attribute_spec.
11835 * final.c (rest_of_handle_final): Disable IPA RA for functions with
11837 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
11838 for functions with noipa attribute.
11839 (cgraph_externally_visible_p): Return true for functions with noipa
11841 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
11842 for functions with noipa attribute.
11843 * doc/extend.texi: Document noipa function attribute.
11844 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
11845 also for functions with noipa attribute.
11846 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
11848 2017-07-26 Andrew Pinski <apinski@cavium.com>
11850 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
11851 vec_unalign_load_cost and vec_unalign_store_cost.
11853 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11855 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
11856 -mvsx-small-integer option.
11857 (ISA_3_0_MASKS_IEEE): Likewise.
11858 (OTHER_VSX_VECTOR_MASKS): Likewise.
11859 (POWERPC_MASKS): Likewise.
11860 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
11861 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
11862 code, only testing for DImode being allowed in non-VSX floating
11864 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
11865 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
11867 (rs6000_option_override_internal): Delete -mvsx-small-integer.
11868 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
11869 TARGET_P8_VECTOR test.
11870 (rs6000_secondary_reload_simple_move): Likewise.
11871 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
11872 since TARGET_P9_VECTOR was already tested.
11873 (rs6000_opt_masks): Remove -mvsx-small-integer.
11874 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
11875 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11877 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
11878 test for TARGET_VEXTRACTUB was used, and that uses
11880 (p9 extract splitter): Likewise.
11881 (vsx_extract_<mode>_di_p9): Likewise.
11882 (vsx_extract_<mode>_store_p9): Likewise.
11883 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
11884 for TARGET_P9_VECTOR was used. Delete code that is now dead with
11885 the elimination of TARGET_VSX_SMALL_INTEGER.
11886 (vsx_extract_<mode>_p8): Likewise.
11887 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
11888 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
11889 (vsx_set_<mode>_p9): Likewise.
11890 (vsx_set_v4sf_p9): Likewise.
11891 (vsx_set_v4sf_p9_zero): Likewise.
11892 (vsx_insert_extract_v4sf_p9): Likewise.
11893 (vsx_insert_extract_v4sf_p9_2): Likewise.
11894 * config/rs6000/rs6000.md (sign extend splitter): Change
11895 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
11896 (floatsi<mode>2_lfiwax_mem): Likewise.
11897 (floatunssi<mode>2_lfiwzx_mem): Likewise.
11898 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
11899 since a test for TARGET_P9_VECTOR was used.
11900 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11901 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
11902 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11903 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11904 TARGET_P8_VECTOR test.
11905 (fix_trunc<mode>si2_stfiwx): Likewise.
11906 (fix_trunc<mode>si2_internal): Likewise.
11907 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
11908 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11910 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11911 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11912 TARGET_P8_VECTOR test.
11913 (fixuns_trunc<mode>si2_stfiwx): Likewise.
11914 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
11915 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11917 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11918 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
11919 since a test for TARGET_P9_VECTOR was used.
11920 (splitter for loading small constants): Likewise.
11922 2017-07-26 Andrew Pinski <apinski@cavium.com>
11924 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
11927 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
11930 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
11931 (fp_valid_at): Likewise.
11933 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11935 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
11936 (qdf24xx_addrcost_table): Likewise.
11937 (cortexa57_tunings): Update to use generic_branch_cost.
11938 (cortexa72_tunings): Likewise.
11939 (cortexa73_tunings): Likewise.
11940 (qdf24xx_tunings): Likewise.
11942 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11944 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
11945 (thunderx2t99_branch_cost): Likewise.
11946 (cortexa35_tunings): Update to use generic_branch_cost.
11947 (cortexa53_tunings): Likewise.
11948 (cortexa57_tunings): Likewise.
11949 (cortexa72_tunings): Likewise.
11950 (cortexa73_tunings): Likewise.
11951 (thunderx2t99_tunings): Likewise.
11953 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11955 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
11956 (sparc_option_override): Honour MASK_FSMULD.
11957 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
11958 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
11959 * config/sparc/sparc.opt (mfsmuld): New option.
11960 * doc/invoke.texi (mfsmuld): Document option.
11962 2017-07-26 Marek Polacek <polacek@redhat.com>
11964 PR middle-end/70992
11965 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
11967 2017-07-26 Richard Biener <rguenther@suse.de>
11969 * gimple-match-head.c (do_valueize): Return OP if valueize
11971 (get_def): New helper to get at the def stmt of a SSA name
11972 if valueize allows.
11973 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
11974 do_valueize to get at the def stmt.
11975 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
11977 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
11979 PR middle-end/46932
11980 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
11982 2017-07-26 Martin Liska <mliska@suse.cz>
11985 * function.c (expand_function_start): Make expansion of
11986 nonlocal_goto_save_area after parm_birth_insn.
11988 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11990 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
11991 from all CPU target flags enable members.
11993 2017-07-26 Richard Biener <rguenther@suse.de>
11995 * genmatch.c (dt_simplify::gen): Make iterator vars const.
11996 (decision_tree::gen): Make 'type' const.
11997 (write_predicate): Likewise.
11999 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12001 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
12002 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
12003 (rs6000_option_override_internal): Likewise.
12004 (rs6000_expand_vector_set): Likewise.
12005 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
12006 (TARGET_UPPER_REGS_SF): Likewise.
12007 (TARGET_UPPER_REGS_DI): Likewise.
12008 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
12009 (TARGET_DIRECT_MOVE_64BIT): Likewise.
12010 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
12011 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12012 (Splitters for DI constants in Altivec registers): Likewise.
12013 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
12014 (vsx_set_v4sf_p9): Likewise.
12015 (vsx_set_v4sf_p9_zero): Likewise.
12016 (vsx_insert_extract_v4sf_p9): Likewise.
12017 (vsx_insert_extract_v4sf_p9_2): Likewise.
12019 2017-07-25 Carl Love <cel@us.ibm.com>
12021 * doc/extend.texi: Update the built-in documentation file for the
12022 existing built-in functions
12023 vector signed char vec_cnttz (vector signed char);
12024 vector unsigned char vec_cnttz (vector unsigned char);
12025 vector signed short vec_cnttz (vector signed short);
12026 vector unsigned short vec_cnttz (vector unsigned short);
12027 vector signed int vec_cnttz (vector signed int);
12028 vector unsigned int vec_cnttz (vector unsigned int);
12029 vector signed long long vec_cnttz (vector signed long long);
12030 vector unsigned long long vec_cnttz (vector unsigned long long);
12032 2017-07-25 Andrew Pinski <apinski@cavium.com>
12034 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
12035 accesses where the use is for the first operand of a BIT_INSERT.
12037 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12040 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12041 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12043 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12045 * config/i386/gstabs.h: Delete.
12046 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12048 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
12050 * config/i386/i386.c (ix86_decompose_address): Do not check for
12051 register RTX when looking at index_reg or base_reg.
12052 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
12054 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
12056 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
12057 to update EH info here.
12059 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12061 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
12063 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12065 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12067 2017-07-25 Torsten Duwe <duwe@suse.de>
12069 * common.opt: Introduce -fpatchable-function-entry
12070 command line option, and its variables function_entry_patch_area_size
12071 and function_entry_patch_area_start.
12072 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12073 including a two-value parser.
12074 * target.def (print_patchable_function_entry): New target hook.
12075 * targhooks.h (default_print_patchable_function_entry): New function.
12076 * targhooks.c (default_print_patchable_function_entry): Likewise.
12077 * toplev.c (process_options): Switch off IPA-RA if
12078 patchable function entries are being generated.
12079 * varasm.c (assemble_start_function): Look at the
12080 patchable-function-entry command line switch and current
12081 function attributes and maybe generate NOP instructions by
12082 calling the print_patchable_function_entry hook.
12083 * doc/extend.texi: Document patchable_function_entry attribute.
12084 * doc/invoke.texi: Document -fpatchable_function_entry
12085 command line option.
12086 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12088 * doc/tm.texi: Re-generate.
12090 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12093 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12094 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12096 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12098 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12100 2017-07-25 Richard Biener <rguenther@suse.de>
12102 PR tree-optimization/81455
12103 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12104 not walk in cycles when looking for guards.
12106 2017-07-25 Richard Biener <rguenther@suse.de>
12108 PR tree-optimization/81529
12109 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12110 when optimizing backedge uses.
12112 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12114 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12116 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12117 to dl_section_ref. On AIX, append an expression to subtract
12118 the size of the section length to dl_section_ref.
12120 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12122 * configure.ac: If any of the config.* scripts fail, exit 1.
12123 * configure: Regenerate.
12125 2017-07-25 Richard Biener <rguenther@suse.de>
12127 PR middle-end/81546
12128 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12129 of immediate uses, be more verbose on errors.
12131 2017-07-25 Richard Biener <rguenther@suse.de>
12133 PR tree-optimization/81510
12134 * tree-vect-loop.c (vect_is_simple_reduction): When the
12135 reduction stmt is not inside the loop bail out.
12137 2017-07-25 Richard Biener <rguenther@suse.de>
12139 PR tree-optimization/81303
12140 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12141 profitability check against LOOP_VINFO_NITERSM1.
12143 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12145 * domwalk.c (cmp_bb_postorder): Simplify.
12146 (sort_bbs_postorder): New function. Use it...
12147 (dom_walker::walk): ...here to optimize common cases.
12149 2017-07-25 Martin Liska <mliska@suse.cz>
12152 * ipa-visibility.c (function_and_variable_visibility): Make the
12153 redirection just on target that supports aliasing.
12154 Fix GNU coding style.
12156 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
12159 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
12161 * config/arm/rtems.h: Likewise.
12162 * config/bfin/rtems.h: Likewise.
12163 * config/i386/rtemself.h: Likewise.
12164 * config/lm32/rtems.h: Likewise.
12165 * config/m32c/rtems.h: Likewise.
12166 * config/m68k/rtemself.h: Likewise.
12167 * config/microblaze/rtems.h: Likewise.
12168 * config/mips/rtems.h: Likewise.
12169 * config/moxie/rtems.h: Likewise.
12170 * config/nios2/rtems.h: Likewise.
12171 * config/powerpcspe/rtems.h: Likewise.
12172 * config/rs6000/rtems.h: Likewise.
12173 * config/rtems.h: Likewise.
12174 * config/sh/rtems.h: Likewise.
12175 * config/sh/rtemself.h: Likewise.
12176 * config/sparc/rtemself.h: Likewise.
12178 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12181 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
12182 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
12183 * tree-ssa-structalias.c (alias_get_name): Same.
12185 2017-07-25 Bin Cheng <bin.cheng@arm.com>
12188 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
12189 instructions if no du chain is found.
12191 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12193 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
12195 2017-07-25 Richard Biener <rguenther@suse.de>
12197 PR middle-end/81505
12198 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
12201 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12203 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
12204 upper-regs options.
12205 (ISA_2_7_MASKS_SERVER): Likewise.
12206 (ISA_3_0_MASKS_IEEE): Likewise.
12207 (OTHER_P8_VECTOR_MASKS): Likewise.
12208 (OTHER_VSX_VECTOR_MASKS): Likewise.
12209 (POWERPC_MASKS): Likewise.
12210 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
12211 duplicate list of options.
12212 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
12213 explicit -mupper-regs options.
12214 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
12215 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
12216 alias for -mupper-regs-df.
12217 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12218 (rs6000_init_hard_regno_mode_ok): Likewise.
12219 (rs6000_option_override_internal): Likewise.
12220 (rs6000_opt_masks): Likewise.
12221 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
12222 options in terms of whether -mvsx or -mpower8-vector was used.
12223 (TARGET_UPPER_REGS_DI): Likewise.
12224 (TARGET_UPPER_REGS_SF): Likewise.
12225 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
12226 -mupper-regs-* options.
12228 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12230 * passes.c (emergency_dump_function): Print some empty lines and a
12231 header before the RTL dump.
12233 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12235 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
12237 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
12240 * config/aarch64/aarch64.c (aarch64_classify_symbol):
12241 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
12243 2017-07-24 Carl Love <cel@us.ibm.com>
12245 * config/rs6000/rs6000-c.c: Add support for built-in functions
12246 vector float vec_extract_fp32_from_shorth (vector unsigned short);
12247 vector float vec_extract_fp32_from_shortl (vector unsigned short);
12248 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
12249 vec_extract_fp_from_shortl): Add defines for the two builtins.
12250 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
12251 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
12253 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
12254 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
12255 * doc/extend.texi: Update the built-in documentation file for the
12256 new built-in function.
12258 2017-07-24 Jakub Jelinek <jakub@redhat.com>
12261 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
12263 * doc/generic.texi: Likewise.
12264 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12265 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12267 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
12269 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
12270 (aarch64_mls_elt_merge<mode>): Likewise.
12272 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
12274 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12275 having __cxa_atexit.
12277 2017-07-23 Michael Collison <michael.collison@arm.com>
12279 * config/arm/arm.c (arm_option_override): Deprecate
12280 use of -mstructure-size-boundary.
12281 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
12282 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
12284 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12287 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12288 Reduce cost estimate for direct moves.
12290 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
12293 * config/i386/i386.c (ix86_option_override_internal): Disable
12294 BMI, BMI2 and TBM instructions for -m16.
12296 2017-07-21 Carl Love <cel@us.ibm.com>
12298 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12299 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12300 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12301 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12302 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12303 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12304 VMULOSW): New enum "unspec" values.
12305 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12306 altivec_vmulosw): New patterns.
12307 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12308 VMULOSW): Add definitions.
12310 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
12312 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
12313 (qdf24xx): Likewise.
12314 * config/aarch64/aarch64-options-extensions.def (rdma); New.
12315 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
12316 (AARCH64_FL_V8_1): Renumber.
12317 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
12318 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
12319 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
12320 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
12321 rdma to feature modifiers list.
12323 2017-07-21 Yury Gribov <tetra2005@gmail.com>
12325 PR middle-end/56727
12326 * ipa-visibility (function_and_variable_visibility): Convert
12327 recursive PLT call to direct call if appropriate.
12329 2017-07-21 Andrew Pinski <apinski@cavium.com>
12331 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
12332 operand 1 to see if the types precision matches.
12333 * fold-const.c (operand_equal_p): Likewise.
12335 2017-07-21 Richard Biener <rguenther@suse.de>
12337 PR tree-optimization/81303
12338 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
12339 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
12340 (vect_peeling_hash_get_lowest_cost): Adjust.
12341 (vect_enhance_data_refs_alignment): Likewise. Use
12342 vect_get_peeling_costs_all_drs to compute the penalty for no
12343 peeling to match up costs.
12345 2017-07-21 Richard Biener <rguenther@suse.de>
12347 PR tree-optimization/81500
12348 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
12349 we didn't identify a reduction path.
12351 2017-07-21 Tom de Vries <tom@codesourcery.com>
12352 Cesar Philippidis <cesar@codesourcery.com>
12354 PR gcov-profile/81442
12355 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
12358 2017-07-21 Tom de Vries <tom@codesourcery.com>
12361 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
12363 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
12364 nvptx_override_options_after_change.
12366 2017-07-21 Ulrich Drepper <drepper@redhat.com>
12368 * dwarf2out.c (output_file_names): Avoid double testing for
12369 dwarf_version >= 5.
12371 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
12373 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
12375 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12377 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
12379 (try_crossjump_to_edge): Do not punt on partitioned functions.
12381 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12383 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
12384 Put all BBs reachable only via paths crossing cold region to cold
12386 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
12388 2016-07-21 Richard Biener <rguenther@suse.de>
12390 PR tree-optimization/81303
12391 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
12392 into account prologue and epilogue iterations when raising
12393 min_profitable_iters to sth at least covering one vector iteration.
12395 2017-07-21 Tamar Christina <tamar.christina@arm.com>
12397 * config/arm/arm.c (arm_test_cpu_arch_dat):
12400 2017-07-20 Nathan Sidwell <nathan@acm.org>
12402 Remove TYPE_METHODS.
12403 * tree.h (TYPE_METHODS): Delete.
12404 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
12405 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
12406 (dbxout_type_methods): Scan TYPE_FIELDS.
12407 (dbxout_type): Don't check TYPE_METHODS here.
12408 * function.c (use_register_for_decl): Always ignore register for
12409 class types when not optimizing.
12410 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
12411 * tree.c (free_lang_data_in_type): Stitch out member functions and
12412 templates from TYPE_FIELDS.
12413 (build_distinct_type_copy, verify_type_variant,
12414 verify_type): Member fns are on TYPE_FIELDS.
12415 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
12416 * tree-pretty-print.c (dump_generic_node): Likewise.
12418 2017-07-20 Jakub Jelinek <jakub@redhat.com>
12421 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
12422 V2TImode and V4TImode.
12423 (ix86_expand_vector_extract): Likewise.
12424 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
12425 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
12426 (ssescalarmode): Handle V4TImode and V2TImode.
12427 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
12428 (*vec_extractv2ti, *vec_extractv4ti): New insns.
12429 (VEXTRACTI128_MODE): New mode iterator.
12430 (splitter for *vec_extractv?ti first element): New.
12431 (VEC_INIT_MODE): New mode iterator.
12432 (vec_init<mode>): Consolidate 3 expanders into one using
12433 VEC_INIT_MODE mode iterator.
12435 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12437 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
12438 non_spilled_static_chain_regno_p.
12440 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12442 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
12444 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
12446 * bb-reorder.c (connect_traces): Allow copying of blocks within
12449 2017-07-20 Richard Biener <rguenther@suse.de>
12451 * gimple.h (gimple_phi_result): Add gphi * overload.
12452 (gimple_phi_result_ptr): Likewise.
12453 (gimple_phi_arg): Likewise. Adjust index assert to only
12454 allow actual argument accesses rather than all slots available
12456 (gimple_phi_arg_def): Add gphi * overload.
12457 * tree-phinodes.c (make_phi_node): Initialize only actual
12459 (resize_phi_node): Clear memory not covered by old node,
12460 do not initialize excess argument slots.
12461 (reserve_phi_args_for_new_edge): Initialize new argument slot
12464 2017-07-20 Bin Cheng <bin.cheng@arm.com>
12466 PR tree-optimization/81388
12468 2016-07-21 Bin Cheng <bin.cheng@arm.com>
12470 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
12471 by removing computation of may_be_zero.
12473 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12474 Tom de Vries <tom@codesourcery.com>
12476 PR middle-end/81030
12477 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
12478 when gimple level profile disagrees with what RTL expander did.
12480 2017-07-20 Richard Biener <rguenther@suse.de>
12482 PR tree-optimization/61171
12483 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
12484 (vect_analyze_stmt): Add slp instance parameter.
12485 (vectorizable_reduction): Likewise.
12486 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
12487 (vect_is_simple_reduction): Deal with chains not detected
12488 as SLP reduction chain, specifically not properly associated
12489 chains containing a mix of plus/minus.
12490 (get_reduction_op): Remove.
12491 (get_initial_defs_for_reduction): Simplify, pass in whether
12492 this is a reduction chain, pass in the SLP node for the PHIs.
12493 (vect_create_epilog_for_reduction): Get the SLP instance as
12495 (vectorizable_reduction): Get the SLP instance as arg.
12496 During analysis remember the SLP node with the PHIs in the
12497 instance. Simplify getting at the vectorized reduction PHIs.
12498 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
12499 through SLP instance.
12500 (vect_slp_analyze_operations): Likewise.
12501 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
12502 (vect_transform_stmt): Likewise.
12504 2017-07-20 Tom de Vries <tom@codesourcery.com>
12506 PR tree-optimization/81489
12507 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
12508 read of phi arg location to before loop that modifies phi.
12510 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12512 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
12515 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12517 PR middle-end/81331
12518 * except.c (execute): Fix ordering issue.
12520 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12522 PR rtl-optimization/81423
12523 * combine.c (make_compound_operation_int): Don't try to optimize
12524 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
12526 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12528 PR rtl-optimization/81423
12529 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
12530 with a constant that is -1 in the truncated to mode.
12532 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12534 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
12535 (determine_unlikely_bbs): ... here.
12536 * predict.h (propagate_unlikely_bbs_forward): Declare.
12537 * cfgexpand.c (pass_expand::execute): Use it.
12538 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
12540 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
12541 propagate_unlikely_bbs_forward.
12543 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12545 PR middle-end/81331
12546 * except.c (maybe_add_nop_after_section_switch): New function.
12549 2017-07-19 Tom de Vries <tom@codesourcery.com>
12551 * gimple.h (gimple_phi_set_arg): Make assert more strict.
12553 2017-07-19 Tom de Vries <tom@codesourcery.com>
12555 * gimple.h (gimple_phi_arg): Make assert more strict.
12557 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
12559 * config.gcc (powerpc*-*-*): Add mmintrin.h.
12560 * config/rs6000/mmintrin.h: New file.
12561 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
12563 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12565 PR tree-optimization/81346
12566 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
12568 2017-07-19 Tom de Vries <tom@codesourcery.com>
12570 * config/nvptx/nvptx.md (VECIM): Add V2DI.
12572 2017-07-19 Tom de Vries <tom@codesourcery.com>
12574 * config/nvptx/nvptx-modes.def: Add V2DImode.
12575 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
12576 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
12577 (nvptx_output_mov_insn): Handle lack of mov.b128.
12578 (nvptx_print_operand): Handle 'H' and 'L' codes.
12579 (nvptx_vector_mode_supported): Allow V2DImode.
12580 (nvptx_preferred_simd_mode): New function.
12581 (nvptx_data_alignment): New function.
12582 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
12583 nvptx_preferred_simd_mode.
12584 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
12586 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
12588 2017-07-19 Tom de Vries <tom@codesourcery.com>
12590 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
12591 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
12592 (nvptx_vector_mode_supported): New function. Allow V2SImode.
12593 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
12594 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
12595 (mov<VECIM>_insn): New define_insn.
12596 (define_expand "mov<VECIM>): New define_expand.
12598 2017-07-19 Tom de Vries <tom@codesourcery.com>
12600 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
12602 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12604 PR tree-optimization/81346
12605 * fold-const.h (fold_div_compare, range_check_type): Declare.
12606 * fold-const.c (range_check_type): New function.
12607 (build_range_check): Use range_check_type.
12608 (fold_div_compare): No longer static, rewritten into
12609 a match.pd helper function.
12610 (fold_comparison): Don't call fold_div_compare here.
12611 * match.pd (X / C1 op C2): New optimization using fold_div_compare
12612 as helper function.
12614 2017-07-19 Nathan Sidwell <nathan@acm.org>
12616 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
12617 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
12618 * tree.c (find_decls_types_r, verify_type): Use
12619 TYPE_{MIN,MAX}_VALUE_RAW.
12620 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
12621 (hash_tree): Likewise.
12622 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12624 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12627 2017-07-18 Tom de Vries <tom@codesourcery.com>
12629 PR middle-end/81464
12630 * omp-expand.c (expand_omp_for_static_chunk): Handle
12631 equal-argument loop exit phi.
12633 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
12636 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
12637 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
12638 operand 2 predicate.
12639 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
12640 operand 2 predicate.
12641 (ror,rol -> rorx splitters): Use const_int_operand as
12642 operand 2 predicate.
12644 2017-06-18 Richard Biener <rguenther@suse.de>
12646 PR tree-optimization/81410
12647 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
12648 the gap in the ! slp_perm SLP case after each group.
12650 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12652 PR middle-end/81463
12653 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
12656 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12658 PR middle-end/81462
12659 * predict.c (set_even_probabilities): Cleanup; do not affect
12660 probabilities that are already known.
12661 (combine_predictions_for_bb): Call even when count is set.
12663 2017-07-18 Nathan Sidwell <nathan@acm.org>
12665 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
12668 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12671 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
12672 optimization for loop niter analysis.
12674 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
12677 * config/avr/avr.c (avr_optimize_casesi): Don't use
12678 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
12680 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12682 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
12683 body_cost_vec from _vect_peel_extended_info.
12684 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
12685 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
12688 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12690 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
12692 2017-07-18 Richard Biener <rguenther@suse.de>
12694 PR tree-optimization/80620
12695 PR tree-optimization/81403
12696 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
12697 info when re-using a VN table entry.
12699 2017-07-18 Richard Biener <rguenther@suse.de>
12701 PR tree-optimization/81418
12702 * tree-vect-loop.c (vectorizable_reduction): Properly compute
12703 vectype_in. Verify that with lane-reducing reduction operations
12704 we have a single def-use cycle.
12706 2017-07-17 Carl Love <cel@us.ibm.com>
12708 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
12710 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12711 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12712 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12713 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12714 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12715 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12716 VMULOSW): New enum "unspec" values.
12717 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12718 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12719 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12720 altivec_vmulosw): New patterns.
12721 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12722 VMULOSW): Add definitions.
12723 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
12725 * config/alpha/alpha.c: Include predict.h.
12727 2017-07-17 Yury Gribov <tetra2005@gmail.com>
12729 * tree-vrp.c (compare_assert_loc): Fix comparison function
12730 to return predictable results.
12732 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12734 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
12736 (subdi3): Likewise.
12737 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
12738 * doc/invoke.texi (mexpand-adddi): Update text.
12740 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12742 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
12743 that also clobbers the CC register. The old expand code is moved
12745 (*arc_clzsi2): ... here.
12746 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
12747 the CC register. The old expand code is moved to ...
12748 (arc_ctzsi2): ... here.
12750 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12752 * config/arc/arc.opt (mindexed-loads): Use initial value
12753 TARGET_INDEXED_LOADS_DEFAULT.
12754 (mauto-modify-reg): Use initial value
12755 TARGET_AUTO_MODIFY_REG_DEFAULT.
12756 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12757 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12758 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12759 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12761 2017-07-17 Martin Liska <mliska@suse.cz>
12764 * opts.c (finish_options): Do not allow -fgnu-tm
12765 w/ -fsanitize={kernel-,}address. Say sorry.
12767 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12770 * tree-loop-distribution.c (classify_partition): Only assert on
12771 numer of iterations.
12772 (merge_dep_scc_partitions): Delete prameter. Update function call.
12773 (distribute_loop): Remove code handling loop with unknown niters.
12774 (pass_loop_distribution::execute): Skip loop with unknown niters.
12776 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12779 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
12780 function sort_partitions_by_post_order.
12782 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12784 PR tree-optimization/81374
12785 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
12786 the max index of basic blocks, rather than number of basic blocks.
12788 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12790 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
12792 (arc_legitimate_pic_operand_p): Likewise.
12793 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
12795 (arc_needs_pcl_p): Likewise.
12796 (arc_legitimate_pc_offset_p): Likewise.
12797 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
12798 function is also used in constrains.md.
12799 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
12800 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
12801 PLUS. Only return true/false in known cases, otherwise assert.
12802 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
12803 is already called in arc_legitimate_constant_p.
12804 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
12806 (LEGITIMATE_PIC_OPERAND_P): Use
12807 arc_raw_symbolic_reference_mentioned_p function.
12808 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
12813 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12814 Andrew Burgess <andrew.burgess@embecosm.com>
12816 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
12817 (arc_return_address_register): New function.
12818 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
12819 (arc_handle_fndecl_attribute): Add naked attribute.
12820 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12821 (TARGET_WARN_FUNC_RETURN): Likewise.
12822 (arc_allocate_stack_slots_for_args): New function.
12823 (arc_warn_func_return): Likewise.
12824 (machine_function): Change type fn_type.
12825 (arc_compute_function_type): Consider new naked function type,
12826 change function return type.
12827 (arc_must_save_register): Adapt to handle new
12828 arc_compute_function_type's return type.
12829 (arc_expand_prologue): Likewise.
12830 (arc_expand_epilogue): Likewise.
12831 (arc_return_address_regs): Delete.
12832 (arc_return_address_register): New function.
12833 (arc_epilogue_uses): Use above function.
12834 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
12835 (arc_function_type): Change encoding, add naked type.
12836 (ARC_INTERRUPT_P): Change to handle the new encoding.
12837 (ARC_FAST_INTERRUPT_P): Likewise.
12838 (ARC_NORMAL_P): Define.
12839 (ARC_NAKED_P): Likewise.
12840 (arc_compute_function_type): Delete prototype.
12841 * config/arc/arc.md (in_ret_delay_slot): Use
12842 arc_return_address_register function.
12843 (simple_return): Likewise.
12844 (p_return_i): Likewise.
12846 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12848 PR tree-optimization/81428
12849 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
12850 can't be built for those types.
12852 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12854 Remove stuff dead since r239246.
12856 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
12857 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
12858 (avr_inform_devices): Remove dead stuff.
12860 2017-07-17 Tamar Christina <tamar.christina@arm.com>
12862 * config/arm/arm_neon.h: Fix softp typo.
12864 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12866 PR tree-optimization/81365
12867 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
12868 aggregate moves onto bb predecessor edges, make sure there are no
12869 loads that could alias the lhs in between the start of bb and the
12872 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12875 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
12876 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
12877 [LSHIFTRT, outer_code = TRUNCATE]: Same.
12879 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12881 PR tree-optimization/81396
12882 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
12883 (init_symbolic_number): Initialize it to 1.
12884 (perform_symbolic_merge): Add n_ops from both operands into the new
12886 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
12887 without base_addr as useless if they need more than one operation.
12888 (bswap_replace): Handle !bswap case for NULL base_addr.
12890 2017-07-17 Tom de Vries <tom@codesourcery.com>
12893 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
12896 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
12898 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
12899 conditional builtin define __FIX_LEON3FT_B2BST.
12901 2017-07-17 Daniel Cederman <cederman@gaisler.com>
12903 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
12904 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
12907 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12909 PR rtl-optimization/81424
12910 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
12911 to remove potential trapping from operands if -fnon-call-exceptions.
12913 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12915 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
12916 profile_proability for scalling.
12917 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
12919 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12921 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
12923 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12925 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
12926 fixpoint arithmetics.
12928 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12930 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
12931 fixpoint arithmetics.
12933 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12935 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
12936 fixpoint arithmetics.
12938 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12940 * profile-count.h (profile_probability::from_reg_br_prob_note,
12941 profile_probability::to_reg_br_prob_note): New functions.
12942 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
12943 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
12944 * predict.c (probability_reliable_p): Update.
12945 (edge_probability_reliable_p): Update.
12946 (br_prob_note_reliable_p): Update.
12947 (invert_br_probabilities): Update.
12948 (add_reg_br_prob_note): New function.
12949 (combine_predictions_for_insn): Update.
12950 * asan.c (asan_clear_shadow): Update.
12951 * cfgbuild.c (compute_outgoing_frequencies): Update.
12952 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12953 (update_br_prob_note): Update.
12954 (rtl_verify_edges): Update.
12955 (purge_dead_edges): Update.
12956 (fixup_reorder_chain): Update.
12957 * emit-rtl.c (try_split): Update.
12958 * ifcvt.c (cond_exec_process_insns): Update.
12959 (cond_exec_process_if_block): Update.
12960 (dead_or_predicable): Update.
12961 * internal-fn.c (expand_addsub_overflow): Update.
12962 (expand_neg_overflow): Update.
12963 (expand_mul_overflow): Update.
12964 * loop-doloop.c (doloop_modify): Update.
12965 * loop-unroll.c (compare_and_jump_seq): Update.
12966 * optabs.c (emit_cmp_and_jump_insn_1): Update.
12967 * predict.h: Update.
12968 * reorg.c (mostly_true_jump): Update.
12970 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
12971 * config/alpha/alpha.c (emit_unlikely_jump): Update.
12972 * config/arc/arc.c: (emit_unlikely_jump): Update.
12973 * config/arm/arm.c: (emit_unlikely_jump): Update.
12974 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
12975 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
12976 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
12977 (ix86_print_operand): Update.
12978 (ix86_split_fp_branch): Update.
12979 (predict_jump): Update.
12980 * config/ia64/ia64.c (ia64_print_operand): Update.
12981 * config/mmix/mmix.c (mmix_print_operand): Update.
12982 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
12983 (rs6000_expand_split_stack_prologue): Update.
12984 * config/rs6000/rs6000.c: Update.
12985 * config/s390/s390.c (s390_expand_vec_strlen): Update.
12986 (s390_expand_vec_movstr): Update.
12987 (s390_expand_cs_tdsi): Update.
12988 (s390_expand_split_stack_prologue): Update.
12989 * config/sh/sh.c (sh_print_operand): Update.
12990 (expand_cbranchsi4): Update.
12991 (expand_cbranchdi4): Update.
12992 * config/sparc/sparc.c (output_v9branch): Update.
12993 * config/spu/spu.c (get_branch_target): Update.
12994 (ea_load_store_inline): Update.
12995 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
12996 * config/tilepro/tilepro.c: Update.
12998 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13000 * gimplify.c (mostly_copy_tree_r): Revert latest change.
13001 (gimplify_save_expr): Likewise.
13003 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13005 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
13007 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13009 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
13011 * timevar.def (TV_IPA_FNSUMMARY): Define.
13013 2017-07-16 Daniel Cederman <cederman@gaisler.com>
13015 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
13016 to back store errata sensitive sequence from being generated.
13017 (sqrtdf2_fix): Likewise.
13019 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13021 * tree-ssa-threadupdate.c (compute_path_counts,
13022 update_joiner_offpath_counts): Use profile_probability.
13024 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
13027 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13029 * config/arm/arm-c.c (arm_cpu_builtins): Define
13030 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13032 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13034 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13035 array entries to represent __ieee128 versions of the
13036 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
13037 scalar_extract_sig, and scalar_insert_exp built-in functions.
13038 (altivec_resolve_overloaded_builtin): Add special case handling
13039 for the __builtin_scalar_insert_exp function, as represented by
13040 the P9V_BUILTIN_VEC_VSIEDP constant.
13041 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
13042 exponent support for __ieee128 argument.
13043 (VSESQP): Add scalar extract signature support for __ieee128
13045 (VSTDCNQP): Add scalar test negative support for __ieee128
13047 (VSIEQP): Add scalar insert exponent support for __int128 argument
13048 with __ieee128 result.
13049 (VSIEQPF): Add scalar insert exponent support for __ieee128
13050 argument with __ieee128 result.
13051 (VSTDCQP): Add scalar test data class support for __ieee128
13053 (VSTDCNQP): Add overload support for scalar test negative with
13054 __ieee128 argument.
13055 (VSTDCQP): Add overload support for scalar test data class
13056 __ieee128 argument.
13057 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
13058 UNSPEC_VSX_SXSIGDP.
13059 (UNSPEC_VSX_SIEXPQP): New constant.
13060 (xsxexpqp): New insn for VSX scalar extract exponent quad
13062 (xsxsigqp): New insn for VSX scalar extract significand quad
13064 (xsiexpqpf): New insn for VSX scalar insert exponent quad
13065 precision with floating point argument.
13066 (xststdcqp): New expand for VSX scalar test data class quad
13068 (xststdcnegqp): New expand for VSX scalar test negative quad
13070 (xststdcqp): New insn to match expansions for VSX scalar test data
13071 class quad precision and VSX scalar test negative quad precision.
13072 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13073 special case operand checking to enforce that second operand of
13074 VSX scalar test data class with quad precision argument is a 7-bit
13076 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13077 prototypes and descriptions of __ieee128 versions of
13078 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13079 scalar_test_data_class, and scalar_test_neg built-in functions.
13081 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13083 PR tree-optimization/81162
13084 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13085 replace a negate with an add.
13087 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13089 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13091 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13093 * config/arm/arm-c.c (arm_cpu_builtins): Define
13094 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13096 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13098 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13099 (armv8-r): Set ARM Cortex-R52 as default CPU.
13100 * config/arm/arm-tables.opt: Regenerate.
13101 * config/arm/arm-tune.md: Regenerate.
13102 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13104 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13105 extension for -mcpu=cortex-r52.
13107 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13109 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13110 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13111 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13112 (fp-armv8): Define it as FP_ARMv8 only.
13113 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13114 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13116 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13117 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13118 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13119 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13120 than TARGET_FPU_ARMV8.
13121 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13122 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13123 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13125 * config/arm/neon.md (neon_vrint): Likewise.
13126 (neon_vcvt): Likewise.
13127 (neon_<fmaxmin_op><mode>): Likewise.
13128 (<fmaxmin><mode>3): Likewise.
13129 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13130 * config/arm/predicates.md (arm_cond_move_operator): Check against
13131 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13133 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13135 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13136 to top of function.
13138 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13140 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13141 loop in comment with memset.
13143 2017-07-14 Martin Liska <mliska@suse.cz>
13145 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
13146 * dwarf2out.c (is_java): Remove the function.
13147 (output_pubname): Remove usage of the function.
13148 (lower_bound_default): Remove usage of DW_LANG_Java.
13149 (gen_compile_unit_die): Likewise.
13150 * gcc.c: Remove compiler defaults for .java and .zip files.
13151 * gimple-expr.c (remove_suffix): Change as there's no longer
13152 extension than 4-letter one.
13153 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
13154 (gimplify_save_expr): Likewise.
13155 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
13156 as it's possible even for other languages than Java.
13157 * langhooks.h (struct lang_hooks): Remove Java from a comment.
13158 * lto-opts.c (lto_write_options): Remove reference to Java.
13159 * opts.c (strip_off_ending): Update file extension handling.
13160 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
13161 * tree-eh.c (lower_resx): Likewise.
13162 * tree.c (free_lang_data_in_type): Remove dead code.
13163 (find_decls_types_r): Likewise.
13164 (build_common_builtin_nodes): Remove Java from a comment.
13165 (verify_type): Remove dead code.
13166 * varasm.c (assemble_external): Remove Java from a comment.
13168 2017-07-14 Martin Liska <mliska@suse.cz>
13170 * opts.c (finish_options): Add quotes.
13171 (common_handle_option): Likewise.
13173 2017-07-14 Martin Liska <mliska@suse.cz>
13175 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
13176 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
13177 Remove N_SO_PASCAL.
13178 * dwarf2out.c (lower_bound_default): Do not handle
13180 (gen_compile_unit_die): Likewise.
13181 * gcc.c: Remove default extension binding for GNU Pascal.
13182 * stmt.c: Remove Pascal language from a comment.
13183 * xcoffout.c: Likewise.
13185 2017-07-13 David Malcolm <dmalcolm@redhat.com>
13188 * diagnostic-show-locus.c (fixit_cmp): New function.
13189 (layout::layout): Sort m_fixit_hints.
13190 (column_range::column_range): Assert that the values are valid.
13191 (struct char_span): New struct.
13192 (correction::overwrite): New method.
13193 (struct source_line): New struct.
13194 (line_corrections::add_hint): Add assertions. Reimplement memcpy
13195 calls in terms of classes source_line and char_span, and
13196 correction::overwrite.
13197 (selftest::test_overlapped_fixit_printing_2): New function.
13198 (selftest::diagnostic_show_locus_c_tests): Call it.
13200 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
13202 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
13203 early if there is no lhs.
13205 2017-07-13 Martin Liska <mliska@suse.cz>
13207 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
13208 (gen_reference_type_die): Likewise.
13209 * stor-layout.c: Remove Pascal-related comment.
13211 2017-07-13 Martin Liska <mliska@suse.cz>
13213 * opts.c (finish_options): Add quotes to error messages.
13214 (parse_sanitizer_options): Likewise.
13216 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
13218 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
13220 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
13222 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
13224 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
13226 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
13228 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
13230 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
13233 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
13234 provides the hardware capability bits, define the macro
13235 __BUILTIN_CPU_SUPPORTS__.
13236 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
13237 if GLIBC does not provide the hardware capability bits. Add a
13238 gcc_unreachable call if the built-in cpu function is neither
13239 __builtin_cpu_is nor __builtin_cpu_supports.
13240 (rs6000_get_function_versions_dispatcher): Change the warning
13241 that an old GLIBC is used which does not export the capability
13242 bits to be an error.
13243 * doc/extend.texi (target_clones attribute): Document the
13244 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
13245 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
13246 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
13247 the macros defined by GCC if the newer GLIBC is available.
13249 2017-07-12 Jeff Law <law@redhat.com>
13251 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
13252 remaining includes slightly.
13253 * config/riscv/riscv-builtins.c: Include profile-count.h.
13255 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13258 * config/avr/avr.c (avr_set_current_function): In diagnostic
13259 messages: Quote keywords and (parts of) identifiers.
13260 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
13263 2017-07-12 Carl Love <cel@us.ibm.com>
13265 * config/rs6000/rs6000-c.c: Add support for built-in functions
13266 vector bool char vec_revb (vector bool char);
13267 vector bool short vec_revb (vector short char);
13268 vector bool int vec_revb (vector bool int);
13269 vector bool long long vec_revb (vector bool long long);
13270 * doc/extend.texi: Update the built-in documentation file for the
13271 new built-in functions.
13273 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13275 * config/s390/s390.md: Remove movcc splitter.
13277 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13279 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
13280 load/store on condition.
13282 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13285 * config/avr/avr.c (avr_encode_section_info)
13286 [progmem && !TREE_READONLY]: Error if progmem object needs
13289 2017-07-11 Michael Collison <michael.collison@arm.com>
13291 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
13294 2017-07-11 Carl Love <cel@us.ibm.com>
13296 * config/rs6000/rs6000-c.c: Add support for builtins
13297 vector unsigned int vec_parity_lsbb (vector signed int);
13298 vector unsigned int vec_parity_lsbb (vector unsigned int);
13299 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
13300 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
13301 vector unsigned long long vec_parity_lsbb (vector signed long long);
13302 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
13303 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
13304 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
13305 * doc/extend.texi: Update the built-in documentation file for the
13306 new built-in functions.
13308 2017-07-11 David Malcolm <dmalcolm@redhat.com>
13310 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
13311 (layout::m_primary_loc): New field.
13312 (layout::layout): Initialize new field. Move location filtering
13313 logic from here to...
13314 (layout::maybe_add_location_range): ...this new method. Add
13315 support for filtering to just the lines already specified by other
13317 (layout::will_show_line_p): New method.
13318 (gcc_rich_location::add_location_if_nearby): New method.
13319 (selftest::test_add_location_if_nearby): New test function.
13320 (selftest::diagnostic_show_locus_c_tests): Call it.
13321 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
13324 2017-07-11 Tom de Vries <tom@codesourcery.com>
13326 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
13327 (bb_first_real_insn): New function.
13328 (nvptx_single): Add extra initialization of broadcasted condition
13331 2017-07-11 Nathan Sidwell <nathan@acm.org>
13333 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
13335 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
13337 * doc/extend.texi (AVR Function Attributes): Remove weblink to
13338 Binutils doc as TEXI will mess them up.
13339 * doc/invoke.texi (AVR Options): Same here.
13341 2017-07-11 Daniel Cederman <cederman@gaisler.com>
13343 * config/sparc/sparc.opt (mfix-ut700): New option.
13344 (mfix-gr712rc): Likewise.
13345 (sparc_fix_b2bst): New variable.
13346 * doc/invoke.texi (SPARC options): Document them.
13347 (ARM options): Fix warnings.
13348 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
13349 instructions to prevent sequences that can trigger the store-store
13350 errata for certain LEON3FT processors.
13351 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
13352 (sparc_option_override): Set sparc_fix_b2bst appropriately.
13353 * config/sparc/sparc.md (fix_b2bst): New attribute.
13354 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
13356 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
13359 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
13361 (*rsqrtsf2_sse): Ditto.
13363 (div<mode>3): Macroize insn from divdf3 and divsf3
13364 using MODEF mode iterator.
13366 2017-07-10 Martin Sebor <msebor@redhat.com>
13368 PR tree-optimization/80397
13369 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
13370 instead of testing for equality to INTEGER_TYPE.
13372 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
13374 * config.gcc: Remove uclibc from arc target spec.
13376 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
13378 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
13380 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13383 * lto-wrapper.c (remove_option): New function.
13384 (merge_and_complain): Merge PIC/PIE options more realistically.
13386 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13388 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
13392 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
13393 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
13394 * config.in: Regenerate.
13395 * configure: Regenerate.
13396 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
13397 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
13398 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
13399 (TARGET_GASISR_PROLOGUES): ...target mask.
13400 * common/config/avr/avr-common.c
13401 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
13402 Set -mgas-isr-prologues.
13403 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
13404 INSERT_PASS_BEFORE for it.
13405 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
13406 * config/avr/avr.c (avr_option_override)
13407 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
13408 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
13409 (avr_attribute_table) <no_gccisr>: Add new function attribute.
13410 (avr_set_current_function) <is_no_gccisr>: Init machine field.
13411 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
13413 (make_avr_pass_pre_proep): New function.
13414 (emit_push_sfr) <treg>: Add argument to function and use it
13415 instead of TMP_REG.
13416 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
13417 and set machine->gasisr.yes.
13418 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
13419 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
13420 __gcc_isr.n_pushed to .L__stack_usage.
13421 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
13422 (avr_asm_final_postscan_insn): ...this new static function.
13423 * config/avr/avr.h (machine_function)
13424 <is_no_gccisr, use_L__stack_usage>: New fields.
13425 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
13426 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
13427 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
13428 (gasisr, *gasisr): New expander and insn.
13429 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13430 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
13431 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
13433 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
13435 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
13438 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13440 Move jump-tables out of .text again.
13443 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
13444 (ASM_OUTPUT_ADDR_VEC): New function.
13445 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
13446 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
13447 INSN_ADDRESSes as asm comment.
13448 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
13449 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
13450 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
13451 * config/avr/avr.md (*tablejump): Adjust comment.
13452 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
13453 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
13455 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
13456 (avr_output_addr_vec): New proto.
13457 (avr_log_t) <insn_addresses>: New field.
13459 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
13462 * config/i386/i386.c (ix86_function_arg_advance): Set
13463 outgoing_args_on_stack to true if there are outgoing arguments
13465 (ix86_function_arg): Likewise.
13466 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
13467 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
13468 * config/i386/i386.h (machine_function): Add
13469 outgoing_args_on_stack.
13471 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
13473 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13474 supporting pthreds.
13475 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
13477 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
13479 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
13480 (REAL_H): Remove $(MACHMODE_H).
13481 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
13483 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
13484 $(MACHMODE_H) and double-int.h.
13485 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
13487 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
13490 2017-07-07 Andrew Pinski <apinski@cavium.com>
13492 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
13493 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
13495 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13497 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
13498 Add warning if GCC was not configured to link against a GLIBC that
13499 exports the hardware capability bits.
13500 (make_resolver_func): Make resolver function private and not a
13501 COMDAT function. Create the name with clone_function_name instead
13502 of make_unique_name.
13505 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
13506 correct operand in doing the split.
13508 2017-07-07 Carl Love <cel@us.ibm.com>
13510 * config/rs6000/rs6000-c: Add support for built-in function
13511 vector unsigned short vec_pack_to_short_fp32 (vector float,
13513 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
13514 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
13515 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
13516 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
13517 (convert_4f32_8i16): Add define_expand.
13518 * doc/extend.texi: Update the built-in documentation file for the
13519 new built-in function.
13521 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13523 * config/sparc/m8.md: New file.
13524 * config/sparc/sparc.md: Include m8.md.
13526 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13528 * config/sparc/sparc.opt: New option -mvis4b.
13529 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
13530 (sparc_option_override): Handle VIS4B.
13531 (enum sparc_builtins): Define
13532 SPARC_BUILTIN_DICTUNPACK{8,16,32},
13533 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
13534 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
13535 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
13536 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
13537 (check_constant_argument): New function.
13538 (sparc_vis_init_builtins): Define builtins
13539 __builtin_vis_dictunpack{8,16,32},
13540 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
13541 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
13542 __builtin_vis_fpcmpde{8,16,32}shl and
13543 __builtin_vis_fpcmpur{8,16,32}shl.
13544 (sparc_expand_builtin): Check that the constant operands to
13545 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
13546 constant and in range.
13547 * config/sparc/sparc-c.c (sparc_target_macros): Handle
13549 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
13550 (SPARC_IMM5_P): Likewise.
13551 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
13552 (enabled): Handle vis4b.
13553 (UNSPEC_DICTUNPACK): New unspec.
13554 (UNSPEC_FPCMPSHL): Likewise.
13555 (UNSPEC_FPUCMPSHL): Likewise.
13556 (UNSPEC_FPCMPDESHL): Likewise.
13557 (UNSPEC_FPCMPURSHL): Likewise.
13558 (cpu_feature): New CPU feature `vis4b'.
13559 (dictunpack{8,16,32}): New insns.
13560 (FPCSMODE): New mode iterator.
13561 (fpcscond): New code iterator.
13562 (fpcsucond): Likewise.
13563 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
13564 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
13565 (fpcmpde{8,16,32}{si,di}shl): Likewise.
13566 (fpcmpur{8,16,32}{si,di}shl): Likewise.
13567 * config/sparc/constraints.md: Define constraints `q' for unsigned
13568 2-bit integer constants and `t' for unsigned 5-bit integer
13570 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
13572 (imm5_operand_dictunpack16): Likewise.
13573 (imm5_operand_dictunpack32): Likewise.
13574 (imm2_operand): Likewise.
13575 * doc/invoke.texi (SPARC Options): Document -mvis4b.
13576 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
13577 ditunpack* and fpcmp*shl builtins.
13579 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13581 * config.gcc: Handle m8 in --with-{cpu,tune} options.
13582 * config.in: Add HAVE_AS_SPARC6 define.
13583 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
13585 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
13587 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
13588 (CPP_CPU_SPEC): Handle m8.
13589 (ASM_CPU_SPEC): Likewise.
13590 * config/sparc/sparc-opts.h (enum processor_type): Add
13592 * config/sparc/sparc.c (m8_costs): New struct.
13593 (sparc_option_override): Handle TARGET_CPU_m8.
13594 (sparc32_initialize_trampoline): Likewise.
13595 (sparc64_initialize_trampoline): Likewise.
13596 (sparc_issue_rate): Likewise.
13597 (sparc_register_move_cost): Likewise.
13598 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
13599 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
13600 (ASM_CPU64_DEFAULT_SPEC): Likewise.
13601 (CPP_CPU_SPEC): Handle M8.
13602 (ASM_CPU_SPEC): Likewise.
13603 (AS_M8_FLAG): Define.
13604 * config/sparc/sparc.md: Add m8 to the cpu attribute.
13605 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
13606 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
13608 * configure: Regenerate.
13609 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
13612 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13614 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
13616 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
13617 ("*movdi_insn_sp32"): Do not set v3pipe.
13618 ("*movsi_insn"): Likewise.
13619 ("*movdi_insn_sp64"): Likewise.
13620 ("*movsf_insn"): Likewise.
13621 ("*movdf_insn_sp32"): Likewise.
13622 ("*movdf_insn_sp64"): Likewise.
13623 ("*zero_extendsidi2_insn_sp64"): Likewise.
13624 ("*sign_extendsidi2_insn"): Likewise.
13625 ("*mov<VM32:mode>_insn"): Likewise.
13626 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13627 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13628 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13629 ("<vlop:code><VL:mode>3"): Likewise.
13630 ("*not_<vlop:code><VL:mode>3"): Likewise.
13631 ("*nand<VL:mode>_vis"): Likewise.
13632 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
13633 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
13634 ("one_cmpl<VL:mode>2"): Likewise.
13635 ("faligndata<VM64:mode>_vis"): Likewise.
13636 ("alignaddrsi_vis"): Likewise.
13637 ("alignaddrdi_vis"): Likweise.
13638 ("alignaddrlsi_vis"): Likewise.
13639 ("alignaddrldi_vis"): Likewise.
13640 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13641 ("bmaskdi_vis"): Likewise.
13642 ("bmasksi_vis"): Likewise.
13643 ("bshuffle<VM64:mode>_vis"): Likewise.
13644 ("cmask8<P:mode>_vis"): Likewise.
13645 ("cmask16<P:mode>_vis"): Likewise.
13646 ("cmask32<P:mode>_vis"): Likewise.
13647 ("pdistn<P:mode>_vis"): Likewise.
13648 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13650 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13652 * config/sparc/sparc.md ("subtype"): New insn attribute.
13653 ("*wrgsr_sp64"): Set insn subtype.
13654 ("*rdgsr_sp64"): Likewise.
13655 ("alignaddrsi_vis"): Likewise.
13656 ("alignaddrdi_vis"): Likewise.
13657 ("alignaddrlsi_vis"): Likewise.
13658 ("alignaddrldi_vis"): Likewise.
13659 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13660 ("fexpand_vis"): Likewise.
13661 ("fpmerge_vis"): Likewise.
13662 ("faligndata<VM64:mode>_vis"): Likewise.
13663 ("bshuffle<VM64:mode>_vis"): Likewise.
13664 ("cmask8<P:mode>_vis"): Likewise.
13665 ("cmask16<P:mode>_vis"): Likewise.
13666 ("cmask32<P:mode>_vis"): Likewise.
13667 ("fchksm16_vis"): Likewise.
13668 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
13669 ("fmean16_vis"): Likewise.
13670 ("fp<plusminus_insn>64_vis"): Likewise.
13671 ("<plusminus_insn>v8qi3"): Likewise.
13672 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13673 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
13674 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
13675 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
13676 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
13677 ("*movqi_insn"): Likewise.
13678 ("*movhi_insn"): Likewise.
13679 ("*movsi_insn"): Likewise.
13680 ("movsi_pic_gotdata_op"): Likewise.
13681 ("*movdi_insn_sp32"): Likewise.
13682 ("*movdi_insn_sp64"): Likewise.
13683 ("movdi_pic_gotdata_op"): Likewise.
13684 ("*movsf_insn"): Likewise.
13685 ("*movdf_insn_sp32"): Likewise.
13686 ("*movdf_insn_sp64"): Likewise.
13687 ("*zero_extendhisi2_insn"): Likewise.
13688 ("*zero_extendqihi2_insn"): Likewise.
13689 ("*zero_extendqisi2_insn"): Likewise.
13690 ("*zero_extendqidi2_insn"): Likewise.
13691 ("*zero_extendhidi2_insn"): Likewise.
13692 ("*zero_extendsidi2_insn_sp64"): Likewise.
13693 ("ldfsr"): Likewise.
13694 ("prefetch_64"): Likewise.
13695 ("prefetch_32"): Likewise.
13696 ("tie_ld32"): Likewise.
13697 ("tie_ld64"): Likewise.
13698 ("*tldo_ldub_sp32"): Likewise.
13699 ("*tldo_ldub1_sp32"): Likewise.
13700 ("*tldo_ldub2_sp32"): Likewise.
13701 ("*tldo_ldub_sp64"): Likewise.
13702 ("*tldo_ldub1_sp64"): Likewise.
13703 ("*tldo_ldub2_sp64"): Likewise.
13704 ("*tldo_ldub3_sp64"): Likewise.
13705 ("*tldo_lduh_sp32"): Likewise.
13706 ("*tldo_lduh1_sp32"): Likewise.
13707 ("*tldo_lduh_sp64"): Likewise.
13708 ("*tldo_lduh1_sp64"): Likewise.
13709 ("*tldo_lduh2_sp64"): Likewise.
13710 ("*tldo_lduw_sp32"): Likewise.
13711 ("*tldo_lduw_sp64"): Likewise.
13712 ("*tldo_lduw1_sp64"): Likewise.
13713 ("*tldo_ldx_sp64"): Likewise.
13714 ("*mov<VM32:mode>_insn"): Likewise.
13715 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13716 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13718 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13720 * config/sparc/sparc.md ("type"): New insn type viscmp.
13721 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
13723 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
13724 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
13725 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13726 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
13728 ("n7_vis_logical_11cycle"): Likewise.
13729 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
13730 * config/sparc/niagara2.md ("niag3_vis": Likewise.
13731 * config/sparc/niagara.md ("niag_vis"): Likewise.
13732 * config/sparc/ultra3.md ("us3_fga"): Likewise.
13733 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
13735 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13737 * config/sparc/sparc.md: New instruction type `bmask'.
13738 (bmaskdi_vis): Use the `bmask' type.
13739 (bmasksi_vis): Likewise.
13740 * config/sparc/ultra3.md (us3_array): Likewise.
13741 * config/sparc/niagara7.md (n7_array): Likewise.
13742 * config/sparc/niagara4.md (n4_array): Likewise.
13743 * config/sparc/niagara2.md (niag2_vis): Likewise.
13744 (niag3_vis): Likewise.
13745 * config/sparc/niagara.md (niag_vis): Likewise.
13747 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13749 * ipa-comdats.c: Remove optimize check from gate.
13750 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
13751 for functions not optimized.
13752 (ipa_fn_summary_read): Skip optimize check.
13753 (ipa_fn_summary_write): Likewise.
13754 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
13756 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
13758 (can_inline_edge_p): Check flag_pcc_struct_return for match.
13759 (check_callers): Give up on caller which is not optimized.
13760 (inline_small_functions): Likewise.
13761 (ipa_inline): Do not give up when not optimizing.
13762 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
13763 away unoptimizes cdtors.
13764 (whole_program_function_and_variable_visibility): Do
13765 ipa_discover_readonly_nonaddressable_vars in LTO mode.
13766 * ipa.c (process_references): Do not check optimize.
13767 (symbol_table::remove_unreachable_nodes): Update optimize check.
13768 (set_writeonly_bit): Update optimize check.
13769 (pass_ipa_cdtor_merge::gate): Do not check optimize.
13770 (pass_ipa_single_use::gate): Remove.
13772 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13774 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
13775 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
13776 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
13777 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
13778 permute_load, permute_store, adjust_extract, adjust_splat,
13779 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
13780 replace_swap_with_copy, dump_swap_insn_table,
13781 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
13782 recombine_lvx_pattern, recombine_stvx_pattern,
13783 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
13784 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
13785 to file rs6000-p8swap.c.
13786 * config/rs6000/rs6000-p8swap.c: New file.
13787 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
13788 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
13789 and rs6000*-*-* targets.
13791 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13793 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
13795 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13797 * lto-wrapper.c (merge_and_complain): Do not merge
13798 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
13799 fsigned_zeros, ftrapping_math, fwrapv.
13800 (append_compiler_options): Do not track these options.
13801 (append_linker_options): Likewie
13803 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13805 * cgraphunit.c (cgraph_node::finalize_function): When
13806 !flag_toplevel_reorde set no_reorder flag.
13807 (varpool_node::finalize_decl): Likewise.
13808 (symbol_table::compile): Drop no toplevel reorder path.
13810 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13812 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
13813 edges; zero probability is not better than uninitialized.
13815 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13817 * asan.h (asan_sanitize_allocas_p): Declare.
13818 * asan.c (asan_sanitize_allocas_p): New function.
13819 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
13820 (handle_builtin_alloca): Likewise.
13821 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
13822 if !asan_sanitize_allocas_p.
13823 * params.def (asan-instrument-allocas): Add new option.
13824 * params.h (ASAN_PROTECT_ALLOCAS): Define.
13825 * opts.c (common_handle_option): Disable allocas sanitization for
13828 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13830 * asan.c: Include gimple-fold.h.
13831 (get_last_alloca_addr): New function.
13832 (handle_builtin_stackrestore): Likewise.
13833 (handle_builtin_alloca): Likewise.
13834 (asan_emit_allocas_unpoison): Likewise.
13835 (get_mem_refs_of_builtin_call): Add new parameter, remove const
13836 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
13837 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
13838 (instrument_builtin_call): Pass gimple iterator to
13839 get_mem_refs_of_builtin_call.
13840 (last_alloca_addr): New global.
13841 * asan.h (asan_emit_allocas_unpoison): Declare.
13842 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
13843 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
13844 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
13845 if function calls alloca.
13846 * gimple-fold.c (replace_call_with_value): Remove static keyword.
13847 * gimple-fold.h (replace_call_with_value): Declare.
13848 * internal-fn.c: Include asan.h.
13849 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
13850 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
13852 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13854 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
13855 (C_SELFTEST_FLAGS): New.
13856 (CPP_SELFTEST_FLAGS): New.
13857 (SELFTEST_DEPS): New, from deps of s-selftest.
13858 (C_SELFTEST_DEPS): New, from deps of s-selftest.
13859 (CPP_SELFTEST_DEPS): New.
13860 (selftest): Add dependency on s-selftest-c++.
13861 (s-selftest): Rename to...
13862 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
13863 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
13864 than SELFTEST_FLAGS.
13865 (selftest-gdb): Rename to...
13866 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
13868 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
13869 (selftest-valgrind): Rename to...
13870 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
13872 (selftest-valgrind): Reintroduce as an alias for
13873 selftest-c-valgrind.
13874 (s-selftest-c++): New.
13875 (selftest-c++-gdb): New.
13876 (selftest-c++-valgrind): New.
13878 2017-07-06 Olivier Hainque <hainque@adacore.com>
13880 * gcc.c (process_command): When deciding if undefined variables
13881 should be ignored when processing specs, accept "gcc -v" as well.
13883 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13885 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
13886 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
13888 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13890 * config/arm/arm-cpus.in (armv8-r): Add new entry.
13891 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
13892 * config/arm/arm-tables.opt: Regenerate.
13893 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
13895 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
13897 2017-07-06 Carl Love <cel@us.ibm.com>
13899 * ChangeLog: Clean up from mid air collision
13901 2017-07-06 Carl Love <cel@us.ibm.com>
13903 * config/rs6000/rs6000-c.c: Add support for built-in functions
13904 vector signed int vec_subc (vector signed int, vector signed int);
13905 vector signed __int128 vec_subc (vector signed __int128,
13906 vector signed __int128);
13907 vector unsigned __int128 vec_subc (vector unsigned __int128,
13908 vector unsigned __int128);
13909 vector signed int vec_sube (vector signed int, vector signed int,
13910 vector signed int);
13911 vector unsigned int vec_sube (vector unsigned int,
13912 vector unsigned int,
13913 vector unsigned int);
13914 vector signed __int128 vec_sube (vector signed __int128,
13915 vector signed __int128,
13916 vector signed__int128);
13917 vector unsigned __int128 vec_sube (vector unsigned __int128,
13918 vector unsigned __int128,
13919 vector unsigned __int128);
13920 vector signed int vec_subec (vector signed int, vector signed int,
13921 vector signed int);
13922 vector unsigned int vec_subec (vector unsigned int,
13923 vector unsigned int,
13924 vector unsigned int);
13925 vector signed __int128 vec_subec (vector signed __int128,
13926 vector signed __int128,
13927 vector signed__int128);
13928 vector unsigned __int128 vec_subec (vector unsigned __int128,
13929 vector unsigned __int128,
13930 vector unsigned __int128);
13931 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
13932 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
13933 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
13934 BU_ALTIVEC_OVERLOAD_X definitions.
13935 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
13936 * doc/extend.texi: Update the built-in documentation file for the new
13937 built-in functions.
13939 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13942 * diagnostic-show-locus.c (layout::layout): Use start and finish
13943 spelling location for the start and finish of each range.
13944 * genmatch.c (linemap_client_expand_location_to_spelling_point):
13945 Add unused aspect param.
13946 * input.c (expand_location_1): Add "aspect" param, and use it
13947 to access the correct part of the location.
13948 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
13950 (expand_location_to_spelling_point): Likewise.
13951 (linemap_client_expand_location_to_spelling_point): Add "aspect"
13952 param, and pass it to expand_location_1.
13954 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
13956 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
13957 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
13958 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
13959 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
13960 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
13961 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
13962 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
13963 _mm_maskz_getmant_ss): New intrinsics.
13964 (__builtin_ia32_getexpss128_mask): Changed to ...
13965 __builtin_ia32_getexpss128_round ... this.
13966 (__builtin_ia32_getexpsd128_mask): Changed to ...
13967 __builtin_ia32_getexpsd128_round ... this.
13968 * config/i386/i386-builtin-types.def
13969 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
13970 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
13971 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
13972 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
13973 __builtin_ia32_getmantss_mask_round): New builtins.
13974 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
13975 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
13976 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
13977 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
13978 * config/i386/sse.md
13979 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
13980 avx512f_sgetexp<mode><mask_scalar_name>
13981 <round_saeonly_scalar_name> ... this.
13982 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
13983 %0, %1, %2<round_saeonly_op3>}): Changed to ...
13984 vgetexp<ssescalarmodesuffix>
13985 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13986 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
13987 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
13988 avx512f_vgetmant<mode><mask_scalar_name>
13989 <round_saeonly_scalar_name> ... this.
13990 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
13991 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
13992 vgetmant<ssescalarmodesuffix>
13993 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
13994 %0<mask_scalar_operand4>, %1, %2
13995 <round_saeonly_scalar_mask_op4>, %3} ... this.
13996 * config/i386/subst.md (mask_scalar_operand4,
13997 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
13998 round_saeonly_scalar_nimm_predicate): New subst attributes.
14000 2017-07-06 Julia Koval <julia.koval@intel.com>
14002 * config/i386/i386.c (ix86_erase_embedded_rounding):
14003 Remove code for old rounding pattern.
14005 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
14007 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
14009 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
14011 * doc/sourcebuild.texi (Test Directives, Variants of
14012 dg-require-support): Add documentation for dg-require-stack-check.
14014 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
14016 * config/i386/subst.md (mask_scalar, round_scalar,
14017 round_saeonly_scalar): New meta-templates.
14018 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
14019 round_scalar_mask_operand3, round_scalar_mask_op3,
14020 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
14021 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
14022 round_saeonly_scalar_constraint,
14023 round_saeonly_scalar_prefix): New subst attribute.
14024 * config/i386/sse.md
14025 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
14026 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
14027 <round_scalar_name> ... this.
14028 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
14029 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
14030 <round_scalar_name> ... this.
14031 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
14032 <sse>_vm<code><mode>3<mask_scalar_name>
14033 <round_saeonly_scalar_name> ... this.
14034 (v<plusminus_mnemonic><ssescalarmodesuffix>
14035 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14036 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14037 v<plusminus_mnemonic><ssescalarmodesuffix>
14038 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14039 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14040 (v<multdiv_mnemonic><ssescalarmodesuffix>
14041 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14042 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14043 v<multdiv_mnemonic><ssescalarmodesuffix>
14044 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14045 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14046 (v<maxmin_float><ssescalarmodesuffix>
14047 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
14048 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
14049 v<maxmin_float><ssescalarmodesuffix>
14050 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14051 %0<mask_scalar_operand3>, %1, %<iptr>2
14052 <round_saeonly_scalar_mask_op3>} ... this.
14054 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
14056 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
14057 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
14059 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
14060 Alan Hayward <alan.hayward@arm.com>
14061 David Sherwood <david.sherwood@arm.com>
14063 * combine.c (simplify_if_then_else): Remove "enum" before
14065 * compare-elim.c (can_eliminate_compare): Likewise.
14066 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14068 (aarch64_lookup_simd_builtin_type): Likewise.
14069 (aarch64_simd_builtin_type): Likewise.
14070 (aarch64_init_simd_builtin_types): Likewise.
14071 (aarch64_simd_expand_args): Likewise.
14072 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14074 (aarch64_reverse_mask): Likewise.
14075 (aarch64_simd_emit_reg_reg_move): Likewise.
14076 (aarch64_gen_adjusted_ldpstp): Likewise.
14077 (aarch64_ccmp_mode_to_code): Likewise.
14078 (aarch64_operands_ok_for_ldpstp): Likewise.
14079 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14080 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14082 (aarch64_min_divisions_for_recip_mul): Likewise.
14083 (aarch64_reassociation_width): Likewise.
14084 (aarch64_get_condition_code_1): Likewise.
14085 (aarch64_simd_emit_reg_reg_move): Likewise.
14086 (aarch64_simd_attr_length_rglist): Likewise.
14087 (aarch64_reverse_mask): Likewise.
14088 (aarch64_operands_ok_for_ldpstp): Likewise.
14089 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14090 (aarch64_gen_adjusted_ldpstp): Likewise.
14091 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14093 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14094 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14095 (arm_lookup_simd_builtin_type): Likewise.
14096 (arm_simd_builtin_type): Likewise.
14097 (arm_init_simd_builtin_types): Likewise.
14098 (arm_expand_builtin_args): Likewise.
14099 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14100 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14101 (ft32_setup_incoming_varargs): Likewise.
14102 (ft32_function_arg): Likewise.
14103 (ft32_function_arg_advance): Likewise.
14104 (ft32_pass_by_reference): Likewise.
14105 (ft32_arg_partial_bytes): Likewise.
14106 (ft32_valid_pointer_mode): Likewise.
14107 (ft32_addr_space_pointer_mode): Likewise.
14108 (ft32_addr_space_legitimate_address_p): Likewise.
14109 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14111 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14112 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14113 (iamcu_alignment): Likewise.
14114 (canonicalize_vector_int_perm): Likewise.
14115 (ix86_noce_conversion_profitable_p): Likewise.
14116 (ix86_mpx_bound_mode): Likewise.
14117 (ix86_operands_ok_for_move_multiple): Likewise.
14118 * config/microblaze/microblaze-protos.h
14119 (microblaze_expand_conditional_branch_reg): Likewise.
14120 * config/microblaze/microblaze.c
14121 (microblaze_expand_conditional_branch_reg): Likewise.
14122 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14124 (rs6000_reassociation_width): Likewise.
14125 (rs6000_invalid_binary_op): Likewise.
14126 (fusion_p9_p): Likewise.
14127 (emit_fusion_p9_load): Likewise.
14128 (emit_fusion_p9_store): Likewise.
14129 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14131 (riscv_hard_regno_mode_ok_p): Likewise.
14132 (riscv_address_insns): Likewise.
14133 (riscv_split_symbol): Likewise.
14134 (riscv_legitimize_move): Likewise.
14135 (riscv_function_value): Likewise.
14136 (riscv_hard_regno_nregs): Likewise.
14137 (riscv_expand_builtin): Likewise.
14138 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14139 (riscv_build_integer): Likewise.
14140 (riscv_split_integer): Likewise.
14141 (riscv_legitimate_constant_p): Likewise.
14142 (riscv_cannot_force_const_mem): Likewise.
14143 (riscv_regno_mode_ok_for_base_p): Likewise.
14144 (riscv_valid_base_register_p): Likewise.
14145 (riscv_valid_offset_p): Likewise.
14146 (riscv_valid_lo_sum_p): Likewise.
14147 (riscv_classify_address): Likewise.
14148 (riscv_legitimate_address_p): Likewise.
14149 (riscv_address_insns): Likewise.
14150 (riscv_load_store_insns): Likewise.
14151 (riscv_force_binary): Likewise.
14152 (riscv_split_symbol): Likewise.
14153 (riscv_force_address): Likewise.
14154 (riscv_legitimize_address): Likewise.
14155 (riscv_move_integer): Likewise.
14156 (riscv_legitimize_const_move): Likewise.
14157 (riscv_legitimize_move): Likewise.
14158 (riscv_address_cost): Likewise.
14159 (riscv_subword): Likewise.
14160 (riscv_output_move): Likewise.
14161 (riscv_canonicalize_int_order_test): Likewise.
14162 (riscv_emit_int_order_test): Likewise.
14163 (riscv_function_arg_boundary): Likewise.
14164 (riscv_pass_mode_in_fpr_p): Likewise.
14165 (riscv_pass_fpr_single): Likewise.
14166 (riscv_pass_fpr_pair): Likewise.
14167 (riscv_get_arg_info): Likewise.
14168 (riscv_function_arg): Likewise.
14169 (riscv_function_arg_advance): Likewise.
14170 (riscv_arg_partial_bytes): Likewise.
14171 (riscv_function_value): Likewise.
14172 (riscv_pass_by_reference): Likewise.
14173 (riscv_setup_incoming_varargs): Likewise.
14174 (riscv_print_operand): Likewise.
14175 (riscv_elf_select_rtx_section): Likewise.
14176 (riscv_save_restore_reg): Likewise.
14177 (riscv_for_each_saved_reg): Likewise.
14178 (riscv_register_move_cost): Likewise.
14179 (riscv_hard_regno_mode_ok_p): Likewise.
14180 (riscv_hard_regno_nregs): Likewise.
14181 (riscv_class_max_nregs): Likewise.
14182 (riscv_memory_move_cost): Likewise.
14183 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
14184 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
14185 (rl78_addr_space_address_mode): Likewise.
14186 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14188 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14189 (rs6000_reassociation_width): Likewise.
14190 (rs6000_invalid_binary_op): Likewise.
14191 (fusion_p9_p): Likewise.
14192 (emit_fusion_p9_load): Likewise.
14193 (emit_fusion_p9_store): Likewise.
14194 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
14195 (ok_for_simple_move_operands): Likewise.
14196 (ok_for_simple_move_strict_operands): Likewise.
14197 (ok_for_simple_arith_logic_operands): Likewise.
14198 (visium_legitimize_reload_address): Likewise.
14199 (visium_select_cc_mode): Likewise.
14200 (output_cbranch): Likewise.
14201 (visium_split_double_move): Likewise.
14202 (visium_expand_copysign): Likewise.
14203 (visium_expand_int_cstore): Likewise.
14204 (visium_expand_fp_cstore): Likewise.
14205 * config/visium/visium.c (visium_pass_by_reference): Likewise.
14206 (visium_function_arg): Likewise.
14207 (visium_function_arg_advance): Likewise.
14208 (visium_libcall_value): Likewise.
14209 (visium_setup_incoming_varargs): Likewise.
14210 (visium_legitimate_constant_p): Likewise.
14211 (visium_legitimate_address_p): Likewise.
14212 (visium_legitimize_address): Likewise.
14213 (visium_secondary_reload): Likewise.
14214 (visium_register_move_cost): Likewise.
14215 (visium_memory_move_cost): Likewise.
14216 (prepare_move_operands): Likewise.
14217 (ok_for_simple_move_operands): Likewise.
14218 (ok_for_simple_move_strict_operands): Likewise.
14219 (ok_for_simple_arith_logic_operands): Likewise.
14220 (visium_function_value_1): Likewise.
14221 (rtx_ok_for_offset_p): Likewise.
14222 (visium_legitimize_reload_address): Likewise.
14223 (visium_split_double_move): Likewise.
14224 (visium_expand_copysign): Likewise.
14225 (visium_expand_int_cstore): Likewise.
14226 (visium_expand_fp_cstore): Likewise.
14227 (visium_split_cstore): Likewise.
14228 (visium_select_cc_mode): Likewise.
14229 (visium_split_cbranch): Likewise.
14230 (output_cbranch): Likewise.
14231 (visium_print_operand_address): Likewise.
14232 * expmed.c (flip_storage_order): Likewise.
14233 * expmed.h (emit_cstore): Likewise.
14234 (flip_storage_order): Likewise.
14235 * genrecog.c (validate_pattern): Likewise.
14236 * hsa-gen.c (gen_hsa_addr): Likewise.
14237 * internal-fn.c (expand_arith_overflow): Likewise.
14238 * ira-color.c (allocno_copy_cost_saving): Likewise.
14239 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14240 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
14241 (process_invariant_for_inheritance): Likewise.
14242 * lra-eliminations.c (move_plus_up): Likewise.
14243 * omp-low.c (lower_oacc_reductions): Likewise.
14244 * simplify-rtx.c (simplify_subreg): Likewise.
14245 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
14246 (TARGET_CHKP_BOUND_MODE): Likewise..
14247 * targhooks.c (default_chkp_bound_mode): Likewise.
14248 (default_setup_incoming_vararg_bounds): Likewise.
14249 * targhooks.h (default_chkp_bound_mode): Likewise.
14250 (default_setup_incoming_vararg_bounds): Likewise.
14251 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
14252 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
14253 (have_whole_vector_shift): Likewise.
14254 * tree-vect-stmts.c (vectorizable_load): Likewise.
14255 * doc/tm.texi: Regenerate.
14257 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14259 Graceful degrade if Binutils PR21472 is not available.
14262 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
14263 .rodata in flash test fails.
14264 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
14265 * confgure: Regenerate.
14266 * config.in: Regenerate.
14267 * config/avr/avr.c (avr_asm_named_section)
14268 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
14269 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
14270 (avr_asm_init_sections): Same.
14272 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14274 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
14275 (fma<VH:mode>4_intrinsic): Likewise.
14276 (*fmsub<VCVTF:mode>4): Likewise.
14277 (*fmsub<VH:mode>4_intrinsic): Likewise.
14279 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14282 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
14283 Don't depend on "optimize > 0".
14284 (out_movhi_r_mr, out_movqi_mr_r): Same.
14285 (out_movhi_mr_r, out_movqi_r_mr): Same.
14286 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
14287 io_address_operand on "optimize > 0".
14289 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14291 * tree-loop-distribution.c: Add general explanantion on the pass.
14292 (generate_loops_for_partition): Mark distributed loop.
14293 (pg_add_dependence_edges): New parameter. Handle alias data
14294 dependence specially and record it in the parameter if asked.
14295 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
14296 (init_partition_graph_vertices, add_partition_graph_edge): New.
14297 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
14298 (free_partition_graph_vdata, build_partition_graph): New.
14299 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
14300 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
14301 (data_ref_segment_size, latch_dominated_by_data_ref): New.
14302 (compute_alias_check_pairs, version_loop_by_alias_check): New.
14303 (version_for_distribution_p, finalize_partitions): New.
14304 (distribute_loop): Handle alias data dependence specially. Factor
14305 out loop fusion code as functions and call these functions.
14307 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14309 * tree-loop-distribution.c (classify_partition): New parameter and
14310 better handle reduction statement.
14311 (rdg_build_partitions): Revise comment.
14312 (distribute_loop): Compute statements in all partitions and pass it
14313 to classify_partition.
14315 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14317 * tree-loop-distribution.c (enum partition_type): New.
14318 (struct partition): New field type.
14319 (partition_merge_into): Add parameter. Update partition type.
14320 (data_dep_in_cycle_p, update_type_for_merge): New functions.
14321 (build_rdg_partition_for_vertex): Compute partition type.
14322 (rdg_build_partitions): Dump partition type.
14323 (distribute_loop): Update calls to partition_merge_into.
14325 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14327 * tree-loop-distribution.c (struct ddr_hasher): New.
14328 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
14330 (classify_partition): Call get_data_dependence.
14331 (pg_add_dependence_edges): Ditto.
14332 (distribute_loop): Release data dependence hash table.
14334 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14336 * tree-loop-distribution.c (ref_base_address): Delete.
14337 (similar_memory_accesses): Rename ...
14338 (share_memory_accesses): ... to this. Check if partitions access
14339 the same memory reference.
14340 (distribute_loop): Call share_memory_accesses.
14342 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14344 * tree-loop-distribution.c (struct partition): New field recording
14345 its data reference.
14346 (partition_alloc, partition_free): Init and release data refs.
14347 (partition_merge_into): Merge data refs.
14348 (build_rdg_partition_for_vertex): Collect data refs for partition.
14349 (pg_add_dependence_edges): Change parameters from vector to bitmap.
14351 (distribute_loop): Remve data refs from vertice data of partition
14354 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14356 * tree-loop-distribution.c (params.h): Include header file.
14357 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
14358 (datarefs_vec): New global var.
14359 (create_rdg_vertices): Use datarefs_vec directly.
14360 (free_rdg): Don't free data references.
14361 (build_rdg): Update use. Don't free data references.
14362 (distribute_loop): Compute global variable for data references.
14363 Bail out if there are too many data references.
14365 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14367 * tree-loop-distribution.c (loop_nest): New global var.
14368 (build_rdg): Use loop directly, rather than loop nest.
14369 (pg_add_dependence_edges): Remove loop nest parameter. Use global
14371 (distribute_loop): Compute global variable loop nest. Update use.
14373 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14375 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
14376 (partition_merge_into): New parameter. Dump reason for fusion.
14377 (distribute_loop): Update use of partition_merge_into.
14379 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14381 * tree-loop-distribution.c (bb_top_order_index): New.
14382 (bb_top_order_index_size, bb_top_order_cmp): New.
14383 (stmts_from_loop): Use topological order.
14384 (pass_loop_distribution::execute): Compute and release topological
14385 order for basic blocks.
14387 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14389 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
14392 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14394 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
14395 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
14396 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
14397 * internal-fn.def (LOOP_DIST_ALIAS): New.
14398 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
14399 (fold_loop_internal_call): ... this.
14400 (vect_loop_dist_alias_call): New function.
14401 (set_uid_loop_bbs): Call fold_loop_internal_call.
14402 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
14405 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14408 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
14409 Require dead FLAGS_REG at the beginning of a peephole.
14411 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14414 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
14415 arguments in the call to __builtin_ia32_sbb_u32.
14416 (_subborrow_u64): Swap _X and _Y arguments in the call to
14417 __builtin_ia32_sbb_u64.
14419 2017-07-04 Jakub Jelinek <jakub@redhat.com>
14422 * tree-vrp.c (compare_assert_loc): Turn into a function template
14423 with stable template parameter. Only test if a->e is NULL,
14424 !a->e == !b->e has been verified already. Use e == NULL or
14425 e != NULL instead of e or ! e tests. If stable is true, don't use
14426 iterative_hash_expr, on the other side allow a or b or both NULL
14427 and sort the NULLs last.
14428 (process_assert_insertions): Sort using compare_assert_loc<false>
14429 instead of compare_assert_loc, later sort using
14430 compare_assert_loc<true> before calling process_assert_insertions_for
14431 in a loop. Use break instead of continue once seen NULL pointer.
14433 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14435 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14436 Cortex-R7 and Cortex-R8 processors.
14438 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14440 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
14441 uninitialized while src is not.
14443 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
14445 * common/config/arm/arm-common.c: Adjust include path for
14447 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
14448 (arm-cpu.h): Create in build directory. Adjust dependency rules.
14449 (arm-cpu-data.h): Likewise.
14450 (arm-cpu-cdata.h): Likewise.
14451 * config/arm/arm-cpu.h: Delete.
14452 * config/arm/arm-cpu-cdata.h: Delete.
14453 * config/arm/arm-cpu-data.h: Delete.
14455 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
14457 * config/arm/arm-cpus.in (cortex-a55): New.
14458 (cortex-a75): Likewise.
14459 (cortex-a75.cortex-a55): Likewise.
14460 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
14462 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
14463 * config/arm/arm-cpu-cdata.h: Regenerate.
14464 * config/arm/arm-cpu-data.h: Regenerate.
14465 * config/arm/arm-cpu.h: Regenerate.
14466 * config/arm/arm-tables.opt: Regenerate.
14467 * config/arm/arm-tune.md: Regenerate.
14469 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14471 * haifa-sched.c (sched_create_recovery_edges): Update profile.
14473 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14475 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
14478 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
14480 PR tree-optimization/81292
14481 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
14482 full_string_p, also call adjust_related_strinfos if the adjustment
14483 is simple, otherwise invalidate related strinfos.
14485 2017-07-04 Martin Liska <mliska@suse.cz>
14488 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
14489 newly created variable as DECL_IGNORED_P.
14491 2017-07-04 Martin Liska <mliska@suse.cz>
14494 * ipa-inline.c (inline_small_functions):
14495 Use xstrdup_for_dump.
14497 2017-07-04 Tom de Vries <tom@codesourcery.com>
14499 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
14501 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14504 * config/darwin.c (darwin_function_switched_text_sections):
14507 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
14509 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
14511 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14513 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
14515 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14517 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
14518 min_profitable_iters, and th as inclusive lower bounds.
14519 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
14520 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
14521 for min_profitable_iters and min_profitable_estimate.
14522 (vect_transform_loop): Treat th as an inclusive lower bound.
14523 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
14525 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14528 * config/darwin.c (darwin_function_switched_text_sections):
14529 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
14530 in two pieces, and suppress the use of buf.
14532 2017-07-03 Nathan Sidwell <nathan@acm.org>
14534 * hash-table.h (hash_table_mod1): Fix indentation.
14536 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14538 PR middle-end/81290
14539 * predict.c (force_edge_cold): Be more careful about propagation
14541 * profile-count.h (profile_probability::guessed,
14542 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
14544 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
14546 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
14548 * doc/invoke.texi (rcpc architecture extension): Document it.
14550 2017-07-03 Richard Biener <rguenther@suse.de>
14552 PR tree-optimization/60510
14553 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
14554 the scalar reduction PHI and use it.
14555 (vectorizable_reduction): Properly guard the single_defuse_cycle
14556 path for non-SLP reduction chains where we cannot use it.
14557 Rework reduc_def/index and vector type deduction. Rework
14558 vector operand gathering during reduction op code-gen.
14559 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
14560 chains dissolve the chain and leave it to non-SLP reduction
14563 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14565 * tree-data-ref.h (dr_alignment): Declare.
14566 * tree-data-ref.c (dr_alignment): New function.
14567 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
14568 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14570 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
14572 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14574 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
14575 and base_misalignment fields.
14576 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
14577 * tree-data-ref.c: Include builtins.h.
14578 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
14579 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
14580 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
14581 * tree-vect-data-refs.c: Include tree-cfg.h.
14582 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
14583 fields instead of calculating an alignment here.
14584 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
14585 innermost_loop_behavior fields.
14587 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14589 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
14591 (DR_STEP_ALIGNMENT): New macro.
14592 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
14593 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
14594 (create_data_ref): Print it.
14595 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
14596 to tell whether the step preserves vector (mis)alignment.
14597 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14598 Move the check for an integer step and generalise to all INTEGER_CST.
14599 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
14600 Print the outer step alignment.
14602 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14604 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
14605 with offset_alignment.
14606 (DR_ALIGNED_TO): Delete.
14607 (DR_OFFSET_ALIGNMENT): New macro.
14608 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
14609 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
14610 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
14611 (create_data_ref): Likewise.
14612 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14613 (vect_analyze_data_refs): Likewise.
14614 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
14615 creating dummy innermost behavior.
14617 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14619 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
14620 with a "innermost_loop_behavior *" and refeence tree.
14621 * tree-data-ref.c (dr_analyze_innermost): Likewise.
14622 (create_data_ref): Update call accordingly.
14623 * tree-predcom.c (find_looparound_phi): Likewise.
14625 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14627 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
14628 fields with dr_wrt_vec_loop.
14629 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
14630 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
14631 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
14632 (vect_dr_behavior): New function.
14633 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14634 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
14635 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
14636 track whether the step preserves the misalignment.
14637 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14638 Use vect_dr_behavior.
14639 (vect_setup_realignment): Update call accordingly.
14640 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
14641 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
14642 call to vect_create_addr_base_for_vector_ref.
14643 (vect_create_cond_for_align_checks): Likewise.
14644 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
14645 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
14646 (vect_recog_mask_conversion_pattern): Likewise.
14647 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
14648 (new_stmt_vec_info): Remove redundant zeroing.
14650 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14652 * common/config/arm/arm-common.c (arm_be8_option): New function.
14653 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
14654 (ISA_ARMv6): Add isa_bit_be8.
14655 * config/arm/arm.h (arm_be8_option): Add prototype.
14656 (BE8_SPEC_FUNCTION): New define.
14657 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
14658 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
14659 (mlittle-endian): Similarly.
14660 (mbe8, mbe32): New options.
14661 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
14662 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
14664 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14666 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
14668 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14670 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
14671 (cleanup_tree_cfg_bb): Use it.
14672 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
14674 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
14676 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14679 * loop-doloop.c (add_test): Update profile.
14681 2017-07-03 Martin Liska <mliska@suse.cz>
14684 * sanopt.c (rewrite_usage_of_param): New function.
14685 (sanitize_rewrite_addressable_params): Likewise.
14686 (pass_sanopt::execute): Call rewrite_usage_of_param.
14688 2017-07-03 Richard Biener <rguenther@suse.de>
14690 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
14691 back to using VIEW_CONVERT_EXPR.
14693 2017-07-03 Martin Liska <mliska@suse.cz>
14696 * doc/extend.texi: Document when a resolver function is
14697 generated for target_clones.
14699 2017-07-03 Martin Liska <mliska@suse.cz>
14701 * asan.c (asan_emit_stack_protection): Unpoison just red zones
14702 and shadow memory of auto variables which are subject of
14703 use-after-scope sanitization.
14704 (asan_expand_mark_ifn): Add do set only when is_poison.
14706 2016-07-03 Richard Biener <rguenther@suse.de>
14708 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
14710 (vect_force_simple_reduction): Record reduction def -> phi mapping.
14711 (vectorizable_reduction): Perform reduction PHI creation when
14712 visiting a reduction PHI and adjust and simplify code generation
14713 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
14714 (vect_transform_loop): Visit reduction PHIs.
14715 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
14716 defs into the SLP tree.
14717 (vect_build_slp_tree): Reduction defs terminate the recursion.
14718 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
14720 (vect_get_vec_defs_for_stmt_copy): Export.
14721 (vect_get_vec_defs): Likewise.
14722 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
14724 (vect_get_vec_defs_for_stmt_copy): Declare.
14725 (vect_get_vec_defs): Likewise.
14727 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14729 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
14730 parameter with a "loop" parameter and use it instead of the
14731 loop containing DR_STMT. Don't check simple_iv when doing
14732 BB analysis. Describe the two analysis modes in the comment.
14734 2017-07-03 Tom de Vries <tom@codesourcery.com>
14736 PR tree-optimization/69468
14737 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
14738 (find_same_succ_bb): Handle ignore_edge_flags.
14740 2017-07-03 Tom de Vries <tom@codesourcery.com>
14742 PR tree-optimization/81192
14743 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
14745 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
14747 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
14749 2017-07-03 Tom de Vries <tom@codesourcery.com>
14751 PR tree-optimization/81192
14752 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
14753 BB_SAME_SUCC (bb) == NULL.
14755 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14757 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
14760 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14762 * dumpfile.c: Include profile-count.h
14763 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
14765 (insert_cond_bb): Update profile.
14766 * tree-cfg.h (insert_cond_bb): Update prototype.
14767 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
14768 * tree-dump.c: Do not include tree-cfg.
14770 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14772 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
14774 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14776 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
14779 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14781 * tree-complex.c (expand_complex_div_wide): update profile.
14783 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14784 Alan Hayward <alan.hayward@arm.com>
14785 David Sherwood <david.sherwood@arm.com>
14787 * Makefile.in (MACHMODE_H): Remove insn-modes.h
14788 (CORETYPES_H): New define.
14789 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
14790 (insn-modes-inline.h, s-modes-inline-h): New rules.
14791 (generated_files): Add insn-modes-inline.h.
14792 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
14793 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
14794 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
14795 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
14796 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
14797 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
14798 (build/gencodes.o, build/genconditions.o): Likewise.
14799 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
14800 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
14801 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
14802 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
14803 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
14804 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
14805 * coretypes.h: Include everything up to real.h for generators.
14806 Include insn-modes.h first. Include wide-int-print.h after
14807 wide-int.h. Include insn-modes-inline.h and then machmode.h.
14808 * machmode.h: Don't include insn-modes.h here.
14809 * function-tests.c: Remove includes of signop.h, machmode.h,
14810 double-int.h and wide-int.h.
14812 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
14814 * optc-save-gen.awk: Likewise.
14815 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
14816 * godump.c: Remove include of wide-int-print.h.
14817 * pretty-print.h: Likewise.
14818 * wide-int-print.cc: Likewise.
14819 * wide-int.cc: Likewise.
14820 * hash-map-tests.c: Remove include of signop.h.
14821 * hash-set-tests.c: Likewise.
14822 * rtl-tests.c: Likewise.
14823 * mkconfig.sh: Remove include of machmode.h.
14824 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
14826 (emit_insn_modes_inline_h): ...this new function. Emit the code
14827 into an insn-modes-inline.h header file, adding appropriate
14828 include guards and end comments.
14829 (emit_insn_modes_c_header): Remove include of machmode.h.
14830 (emit_min_insn_modes_c_header): Include coretypes.h rather than
14832 (main): Handle -i flag and call emit_insn_modes_inline_h when
14835 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14837 * tree-ssa-strlen.c (strinfo): Rename the length field to
14838 nonzero_chars. Add a full_string_p field.
14839 (compare_nonzero_chars, zero_length_string_p): New functions.
14840 (get_addr_stridx): Add an offset_out parameter.
14841 Use compare_nonzero_chars.
14842 (get_stridx): Update accordingly. Use compare_nonzero_chars.
14843 (new_strinfo): Update after above changes to strinfo.
14844 (set_endptr_and_length): Set full_string_p.
14845 (get_string_length): Update after above changes to strinfo.
14846 (unshare_strinfo): Update call to new_strinfo.
14847 (maybe_invalidate): Likewise.
14848 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
14849 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
14850 as a uhwi instead of an shwi. Update after above changes to
14851 strinfo and new_strinfo.
14852 (zero_length_string): Assert that chainsi contains full strings.
14853 Use zero_length_string_p. Update call to new_strinfo.
14854 (adjust_related_strinfos): Update after above changes to strinfo.
14855 Copy full_string_p from origsi.
14856 (adjust_last_stmt): Use zero_length_string_p.
14857 (handle_builtin_strlen): Update after above changes to strinfo and
14858 new_strinfo. Install the lhs as the string length if the previous
14859 entry didn't describe a full string.
14860 (handle_builtin_strchr): Update after above changes to strinfo
14862 (handle_builtin_strcpy): Likewise.
14863 (handle_builtin_strcat): Likewise.
14864 (handle_builtin_malloc): Likewise.
14865 (handle_pointer_plus): Likewise.
14866 (handle_builtin_memcpy): Likewise. Track nonzero characters
14867 that aren't necessarily followed by a nul terminator.
14868 (handle_char_store): Likewise.
14870 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14872 PR tree-optimization/80769
14873 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
14874 for malloc and calloc. Document the new invariant that all related
14875 strinfos have delayed lengths or none do.
14876 (verify_related_strinfos): Move earlier in file.
14877 (set_endptr_and_length): New function, split out from...
14878 (get_string_length): ...here. Also set the lengths of related
14880 (zero_length_string): Assert that chainsi has known (rather than
14882 (adjust_related_strinfos): Likewise.
14884 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14886 PR tree-optimization/81136
14887 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
14888 assert that two references with the same misalignment have the same
14889 compile-time misalignment if those compile-time misalignments
14892 2017-07-01 Andi Kleen <ak@linux.intel.com>
14894 * print-tree.c (print_node): Print all attributes.
14896 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14898 * cfg.c (scale_bbs_frequencies): New function.
14899 * cfg.h (scale_bbs_frequencies): Declare it.
14900 * cfgloopanal.c (single_likely_exit): Cleanup.
14901 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
14903 (scale_loop_profile): Likewise.
14904 (loop_version): Likewise.
14905 (create_empty_loop_on_edge): Update.
14906 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
14907 scale_loop_frequencies, scale_loop_profile, loopify,
14908 loop_version): Update prototypes.
14909 * modulo-sched.c (sms_schedule): Update.
14910 * predict.c (unlikely_executed_edge_p): Also check probability.
14911 (probably_never_executed_edge_p): Fix typo.
14912 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14913 * tree-parloops.c (gen_parallel_loop): Update.
14914 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
14915 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14916 * tree-ssa-loop-split.c (split_loop): Update.
14917 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14918 * tree-vect-loop-manip.c (vect_do_peeling): Update.
14919 (vect_loop_versioning): Update.
14920 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14922 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14924 * trans-mem.c (split_bb_make_tm_edge): Update profile.
14926 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14928 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
14929 to keep profile consistent.
14931 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14933 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
14934 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
14935 * profile-count.h (max_safe_multiplier): Make unsigned.
14936 (profile_count::guessed_zero): New.
14938 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14940 * bb-reorder.c (fix_up_crossing_landing_pad,
14941 fix_crossing_conditional_branches): Use make_single_succ_edge
14942 to keep profile consistent.
14944 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14946 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
14949 2017-07-01 Jakub Jelinek <jakub@redhat.com>
14952 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
14953 the right scopes, make sure cond_jump isn't preserved between multiple
14954 iterations. Search for fallthru edge whenever there are 3+ edges and
14955 use find_fallthru_edge for it.
14957 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14959 Patch by Alexander Monakov <amonakov@ispras.ru>
14960 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
14961 probabilities consistently.
14963 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14965 * pa.c (pa_expand_compare_and_swap_loop): Update call of
14966 emit_cmp_and_jump_insns.
14968 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14971 * tree-inline.c (expand_call_inline): Combine profile statuses.
14973 2017-06-30 Andrew Pinski <apinski@cavium.com>
14975 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
14976 fold_stmt returned true.
14978 2017-06-30 Nathan Sidwell <nathan@acm.org>
14980 * ggc.h (empty_string): Delete.
14981 * cfgexpand.c (expand_asm_stmt): Use plain "".
14982 * optabs.c (expand_asm_memory_barrier): Likewise.
14983 * stringpool.c (empty_string): Delete.
14984 (digit_vector, digit_string): Delete.
14985 (ggc_alloc_string): Use plain "", don't optimize single digit
14986 strings. Use ggc_alloc_atomic.
14988 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
14990 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
14991 comparison set and one other set, use the cost of the non-comparison
14994 2017-06-30 Nathan Sidwell <nathan@acm.org>
14996 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
14999 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
15001 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
15002 loops. Remove now unneeded calls to gimple_switch_set_label() that
15003 just set removed labels to NULL_TREE.
15005 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
15007 * tree-ssanames.c (set_range_info_raw): Abstract from ...
15008 (set_range_info): ...here. Only call set_range_info_raw if domain
15010 (set_nonzero_bits): Call set_range_info_raw.
15011 * tree-ssanames.h (set_range_info_raw): New.
15013 2017-06-30 Jakub Jelinek <jakub@redhat.com>
15016 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
15017 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
15018 of nonimmediate_operand and <store_mask_constraint> instead of m
15019 for the input operand. For V8FI iterator, always split if input
15020 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
15021 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
15022 <store_mask_predicate> instead of register_operand and
15023 <store_mask_constraint> instead of v for the input operand. Make
15024 sure both operands aren't MEMs for if not <mask_applied>.
15026 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
15028 * lto-wrapper.c (copy_file) Close both file descriptors before
15031 2017-06-30 Martin Liska <mliska@suse.cz>
15034 * multiple_target.c (create_dispatcher_calls): Make ifunc
15035 also for function that don't have calls or are not referenced.
15037 2017-06-30 Richard Biener <rguenther@suse.de>
15039 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
15040 analyze the first scalar stmt. Move vector type computation
15041 for the BB case here from ...
15042 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
15043 live operation processing in the SLP case properly.
15045 2017-06-30 Richard Biener <rguenther@suse.de>
15047 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
15049 2017-06-30 Martin Liska <mliska@suse.cz>
15052 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
15053 before BUILT_IN_UNWIND_RESUME when ASAN is used.
15055 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
15057 * doc/invoke.texi (AArch64): Add missing options and remove redundant
15060 2017-06-30 Richard Biener <rguenther@suse.de>
15062 PR tree-optimization/81249
15063 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
15064 condition reduction result to original scalar type.
15066 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15068 * profile-count.h (enum profile_quality): Fix typos and whitespace
15071 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15073 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15074 type for branch probabilities.
15076 2017-06-29 Julian Brown <julian@codesourcery.com>
15077 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15079 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15080 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15081 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15082 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15084 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15086 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15087 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15088 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15089 CC usage from generic code to here.
15090 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15091 CC usage into the target macros.
15093 2017-06-29 Maya Rashish <coypu@sdf.org>
15095 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15098 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15100 * arm/arm-builtins.c: Include profile-count.h
15101 * except.c (sjlj_emit_function_enter): Use
15102 profile_probability::unlikely.
15104 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15106 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15107 and tocrel_offset be pointer args rather than implicitly using
15109 (legitimate_constant_pool_address_p, rs6000_emit_move,
15110 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15111 tocrel_offset and use in toc_relative_expr_p call.
15112 (print_operand, print_operand_address): Use static tocrel_base_oac
15113 and tocrel_offset_oac.
15114 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15117 2017-06-29 Maya Rashish <coypu@sdf.org>
15119 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15121 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15123 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15124 objects, take into account only the alignment of 'op0' and 'mode1' if
15127 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15129 * ccmp.c (ccmp_tree_comparison_p): New function.
15130 (ccmp_candidate_p): Update to use above function.
15131 (get_compare_parts): New function.
15132 (expand_ccmp_next): Update to use new functions.
15133 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15135 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15136 take mode as argument.
15137 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15138 * expr.c (expand_expr_real_1): Pass mode as argument.
15140 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15142 * combine.c (combine_instructions): Print insns to dump_file, together
15145 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15147 * asan.c (asan_emit_stack_protection): Update.
15148 (create_cond_insert_point): Update.
15149 * auto-profile.c (afdo_propagate_circuit): Update.
15150 * basic-block.h (struct edge_def): Turn probability to
15151 profile_probability.
15152 (EDGE_FREQUENCY): Update.
15153 * bb-reorder.c (find_traces_1_round): Update.
15154 (better_edge_p): Update.
15155 (sanitize_hot_paths): Update.
15156 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
15157 (make_single_succ_edge): Update.
15158 (check_bb_profile): Update.
15159 (dump_edge_info): Update.
15160 (update_bb_profile_for_threading): Update.
15161 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
15162 probabilitycount to 0.
15163 * cfgbuild.c (compute_outgoing_frequencies): Update.
15164 * cfgcleanup.c (try_forward_edges): Update.
15165 (outgoing_edges_match): Update.
15166 (try_crossjump_to_edge): Update.
15167 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
15168 (expand_gimple_tailcall): Update.
15169 (construct_init_block): Use make_single_succ_edge.
15170 (construct_exit_block): Use make_single_succ_edge.
15171 * cfghooks.c (verify_flow_info): Update.
15172 (redirect_edge_succ_nodup): Update.
15173 (split_edge): Update.
15174 (account_profile_record): Update.
15175 * cfgloopanal.c (single_likely_exit): Update.
15176 * cfgloopmanip.c (scale_loop_profile): Update.
15177 (set_zero_probability): Remove.
15178 (duplicate_loop_to_header_edge): Update.
15179 * cfgloopmanip.h (loop_version): Update prototype.
15180 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
15181 (force_nonfallthru_and_redirect): Update.
15182 (update_br_prob_note): Update.
15183 (rtl_verify_edges): Update.
15184 (purge_dead_edges): Update.
15185 (rtl_lv_add_condition_to_bb): Update.
15186 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
15187 * cgraphunit.c (init_lowered_empty_function): Update.
15188 (cgraph_node::expand_thunk): Update.
15189 * cilk-common.c: Include profile-count.h
15190 * dojump.c (inv): Remove.
15191 (jumpifnot): Update.
15192 (jumpifnot_1): Update.
15193 (do_jump_1): Update.
15195 (do_jump_by_parts_greater_rtx): Update.
15196 (do_compare_rtx_and_jump): Update.
15197 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
15198 do_jump_1. do_compare_rtx_and_jump): Update prototype.
15199 * dwarf2cfi.c: Include profile-count.h
15200 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
15201 (sjlj_emit_dispatch_table): Likewise.
15202 * explow.c: Include profile-count.h
15203 * expmed.c (emit_store_flag_force): Update.
15204 (do_cmp_and_jump): Update.
15205 * expr.c (compare_by_pieces_d::generate): Update.
15206 (compare_by_pieces_d::finish_mode): Update.
15207 (emit_block_move_via_loop): Update.
15208 (store_expr_with_bounds): Update.
15209 (store_constructor): Update.
15210 (expand_expr_real_2): Update.
15211 (expand_expr_real_1): Update.
15212 * expr.h (try_casesi, try_tablejump): Update prototypes.
15213 * gimple-pretty-print.c (dump_probability): Update.
15214 (dump_profile): New.
15215 (dump_gimple_label): Update.
15216 (dump_gimple_bb_header): Update.
15217 * graph.c (draw_cfg_node_succ_edges): Update.
15218 * hsa-gen.c (convert_switch_statements): Update.
15219 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
15220 (find_if_case_1): Update.
15221 (find_if_case_2): Update.
15222 * internal-fn.c (expand_arith_overflow_result_store): Update.
15223 (expand_addsub_overflow): Update.
15224 (expand_neg_overflow): Update.
15225 (expand_mul_overflow): Update.
15226 (expand_vector_ubsan_overflow): Update.
15227 * ipa-cp.c (good_cloning_opportunity_p): Update.
15228 * ipa-split.c (split_function): Use make_single_succ_edge.
15229 * ipa-utils.c (ipa_merge_profiles): Update.
15230 * loop-doloop.c (add_test): Update.
15231 (doloop_modify): Update.
15232 * loop-unroll.c (compare_and_jump_seq): Update.
15233 (unroll_loop_runtime_iterations): Update.
15234 * lra-constraints.c (lra_inheritance): Update.
15235 * lto-streamer-in.c (input_cfg): Update.
15236 * lto-streamer-out.c (output_cfg): Update.
15237 * mcf.c (adjust_cfg_counts): Update.
15238 * modulo-sched.c (sms_schedule): Update.
15239 * omp-expand.c (expand_omp_for_init_counts): Update.
15240 (extract_omp_for_update_vars): Update.
15241 (expand_omp_ordered_sink): Update.
15242 (expand_omp_for_ordered_loops): Update.
15243 (expand_omp_for_generic): Update.
15244 (expand_omp_for_static_nochunk): Update.
15245 (expand_omp_for_static_chunk): Update.
15246 (expand_cilk_for): Update.
15247 (expand_omp_simd): Update.
15248 (expand_omp_taskloop_for_outer): Update.
15249 (expand_omp_taskloop_for_inner): Update.
15250 * omp-simd-clone.c (simd_clone_adjust): Update.
15251 * optabs.c (expand_doubleword_shift): Update.
15252 (expand_abs): Update.
15253 (emit_cmp_and_jump_insn_1): Update.
15254 (expand_compare_and_swap_loop): Update.
15255 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
15256 * predict.c (predictable_edge_p): Update.
15257 (edge_probability_reliable_p): Update.
15258 (set_even_probabilities): Update.
15259 (combine_predictions_for_insn): Update.
15260 (combine_predictions_for_bb): Update.
15261 (propagate_freq): Update.
15262 (estimate_bb_frequencies): Update.
15263 (force_edge_cold): Update.
15264 * profile-count.c (profile_count::dump): Add missing space into dump.
15265 (profile_count::debug): Add newline.
15266 (profile_count::differs_from_p): Explicitly convert to unsigned.
15267 (profile_count::stream_in): Update.
15268 (profile_probability::dump): New member function.
15269 (profile_probability::debug): New member function.
15270 (profile_probability::differs_from_p): New member function.
15271 (profile_probability::differs_lot_from_p): New member function.
15272 (profile_probability::stream_in): New member function.
15273 (profile_probability::stream_out): New member function.
15274 * profile-count.h (profile_count_quality): Rename to ...
15275 (profile_quality): ... this one.
15276 (profile_probability): New.
15277 (profile_count): Update.
15278 * profile.c (compute_branch_probabilities): Update.
15279 * recog.c (peep2_attempt): Update.
15280 * sched-ebb.c (schedule_ebbs): Update.
15281 * sched-rgn.c (find_single_block_region): Update.
15282 (compute_dom_prob_ps): Update.
15283 (schedule_region): Update.
15284 * sel-sched-ir.c (compute_succs_info): Update.
15285 * stmt.c (struct case_node): Update.
15286 (do_jump_if_equal): Update.
15287 (get_outgoing_edge_probs): Update.
15288 (conditional_probability): Update.
15289 (emit_case_dispatch_table): Update.
15290 (expand_case): Update.
15291 (expand_sjlj_dispatch_table): Update.
15292 (emit_case_nodes): Update.
15293 * targhooks.c: Update.
15294 * tracer.c (better_p): Update.
15295 (find_best_successor): Update.
15296 * trans-mem.c (expand_transaction): Update.
15297 * tree-call-cdce.c: Update.
15298 * tree-cfg.c (gimple_split_edge): Upate.
15299 (move_sese_region_to_fn): Upate.
15300 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
15301 * tree-eh.c (lower_resx): Upate.
15302 (cleanup_empty_eh_move_lp): Upate.
15303 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15304 * tree-inline.c (copy_edges_for_bb): Update.
15305 (copy_cfg_body): Update.
15306 * tree-parloops.c (gen_parallel_loop): Update.
15307 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
15308 (gimple_gen_time_profiler): Update.
15309 * tree-ssa-dce.c (remove_dead_stmt): Update.
15310 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15311 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
15312 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
15313 (unloop_loops): Update.
15314 (try_peel_loop): Update.
15315 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15316 * tree-ssa-loop-split.c (connect_loops): Update.
15317 (split_loop): Update.
15318 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15319 (hoist_guard): Update.
15320 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
15321 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
15322 (value_replacement): Update.
15323 * tree-ssa-reassoc.c (branch_fixup): Update.
15324 * tree-ssa-tail-merge.c (replace_block_by): Update.
15325 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
15326 (create_edge_and_update_destination_phis): Update.
15327 (compute_path_counts): Update.
15328 (recompute_probabilities): Update.
15329 (update_joiner_offpath_counts): Update.
15330 (freqs_to_counts_path): Update.
15331 (duplicate_thread_path): Update.
15332 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
15333 (struct switch_conv_info): Update.
15334 (gen_inbound_check): Update.
15335 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15336 (vect_do_peeling): Update.
15337 (vect_loop_versioning): Update.
15338 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15339 (optimize_mask_stores): Update.
15340 * ubsan.c (ubsan_expand_null_ifn): Update.
15341 * value-prof.c (gimple_divmod_fixed_value): Update.
15342 (gimple_divmod_fixed_value_transform): Update.
15343 (gimple_mod_pow2): Update.
15344 (gimple_mod_pow2_value_transform): Update.
15345 (gimple_mod_subtract): Update.
15346 (gimple_mod_subtract_transform): Update.
15347 (gimple_ic): Update.
15348 (gimple_stringop_fixed_value): Update.
15349 (gimple_stringops_transform): Update.
15350 * value-prof.h: Update.
15352 2017-06-29 Carl Love <cel@us.ibm.com>
15354 * config/rs6000/rs6000-c.c: Add support for built-in functions
15355 vector signed int vec_signed (vector float);
15356 vector signed long long vec_signed (vector double);
15357 vector signed int vec_signed2 (vector double, vector double);
15358 vector signed int vec_signede (vector double);
15359 vector signed int vec_signedo (vector double);
15360 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
15361 instruction generator.
15362 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15363 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
15364 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
15366 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
15367 vunsignede_v2df): Add define_expands.
15368 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
15369 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
15370 VEC_UNSIGNEDO): Add definitions.
15371 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15372 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
15373 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
15374 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
15375 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
15376 * config/rs6000/altivec.h (vec_signed, vec_signed2,
15377 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
15378 vec_unsignede, vec_unsignedo): Add builtin defines.
15379 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
15381 * doc/extend.texi: Update the built-in documentation file for the
15382 new built-in functions.
15384 2017-06-29 Richard Biener <rguenther@suse.de>
15386 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
15387 reduction chains to LOOP_VINFO_REDUCTIONS.
15388 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
15389 SLP reductions after processing reduction chains.
15391 2017-06-29 Nathan Sidwell <nathan@acm.org>
15393 * builtins.c (fold_builtin_FUNCTION): Use
15394 lang_hooks.decl_printable_name.
15396 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
15398 PR middle-end/81194
15399 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
15400 with only one label.
15401 * stmt.c (expand_case): Assert NCASES is greater than one.
15403 2017-06-29 Richard Biener <rguenther@suse.de>
15405 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
15407 (group_case_labels): Likewise.
15408 (find_taken_edge): Push sanity checking on val to workers...
15409 (find_taken_edge_cond_expr): ... here
15410 (find_taken_edge_switch_expr): ... and here, handle cases
15411 with just a default label.
15412 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
15413 (group_case_labels): Likewise.
15414 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
15415 group_case_labels does anything cleanup the CFG again.
15417 2017-06-29 Bin Cheng <bin.cheng@arm.com>
15419 PR tree-optimization/81196
15420 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
15421 exit condition comparing two IVs.
15423 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
15425 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
15426 profile to the dummy entry at the end of the list of architectures.
15427 * config/arm/arm-cpu-cdata.h: Regenerated.
15429 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15430 Michael Collison <michael.collison@arm.com>
15433 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
15435 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
15436 (*aarch64_reg_<mode>3_minus_mask): New pattern.
15437 (*aarch64_<optab>_reg_di3_mask2): New pattern.
15438 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
15439 of shift when the shift amount is masked with constant equal to
15440 the size of the mode.
15441 * config/aarch64/predicates.md (subreg_lowpart_operator): New
15444 2017-06-29 Martin Liska <mliska@suse.cz>
15446 * config/i386/i386.opt: Change range from [1,5] to [0,5].
15448 2017-06-29 Yury Gribov <tetra2005@gmail.com>
15451 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
15454 (edge_growth_cache_entry::edge_growth_cache_entry): New
15456 (reset_edge_growth_cache): Update to use constructor.
15458 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15460 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
15461 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15462 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15464 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
15466 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
15467 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
15469 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
15471 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
15472 (*-linux-uclibc*): Add t-uclibc tmake_file.
15473 * config/t-musl: New.
15474 * config/t-uclibc: New.
15476 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
15478 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
15480 (gen_comm_data): Emit architectural setting of arch_prof.
15481 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
15483 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
15484 (armv8-m.base, armv8-m.main): Likewise.
15485 * arm-protos.h (arm_build_target): Add profile field.
15486 (arch_option): Likewise.
15487 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
15489 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
15490 arm_active_target.profile.
15492 2017-06-28 Richard Biener <rguenther@suse.de>
15494 PR middle-end/81227
15495 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
15496 TYPE_OVERFLOW_WRAPS.
15497 * match.pd (negate_expr_p): Likewise.
15498 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
15499 fold_build2, not fold_binary.
15501 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15503 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
15504 Convert memory address to Pmode.
15505 (aarch64_print_operand): Assert MEM operands are always Pmode.
15507 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15510 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
15511 Remove redundant if.
15512 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
15513 * config/arm/aarch-common-protos.h
15514 (aarch_forward_to_shift_is_not_shifted_re): Remove.
15515 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
15517 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
15520 * multiple_target.c (create_dispatcher_calls): Set the default
15521 clone to be static, not public.
15523 2017-06-28 Richard Biener <rguenther@suse.de>
15525 * tree-vect-loop.c (vectorizable_reduction): Move special
15526 cond reduction IV var creation ...
15527 (vect_create_epilog_for_reduction): ... here. Remove induction_index
15528 parameter. Use STMT_VINFO_VECTYPE.
15529 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
15532 2017-06-28 Martin Liska <mliska@suse.cz>
15535 * ipa-visibility.c (non_local_p): Handle visibility.
15537 2017-06-28 Martin Liska <mliska@suse.cz>
15540 * common.opt: Add IntegerRange to various options.
15541 * opt-functions.awk (integer_range_info): New function.
15542 * optc-gen.awk: Add integer_range_info to cl_options struct.
15543 * opts-common.c (decode_cmdline_option): Handle
15544 CL_ERR_INT_RANGE_ARG.
15545 (cmdline_handle_error): Likewise.
15546 * opts.c (print_filtered_help): Show valid interval in
15547 when --help is provided.
15548 * opts.h (struct cl_option): Add range_min and range_max fields.
15549 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
15551 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
15553 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
15554 (x * C EQ/NE y * C): New transformation.
15556 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
15558 * genmultilib (combination_space): Accept '+' in option names.
15560 2017-06-28 Martin Liska <mliska@suse.cz>
15563 * asan.c (instrument_derefs): Bail out inner references
15564 that are hard register variables.
15566 2017-06-28 Jakub Jelinek <jakub@redhat.com>
15569 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
15570 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
15572 2017-06-28 Richard Biener <rguenther@suse.de>
15574 * tree-vectorizer.h (vect_get_vec_defs): Remove.
15575 (vect_get_slp_defs): Adjust.
15576 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
15578 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
15580 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
15581 get_initial_defs_for_reduction instead of vect_get_vec_defs.
15582 (vectorizable_reduction): Adjust.
15583 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
15585 (vect_get_slp_defs): Likewise.
15586 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
15587 (vectorizable_bswap): Adjust.
15588 (vectorizable_call): Likewise.
15589 (vectorizable_conversion): Likewise.
15590 (vectorizable_assignment): Likewise.
15591 (vectorizable_shift): Likewise.
15592 (vectorizable_operation): Likewise.
15593 (vectorizable_store): Likewise.
15594 (vectorizable_condition): Likewise.
15595 (vectorizable_comparison): Likewise.
15597 2017-06-28 Michael Collison <michael.collison@arm.com>
15600 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
15602 (arm_gen_movmemqi): Removed unused variable 'i'.
15603 Convert 'for' loop into 'while' loop.
15604 (arm_expand_prologue): Remove last unnecessary set of insn.
15605 (thumb_pop): Remove unused variable 'pushed_words'.
15606 (thumb_exit): Remove last unnecessary set of regs_to_pop.
15608 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15610 * config/s390/predicates.md: Use s390_rel_address_ok_p.
15611 * config/s390/s390-protos.h: Add prototype of
15612 s390_rel_address_ok_p.
15613 * config/s390/s390.c (s390_got_symbol): New function.
15614 (s390_rel_address_ok_p): New function.
15615 (legitimize_pic_address): Use s390_rel_address_ok_p.
15616 (s390_load_got): Use s390_got_symbol.
15617 (s390_option_override): Issue error if
15618 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
15619 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
15621 * config/s390/s390.opt: New option mpic-data-is-text-relative.
15623 2017-06-27 Andrew Pinski <apinski@cavium.com>
15625 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
15626 (X * copysign (1.0, X)): New pattern.
15627 (X * copysign (1.0, -X)): New pattern.
15628 (copysign (-1.0, CST)): New pattern.
15630 2017-06-27 Joseph Myers <joseph@codesourcery.com>
15632 * genmultilib (combination_space): Remove variable.
15633 Validate reuse rules against regular expression for any sequence
15634 of multilib options in any order.
15636 2017-06-27 Michael Collison <michael.collison@arm.com>
15638 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
15639 call aarch64_split_simd_combine.
15640 * (aarch64_combine_internal<mode>): Delete pattern.
15641 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
15642 Allow register and subreg operands.
15644 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15646 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
15647 specific need, just fallback on defaults.
15648 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
15650 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15651 Olivier Hainque <hainque@adacore.com>
15653 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
15655 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
15656 targets. Pick a default if no particular attempt applied.
15657 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
15660 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15662 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
15663 (x86_64-wrs-vxworks7): Likewise.
15665 2017-06-27 Marek Polacek <polacek@redhat.com>
15668 * ubsan.c (instrument_null): Check get_base_address's result for null.
15670 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15672 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
15674 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15676 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
15677 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
15678 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
15679 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
15680 New function types.
15681 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
15682 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
15683 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
15684 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
15685 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
15686 BUILT_IN_FEUPDATEENV): New builtins.
15687 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
15688 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
15689 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
15690 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
15692 (builtin_structptr_types): Adjust size.
15693 * tree.c (builtin_structptr_types): Add four entries.
15695 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15696 Olivier Hainque <hainque@adacore.com>
15698 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
15699 (TLS_SYM): New local macro, forcing reference to __tls__ on
15700 link command lines for VxWorks 7 RTPs, triggering initialization
15702 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
15703 OS features TLS support, true for RTPs on VxWorks 7.
15704 * config/vxworks.c (vxworks_override_options): Setup emutls
15707 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15709 * predict.c (test_prediction_value_range): Use -1U instead of -1
15710 to avoid narrowing conversion warning.
15711 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
15712 to avoid narrowing conversion warning.
15713 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
15715 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
15717 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15719 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
15720 64bit configurations.
15721 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
15722 (SIZE_TYPE): Likewise.
15723 * config/vxworks.c (vxworks_emutls_var_fields): Use
15724 long_unsigned_type_node instead of unsigned_type_node as the offset
15725 field type, which is "pointer" mode in emutls.c.
15727 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15730 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
15732 PR middle-end/81207
15733 * gimple-fold.c (replace_call_with_call_and_fold): Handle
15734 gimple_vuse copying separately from gimple_vdef copying.
15736 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15738 * value-prof.c (free_hist): Remove call to memset and the enclosing if
15741 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15742 Olivier Hainque <hainque@adacore.com>
15744 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
15745 for all vxworks7 targets.
15746 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
15747 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
15748 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
15749 variations for VX6/VX7 and 32/64bits later on in ...
15750 (VXWORKS_LIB_SPEC): Leverage new macros.
15751 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
15752 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
15754 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15756 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
15757 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
15759 2017-06-26 Carl Love <cel@us.ibm.com>
15761 * config/rs6000/rs6000-c.c: Add support for built-in functions
15762 vector bool char vec_reve (vector bool char);
15763 vector signed char vec_reve (vector signed char);
15764 vector unsigned char vec_reve (vector unsigned char);
15765 vector bool int vec_reve (vector bool int);
15766 vector signed int vec_reve (vector signed int);
15767 vector unsigned int vec_reve (vector unsigned int);
15768 vector bool long long vec_reve (vector bool long long);
15769 vector signed long long vec_reve (vector signed long long);
15770 vector unsigned long long vec_reve (vector unsigned long long);
15771 vector bool short vec_reve (vector bool short);
15772 vector signed short vec_reve (vector signed short);
15773 vector double vec_reve (vector double);
15774 vector float vec_reve (vector float);
15775 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
15776 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
15777 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
15778 (altivec_vreve): New pattern.
15779 * config/rs6000/altivec.h (vec_reve): New define.
15780 * doc/extend.texi (vec_rev): Update the built-in documentation file
15781 for the new built-in functions.
15783 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15785 PR tree-optimization/71815
15786 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
15788 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
15790 (slsr_process_phi): Likewise.
15791 (replace_uncond_cands_and_profitable_phis): Don't replace a
15792 multiply candidate with a stride of 1 (copy or cast).
15793 (phi_incr_cost): Call uses_consumed_by_stmt rather than
15795 (lowest_cost_path): Likewise.
15796 (total_savings): Likewise.
15798 2017-06-26 Richard Biener <rguenther@suse.de>
15801 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
15802 Use def_builtin_pure for all gather builtins.
15804 2017-06-26 Richard Biener <rguenther@suse.de>
15806 PR tree-optimization/81203
15807 * tree-tailcall.c (find_tail_calls): Do not move stmts into
15808 non-dominating BBs.
15810 2017-06-26 Marek Polacek <polacek@redhat.com>
15813 * doc/invoke.texi: Document -Wmultistatement-macros.
15815 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
15817 * doc/sourcebuild.texi (ARM-specific attributes): Document new
15818 arm_neon_ok_no_float_abi effective target.
15820 2017-06-26 Richard Biener <rguenther@suse.de>
15822 PR tree-optimization/80928
15823 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
15824 (copy_bbs): Set BB_DUPLICATED flag early.
15825 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
15827 (execute_on_shrinking_pred): Likewise.
15828 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
15829 BB_DUPLICATED blocks.
15830 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
15831 iterate over all PHIs considering removal of *gsi.
15833 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
15835 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
15838 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15840 * config/rs6000/rs6000-string.c: (expand_block_clear,
15841 do_load_for_compare, select_block_compare_mode,
15842 compute_current_alignment, expand_block_compare,
15843 expand_strncmp_align_check, expand_strn_compare,
15844 expand_block_move, rs6000_output_load_multiple)
15845 Move functions related to string/block move/compare
15846 to a separate file.
15847 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
15848 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
15849 for this function which is now used in two files.
15850 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
15851 * config.gcc: Add rs6000-string.o to extra_objs for
15852 targets powerpc*-*-* and rs6000*-*-*.
15854 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
15857 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
15858 32-bit, since indexed is not valid for DImode.
15859 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
15860 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
15861 (define_peephole2 for Altivec d-form load): Add 32-bit support.
15862 (define_peephole2 for Altivec d-form store): Likewise.
15865 * multiple_target.c (create_dispatcher_calls): Only create the
15866 dispatcher call if the function is the default clone of a
15867 versioned function.
15869 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
15871 PR middle-end/80902
15872 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
15873 a call, force the call to not be a tail call.
15875 2017-06-23 Jeff Law <law@redhat.com>
15877 * doc/contrib.texi: Add entry for Steven Pemberton's work on
15880 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
15882 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
15883 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
15884 handling for early expansion of vector shifts (sl,sr,sra,rl).
15885 (builtin_function_type): Add vector shift right instructions
15886 to the unsigned argument list.
15888 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
15890 rtl-optimizatoin/79286
15891 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
15892 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
15893 trap. PIC register plus a const unspec without offset can never trap.
15895 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
15897 * tree.h (builtin_structptr_type): New type.
15898 (builtin_structptr_types): Declare new array.
15899 * tree.c (builtin_structptr_types): New array.
15900 (free_lang_data, build_common_tree_nodes): Use it.
15902 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
15905 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
15908 2017-06-22 Matt Turner <mattst88@gmail.com>
15910 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
15911 Lake models to skylake case. Assume skylake for unknown
15912 models with clflushopt.
15914 2017-06-22 Jeff Law <law@redhat.com>
15916 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
15917 frame sizes that do not satisfy aarch64_uimm12_shift.
15919 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
15921 * profile-count.h (apply_probability,
15922 apply_scale, probability_in): Fix checks for zero.
15924 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15926 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
15927 * doc/cppdiropts.texi (-I @var{dir}): Document it.
15929 2016-06-22 Richard Biener <rguenther@suse.de>
15931 * tree-vect-loop.c (vect_model_reduction_cost): Handle
15932 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
15933 REDUC_MAX_EXPR support.
15934 (vectorizable_reduction): Likewise.
15935 (vect_create_epilog_for_reduction): Likewise.
15937 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
15939 * match.pd (A / (1 << B) -> A >> B): New.
15940 * generic-match-head.c: Include optabs-tree.h.
15941 * gimple-match-head.c: Likewise.
15942 * optabs-tree.h (target_supports_op_p): New.
15943 * optabs-tree.c (target_supports_op_p): New.
15945 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15947 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
15948 $gcc_cv_ld --help output.
15949 (gcc_cv_ld_demangle): Likewise.
15950 (gcc_cv_ld_eh_frame_hdr): Likewise.
15951 (gcc_cv_ld_pie): Likewise.
15952 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
15953 (gcc_cv_ld_buildid): Likewise.
15954 (gcc_cv_ld_sysroot): Likewise.
15955 (ld_bndplt_support): Likewise.
15956 (ld_pushpopstate_support): Likewise.
15957 * configure: Regenerate.
15958 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
15960 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15963 * config/i386/sse.md (round<mode>2): Renumber match_dup and
15964 operands indexes to avoid gap between operands and match_dups.
15966 2017-06-21 Andrew Pinski <apinski@cavium.com>
15968 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
15969 Increment Arith_shift and Arith_shift_reg by 1.
15970 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
15972 * config/aarch64/aarch64.c (thunderx_tunings): Enable
15973 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
15974 (aarch64_strip_extend): Add new argument and test for it.
15975 (aarch64_cheap_mult_shift_p): New function.
15976 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
15977 add a cost if it is true.
15978 Update calls to aarch64_strip_extend.
15979 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
15981 2017-06-21 Andrew Pinski <apinski@cavium.com>
15983 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
15985 (thunderxt88): Likewise.
15986 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
15987 (thunderx_prefetch_tune): New variable.
15988 (thunderx2t99_prefetch_tune): Update for the correct values.
15989 (thunderxt88_tunings): New variable.
15990 (thunderx_tunings): Use thunderx_prefetch_tune instead of
15991 generic_prefetch_tune.
15992 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
15994 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15996 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
15997 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
15998 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
15999 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
16000 (aarch64_atomic_cas<mode>, GPI): Likewise.
16002 2017-06-21 Martin Liska <mliska@suse.cz>
16004 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
16005 statements on cold and hot labels.
16006 * predict.c (tree_estimate_probability_bb): Remove the
16007 prediction from this place.
16009 2017-06-21 Martin Liska <mliska@suse.cz>
16011 PR tree-optimization/79489
16012 * gimplify.c (maybe_add_early_return_predict_stmt): New
16014 (gimplify_return_expr): Call the function.
16015 * predict.c (tree_estimate_probability_bb): Remove handling
16017 * predict.def: Update comment about early return predictor.
16018 * gimple-predict.h (is_gimple_predict): New function.
16019 * predict.def: Change default value of early return to 66.
16020 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
16022 * passes.def: Put pass_strip_predict_hints to the beginning of
16025 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
16027 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
16028 FUNCTION_DECL declarations.
16029 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
16031 (dwaf2out_decl): Likewise.
16032 * godump.c (go_early_global_decl): Skip call to the real debug hook
16033 for FUNCTION_DECL declarations.
16034 * passes.c (rest_of_decl_compilation): Skip call to the
16035 early_global_decl debug hook for FUNCTION_DECL declarations, unless
16036 -fdump-go-spec is passed.
16038 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16040 * config/i386/i386.c (struct builtin_isa): New field pure_p.
16041 Reorder for compactness.
16042 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
16043 (def_builtin_pure, def_builtin_pure2): New functions.
16044 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
16046 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16048 * match.pd (nop_convert): New predicate.
16049 ((A +- CST1) +- CST2): Allow some NOP conversions.
16051 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16054 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
16055 with ctors/dtors if GOVD_SHARED is set.
16057 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16059 * config/aarch64/aarch64.md (movti_aarch64):
16060 Emit mov rather than orr.
16061 (movtf_aarch64): Likewise.
16062 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
16063 Emit mov rather than orr.
16065 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16067 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16068 Swap alternatives, make integer dup more expensive.
16070 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16072 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16073 Return true for non-tls symbols.
16075 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16077 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16078 (cortex-a75): Likewise.
16079 (cortex-a75.cortex-a55): Likewise.
16080 * config/aarch64/aarch64-tune.md: Regenerate.
16081 * doc/invoke.texi (-mtune): Document new values for -mtune.
16083 2017-06-21 Tom de Vries <tom@codesourcery.com>
16085 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16086 stack_size feature.
16087 (Effective-Target Keywords, Other attributes): Suggest using
16088 dg-add-options stack_size feature to get stack limit in stack_size
16089 effective target documentation.
16091 2017-06-21 Julian Brown <julian@codesourcery.com>
16092 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16094 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16095 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16096 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16098 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16099 attribute type list for neon_multiply.
16100 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16101 attribute type list for neon_multiply.
16102 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16103 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16104 attribute type list for neon_multiply.
16105 * config/arm/types.md (crypto_pmull): Add.
16106 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16107 attribute type list.
16109 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16111 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16114 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16116 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16117 to make sure not to dereference a NULL cost_classes_ptr pointer.
16119 2017-06-20 Carl Love <cel@us.ibm.com>
16121 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16122 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16123 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16124 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16125 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16126 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16127 VMULOSW): New enum "unspec" values.
16128 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16129 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16130 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16131 altivec_vmulosw): New patterns.
16132 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16133 VMULOSW): Add definitions.
16135 2017-06-20 Julia Koval <julia.koval@intel.com>
16137 * config/i386/i386.c: Fix rounding expand for new pattern.
16138 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16140 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16142 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16143 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
16145 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16147 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
16150 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16152 * config/aarch64/aarch64-cores.def: Rearrange to sort by
16153 architecture, then by implementer ID.
16154 * config/aarch64/aarch64-tune.md: Regenerate.
16156 2017-06-20 Richard Biener <rguenther@suse.de>
16158 PR middle-end/81097
16159 * fold-const.c (split_tree): Fold to type before negating.
16161 2017-06-20 David Malcolm <dmalcolm@redhat.com>
16163 * diagnostic-show-locus.c
16164 (selftest::test_fixit_deletion_affecting_newline): New function.
16165 (selftest::diagnostic_show_locus_c_tests): Call it.
16167 2017-06-20 Andreas Schwab <schwab@suse.de>
16170 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
16173 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16175 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
16176 __ARM_FEATURE_COPROC according to support.
16178 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16180 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16181 Rewritten to avoid overflow for > 32-bit pointers.
16184 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
16185 by removing enum keyword.
16186 (ubsan_type_descriptor): Likewise. Formatting fix.
16189 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
16190 splitter): Require TARGET_SSE2 in the condition.
16192 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
16195 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
16196 for doing vector set of SFmode on ISA 3.0.
16197 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
16198 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
16200 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
16201 SFmode value into a V4SF variable that was extracted from another
16202 V4SF variable without converting the element to double precision
16203 and back to single precision vector format.
16204 (vsx_insert_extract_v4sf_p9_2): Likewise.
16206 2017-06-19 Jakub Jelinek <jakub@redhat.com>
16208 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
16209 in UWHI to avoid undefined overflow.
16212 * ubsan.h (enum ubsan_encode_value_phase): New.
16213 (ubsan_encode_value): Change second argument to
16214 enum ubsan_encode_value_phase with default value of
16215 UBSAN_ENCODE_VALUE_GENERIC.
16216 * ubsan.c (ubsan_encode_value): Change second argument to
16217 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
16218 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
16219 create_tmp_var_raw instead of create_tmp_var and use a
16221 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
16222 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
16223 ubsan_encode_value callers.
16226 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
16227 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
16228 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
16230 2017-06-19 Richard Biener <rguenther@suse.de>
16232 PR middle-end/81118
16233 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
16234 estimates if we changed anything.
16236 2017-06-19 Richard Biener <rguenther@suse.de>
16238 PR tree-optimization/80887
16239 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
16240 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
16241 simplified lookups, then reset mprts_hook.
16242 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
16244 (try_to_simplify): Likewise.
16246 2017-06-19 Martin Liska <mliska@suse.cz>
16249 * gimplify.c (gimplify_switch_expr):
16250 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
16252 2017-06-19 Martin Liska <mliska@suse.cz>
16254 * doc/install.texi: Document that PGO runs in 4 stages.
16256 2017-06-19 Martin Liska <mliska@suse.cz>
16259 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
16260 to dispatcher function name.
16261 * multiple_target.c (replace_function_decl): New function.
16262 (create_dispatcher_calls): Redirect both edges and references.
16264 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
16266 * profile-count.c (profile_count::dump): Dump quality.
16267 (profile_count::differs_from_p): Update for unsigned val.
16268 * profile-count.h (profile_count_quality): New enum.
16269 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
16271 2017-06-19 Richard Biener <rguenther@suse.de>
16273 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
16274 struct function as arg.
16275 (estimate_numbers_of_iterations): Export overload with loop arg.
16276 (free_numbers_of_iterations_estimates_loop): Use an overload of
16277 free_numbers_of_iterations_estimates instead.
16278 * tree-cfg.c (remove_bb): Adjust.
16279 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
16280 * tree-parloops.c (gen_parallel_loop): Likewise.
16281 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
16283 (tree_unroll_loops_completely): Likewise.
16284 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
16285 Use an overload instead and export.
16286 (estimated_loop_iterations): Adjust.
16287 (max_loop_iterations): Likewise.
16288 (likely_max_loop_iterations): Likewise.
16289 (estimate_numbers_of_iterations): Take struct function as arg
16291 (loop_exits_before_overflow): Adjust.
16292 (free_numbers_of_iterations_estimates_loop): Use an overload.
16293 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
16294 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
16296 2017-06-19 Richard Biener <rguenther@suse.de>
16299 * ipa-prop.c (find_constructor_constant_at_offset): Handle
16300 RANGE_EXPR conservatively.
16302 2017-06-16 Carl Love <cel@us.ibm.com>
16304 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16305 definitions for vec_float, vec_float2, vec_floato,
16306 vec_floate built-ins.
16307 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
16308 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
16310 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
16311 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
16312 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
16313 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
16314 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
16315 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
16316 vec_floato): Add builtin defines.
16317 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
16318 Update the built-in documentation file for the new built-in
16321 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16323 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
16324 (mthumb): Mark as the negative of -marm.
16326 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16328 * doc/invoke.texi (ARM Options, -mcpu): Document supported
16330 (ARM Options, -mtune): Document that this accepts the same
16331 extension options as -mcpu.
16332 (ARM Options, -mfpu): Document addition of -mfpu=auto.
16334 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16336 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
16337 permitted extensions.
16339 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16341 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
16342 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
16343 (armv8-m.main): Add option +nodsp.
16344 * config/arm/arm-cpu-cdata.h: Regenerated.
16346 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16348 * config/arm/t-fuchsia: New file.
16349 * config.gcc (arm*-*-fuchsia*): Use it.
16351 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16353 * config/arm/t-symbian: Rewrite for new option infrastructure.
16355 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16357 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
16358 (MULTILIB_REQUIRED): Likewise.
16360 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16362 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
16363 (MULTILIB_RESUE): Likewise.
16364 (MULTILIB_MATCHES): Likewise.
16365 (MULTLIB_REQUIRED): Likewise.
16367 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16369 * config/arm/t-rtems: Rewrite for new option framework.
16371 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16373 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
16374 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
16375 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
16376 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
16377 * config/arm/t-multilib: ... here.
16378 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
16379 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
16380 armv7-a and armv8*-a when A-profile libraries have not been built.
16381 * config/arm/t-rmprofile: Rewrite.
16383 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16385 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
16386 with a backslash. Remove the backslash after substituting unescaped
16388 * doc/fragments.texi (MULTILIB_REUSE): Document it.
16390 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16392 * config.gcc: (arm*-*-*): When building a-profile libraries, force
16393 the driver to pass through the default setting of -mfloat-abi.
16394 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
16396 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
16397 (all_feat_combs): New rule.
16398 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
16400 * config/arm/t-aprofile: Rewrite.
16402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16404 * config/arm/arm.h (FPUTYPE_AUTO): Define.
16405 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
16406 fpu is not specified by the user/command-line.
16407 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
16408 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
16409 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
16410 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
16411 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
16412 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
16414 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16416 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
16417 * config/arm/t-arm-elf: Rewritten.
16419 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16421 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
16422 have some floating-point instructions.
16423 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
16424 (TARGET_MAYBE_HARD_FLOAT): New macro.
16425 * config/arm/arm-builtins.c (arm_init_builtins): Use
16426 TARGET_MAYBE_HARD_FLOAT.
16427 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
16429 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16431 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
16432 (configargs.h): Include it.
16433 (arm_print_hint_for_fpu_option): New function.
16434 (arm_parse_fpu_option): New function.
16435 (candidate_extension): New class.
16436 (arm_canon_for_multilib): New function.
16437 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
16438 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
16439 (ARCH_CANONICAL_SPECS): New macro.
16440 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
16442 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16444 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
16445 are set after handling multilib fragments. Set target_cpu_default2
16448 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16450 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
16452 (arm*-*-*): Set target_cpu_default2 to a quoted string.
16453 * config/arm/parsecpu.awk (check_cpu): Validate any extension
16455 (check_arch): Likewise.
16456 * config/arm/arm.c (arm_configure_build_target): Handle
16457 TARGET_CPU_DEFAULT being a string constant. Scan any feature
16458 options in the default.
16460 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16462 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
16463 when an option is an alias of another.
16464 * config/arm/parsecpu.awk (optalias): New parser token.
16465 (gen_comm_data): Mark non-alias options as such. Emit entries
16466 for extension aliases.
16467 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
16468 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
16469 (armv6kz, armv6zk, armv6t2): Likewise.
16470 (armv7): Make vfpv3-d16 an alias.
16471 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
16473 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
16474 Sort in canonical order.
16475 (armv8-a): Sort in canonical order.
16476 (armv8.1-a, armv8.2-a): Likewise.
16477 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
16479 (cortex-a9): Sort in canonical order.
16480 * config/arm/arm.c (selftests.h): Include it.
16481 (arm_test_cpu_arch_data): New function.
16482 (arm_run_self_tests): New function.
16483 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
16484 (targetm): Move declaration to the end of the file.
16485 * arm-cpu-cdata.h: Regenerated.
16487 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16489 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
16490 call to target_mode_check describing the type of option passed.
16491 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
16492 (arm_target_thumb_only): Use arm_parse_arch_option_name or
16493 arm_parse_cpu_option_name to match parameters against list of
16495 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
16496 arm_arch_core_flags data structure.
16497 * config/arm/arm-cpu_cdata.h: Regenerated.
16499 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16501 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
16503 (arm_print_hint_for_cpu_option): Likewise.
16504 (arm_print_hint_for_arch_option): Likewise.
16505 (arm_parse_cpu_option_name): Likewise.
16506 (arm_parse_arch_option_name): Likewise.
16507 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
16508 of entries in the all_fpus list.
16509 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
16510 (arm_parse_cpu_option_name): Declare.
16511 (arm_parse_arch_option_name): Declare.
16512 (arm_parse_option_features): Declare.
16513 (arm_intialize_isa): Declare.
16514 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
16516 (gen_comm_data): ... here. Make definitions non-static.
16517 * config/arm/arm-cpu-data.h: Regenerated.
16518 * config/arm/arm-cpu-cdata.h: Regenerated.
16520 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16522 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
16523 (cpu_arch_extension): New structure.
16524 (cpu_arch_option, arch_option, cpu_option): New structures.
16525 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
16526 architecture types.
16527 (gen_data): Generate new format data tables.
16528 * config/arm/arm.c (cpu_tune): New structure.
16529 (cpu_option, processors): Delete.
16530 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
16531 (arm_print_hint_for_cpu_option): ... this and ...
16532 (arm_print_hint_for_arch_option): ... this.
16533 (arm_parse_arch_cpu_name): Delete. Replace with ...
16534 (arm_parse_cpu_option_name): ... this and ...
16535 (arm_parse_arch_option_name): ... this.
16536 (arm_unrecognized_feature): Change type of target parameter to
16538 (arm_parse_arch_cpu_features): Delete. Replace with ...
16539 (arm_parse_option_features): ... this.
16540 (arm_configure_build_target): Rework to use new configuration data
16542 (arm_print_tune_info): Rework for new configuration data tables.
16543 * config/arm/arm-cpu-data.h: Regenerated.
16544 * config/arm/arm-cpu.h: Regenerated.
16546 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16548 * Makefile.in (OBJS): Move sbitmap.o from here ...
16549 (OBJS-libcommon): ... to here.
16551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16553 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
16554 (ISA_ALL_CRYPTO): New macro.
16555 (ISA_ALL_SIMD): New macro
16556 (ISA_ALL_FP): New macro.
16557 * config/arm/arm.c (fpu_bitlist): Update initializer.
16558 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
16560 (arm9e): Add fpu. Add option for nofp
16561 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
16562 (arm926ej-s, arm1026ej-s): Likewise.
16563 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
16564 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
16565 neon-fp16, neon-vfpv4, nofp and nosimd.
16566 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
16567 (cortex-a8): Add fpu. Add option for nofp.
16568 (cortex-a9): Add fpu. Add options for nosimd and nofp.
16569 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
16570 (cortex-r4f): Add fpu.
16571 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
16572 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
16574 (cortex-r8): Likewise.
16575 (cortex-m4): Add fpu. Add option for nofp.
16576 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
16577 (cortex-a17.cortex-a7): Likewise.
16578 (cortex-a32): Add fpu. Add options for crypto and nofp.
16579 (cortex-a35, cortex-a53): Likewise.
16580 (cortex-a57): Add fpu. Add option for crypto.
16581 (cortex-a72, cortex-a73): Likewise.
16582 (exynos-m1): Likewise.
16583 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
16584 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
16585 (cortex-m33): Add fpu. Add option for nofp.
16586 * config/arm/arm-cpu-cdata.h: Regenerated
16587 * config/arm/arm-cpu-data.h: Regenerated.
16589 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16591 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
16592 (armv5te, armv5tej): Likewise.
16593 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
16594 (armv7): Add options fp and vfpv3-d16.
16595 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
16596 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
16598 (armv7ve): Likewise.
16599 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
16600 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
16601 (armv8-a): Add nocrypto option.
16602 (armv8.1-a, armv8.2-a): Likewise.
16603 (armv8-m.main): add options fp, fp.dp and nofp.
16605 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16607 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
16609 (armv8-a+crc): Delete.
16610 (armv8.1-a): Add options simd, crypto and nofp.
16611 (armv8.2-a): Add options fp16, simd, crypto and nofp.
16612 (armv8.2-a+fp16): Delete.
16613 (armv8-m.main): Add option dsp.
16614 (armv8-m.main+dsp): Delete.
16615 (cortex-a8): Add fpu. Add nofp option.
16616 (cortex-a9): Add fpu. Add nofp and nosimd options.
16617 * config/arm/parsecpu.awk (gen_data): Generate option tables and
16618 link to main cpu and architecture data structures.
16619 (gen_comm_data): Only put isa attributes from the main architecture
16621 (option): New statement for architecture and CPU entries.
16622 * arm.c (struct cpu_option): New structure.
16623 (struct processors): Add entry for options.
16624 (arm_unrecognized_feature): New function.
16625 (arm_parse_arch_cpu_name): Ignore any characters after the first
16627 (arm_parse_arch_cpu_feature): New function.
16628 (arm_configure_build_target): Separate out any CPU and architecture
16629 features and parse separately. Don't error out if -mfpu=auto is
16630 used with only an architecture string.
16631 (arm_print_asm_arch_directives): New function.
16632 (arm_file_start): Call it.
16633 * config/arm/arm-cpu-cdata.h: Regenerated.
16634 * config/arm/arm-cpu-data.h: Likewise.
16635 * config/arm/arm-tables.opt: Likewise.
16637 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16639 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
16640 assembler when it is not -mfpu=auto.
16642 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16644 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
16645 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
16646 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
16647 (ASM_CPU_SPEC): Rewrite.
16648 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
16649 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
16650 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
16651 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
16652 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
16653 copied string is NUL-terminated. Also strip any characters prefixed
16655 (arm_rewrite_selected_arch): New function.
16656 (arm_rewrite_march): New function.
16658 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16660 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
16661 (x_arm_cpu_string, x_arm_tune_string): Likewise.
16662 (march, mcpu, mtune): Convert to string-based options.
16663 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
16664 (arm_parse_arch_cpu_name): New function.
16665 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
16666 identify selected architecture or CPU.
16667 (arm_option_save): New function.
16668 (TARGET_OPTION_SAVE): Redefine.
16669 (arm_option_restore): Restore string options.
16670 (arm_option_print): Print string options.
16672 2017-06-16 Martin Sebor <msebor@redhat.com>
16674 PR tree-optimization/80933
16675 PR tree-optimization/80934
16676 * builtins.c (fold_builtin_3): Do not handle bcmp here.
16677 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
16678 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
16679 (gimple_fold_builtin): Call them.
16681 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16683 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
16684 as unlikely; update profile.
16686 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16688 * predict.c (force_edge_cold): Handle declaring edges impossible
16691 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16693 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
16695 (try_unroll_loop_completely): Fix reporting.
16697 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16699 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
16701 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
16704 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
16705 if given a non-constant argument for an intrinsic which requires a
16708 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16710 * profile.c (compare_freqs): New function.
16711 (branch_prob): Sort edge list.
16712 (find_spanning_tree): Assume that the list is priority sorted.
16714 2017-06-16 Richard Biener <rguenther@suse.de>
16716 PR tree-optimization/81090
16717 * passes.def (pass_record_bounds): Remove.
16718 * tree-pass.h (make_pass_record_bounds): Likewise.
16719 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
16720 make_pass_record_bounds): Likewise.
16721 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
16722 not free niter estimates at the beginning but at the end.
16723 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
16725 2017-06-16 Richard Biener <rguenther@suse.de>
16727 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
16728 initializer to workaround ICE in host GCC 4.8.
16730 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16732 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
16734 (clone_inlined_nodes): Update.
16736 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16738 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
16739 prefetch settings, and enable prefetching by default at -O3.
16741 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16743 * config/aarch64/aarch64.c (aarch64_override_options_internal):
16744 Set flag_prefetch_loop_arrays according to tuning data.
16746 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16748 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
16749 New tune structure.
16750 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
16751 [Unrelated to main purpose of the patch] Place the pointer field last
16752 to enable type checking errors when tune structure are wrongly merged.
16753 * config/aarch64/aarch64.c (generic_prefetch_tune,)
16754 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
16755 (thunderx2t99_prefetch_tune): New tune constants.
16756 (tune_params *_tunings): Update all tunings (no functional change).
16757 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
16758 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
16759 from tunings structures.
16761 2017-06-16 Jakub Jelinek <jakub@redhat.com>
16764 * ubsan.c (instrument_null): Add T argument, use it instead
16765 of computing it based on IS_LHS.
16766 (instrument_object_size): Likewise.
16767 (pass_ubsan::execute): Adjust instrument_null and
16768 instrument_object_size callers to pass gimple_get_lhs or
16769 gimple_assign_rhs1 result to it. Use instrument_null instead of
16770 calling get_base_address and instrument_mem_ref. Handle
16771 aggregate call arguments for object-size sanitization.
16773 2017-06-16 Yury Gribov <tetra2005@gmail.com>
16775 PR tree-optimization/81089
16776 * tree-vrp.c (is_masked_range_test): Validate operands of
16779 2017-06-15 Martin Sebor <msebor@redhat.com>
16782 * dumpfile.c (dump_register): Avoid calling memset to initialize
16783 a class with a default ctor.
16784 * gcc.c (struct compiler): Remove const qualification.
16785 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
16786 * hash-table.h: Ditto.
16787 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
16789 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
16790 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
16792 * params.h (struct param_info): Make struct members non-const.
16793 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
16794 with default initialization.
16795 * vec.h (vec_copy_construct, vec_default_construct): New helper
16797 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
16798 with vec_copy_construct.
16799 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
16800 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
16801 * doc/invoke.texi (-Wclass-memaccess): Document.
16803 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16805 * emit-rtl.h (is_leaf): Update comment about local
16806 register allocator.
16808 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16811 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
16812 for a variable to have a section before checking if the section has a
16814 Set section to.persistent if persistent attribute is set.
16815 Warn if .persistent attribute is used on an automatic variable.
16817 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
16819 PR rtl-optimization/80474
16820 * reorg.c (update_block): Do not ignore instructions in a delay slot.
16822 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
16824 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
16827 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
16829 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
16830 (casesi): Emit bounds checking as RTL.
16831 (casesi_internal_mips16_<mode>): Remove bounds checking.
16833 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
16835 * config/xtensa/xtensa.c (xtensa_option_override): Append
16836 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
16837 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
16838 xtensa_doloop_hooks): Define unconditionally.
16839 (xtensa_reorg_loops): Only call reorg_loops in the presence of
16841 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
16842 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
16843 for it in xtensa_option_override.
16844 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
16845 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
16847 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
16849 * doc/cppopts.texi: Document '-' special value to -MF.
16851 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
16853 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
16854 (cortex_a53_fconst): Likewise.
16855 (cortex_a53_fpmul): Likewise.
16856 (cortex_a53_f_load_64): Likewise.
16857 (cortex_a53_f_load_many): Likewise.
16858 (cortex_a53_advsimd_alu): Likewise.
16859 (cortex_a53_advsimd_alu_q): Likewise.
16860 (cortex_a53_advsimd_mul): Likewise.
16861 (cortex_a53_advsimd_mul_q): Likewise.
16862 (fpmac bypass): Add new bypass for fpmac-fpmac case.
16863 Add missing fmul, r2f_cvt and fconst cases.
16865 2017-06-14 Richard Biener <rguenther@suse.de>
16867 PR middle-end/81088
16868 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
16870 (fold_binary_loc): When associating do not treat pre-existing
16871 TREE_OVERFLOW on literal constants as a reason to allow
16872 TREE_OVERFLOW on associated literal constants.
16874 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16876 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
16877 (MASK_FEATURES): New macro.
16878 * config/sparc/sparc.c (sparc_option_override): Remove the special
16879 handling of -mfpu and generalize it to all MASK_FEATURES switches.
16881 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16883 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
16884 a division of 0 if non-call exceptions are enabled.
16886 2017-06-14 Andrew Pinski <apinski@cavium.com>
16887 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16890 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
16891 Improve vector initialization code gen for only variable case.
16893 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16895 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
16897 2017-06-14 Richard Biener <rguenther@suse.de>
16899 PR tree-optimization/81083
16900 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
16903 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16905 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
16906 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
16907 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
16908 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
16909 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
16910 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
16912 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16914 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
16915 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
16917 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16919 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
16921 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16923 * config/rs6000/t-rtems: Don't handle SPE.
16925 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16927 * config/rs6000/t-linux: Don't handle SPE.
16929 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16931 * config/rs6000/eabispe.h: Delete file.
16933 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16935 * config/rs6000/t-spe: Delete file.
16937 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16939 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
16940 (rs6000_legitimate_offset_address_p): Return false for anything in
16941 V2SImode or V2SFmode.
16943 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16945 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
16946 except V2SF and V2SI. Rearrange the vector modes, and add comments.
16947 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
16949 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
16950 (rs6000_legitimate_offset_address_p): Ditto.
16951 (rs6000_emit_move): Ditto.
16952 (rs6000_init_builtins): Remove V4HI_type_node.
16954 2017-06-13 Martin Liska <mliska@suse.cz>
16957 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
16958 (gate_asan): Likewise.
16959 * asan.h (asan_no_sanitize_address_p): Remove the function.
16960 (sanitize_flags_p): New function.
16961 * builtins.def: Fix coding style.
16962 * common.opt: Use renamed enum value.
16963 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
16964 * doc/extend.texi: Document no_sanitize attribute.
16965 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
16966 to SANITIZE_UNDEFINED_NONDEFAULT.
16967 * gcc.c (sanitize_spec_function): Use the renamed enum value.
16968 * gimple-fold.c (optimize_atomic_compare_exchange_p):
16969 Use sanitize_flags_p.
16970 * gimplify.c (gimplify_function_tree): Likewise.
16971 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
16972 * opts.c (parse_no_sanitize_attribute): New function.
16973 (common_handle_option): Use renamed enum value.
16974 * opts.h (parse_no_sanitize_attribute): Declare.
16975 * tree.c (sanitize_flags_p): New function.
16976 * tree.h: Declared here.
16977 * tsan.c: Use sanitize_flags_p.
16978 * ubsan.c (ubsan_expand_null_ifn): Likewise.
16979 (instrument_mem_ref): Likewise.
16980 (instrument_bool_enum_load): Likewise.
16981 (do_ubsan_in_current_function): Remove the function.
16982 (pass_ubsan::execute): Use sanitize_flags_p.
16983 * ubsan.h: Remove do_ubsan_in_current_function
16984 * tree-cfg.c (print_no_sanitize_attr_value): New function.
16985 (dump_function_to_file): Use it here.
16987 2017-06-13 Martin Jambor <mjambor@suse.cz>
16989 PR tree-optimization/80803
16990 PR tree-optimization/81063
16991 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
16992 (propagate_subaccesses_across_link): Enqueue subtree whenever
16993 necessary instead of relying on the caller.
16995 2017-06-13 Martin Jambor <mjambor@suse.cz>
16997 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
16998 that have a first_link.
16999 (sort_and_splice_var_accesses): Do not check first_link before
17001 (subtree_mark_written_and_enqueue): Likewise.
17002 (propagate_all_subaccesses): Likewise and do not stop at first
17003 parent with a first_link.
17005 2017-06-13 Martin Jambor <mjambor@suse.cz>
17007 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
17010 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17012 * match.pd: New pattern.
17014 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17016 * tree-vrp.c (is_masked_range_test): New function.
17017 (register_edge_assert_for): Determine ranges for
17020 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17022 PR tree-optimization/67328
17023 * fold-const.c (maskable_range_p): New function.
17024 (build_range_check): Generate bittests if possible.
17026 2017-06-13 Martin Liska <mliska@suse.cz>
17028 * gimple-pretty-print.c (dump_probability): Add new argument.
17029 (dump_edge_probability): Dump both probability and count.
17030 (dump_gimple_label): Likewise.
17031 (dump_gimple_bb_header): Likewise.
17033 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
17036 * config/avr/avr-devices.c: Fix indentation.
17037 * config/avr/gen-avr-mmcu-specs.c: Dito.
17039 2017-06-13 Richard Biener <rguenther@suse.de>
17041 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
17042 instead get vector type from stmt_info.
17043 (vectorizable_reduction): Adjust. Remove dead code.
17045 2017-06-13 Richard Biener <rguenther@suse.de>
17047 PR middle-end/81065
17048 * fold-const.c (extract_muldiv_1): Remove bogus distribution
17049 case of C * (x * C2 + C3).
17050 (fold_addr_of_array_ref_difference): Properly fold index difference.
17052 2017-06-12 David S. Miller <davem@davemloft.net>
17055 * config/sparc/sparc.md (return expander): Emit frame blockage if
17056 function uses alloca.
17058 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17060 * combine.c (make_field_assignment): Check len rather than the mode
17061 precision when calling force_to_mode.
17063 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
17065 Support multilibs and devices that see flash in RAM address range.
17068 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17069 (avr_mcu_t) <flash_pm_offset>: New field.
17070 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17071 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17072 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17073 (AVR_TINY_PM_OFFSET): Remove macro.
17074 * config/avr/avr.opt (-mshort-calls): New option.
17075 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17076 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17077 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17078 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17079 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17080 instead of avr_arch->have_jmp_call.
17081 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17082 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17083 avr_arch->flash_pm_offset to define.
17084 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17085 new field flash_pm_offset. Add entry for avrxmega3.
17086 (avr_texinfo): Add entry for avrxmega3.
17087 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17088 attiny212, attiny214,
17089 attiny412, attiny414, attiny416, attiny417,
17090 attiny814, attiny816, attiny817,
17091 attiny1614, attiny1616, attiny1617,
17092 attiny3214, attiny3216, attiny3217.
17093 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17094 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17095 (avr_print_operand_address) [AVR_TINY]: Same.
17096 (avr_asm_init_sections) <readonly_data_section>: Only patch
17097 callback if avr_arch->flash_pm_offset = 0.
17098 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17099 for rodata if avr_arch->flash_pm_offset != 0.
17100 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17101 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17102 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17103 (m_options): Append opt_rcall.
17104 (m_dirnames): Append dir_rcall.
17105 * config/avr/t-multilib: Regenerate.
17107 * configure.ac [target=avr]: Check whether avrxmega3 default
17108 linker description file works as needed.
17109 * configure: Regenerate.
17110 * doc/avr-mmcu.texi: Regenerate.
17111 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17112 <__AVR_ARCH__>: Document avrxmega3 and 103.
17113 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17114 <__AVR_SHORT_CALLS__>: Document it.
17115 <__AVR_PM_BASE_ADDRESS__>: Document it.
17116 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17117 (AVR Variable Attributes) <progmem>: Document this is
17118 not needed for avrxmega3.
17119 (AVR Named Address Spaces) <__flash>: Dito.
17121 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17123 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17125 2017-06-12 Doug Rupp <rupp@adacore.com>
17127 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17128 Append vxworks-stdint.h to the tm_file list.
17129 * config/vxworks-stdint.h: New file.
17131 2017-06-12 Martin Liska <mliska@suse.cz>
17133 PR tree-optimization/81041
17134 * tree-profile.c (gimple_gen_ic_func_profiler):
17135 Create an extra BB in profile-generate
17136 (gimple_gen_time_profiler): Likewise.
17138 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17140 PR tree-optimization/81003
17141 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17142 (update_range_test): Use it instead of force_gimple_operand_gsi.
17144 2017-06-12 Richard Biener <rguenther@suse.de>
17146 PR tree-optimization/81053
17147 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
17148 with backedge value not defined in loop. Simplify def stmt
17151 2017-06-11 Tom de Vries <tom@codesourcery.com>
17154 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
17156 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
17157 nvptx_cannot_force_const_mem.
17159 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17161 * opts.c (finish_options): Move test for flag_split_stack after
17162 it has been initialized.
17164 2017-06-11 Jason Merrill <jason@redhat.com>
17166 * tree.h (id_equal): New.
17167 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
17168 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
17169 instead of strcmp of IDENTIFIER_POINTER.
17171 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17173 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
17174 (mark_all_inlined_calls_cdtor): Fix formating.
17175 (inline_transform): Rescale profile before inlining.
17177 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17179 * cgraph.h (cgraph_edge::clone): Update prototype.
17180 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
17181 (cgraph_node::create_clone): Update.
17182 (cgraph_node::create_version_clone): Update.
17183 * tree-inline.c (copy_bb): Update.
17184 (expand_call_inline): Update.
17186 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17188 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
17189 factored out from ...
17190 (rs6000_emit_prologue): ... here.
17192 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17194 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
17195 factored out from ...
17196 (rs6000_emit_prologue): ... here.
17198 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17200 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
17202 (handle_missing_profiles): Fix computation of tp_first_run.
17203 (counts_to_freqs): Do not touch freqs when count is 0.
17205 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17207 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
17210 2017-06-10 Tom de Vries <tom@codesourcery.com>
17212 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17213 attributes): Document signal effective target.
17215 2017-06-10 Tom de Vries <tom@codesourcery.com>
17217 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17218 Document effective target stack_size.
17220 2017-06-09 David Malcolm <dmalcolm@redhat.com>
17222 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
17223 to the edit_context if they can be auto-applied.
17225 2017-06-9 Ian Lance Taylor <iant@golang.org>
17227 * opts.c (finish_options): If -fsplit-stack, disable implicit
17228 -forder-blocks-and-partition.
17229 * doc/invoke.texi (Optimize Options): Document that when using
17230 -fsplit-stack -forder-blocks-and-partition is not implicitly
17233 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17235 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
17236 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
17237 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
17238 * builtins.def (abort, trap, unreachable): Declare cold.
17239 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
17240 * tree-core.h (ECF_COLD): New.
17241 * tree.c (set_call_expr_flags): Handle ECF_COLD.
17242 (build_common_builtin_nodes): Mark unreachable and abort as cold.
17244 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17246 * predict.c (unlikely_executed_stmt_p): Cleanup.
17248 2017-06-09 Richard Biener <rguenther@suse.de>
17250 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
17251 model if the ref is always written to.
17253 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17255 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
17257 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17259 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
17262 2017-06-09 Tom de Vries <tom@codesourcery.com>
17265 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
17266 "target cannot support label values" when encountering LABEL_REF.
17268 2017-06-09 Martin Liska <mliska@suse.cz>
17270 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
17271 (gimple_gen_ic_func_profiler): Emit direct comparison
17272 of __gcov_indirect_call_callee with NULL.
17273 (gimple_gen_time_profiler): Change probability from
17274 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
17276 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17278 * profile.c (edge_gcov_counts): Turn to pointer.
17279 (compute_branch_probabilities, compute_branch_probabilities): Update.
17280 (branch_prob): Do not clear edge_gcov_count.
17281 * profile.h (edge_gcov_counts): Turn to pointer.
17282 (edge_gcov_count): Update.
17284 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17286 * gimple.h (gimple_check_failed): Mark cold.
17288 2017-06-09 Richard Biener <rguenther@suse.de>
17290 PR tree-optimization/66623
17291 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
17292 refactor check_reduction into two parts, properly computing
17293 whether we have to check reduction validity for outer loop
17296 2017-06-09 Richard Biener <rguenther@suse.de>
17298 PR tree-optimization/79483
17299 * graphite-scop-detection.c (order): New global.
17300 (get_order): Compute bb to order mapping that satisfies code
17301 generation constraints.
17302 (cmp_pbbs): New helper.
17303 (build_scops): Start domwalk at entry block, sort generated
17306 2017-06-09 Richard Biener <rguenther@suse.de>
17308 PR middle-end/81007
17309 * ipa-polymorphic-call.c
17310 (ipa_polymorphic_call_context::restrict_to_inner_class):
17311 Skip FIELD_DECLs with error_mark_node type.
17312 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
17315 2017-06-09 Martin Liska <mliska@suse.cz>
17317 * predict.c (struct branch_predictor): New struct.
17318 (test_prediction_value_range): New test.
17319 (predict_c_tests): New function.
17320 * selftest-run-tests.c (selftest::run_tests): Run the function.
17321 * selftest.h: Declare new tests.
17323 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
17326 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
17327 gen_add3_insn did not fail.
17328 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
17329 r0, construct that number in a temporary reg and add that reg to r0.
17330 If asked to put the result in r0 as well, fail.
17332 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
17334 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17335 for early expansion of vec_eqv.
17337 2017-06-08 Jakub Jelinek <jakub@redhat.com>
17339 PR middle-end/81005
17340 * ubsan.c (instrument_null): Avoid pointless code temporary.
17341 (pass_ubsan::execute): Instrument aggregate arguments of calls.
17343 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
17347 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
17350 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
17353 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17355 * predict.c (unlikely_executed_edge_p): Move ahead.
17356 (probably_never_executed_edge_p): Use it.
17358 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17360 PR middle-end/79988
17361 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
17362 gimple_call_builtin_p call.
17364 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17366 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
17367 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
17368 rtl_check_failed_type2, rtl_check_failed_code1,
17369 rtl_check_failed_code2, rtl_check_failed_code_mode,
17370 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
17371 rtvec_check_failed_bounds, rtl_check_failed_flag,
17372 _fatal_insn_not_found, _fatal_insn): Likewise.
17373 * tree.h (tree_contains_struct_check_failed,
17374 tree_check_failed, tree_not_check_failed,
17375 tree_class_check_failed, tree_range_check_failed,
17376 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
17377 tree_vec_elt_check_failed, phi_node_elt_check_failed,
17378 tree_operand_check_failed, omp_clause_check_failed,
17379 omp_clause_operand_check_failed, omp_clause_range_check_failed):
17382 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17384 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
17385 flag_branch_probabilities.
17386 * ipa-inline.c (edge_badness): Likewise.
17387 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17388 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
17389 * predict.c (maybe_hot_frequency_p): Likewise.
17390 (probably_never_executed): Likewise.
17391 * sched-ebb.c (schedule_ebbs): Likewise.
17392 * sched-rgn.c (find_single_block_region): Likewise.
17393 * tracer.c (tail_duplicate): Likewise.
17395 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17397 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
17398 longer requires x_flag_profile_use.
17400 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17402 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
17403 instead of flag_reorder_blocks_and_partition.
17404 * dbxout.c (dbxout_function_end): Likewise.
17405 * dwarf2out.c (gen_subprogram_die): Likewise.
17406 * haifa-sched.c (sched_create_recovery_edges): Likewise.
17407 * hw-doloop.c (reorg_loops): Likewise.
17408 * varasm.c (assemble_start_function,
17409 assemble_end_function): Likewise.
17410 (decide_function_section): Do not check for
17411 flag_reorder_blocks_and_partition.
17413 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17415 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17417 (chkp_get_hard_register_fake_addr_expr): Ditto.
17418 (chkp_build_addr_expr): Add check for hard reg case.
17419 (chkp_parse_array_and_component_ref): Ditto.
17420 (chkp_find_bounds_1): Ditto.
17421 (chkp_process_stmt): Don't generate bounds store for
17424 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17426 * predict.c (maybe_hot_bb_p): Do not check profile status.
17427 (maybe_hot_edge_p): Likewise.
17428 (probably_never_executed): Check for zero counts even if profile
17430 (unlikely_executed_edge_p): New function.
17431 (unlikely_executed_stmt_p): New function.
17432 (unlikely_executed_bb_p): New function.
17433 (set_even_probabilities): Use unlikely predicates.
17434 (combine_predictions_for_bb): Likewise.
17435 (predict_paths_for_bb): Likewise.
17436 (predict_paths_leading_to_edge): Likewise.
17437 (determine_unlikely_bbs): New function.
17438 (estimate_bb_frequencies): Use it.
17439 (compute_function_frequency): Use zero counts even if profile is
17441 * profile-count.h: Fix typo.
17443 2017-08-08 Julia Koval <julia.koval@intel.com>
17445 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
17446 _mm512_mask_cvtsepi16_storeu_epi8,
17447 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
17448 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
17449 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
17450 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
17451 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
17452 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
17453 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
17454 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
17455 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
17456 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
17457 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
17458 __builtin_ia32_pmovuswb256mem_mask,
17459 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
17460 __builtin_ia32_pmovwb512mem_mask): New builtins.
17462 2017-08-08 Julia Koval <julia.koval@intel.com>
17464 PR target/73350,80862
17465 * config/i386/subst.md (round): Fix round pattern.
17466 * config/i386/i386.c (ix86_erase_embedded_rounding):
17467 Fix erasing rounding for the fixed pattern.
17469 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17471 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
17473 2017-06-08 Martin Liska <mliska@suse.cz>
17475 PR gcov-profile/80911
17476 * gcov.c (block_info::block_info): New constructor.
17478 2017-06-07 Carl Love <cel@us.ibm.com>
17480 * config/rs6000/rs6000-c: The return type of the following
17481 built-in functions was implemented as int not long long. Fix sign
17482 of return value for the unsigned version of vec_mulo and vec_mule.
17483 vector unsigned long long vec_bperm (vector unsigned long long,
17484 vector unsigned char)
17485 vector signed long long vec_mule (vector signed int,
17487 vector unsigned long long vec_mule (vector unsigned int,
17488 vector unsigned int)
17489 vector signed long long vec_mulo (vector signed int,
17491 vector unsigned long long vec_mulo (vector unsigned int,
17492 vector unsigned int)
17493 * doc/extend.texi: Fix the documentation for the built-in
17496 2017-06-07 Carl Love <cel@us.ibm.com>
17499 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
17502 2017-06-07 Carl Love <cel@us.ibm.com>
17504 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
17505 support, Generate doublehv for signed int/float for BE case only.
17507 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
17509 * doc/invoke.texi (mcx16): Rewrite.
17511 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17513 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
17514 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
17515 *mov<mode>_softfloat, and an anonymous splitter): Use
17516 nonimmediate_operand instead of rs6000_nonimmediate_operand.
17518 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17520 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
17522 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17523 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
17524 (rs6000_debug_reg_global): Adjust.
17525 (rs6000_init_hard_regno_mode_ok): Adjust.
17526 (rs6000_dbx_register_number): Adjust.
17527 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
17528 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17529 Remove SPE_ACC and SPEFSCR.
17530 (REG_ALLOC_ORDER): Ditto.
17531 (FRAME_POINTER_REGNUM): Change to 111.
17532 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
17533 (REG_CLASS_NAMES): Ditto.
17534 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
17535 (REGISTER_NAMES): Ditto.
17536 (ADDITIONAL_REG_NAMES): Ditto.
17537 (rs6000_reg_names): Ditto.
17538 * config/rs6000/rs6000.md: Renumber some register number
17541 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17543 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
17545 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17546 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
17548 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
17549 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17550 Delete the SPE high registers.
17551 (REG_ALLOC_ORDER): Ditto.
17552 (enum reg_class): Remove SPE_HIGH_REGS.
17553 (REG_CLASS_NAMES): Ditto.
17554 (REG_CLASS_CONTENTS): Delete the SPE high registers.
17555 (REGISTER_NAMES): Ditto.
17556 (rs6000_reg_names): Ditto.
17557 * doc/tm.texi.in: Remove SPE as example.
17558 * doc/tm.texi: Regenerate.
17560 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17562 * config/rs6000/8540.md (ppc8540_brinc): Delete.
17563 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
17564 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
17565 * config/rs6000/rs6000.md (type): Remove "brinc".
17567 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17569 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
17570 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
17571 * config/rs6000/linuxspe.h: Delete file.
17572 * config/rs6000/rs6000.md: Don't include spe.md.
17573 * config/rs6000/spe.h: Delete file.
17574 * config/rs6000/spe.md: Delete file.
17575 * config/rs6000/t-rs6000: Remove spe.md.
17577 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17579 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
17580 (reg_or_none500mem_operand): Delete.
17581 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
17582 instead of reg_or_none500mem_operand.
17584 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17586 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17587 handling of SPE flags.
17588 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
17590 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17592 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17594 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
17595 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
17596 paired_divv2sf3): Similar.
17597 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
17598 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
17599 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
17601 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
17602 Rename the paired_* instruction patterns.
17603 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
17605 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
17606 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
17607 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
17608 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
17609 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
17610 PAIRED_VECTOR_MODE.
17611 (struct machine_function): Delete field spe_insn_chain_scanned_p.
17612 (spe_func_has_64bit_regs_p): Delete.
17613 (spe_expand_predicate_builtin): Delete.
17614 (spe_expand_evsel_builtin): Delete.
17615 (TARGET_DWARF_REGISTER_SPAN): Do not define.
17616 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
17617 (invalid_e500_subreg): Delete.
17618 (rs6000_legitimize_address): Always force_reg op2 as well, for
17619 paired single memory accesses.
17620 (rs6000_member_type_forces_blk): Delete.
17621 (rs6000_spe_function_arg): Delete.
17622 (rs6000_expand_unop_builtin): Delete SPE handling.
17623 (rs6000_expand_binop_builtin): Ditto.
17624 (spe_expand_stv_builtin): Delete.
17625 (bdesc_2arg_spe): Delete.
17626 (spe_expand_builtin): Delete.
17627 (spe_expand_predicate_builtin): Delete.
17628 (spe_expand_evsel_builtin): Delete.
17629 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
17630 (spe_init_builtins): Delete.
17631 (spe_func_has_64bit_regs_p): Delete.
17632 (savres_routine_name): Delete "info" parameter. Adjust callers.
17633 (rs6000_emit_stack_reset): Ditto.
17634 (rs6000_dwarf_register_span): Delete.
17635 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
17636 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
17637 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
17639 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
17641 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
17642 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
17643 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
17644 mulv2sf3, divv2sf3): Delete expanders.
17646 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17648 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
17650 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17652 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
17653 * config/rs6000/rs6000.c: Ditto.
17655 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17657 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
17658 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
17659 comparison_operator.
17661 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17663 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
17664 * config/rs6000/rs6000.opt: Ditto.
17665 * config/rs6000/t-rtems: Ditto.
17667 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17669 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
17670 TARGET_E500_SINGLE by 0, simplify.
17671 * config/rs6000/rs6000.c: Ditto.
17672 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
17673 (spe_build_register_parallel): Delete.
17674 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
17675 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
17676 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
17677 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
17678 (E500_CONVERT): Delete.
17679 * config/rs6000/spe.md: Remove many patterns and all define_constants.
17681 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17683 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
17684 * config/rs6000/dfp.md: Ditto.
17685 (negdd2, *negdd2_fpr): Merge.
17686 (absdd2, *absdd2_fpr): Merge.
17687 (negtd2, *negtd2_fpr): Merge.
17688 (abstd2, *abstd2_fpr): Merge.
17689 * config/rs6000/e500.h: Delete file.
17690 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
17691 TARGET_FPRS by 1 and simplify.
17692 * config/rs6000/rs6000-c.c: Ditto.
17693 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
17694 TARGET_DF_SPE by 0.
17695 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
17697 * config/rs6000/rs6000.md: Ditto.
17698 (floatdidf2, *floatdidf2_fpr): Merge.
17699 (move_from_CR_gt_bit): Delete.
17700 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
17701 (E500_CR_IOR_COMPARE): Delete.
17702 (All patterns that require !TARGET_FPRS): Delete.
17703 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
17705 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17707 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
17709 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17711 * graphds.c (add_edge): Intitialize edge's attached data.
17712 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
17713 pointer parameter. Call pointed function on each edge during
17714 graph traversing. Skip traversing the edge when the function
17716 (graphds_dfs, graphds_scc): Ditto.
17717 (for_each_edge): New parameter. Pass the new parameter to callback
17719 * graphds.h (skip_edge_callback): New function pointer type.
17720 (graphds_dfs, graphds_scc): New function pointer parameter.
17721 (graphds_edge_callback, for_each_edge): New parameter.
17723 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17725 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
17726 out code checking if runtime alias check is possible to below ...
17727 Call the new function.
17728 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
17729 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
17731 2017-06-07 Marek Polacek <polacek@redhat.com>
17734 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
17735 TYPE_OVERFLOW_WRAPS check.
17737 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17739 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
17740 if versioning is required.
17741 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
17742 peeling with the check for versioning.
17744 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17746 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
17747 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
17748 Set true to new parameter if new ssa variable is defined.
17749 (vect_gen_vector_loop_niters): Refactor. Set range information
17750 for the new vector loop bound variable.
17751 (vect_do_peeling): Ditto.
17753 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17755 * tree-affine.c (ssa.h): Include header file.
17756 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
17757 has wrapping overflow behavior.
17759 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17761 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
17763 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17765 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
17766 (tree_to_aff_combination): ... here.
17768 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17770 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
17771 reg_pressure model function.
17772 (ivopts_global_cost_for_size): Delete.
17773 (determine_set_costs, iv_ca_recount_cost): Call new model function
17774 ivopts_estimate_reg_pressure.
17776 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17778 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
17779 expensive than udiv. Remove floating point cases from mod.
17781 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17783 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
17784 Increase idiv cost.
17786 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17788 * config/aarch64/aarch64.md
17789 (copysignsf3): Fix mask generation.
17791 2017-06-07 Jakub Jelinek <jakub@redhat.com>
17793 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
17795 (class dump_manager): Add register_dumps method.
17796 * dumpfile.c: Include langhooks.h.
17797 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
17798 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
17799 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
17800 (dump_manager::dump_register): Start with 512 entries instead of 32.
17801 (dump_manager::register_dumps): New method.
17802 * toplev.c (general_init): Instead of invoking register_dumps
17803 langhook, invoke register_dumps method on the dump manager.
17804 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
17807 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
17809 * doc/md.texi: Clarify the restrictions on a define_insn condition.
17810 Say that # requires an associated define_split to exist, and that
17811 the define_split must be suitable for use after register allocation.
17813 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17815 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
17816 (compute_outgoing_frequencies): Also initialize zero counts.
17817 (find_many_sub_basic_blocks): Do not produce uninitialized profile
17818 around loops; preserve more of profile when nothing changes.
17820 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
17822 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
17824 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
17825 * config/arm/arm-cpu-cdata.h: Regenerate.
17826 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
17827 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
17828 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
17829 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
17830 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
17832 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
17834 * config/arm/t-rmprofile: Likewise.
17835 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
17837 2017-06-06 David S. Miller <davem@davemloft.net>
17840 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
17841 blockage if function uses alloca.
17843 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17845 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
17846 New "uid" fields to hold pretty-print IDs of group and ref.
17847 Memory references are now identified as <group_id>:<ref_id>
17848 instead of using [random] addresses.
17849 (dump_mem_details): Simplify, no functional change.
17850 (dump_mem_ref): Simplify and make output more concise.
17851 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
17852 (find_or_create_group): Initialize group uid.
17853 (record_ref): Initialize ref uid. Improve debug output.
17854 (prune_group_by_reuse, should_issue_prefetch_p,)
17855 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
17856 (mark_nontemporal_store, determine_loop_nest_reuse):
17857 Improve debug output.
17859 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17861 * dbgcnt.def (prefetch): New debug counter.
17862 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
17863 (schedule_prefetches): Stop issueing prefetches if debug counter
17866 2017-06-06 Tom de Vries <tom@codesourcery.com>
17868 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
17869 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
17871 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17873 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
17874 Use aarch64_reg_or_zero predicate for operand 4.
17875 (aarch64_compare_and_swap<mode> define_insn_and_split):
17876 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
17877 (aarch64_store_exclusive<mode>): Likewise for operand 2.
17879 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17881 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
17882 (arm_compute_save_core_reg_mask): This.
17883 (thumb1_compute_save_reg_mask): Rename into ...
17884 (thumb1_compute_save_core_reg_mask): This.
17885 (arm_compute_save_reg0_reg12_mask): Adapt comment.
17886 (arm_compute_frame_layout): Likewise.
17888 2017-06-06 Richard Biener <rguenther@suse.de>
17890 PR tree-optimization/80974
17891 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
17892 keep or clear leaders SSA info.
17894 2017-06-06 Tom de Vries <tom@codesourcery.com>
17896 * config/nvptx/nvptx.c (split_mode_p): New function.
17897 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
17899 2017-06-06 Tom de Vries <tom@codesourcery.com>
17901 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
17903 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17906 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
17909 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17911 * shrink-wrap.c (handle_simple_exit): Update profile.
17912 (try_shrink_wrapping): Upate profile.
17914 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17916 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
17917 (tree_guess_outgoing_edge_probabilities): New.
17918 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
17919 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
17921 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17923 * ipa-split.c (split_function): Initialize return bb profile.
17925 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17927 * profile.c (compute_branch_probabilities): Also initialize
17928 EXIT_BLOCK profile.
17930 2017-06-06 Richard Biener <rguenther@suse.de>
17932 PR tree-optimization/80928
17933 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
17934 (vect_analyze_loop_operations): Properly guard analysis for
17936 (vect_transform_loop): Likewise.
17937 (vect_analyze_loop_2): Also reset SLP type on PHIs.
17938 (vect_model_induction_cost): Do not cost for pure SLP.
17939 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
17940 of induction in inner loop vectorization.
17941 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
17942 (vect_get_and_check_slp_defs): Handle vect_induction_def.
17943 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
17945 (vect_analyze_slp_cost_1): Cost induction.
17946 (vect_detect_hybrid_slp_stmts): Handle PHIs.
17947 (vect_get_slp_vect_defs): Likewise.
17948 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
17949 (vect_transform_stmt): Handle SLP reductions.
17950 * tree-vectorizer.h (vectorizable_induction): Adjust.
17952 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17954 * config/rs6000/rs6000.c (make_resolver_func): Update
17955 init_lowered_empty_function call.
17957 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
17959 * doc/invoke.texi: Document the -fprofile-abs-path option.
17960 * common.opt (fprofile-abs-path): New option.
17961 * gcov-io.h (gcov_write_filename): Declare.
17962 * gcov-io.c (gcov_write_filename): New function.
17963 * coverage.c (coverage_begin_function): Use gcov_write_filename.
17964 * profile.c (output_location): Likewise.
17966 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17968 * shring-wrap.c: Revert accidental commit.
17970 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
17972 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
17974 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17976 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
17978 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
17979 profile in callgraph edge.
17980 * profile-count.h (apply_probability): If THIS is 0, then result is 0
17981 (apply_scale): Likewise.
17982 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
17983 Also scale profile when inlining function with zero profile.
17984 (initialize_cfun): Update exit block profile even when it is zero.
17985 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
17986 when profile is read.
17988 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17990 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
17991 (CLONE_*): New constants to define the processors we can generate
17992 code for with the target_clone attribute.
17993 (rs6000_clone_map): New array to identify which clone processors
17994 the current program is running on.
17995 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
17996 target_clone attribute.
17997 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
17998 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
17999 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
18000 (cpu_expand_builtin): Add support for target_clone attribute.
18001 (rs6000_valid_attribute_p): Allow "default" attribute.
18002 (get_decl_name): New debug function to simplify printing the
18003 current function name in debugging statements.
18004 (rs6000_clone_priority): New functions to support the target_clone
18005 attribute, and be able to generate code to switch between ISA 2.05
18006 through ISA 3.0 (power6 through power9).
18007 (rs6000_compare_version_priority): Likewise.
18008 (rs6000_get_function_versions_dispatcher): Likewise.
18009 (make_resolver_func): Likewise.
18010 (add_condition_to_bb): Likewise.
18011 (dispatch_function_versions): Likewise.
18012 (rs6000_generate_version_dispatcher_body): Likewise.
18013 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
18014 (fusion_gpr_load_p): Fix a spacing issue.
18015 * doc/extend.texi (Common Function Attributes): Document that the
18016 PowerPC supports the target_clone attribute.
18018 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18020 * config/arm/arm.h: explain F symbol found in description of ARM
18021 register allocation in its legend.
18023 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18025 * config/mips/frame-header-opt.c: Include profile-count.h.
18026 * config/riscv/riscv.c: Include profile-count.h
18028 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18030 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
18032 (sm_set_flag_if_changed): Add bbs field.
18033 (execute_sm_if_changed_flag_set): Pass BBS.
18034 (execute_sm): Update.
18036 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18038 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
18041 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18043 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
18044 (peephole2): New peephole2 to emit the above.
18045 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
18047 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18049 * config/aarch64/aarch64.c (define_peephole2 above
18050 *sub_<shift>_<mode>): New peephole.
18052 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18054 * config/i386/i386.c (make_resolver_func): Update.
18055 * Makefile.in: Add profile-count.h and profile-count.o
18056 * auto-profile.c (afdo_indirect_call): Update to new API.
18057 (afdo_set_bb_count): Update.
18058 (afdo_propagate_edge): Update.
18059 (afdo_propagate_circuit): Update.
18060 (afdo_calculate_branch_prob): Update.
18061 (afdo_annotate_cfg): Update.
18062 * basic-block.h: Include profile-count.h
18063 (struct edge_def): Turn count to profile_count.
18064 (struct basic_block_def): Likewie.
18065 (REG_BR_PROB_BASE): Move to profile-count.h
18066 (RDIV): Move to profile-count.h
18067 * bb-reorder.c (max_entry_count): Turn to profile_count.
18068 (find_traces): Update.
18069 (rotate_loop):Update.
18070 (connect_traces):Update.
18071 (sanitize_hot_paths):Update.
18072 * bt-load.c (migrate_btr_defs): Update.
18073 * cfg.c (RDIV): Remove.
18074 (init_flow): Use alloc_block.
18075 (alloc_block): Uninitialize count.
18076 (unchecked_make_edge): Uninitialize count.
18077 (check_bb_profile): Update.
18078 (dump_edge_info): Update.
18079 (dump_bb_info): Update.
18080 (update_bb_profile_for_threading): Update.
18081 (scale_bbs_frequencies_int): Update.
18082 (scale_bbs_frequencies_gcov_type): Update.
18083 (scale_bbs_frequencies_profile_count): New.
18084 * cfg.h (update_bb_profile_for_threading): Update.
18085 (scale_bbs_frequencies_profile_count): Declare.
18086 * cfgbuild.c (compute_outgoing_frequencies): Update.
18087 (find_many_sub_basic_blocks): Update.
18088 * cfgcleanup.c (try_forward_edges): Update.
18089 (try_crossjump_to_edge): Update.
18090 * cfgexpand.c (expand_gimple_tailcall): Update.
18091 (construct_exit_block): Update.
18092 * cfghooks.c (verify_flow_info): Update.
18093 (dump_bb_for_graph): Update.
18094 (split_edge): Update.
18095 (make_forwarder_block): Update.
18096 (duplicate_block): Update.
18097 (account_profile_record): Update.
18098 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18099 (get_estimated_loop_iterations): Update.
18100 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18101 (single_likely_exit): Update.
18102 * cfgloopmanip.c (scale_loop_profile): Update.
18104 (set_zero_probability): Update.
18105 (lv_adjust_loop_entry_edge): Update.
18106 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18107 (purge_dead_edges): Update.
18108 (rtl_account_profile_record): Update.
18109 * cgraph.c (cgraph_node::create): Uninitialize count.
18110 (symbol_table::create_edge): Uninitialize count.
18111 (cgraph_update_edges_for_call_stmt_node): Update.
18112 (cgraph_edge::dump_edge_flags): Update.
18113 (cgraph_node::dump): Update.
18114 (cgraph_edge::maybe_hot_p): Update.
18115 * cgraph.h: Include profile-count.h
18116 (create_clone), create_edge, create_indirect_edge): Update.
18117 (cgraph_node): Turn count to profile_count.
18118 (cgraph_edge0: Likewise.
18119 (make_speculative, clone): Update.
18120 (create_edge): Update.
18121 (init_lowered_empty_function): Update.
18122 * cgraphclones.c (cgraph_edge::clone): Update.
18123 (duplicate_thunk_for_node): Update.
18124 (cgraph_node::create_clone): Update.
18125 * cgraphunit.c (cgraph_node::analyze): Update.
18126 (cgraph_node::expand_thunk): Update.
18127 * final.c (dump_basic_block_info): Update.
18128 * gimple-streamer-in.c (input_bb): Update.
18129 * gimple-streamer-out.c (output_bb): Update.
18130 * graphite.c (print_global_statistics): Update.
18131 (print_graphite_scop_statistics): Update.
18132 * hsa-brig.c: Include basic-block.h.
18133 * hsa-dump.c: Include basic-block.h.
18134 * hsa-gen.c (T sum_slice): Update.
18135 (convert_switch_statements):Update.
18136 * hsa-regalloc.c: Include basic-block.h.
18137 * ipa-chkp.c (chkp_produce_thunks): Update.
18138 * ipa-cp.c (struct caller_statistics): Update.
18139 (init_caller_stats): Update.
18140 (gather_caller_stats): Update.
18141 (ipcp_cloning_candidate_p): Update.
18142 (good_cloning_opportunity_p): Update.
18143 (get_info_about_necessary_edges): Update.
18144 (dump_profile_updates): Update.
18145 (update_profiling_info): Update.
18146 (update_specialized_profile): Update.
18147 (perhaps_add_new_callers): Update.
18148 (decide_about_value): Update.
18149 (ipa_cp_c_finalize): Update.
18150 * ipa-devirt.c (struct odr_type_warn_count): Update.
18151 (struct decl_warn_count): Update.
18152 (struct final_warning_record): Update.
18153 (possible_polymorphic_call_targets): Update.
18154 (ipa_devirt): Update.
18155 * ipa-fnsummary.c (redirect_to_unreachable): Update.
18156 * ipa-icf.c (sem_function::merge): Update.
18157 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
18158 * ipa-inline.c (compute_uninlined_call_time): Update.
18159 (compute_inlined_call_time): Update.
18160 (want_inline_small_function_p): Update.
18161 (want_inline_self_recursive_call_p): Update.
18162 (edge_badness): Update.
18163 (lookup_recursive_calls): Update.
18164 (recursive_inlining): Update.
18165 (inline_small_functions): Update.
18166 (dump_overall_stats): Update.
18167 (dump_inline_stats): Update.
18168 * ipa-profile.c (ipa_profile_generate_summary): Update.
18169 (ipa_propagate_frequency): Update.
18170 (ipa_profile): Update.
18171 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18172 * ipa-utils.c (ipa_merge_profiles): Update.
18173 * loop-doloop.c (doloop_modify): Update.
18174 * loop-unroll.c (report_unroll): Update.
18175 (unroll_loop_runtime_iterations): Update.
18176 * lto-cgraph.c (lto_output_edge): Update.
18177 (lto_output_node): Update.
18178 (input_node): Update.
18179 (input_edge): Update.
18180 (merge_profile_summaries): Update.
18181 * lto-streamer-in.c (input_cfg): Update.
18182 * lto-streamer-out.c (output_cfg): Update.
18183 * mcf.c (create_fixup_graph): Update.
18184 (adjust_cfg_counts): Update.
18185 (sum_edge_counts): Update.
18186 * modulo-sched.c (sms_schedule): Update.
18187 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
18188 * predict.c (maybe_hot_count_p): Update.
18189 (probably_never_executed): Update.
18190 (dump_prediction): Update.
18191 (combine_predictions_for_bb): Update.
18192 (propagate_freq): Update.
18193 (handle_missing_profiles): Update.
18194 (counts_to_freqs): Update.
18195 (rebuild_frequencies): Update.
18196 (force_edge_cold): Update.
18197 * predict.h: Include profile-count.h
18198 (maybe_hot_count_p, counts_to_freqs): UPdate.
18199 * print-rtl-function.c: Do not include cfg.h
18200 * print-rtl.c: Include basic-block.h
18201 * profile-count.c: New file.
18202 * profile-count.h: New file.
18203 * profile.c (is_edge_inconsistent): Update.
18204 (correct_negative_edge_counts): Update.
18205 (is_inconsistent): Update.
18206 (set_bb_counts): Update.
18207 (read_profile_edge_counts): Update.
18208 (compute_frequency_overlap): Update.
18209 (compute_branch_probabilities): Update; Initialize and deinitialize
18211 (branch_prob): Update.
18212 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
18213 (edge_gcov_count): New.
18214 (bb_gcov_count): New.
18215 * shrink-wrap.c (try_shrink_wrapping): Update.
18216 * tracer.c (better_p): Update.
18217 * trans-mem.c (expand_transaction): Update.
18218 (ipa_tm_insert_irr_call): Update.
18219 (ipa_tm_insert_gettmclone_call): Update.
18220 * tree-call-cdce.c: Update.
18221 * tree-cfg.c (gimple_duplicate_sese_region): Update.
18222 (gimple_duplicate_sese_tail): Update.
18223 (gimple_account_profile_record): Update.
18224 (execute_fixup_cfg): Update.
18225 * tree-inline.c (copy_bb): Update.
18226 (copy_edges_for_bb): Update.
18227 (initialize_cfun): Update.
18228 (freqs_to_counts): Update.
18229 (copy_cfg_body): Update.
18230 (expand_call_inline): Update.
18231 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18232 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
18233 (try_unroll_loop_completely): Update.
18234 (try_peel_loop): Update.
18235 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18236 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
18237 * tree-ssa-loop-split.c (connect_loops): Update.
18238 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
18239 * tree-ssa-reassoc.c (branch_fixup): Update.
18240 * tree-ssa-tail-merge.c (replace_block_by): Update.
18241 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
18242 (compute_path_counts): Update.
18243 (update_profile): Update.
18244 (recompute_probabilities): Update.
18245 (update_joiner_offpath_counts): Update.
18246 (estimated_freqs_path): Update.
18247 (freqs_to_counts_path): Update.
18248 (clear_counts_path): Update.
18249 (ssa_fix_duplicate_block_edges): Update.
18250 (duplicate_thread_path): Update.
18251 * tree-switch-conversion.c (case_bit_test_cmp): Update.
18252 (struct switch_conv_info): Update.
18253 * tree-tailcall.c (decrease_profile): Update.
18254 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18255 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18256 * value-prof.c (check_counter): Update.
18257 (gimple_divmod_fixed_value): Update.
18258 (gimple_mod_pow2): Update.
18259 (gimple_mod_subtract): Update.
18260 (gimple_ic_transform): Update.
18261 (gimple_stringop_fixed_value): Update.
18262 * value-prof.h (gimple_ic): Update.
18264 2017-06-02 Carl Love <cel@us.ibm.com>
18266 * config/rs6000/rs6000-c: Add support for built-in functions
18267 vector double vec_doublee (vector signed int);
18268 vector double vec_doublee (vector unsigned int);
18269 vector double vec_doublee (vector float);
18270 vector double vec_doubleh (vector signed int);
18271 vector double vec_doubleh (vector unsigned int);
18272 vector double vec_doubleh (vector float);
18273 vector double vec_doublel (vector signed int);
18274 vector double vec_doublel (vector unsigned int);
18275 vector double vec_doublel (vector float);
18276 vector double vec_doubleo (vector signed int);
18277 vector double vec_doubleo (vector unsigned int);
18278 vector double vec_doubleo (vector float);.
18279 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
18280 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
18282 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
18283 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
18284 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
18286 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
18287 vec_doublel, vec_doubleh.
18288 * doc/extend.texi: Update the built-in documentation file for the
18289 new built-in functions.
18291 2017-06-02 David Malcolm <dmalcolm@redhat.com>
18294 * ipa-inline-analysis.c (free_growth_caches): Set
18295 edge_removal_hook_holder to NULL after removing it.
18297 2017-06-02 Sudakshina Das <sudi.das@arm.com>
18299 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
18300 comparision with zero.
18302 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18303 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18304 for early expansion of vec_min and vec_max builtins.
18305 (builtin_function_type): Add min/max unsigned variants to those
18306 identified as having unsigned arguments.
18308 2017-06-02 Olivier Hainque <hainque@adacore.com>
18310 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
18312 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18314 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
18315 Use VALL_F16 iterator rather than VALL.
18317 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18319 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
18320 Emit CBNZ inside loop when doing a strong exchange and comparing
18321 against zero. Generate the CC flags after the loop.
18323 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
18325 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
18326 (dl_section_ref): New.
18327 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
18328 On AIX, append an expression to subtract the size of the
18329 section length to dl_section_ref.
18331 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18333 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18334 for early expansion of vector absolute builtins.
18336 2017-06-02 Richard Biener <rguenther@suse.de>
18338 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
18339 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
18341 2017-06-02 Richard Biener <rguenther@suse.de>
18343 PR tree-optimization/80948
18344 * tree-tailcall.c (find_tail_calls): Track stmts to move in
18345 stmt order as well.
18347 2017-06-02 Richard Biener <rguenther@suse.de>
18349 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
18351 * tree-vect-stmts.c (process_use): Do not mark backedge defs
18352 for inductions as relevant.
18354 2017-06-02 Richard Biener <rguenther@suse.de>
18356 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
18357 (vectorizable_induction): ... this. Remove dead code.
18359 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
18361 * builtins. (expand_builtin_alloca): Remove second parameter and
18362 infer its value from the first parameter instead.
18363 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
18365 2017-06-02 Jakub Jelinek <jakub@redhat.com>
18367 PR rtl-optimization/80903
18368 * loop-doloop.c (add_test): Unshare sequence.
18370 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
18372 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
18374 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
18376 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
18378 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
18379 xlogue_layout::get_instance, logue_layout::xlogue_layout,
18380 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
18381 (xlogue_layout::get_stub_rtx): Make static.
18382 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
18383 (xlogue_layout::compute_stub_managed_regs): Rename to...
18384 (xlogue_layout::count_stub_managed_regs): ...this.
18385 (xlogue_layout::is_stub_managed_reg): New function.
18386 (xlogue_layout::m_stub_names): Rename to...
18387 (xlogue_layout::s_stub_names): ...this, make static.
18388 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
18389 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
18390 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
18391 xlogue_layout::s_stub_names): Instantiate statics.
18392 (stub_managed_regs): Remove.
18393 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
18394 (disable_call_ms2sysv_xlogues): Rename to...
18395 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
18396 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
18398 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
18399 change after reload_completed.
18400 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
18402 (ix86_expand_prologue): Likewise.
18403 (ix86_expand_epilogue): Likewise.
18404 (ix86_expand_split_stack_prologue): Likewise.
18405 (ix86_compute_frame_layout): Remove frame parameter ...
18406 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
18407 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
18409 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
18410 (ix86_frame): Move from here ...
18411 * config/i386/i386.h (ix86_frame): ... to here.
18412 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
18413 complete ix86_frame data structure instead. Remove some_ld_name.
18415 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18417 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
18418 symbols that hold a DECL_VALUE_EXPR.
18420 2017-06-01 Martin Jambor <mjambor@suse.cz>
18422 PR tree-optimization/80898
18423 * tree-sra.c (process_subtree_disqualification): Removed.
18424 (disqualify_candidate): Do not acll
18425 process_subtree_disqualification.
18426 (subtree_mark_written_and_enqueue): New function.
18427 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
18428 RHS has been disqualified and re-queue LHS if necessary. Apart
18429 from that, ignore disqualified RHS.
18431 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18433 * config/s390/s390.c (s390_emit_epilogue): Disable early return
18434 address fetch for z10 or later.
18436 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18438 * config/arc/arc.md (tst_movb): Add guard when splitting.
18440 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18442 * config/arc/arc.c (arc_can_eliminate): Test against
18443 arc_frame_pointer_needed.
18445 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18447 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
18448 to prevent store reordering.
18449 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
18450 (type): Add block type.
18451 (stack_tie): Define special instruction to be used in
18454 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18456 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
18457 constraint. It is not valid for the pattern.
18458 (noncommutative_binary_comparison): Likewise.
18460 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18462 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
18465 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18467 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
18468 be used by the reg-alloc.
18470 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18472 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
18473 reg-alloc when having mul64 or mul32x16 instructions.
18474 (mulsidi3): Likewise.
18475 (umulsidi3): Likewise.
18476 (mulsi32x16): New pattern.
18477 (mulsi64): Likewise.
18478 (mulsidi64): Likewise.
18479 (umulsidi64): Likewise.
18480 (MUL32x16_REG): Define.
18481 (mul64_600): Use MUL32x16_REG.
18482 (mac64_600): Likewise.
18483 (umul64_600): Likewise.
18484 (umac64_600): Likewise.
18486 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18488 * config/arc/arc.md (mulsi3_700): Make it commutative.
18490 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
18492 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
18494 (*sign_extendsidi2_insn): Likewise for movstosw.
18496 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18498 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
18499 the type of the input discriminant value. Convert the
18500 discriminant value of signedness vary.
18502 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
18504 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
18505 Add to -Wall section.
18507 2017-06-01 Richard Biener <rguenther@suse.de>
18509 PR middle-end/66313
18510 * fold-const.c (fold_plusminus_mult_expr): If the factored
18511 factor may be zero use a wrapping type for the inner operation.
18512 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
18513 and handle moved defs.
18514 (process_assignment): Properly guard the unary op case. Return a
18515 tri-state indicating that moving the stmt before the call may allow
18516 to continue. Pass through to_move.
18517 (find_tail_calls): Handle moving unrelated defs before
18520 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
18523 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
18524 splitter result in the canonical way.
18526 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
18528 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
18529 also for 32bit target. Update insn attributes.
18530 (zero-extendsidi2 splitter): Allow all registers for operand 1.
18532 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
18534 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
18535 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
18536 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
18537 (_mm_maskz_min_ss): New intrinsics.
18539 2017-05-31 Martin Liska <mliska@suse.cz>
18541 * tree-vect-loop.c (vect_create_epilog_for_reduction):
18542 Change comment style to one we normally use.
18543 (vectorizable_reduction): Likewise.
18544 (vectorizable_induction): Likewise.
18545 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18546 (vectorizable_call): Likewise.
18547 (vectorizable_simd_clone_call): Likewise.
18548 (vectorizable_conversion): Likewise.
18549 (vectorizable_assignment): Likewise.
18550 (vectorizable_shift): Likewise.
18551 (vectorizable_operation): Likewise.
18552 (vectorizable_store): Likewise.
18553 (vectorizable_load): Likewise.
18554 * tree-vectorizer.h: Likewise.
18556 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
18558 * passes.c (emergency_dump_function): New.
18559 * tree-pass.h (emergency_dump_function): Declare.
18560 * plugin.c (plugins_internal_error_function): Remove.
18561 * plugin.h (plugins_internal_error_function): Remove declaration.
18562 * toplev.c (internal_error_function): New static function. Use it...
18563 (general_init): ...here.
18565 2017-05-31 Graham Markall <graham.markall@embecosm.com>
18567 * config/arc/arc.c (arc_print_operand): Handle constant operands.
18568 (arc_rtx_costs): Add costs for new patterns.
18569 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
18570 * config/arc/predicates.md: Add _1_2_3_operand predicate.
18572 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18574 * tree-ssa-strlen.c (get_next_strinfo): New function.
18575 (get_stridx_plus_constant): Use it.
18576 (zero_length_string): Likewise.
18577 (adjust_related_strinfos): Likewise.
18578 (adjust_last_stmt): Likewise.
18580 2017-05-31 Richard Biener <rguenther@suse.de>
18583 * config/i386/i386.c (ix86_expand_builtin): Remove assert
18584 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
18586 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18588 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
18589 loop_vinfo argument and use of dependence distance vectors.
18590 Check instead whether the two references differ only in their
18591 initial value and assume that they have the same alignment if the
18592 difference is a multiple of the vector alignment.
18593 (vect_analyze_data_refs_alignment): Update call accordingly.
18595 2017-05-31 Martin Liska <mliska@suse.cz>
18598 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
18600 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18602 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
18603 (create_intersect_range_checks): Move from ...
18604 * tree-data-ref.c (create_intersect_range_checks_index)
18605 (create_intersect_range_checks): ... to here.
18606 (create_runtime_alias_checks): New function factored from ...
18607 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
18608 here. Call above function.
18609 * tree-data-ref.h (create_runtime_alias_checks): New function.
18611 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18613 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
18614 segment length for dr_b and compute it in wide_int.
18616 2017-05-31 Richard Biener <rguenther@suse.de>
18618 PR tree-optimization/80906
18619 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
18620 and pass through iv_map.
18621 (copy_bb_and_scalar_dependences): Adjust.
18622 (translate_pending_phi_nodes): Likewise.
18623 (copy_loop_close_phi_args): Handle code-generating IVs instead
18626 2017-05-30 David Malcolm <dmalcolm@redhat.com>
18628 * diagnostic-color.c (color_dict): Add "type-diff".
18629 (parse_gcc_colors): Update comment.
18630 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
18631 -fdiagnostics-show-template-tree and -fno-elide-type.
18632 (GCC_COLORS): Add type-diff to example.
18634 (-fdiagnostics-show-template-tree): New.
18635 (-fno-elide-type): New.
18636 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
18637 the pp_format_decoder callback. Call any m_format_postprocessor's
18639 (pretty_printer::pretty_printer): Initialize
18640 m_format_postprocessor.
18641 (pretty_printer::~pretty_printer): Delete any
18642 m_format_postprocessor.
18643 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
18644 (class format_postprocessor): New class.
18645 (struct pretty_printer::format_decoder): Document the new parameters.
18646 (struct pretty_printer::m_format_postprocessor): New field.
18647 * tree-diagnostic.c (default_tree_printer): Update for new
18648 bool and const char ** params.
18649 * tree-diagnostic.h (default_tree_printer): Likewise.
18651 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
18653 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
18654 (lwa_operand): Delete rs6000_gen_cell_microcode test.
18655 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18656 rs6000_gen_cell_microcode code.
18657 (rs6000_final_prescan_insn): Delete.
18658 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
18659 "warn-cell-microcode" entries.
18660 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
18661 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
18662 throughout. Change cc_reg_not_micro_cr0_operand to
18663 cc_reg_not_cr0_operand throughout.
18664 (*extendhi<mode>2_noload): Delete.
18665 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
18666 (mwarn-cell-microcode): Delete.
18667 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
18668 -mgen-cell-microcode and -mwarn-cell-microcode.
18670 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
18673 * config/i386/constraints.md (Yd): New constraint.
18675 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
18676 and (?Yd, r) alternatives. Update insn attributes.
18677 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
18678 and (?*Yd, r) alternatives. Update insn attributes.
18679 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
18681 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
18683 * gimplify.c (gimplify_modify_expr): Don't create a
18684 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
18687 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
18689 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
18691 2017-05-30 Richard Biener <rguenther@suse.de>
18693 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
18694 and reduc_def fields.
18695 (STMT_VINFO_REDUC_TYPE): New define.
18696 (STMT_VINFO_REDUC_DEF): Likewise.
18697 (vect_force_simple_reduction): Adjust prototype.
18698 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
18699 (vect_is_simple_reduction): Remove check_reduction argument.
18700 (vect_force_simple_reduction): Adjust and set
18701 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18702 (vectorizable_reduction): Do not re-do reduction analysis
18703 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18704 * tree-parloops.c (gather_scalar_reductions): Adjust.
18706 2017-05-30 Richard Biener <rguenther@suse.de>
18708 PR middle-end/80901
18709 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
18712 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
18714 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
18715 Introduce unknown_misalignment parameter and remove vf.
18716 (vect_peeling_hash_get_lowest_cost):
18717 Pass unknown_misalignment parameter.
18718 (vect_enhance_data_refs_alignment):
18719 Fix unsupportable data ref treatment.
18721 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18723 * tree-vect-data-refs.c (vect_get_data_access_cost):
18724 Workaround for SLP handling.
18725 (vect_enhance_data_refs_alignment):
18726 Compute costs for doing no peeling at all, compare to the best
18727 peeling costs so far and avoid peeling if cheaper.
18729 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18731 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
18732 Return peeling info and set costs to zero for unlimited cost
18734 (vect_enhance_data_refs_alignment): Also inspect all datarefs
18735 with unknown misalignment. Compute and costs for unknown
18736 misalignment, compare them to the costs for known misalignment
18737 and choose the cheapest for peeling.
18739 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18741 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
18742 (vect_get_peeling_costs_all_drs): Create function.
18743 (vect_peeling_hash_get_lowest_cost):
18744 Use vect_get_peeling_costs_all_drs.
18745 (vect_peeling_supportable): Create function.
18746 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
18748 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18750 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
18751 DR_HAS_NEGATIVE_STEP.
18752 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
18753 (vect_enhance_data_refs_alignment): Use.
18754 (vect_duplicate_ssa_name_ptr_info): Use.
18755 * tree-vectorizer.h (dr_misalignment): Use.
18756 (known_alignment_for_access_p): Use.
18758 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18761 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
18763 (has_section_name): New function.
18765 2017-05-30 Martin Liska <mliska@suse.cz>
18768 * auto-profile.c (get_function_decl_from_block): Fix
18771 2017-05-30 Richard Biener <rguenther@suse.de>
18773 PR middle-end/80876
18774 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
18776 2017-05-30 Martin Liska <mliska@suse.cz>
18778 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
18779 * dumpfile.h (struct dump_file_info): Remove ctors.
18781 2017-05-30 Martin Liska <mliska@suse.cz>
18783 * predict.def: Fix GNU coding style.
18785 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18787 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18788 Mark 'to' argument with ATTRIBUTE_UNUSED.
18790 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18792 * config/xtensa/xtensa.c (xtensa_emit_call): Use
18793 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
18794 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
18797 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
18799 * doc/install.texi (Options specification): Restore entry of
18800 --enable-sjlj-exceptions.
18802 2017-05-27 Michael Eager <eager@eagercon.com>
18805 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
18807 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
18809 * config/microblaze/microblaze.h
18810 (FIXED_REGISTERS): Update in macro.
18811 (CALL_USED_REGISTERS): Update in macro.
18813 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
18815 * doc/install.texi: Add links to macOS binary distributions.
18817 2017-05-27 Jakub Jelinek <jakub@redhat.com>
18821 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18823 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18825 2017-05-26 Martin Liska <mliska@suse.cz>
18827 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
18829 2017-05-26 Martin Liska <mliska@suse.cz>
18831 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
18832 always leading ';; '.
18833 (dump_bb_info): Likewise.
18834 (brief_dump_cfg): Likewise.
18835 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
18836 * dumpfile.c: Remove usage of TDF_VERBOSE.
18837 * dumpfile.h (enum dump_kind): Likewise.
18838 (dump_gimple_bb_header): Do not use TDF_COMMENT.
18839 * print-tree.c (debug_verbose): Remove.
18840 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
18841 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
18842 * tree-diagnostic.c (default_tree_printer): Replace
18843 TDF_DIAGNOSTIC with TDF_SLIM.
18845 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18847 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
18848 in parameter loop, rather than loop_vinfo.
18849 (create_intersect_range_checks): Ditto.
18850 (vect_create_cond_for_alias_checks): Update call to above functions.
18852 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18854 PR tree-optimization/80815
18855 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
18856 for merging runtime alias checks. Handle negative DR_STEPs.
18858 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18860 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
18862 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
18863 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
18864 out code pruning runtime alias checks.
18865 * tree-data-ref.c (prune_runtime_alias_test_list): New function
18866 factored out from above.
18867 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18869 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18871 (prune_runtime_alias_test_list): New decalaration.
18873 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18875 * tree-vect-data-refs.c (compare_tree): Rename and move ...
18876 * tree-data-ref.c (data_ref_compare_tree): ... to here.
18877 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
18878 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
18879 (operator==, comp_dr_with_seg_len_pair): Ditto.
18880 (vect_prune_runtime_alias_test_list): Ditto.
18882 2017-05-26 Martin Liska <mliska@suse.cz>
18885 * params.def: Bound partial-inlining-entry-probability param.
18887 2017-05-26 Marek Polacek <polacek@redhat.com>
18890 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
18893 2017-05-26 Richard Biener <rguenther@suse.de>
18895 PR tree-optimization/80842
18896 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
18899 2017-05-26 Richard Biener <rguenther@suse.de>
18901 PR tree-optimization/80844
18902 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
18904 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
18906 * doc/md.texi (Machine Constraints): Update x86 family
18907 machine constraints section to match 'config/i386/constraints.md'.
18909 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
18911 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
18913 2017-05-25 Nathan Sidwell <nathan@acm.org>
18915 * doc/invoke.texi (--enable-languages): Update documentation.
18917 2017-05-25 Martin Liska <mliska@suse.cz>
18919 * dumpfile.c: Add TDF_FOLDING.
18920 * dumpfile.h (enum dump_kind): Likewise.
18921 * genmatch.c (dt_simplify::gen_1): Use it.
18923 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18925 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
18927 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18929 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18930 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
18932 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18934 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
18935 * match.pd (X == C): Rewrite it here.
18936 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
18937 with_certain_nonzero_bits2): New predicates.
18938 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
18940 2017-05-24 Nathan Sidwell <nathan@acm.org>
18942 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
18945 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
18948 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18950 * config/powerpcspe: New port. Files are copied from the rs6000
18951 port, with "rs6000" in filenames replaced by "powerpcspe".
18953 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
18955 PR rtl-optimization/80754
18956 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
18958 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
18960 * config/sparc/sparc.md (length): Return the correct value for -mflat
18961 sibcalls to match output_sibcall.
18963 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18967 * config/rs6000/rs6000.c (struct machine_function): Add new field
18969 (rs6000_get_separate_components): Init that field, use it.
18970 (rs6000_components_for_bb): Use the field.
18972 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
18974 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
18976 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
18978 PR middle-end/80823
18979 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
18981 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18984 * config/s390/s390.c (s390_check_qrst_address): Check incoming
18985 address against address_operand predicate.
18986 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
18988 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
18990 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
18991 parameters passed indirectly.
18993 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
18995 * config/i386/i386.md (*movdi_internal): Remove SSE4
18996 alternative 18 (?r, *v). Update insn attributes.
18997 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
18998 Update insn attributes.
18999 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
19000 Update insn attributes.
19001 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
19002 alternative 1 (r, v). Remove isa attribute.
19003 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
19004 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
19005 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19007 2017-05-23 Tom de Vries <tom@codesourcery.com>
19009 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
19012 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19014 * cgraphunit.c (symbol_table::process_new_functions): Update.
19015 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
19016 (inline_generate_summary): Rename to ...
19017 (ipa_fn_summary_generate): ... this one.
19018 (inline_read_summary): Rename to ...
19019 (ipa_fn_summary_read): ... this one.
19020 (inline_write_summary): Rename to ...
19021 (ipa_fn_summary_write): ... this one.
19022 (inline_free_summary): Rename to ...
19023 (ipa_free_fn_summary): ... this one.
19024 (pass_data_local_fn_summary, pass_local_fn_summary,
19025 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
19026 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
19027 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
19028 make_pass_ipa_fn_summary): New.
19029 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
19030 inline_write_summary, inline_free_summary): Remove.
19031 (ipa_free_fn_summary) : New.
19032 * ipa-inline.c (ipa_inline): Update.
19033 (pass_ipa_inline): Do not generate summaries.
19034 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
19036 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
19037 and add pass_ipa_fn_summary.
19038 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
19040 (make_pass_inline_parameters): Remove.
19042 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
19044 * omp-low.c (struct omp_context): Remove "default_kind" member.
19047 * omp-offload.c (execute_oacc_device_lower): Remove the
19048 parallelism dimensions function attributes for unparallelized
19049 OpenACC kernels constructs.
19051 2017-05-23 Martin Liska <mliska@suse.cz>
19053 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
19055 (cgraph_edge::make_speculative): Likewise.
19056 (cgraph_edge::resolve_speculation): Likewise.
19057 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19058 (cgraph_node::dump): Likewise.
19059 * cgraph.h: Likewise.
19060 * cgraphunit.c (analyze_functions): Likewise.
19061 (symbol_table::compile): Likewise.
19062 * ipa-cp.c (print_all_lattices): Likewise.
19063 (determine_versionability): Likewise.
19064 (initialize_node_lattices): Likewise.
19065 (ipcp_verify_propagated_values): Likewise.
19066 (estimate_local_effects): Likewise.
19067 (update_profiling_info): Likewise.
19068 (create_specialized_node): Likewise.
19069 (perhaps_add_new_callers): Likewise.
19070 (decide_about_value): Likewise.
19071 (decide_whether_version_node): Likewise.
19072 (identify_dead_nodes): Likewise.
19073 (ipcp_store_bits_results): Likewise.
19074 * ipa-devirt.c (dump_targets): Likewise.
19075 (ipa_devirt): Likewise.
19076 * ipa-icf.c (sem_item::dump): Likewise.
19077 (sem_function::equals): Likewise.
19078 (sem_variable::equals): Likewise.
19079 (sem_item_optimizer::read_section): Likewise.
19080 (sem_item_optimizer::execute): Likewise.
19081 (congruence_class::dump): Likewise.
19082 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19083 (dump_inline_summary): Likewise.
19084 (estimate_node_size_and_time): Likewise.
19085 (inline_analyze_function): Likewise.
19086 * ipa-inline-transform.c (inline_call): Likewise.
19087 * ipa-inline.c (report_inline_failed_reason): Likewise.
19088 (want_early_inline_function_p): Likewise.
19089 (edge_badness): Likewise.
19090 (update_edge_key): Likewise.
19091 (inline_small_functions): Likewise.
19092 * ipa-profile.c (ipa_profile): Likewise.
19093 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19094 (ipa_make_edge_direct_to_target): Likewise.
19095 (remove_described_reference): Likewise.
19096 (ipa_impossible_devirt_target): Likewise.
19097 (propagate_controlled_uses): Likewise.
19098 (ipa_print_node_params): Likewise.
19099 (ipcp_transform_function): Likewise.
19100 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19101 (propagate_pure_const): Likewise.
19102 * ipa-reference.c (generate_summary): Likewise.
19103 (read_write_all_from_decl): Likewise.
19104 (propagate): Likewise.
19105 (ipa_reference_read_optimization_summary): Likewise.
19106 * ipa-utils.c (ipa_merge_profiles): Likewise.
19107 * ipa.c (walk_polymorphic_call_targets): Likewise.
19108 (symbol_table::remove_unreachable_nodes): Likewise.
19109 (ipa_single_use): Likewise.
19110 * passes.c (execute_todo): Likewise.
19111 * predict.c (drop_profile): Likewise.
19112 * symtab.c (symtab_node::get_dump_name): New function.
19113 (symtab_node::dump_name): Likewise.
19114 (symtab_node::dump_asm_name): Likewise.
19115 (symtab_node::dump_references): Likewise.
19116 (symtab_node::dump_referring): Likewise.
19117 (symtab_node::dump_base): Likewise.
19118 (symtab_node::debug_symtab): Likewise.
19119 * tree-sra.c (convert_callers_for_node): Likewise.
19120 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19121 * value-prof.c (init_node_map): Likewise.
19123 2017-05-23 Martin Liska <mliska@suse.cz>
19125 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19126 and symtab_node::debug_symtab to symbol_table::debug.
19127 * cgraphunit.c (analyze_functions): Use the renamed function.
19128 (symbol_table::compile): Likewise.
19129 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19130 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19131 * passes.c (execute_todo): Likewise.
19132 * symtab.c (symbol_table::dump): New function.
19133 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19135 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19137 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19138 that nonconst implies exec.
19140 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19142 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19143 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19144 (inline_edge_summary_vec): Turn into ...
19145 (ipa_call_summaries): ... this one.
19146 (redirect_to_unreachable, edge_set_predicate,
19147 evaluate_properties_for_edge, inline_summary_alloc,
19148 reset_ipa_call_summary, reset_inline_summary,
19149 inline_summary_t::duplicate): Update.
19150 (inline_edge_duplication_hook): Turn to ...
19151 (ipa_call_summary_t::duplicate): ... this one.
19152 (inline_edge_removal_hook): Turn to ...
19153 (ipa_call_summary_t::remove): ... this one.
19154 (dump_inline_edge_summary): Turn to ...
19155 (dump_ipa_call_summary): ... this one.
19156 (estimate_function_body_sizes): Update.
19157 (inline_update_callee_summaries): Update.
19158 (remap_edge_change_prob): Update.
19159 (remap_edge_summaries): Update.
19160 (inline_merge_summary): Update.
19161 (do_estimate_edge_time): Update.
19162 (inline_generate_summary): Update.
19163 (inline_read_section): Update.
19164 (inline_read_summary): Update.
19165 (inline_free_summary): Update.
19166 * ipa-inline.c (can_inline_edge_p): Update.
19167 (compute_inlined_call_time): Update.
19168 (want_inline_small_function_p): Update.
19169 (edge_badness): Update.
19170 (early_inliner): Update.
19171 * ipa-inline.h (inline_edge_summary): Turn to ...
19172 (ipa_call_summary): ... this one.
19173 (ipa_call_summary_t): New class.
19174 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19175 (ipa_call_summaries): New.
19176 (inline_edge_summary): Remove.
19177 (estimate_edge_growth): Update.
19178 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19179 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19180 * ipa-split.c (execute_split_functions): Update.
19181 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19183 2017-05-23 Tom de Vries <tom@codesourcery.com>
19185 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19186 attributes): Document rdrand effective target.
19188 2017-05-23 Tom de Vries <tom@codesourcery.com>
19190 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19191 attributes): Sort alphabetically.
19193 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
19195 * config/avr/genmultilib.awk: Use gsub instead of gensub.
19197 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19200 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
19201 V2DF/V2DI splat into two separate patterns, one that handles
19202 registers, and the other that only handles memory. Drop support
19203 for splatting from a GPR on ISA 2.07 and then splitting the
19204 splat into direct move and splat.
19205 (vsx_splat_<mode>_reg): Likewise.
19206 (vsx_splat_<mode>_mem): Likewise.
19208 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
19210 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
19212 2017-05-22 Jakub Jelinek <jakub@redhat.com>
19214 PR middle-end/80809
19215 * omp-low.c (finish_taskreg_remap): New function.
19216 (finish_taskreg_scan): If unit size of ctx->record_type
19217 is non-constant, unshare the size expression and replace
19218 decls in it with possible outer var refs.
19220 PR middle-end/80809
19221 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
19222 GOVD_SHARED rather than GOVD_PRIVATE with it.
19223 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
19224 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
19226 PR middle-end/80853
19227 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
19228 as last argument to build_outer_var_ref for pointer bases of array
19229 section reductions.
19231 2017-05-19 Martin Sebor <msebor@redhat.com>
19233 * print-tree.c (print_node): Print DECL_READ_P flag.
19235 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19237 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
19238 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
19239 * cgraph.c: Likewise.
19240 * cgraphunit.c: Likewise.
19241 * gengtype.c: Likewise.
19242 * ipa-cp.c: Likewise.
19243 * ipa-devirt.c: Likewise.
19244 * ipa-icf.c: Likewise.
19245 * ipa-predicate.c: Likewise.
19246 * ipa-profile.c: Likewise.
19247 * ipa-prop.c: Likewise.
19248 * ipa-split.c: Likewise.
19250 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
19251 edge_predicate_pool, dump_inline_hints,
19252 inline_summary::account_size_time, redirect_to_unreachable,
19253 edge_set_predicate, set_hint_predicate,
19254 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
19255 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
19256 inline_summary_t::remove, remap_hint_predicate_after_duplication,
19257 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
19258 ipa_call_summary_t::remove, initialize_growth_caches,
19259 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
19260 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
19261 mark_modified, unmodified_parm_1, unmodified_parm,
19262 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
19263 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
19264 compute_bb_predicates, will_be_nonconstant_expr_predicate,
19265 will_be_nonconstant_predicate, record_modified_bb_info,
19266 get_minimal_bb, record_modified, param_change_prob,
19267 phi_result_unknown_predicate, predicate_for_phi_result,
19268 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
19269 estimate_function_body_sizes, compute_inline_parameters,
19270 compute_inline_parameters_for_curren, pass_data_inline_parameters,
19271 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
19272 inline_update_callee_summaries, remap_edge_change_prob,
19273 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
19274 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
19275 inline_analyze_function, inline_summary_t::insert,
19276 inline_generate_summary, read_ipa_call_summary, inline_read_section,
19277 inline_read_summary, write_ipa_call_summary, inline_write_summary,
19278 inline_free_summary): Move to ipa-fnsummary.h
19279 (predicate_t): Remove.
19280 * ipa-fnsummary.c: New file.
19281 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
19282 (enum inline_hints_vals, inline_hints, agg_position_info,
19283 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
19284 inline_summaries, ipa_call_summary, ipa_call_summary_t,
19285 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
19286 dump_inline_summary, dump_inline_hints, inline_generate_summary,
19287 inline_read_summary, inline_write_summary, inline_free_summary,
19288 inline_analyze_function, initialize_inline_failed,
19289 inline_merge_summary, inline_update_overall_summary,
19290 compute_inline_parameters): Move to ipa-fnsummary.h
19291 * ipa-fnsummary.h: New file.
19292 * ipa-inline-transform.h: Include ipa-inline.h.
19293 * ipa-inline.c: LIkewise.
19295 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19297 * ipa-inline.c (edge_badness): Use inlined_time instead of
19298 inline_summaries->get.
19300 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19302 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
19304 2017-05-22 Nathan Sidwell <nathan@acm.org>
19306 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
19307 (fdump-lang): Document 'raw' option.
19308 * dumpfile.h (TDI_tu): Delete.
19309 * dumpfile.c (dump_files): Remove translation-unit.
19310 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
19312 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
19314 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
19315 command option from $(AWK) call.
19316 * config/avr/genmultilib.awk: Simplify and rewrite so that it
19317 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
19318 [FORMAT]: Remove handling of variable.
19319 * config/avr/t-multilib: Regenerate.
19321 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19323 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
19325 (dump_inline_summary): Do not print self_time.
19326 (estimate_function_body_sizes): Do not set self_time.
19327 (compute_inline_parameters): Likewise.
19328 (inline_read_section, inline_write_summary): Do not stream self_time.
19329 * ipa-inline.h (inline_summary): Drop self_time.
19331 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19333 * ipa-inline-analysis.c (account_size_time): Rename to ...
19334 (inline_summary::account_size_time): ... this one.
19335 (reset_ipa_call_summary): Turn to ...
19336 (ipa_call_summary::reset): ... this one.
19337 (reset_inline_summary): Turn to ...
19338 (inline_summary::reset): ... this one.
19339 (inline_summary_t::remove): Update.
19340 (inline_summary_t::duplicate): Update.
19341 (ipa_call_summary_t::remove): Update.
19342 (dump_inline_summary): Update.
19343 (estimate_function_body_sizes): Update.
19344 (compute_inline_parameters): Update.
19345 (estimate_node_size_and_time): Update.
19346 (inline_merge_summary): Update.
19347 (inline_update_overall_summary): Update.
19348 (inline_read_section): Update.
19349 (inline_write_summary): Update.
19350 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
19351 add account_size_time and reset member functions.
19352 (ipa_call_summary): Add reset function.
19353 * ipa-predicate.h (predicate::operator &): Constify.
19355 2017-05-22 Richard Biener <rguenther@suse.de>
19357 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
19359 2017-05-19 Jason Merrill <jason@redhat.com>
19361 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
19363 2017-05-19 Marek Polacek <polacek@redhat.com>
19366 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
19367 TYPE_OVERFLOW_WRAPS checks.
19369 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
19371 * tree-core.h (enum omp_clause_default_kind): Add
19372 "OMP_CLAUSE_DEFAULT_PRESENT".
19373 * tree-pretty-print.c (dump_omp_clause): Handle it.
19374 * gimplify.c (enum gimplify_omp_var_data): Add
19375 "GOVD_MAP_FORCE_PRESENT".
19376 (gimplify_adjust_omp_clauses_1): Map it to
19377 "GOMP_MAP_FORCE_PRESENT".
19378 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
19380 * gimplify.c (oacc_default_clause): Clarify.
19382 2017-05-19 Nathan Sidwell <nathan@acm.org>
19384 LANG_HOOK_REGISTER_DUMPS
19385 * toplev.c (general_init): Call register dump lang hook.
19386 * doc/invoke.texi: Document -fdump-lang option family.
19387 * dumpfile.c (dump_files): Remove class dump here.
19388 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
19389 * dumpfile.h (tree_dump_index): Remove TDI_class.
19390 * langhooks-def.h (lhd_register_dumps): Declare.
19391 (LANG_HOOKS_REGISTER_DUMPS): Define.
19392 (LANG_HOOKS_INITIALIZER): Add it.
19393 * langhooks.c (lhd_register_dumps): Define.
19394 * langhooks.h (struct lang_hooks): Add register_dumps.
19396 2017-05-19 Nathan Sidwell <nathan@acm.org>
19398 * context.h (context::set_passes): New.
19399 * context.c (context::context): Do not create pass manager.
19400 * toplev.c (general_init): Create pass manager here.
19402 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
19404 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
19405 use this splitter if two add or or instructions would also work for
19406 the constant we want to generate.
19408 2017-05-19 Richard Biener <rguenther@suse.de>
19411 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
19412 predicate evaluation.
19414 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19416 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
19418 * ipa-inline.c (want_inline_small_function_p): Do not cast to
19421 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19423 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19424 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19425 (inline_edge_summary_vec): Turn into ...
19426 (ipa_call_summaries): ... this one.
19427 (redirect_to_unreachable, edge_set_predicate,
19428 evaluate_properties_for_edge, inline_summary_alloc,
19429 reset_ipa_call_summary, reset_inline_summary,
19430 inline_summary_t::duplicate): Update.
19431 (inline_edge_duplication_hook): Turn to ...
19432 (ipa_call_summary_t::duplicate): ... this one.
19433 (inline_edge_removal_hook): Turn to ...
19434 (ipa_call_summary_t::remove): ... this one.
19435 (dump_inline_edge_summary): Turn to ...
19436 (dump_ipa_call_summary): ... this one.
19437 (estimate_function_body_sizes): Update.
19438 (inline_update_callee_summaries): Update.
19439 (remap_edge_change_prob): Update.
19440 (remap_edge_summaries): Update.
19441 (inline_merge_summary): Update.
19442 (do_estimate_edge_time): Update.
19443 (inline_generate_summary): Update.
19444 (inline_read_section): Update.
19445 (inline_read_summary): Update.
19446 (inline_free_summary): Update.
19447 * ipa-inline.c (can_inline_edge_p): Update.
19448 (compute_inlined_call_time): Update.
19449 (want_inline_small_function_p): Update.
19450 (edge_badness): Update.
19451 (early_inliner): Update.
19452 * ipa-inline.h (inline_edge_summary): Turn to ...
19453 (ipa_call_summary): ... this one.
19454 (ipa_call_summary_t): New class.
19455 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19456 (ipa_call_summaries): New.
19457 (inline_edge_summary): Remove.
19458 (estimate_edge_growth): Update.
19459 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19460 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19461 * ipa-split.c (execute_split_functions): Update.
19462 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19464 2017-05-19 Richard Biener <rguenther@suse.de>
19466 PR middle-end/80764
19467 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
19469 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
19471 * config/rs6000/rs6000.c (struct machine_function): Add field
19472 fpr_is_wrapped_separately.
19473 (rs6000_get_separate_components): Use 64 components. Handle the
19474 new FPR components.
19475 (rs6000_components_for_bb): Handle the FPR components.
19476 (rs6000_emit_prologue_components): Handle the FPR components.
19477 (rs6000_emit_epilogue_components): Handle the FPR components.
19478 (rs6000_set_handled_components): Handle the FPR components.
19479 (rs6000_emit_prologue): Don't output prologue code for those FPRs
19480 that are already separately shrink-wrapped.
19481 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
19482 that are already separately shrink-wrapped.
19484 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19487 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
19490 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
19491 (define_peephole2 for Altivec d-form load): Add peepholes to catch
19492 cases where the register allocator uses a move and an offsettable
19493 memory operation to/from a FPR register on ISA 2.06/2.07.
19494 (define_peephole2 for Altivec d-form store): Likewise.
19496 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
19499 * config/i386/mmx.md (*mov<mode>_internal): Enable
19500 alternatives 11, 12, 13 and 14 also for 32bit targets.
19501 Remove alternatives 15, 16, 17 and 18.
19502 * config/i386/sse.md (vec_concatv2di): Change
19503 alternative (!x, *y) to (x, ?!*Yn).
19505 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
19507 * dumpfile.h (enum dump_kind): Remove stray comma.
19509 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19511 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
19512 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
19513 predicate::num_conditions
19514 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
19515 (CHANGED): turn into predicate::changed.
19516 (agg_position_info): Move to ipa-predicate.h
19517 (add_condition, predicate::add_clause, predicate::operator &=,
19518 predicate::or_with, predicate::evaluate, predicate::probability,
19519 dump_condition, dump_clause, predicate::dump,
19520 predicate::remap_after_duplication, predicate::remap_after_inlining,
19521 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
19522 (evaluate_conditions_for_known_args): Update.
19523 (set_cond_stmt_execution_predicate): Update.
19524 * ipa-inline.h: Include ipa-predicate.h
19525 (condition, inline_param_summary, conditions, agg_position_info,
19526 predicate): Move to ipa-predicate.h
19527 * ipa-predicate.c: New file.
19528 * ipa-predicate.h: New file.
19530 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
19532 * final.c (leaf_function_p): Check we are not in a sequence.
19534 2017-05-18 Martin Liska <mliska@suse.cz>
19536 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
19537 * dumpfile.c (dump_register): Use new enum dump_kind.
19538 (get_dump_file_name): Likewise.
19539 (dump_enable_all): Likewise.
19540 (dump_switch_p_1): Likewise.
19541 (enable_rtl_dump_file): Remove usage of TDF_RTL.
19542 * dumpfile.h (enum dump_kind): New enum type.
19543 (struct dump_file_info): Create constructor and
19544 format fields and comments.
19545 * passes.c (pass_manager::register_one_dump_file):
19547 * statistics.c (statistics_early_init): Likewise.
19548 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
19549 TDF_TREE with TDF_SLIM.
19550 (gather_memory_references_ref): Likewise.
19552 2017-05-18 Martin Liska <mliska@suse.cz>
19554 * vec.h (struct vnull): Use it.
19556 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19558 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
19559 (true_predicate, false_predicate, true_predicate_p,
19560 false_predicate_p): Remove.
19561 (single_cond_predicate, not_inlined_predicate): Turn to member function
19563 (add_condition): Update.
19564 (add_clause): Turn to...
19565 (predicate::add_clause): ... this one; update; allow passing NULL
19567 (and_predicates): Turn to ...
19568 (predicate::operator &=): ... this one.
19569 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
19570 (or_predicates): Turn to ...
19571 (predicate::or_with): ... this one.
19572 (evaluate_predicate): Turn to ...
19573 (predicate::evaluate): ... this one.
19574 (predicate_probability): Turn to ...
19575 (predicate::probability): ... this one.
19576 (dump_condition): Update.
19577 (dump_predicate): Turn to ...
19578 (predicate::dump): ... this one.
19579 (account_size_time): Update.
19580 (edge_set_predicate): Update.
19581 (set_hint_predicate): UPdate.
19582 (evaluate_conditions_for_known_args): Update.
19583 (evaluate_properties_for_edge): Update.
19584 (remap_predicate_after_duplication): Turn to...
19585 (predicate::remap_after_duplication): ... this one.
19586 (remap_hint_predicate_after_duplication): Update.
19587 (inline_summary_t::duplicate): UPdate.
19588 (dump_inline_edge_summary): Update.
19589 (dump_inline_summary): Update.
19590 (set_cond_stmt_execution_predicate): Update.
19591 (set_switch_stmt_execution_predicate): Update.
19592 (compute_bb_predicates): Update.
19593 (will_be_nonconstant_expr_predicate): Update.
19594 (will_be_nonconstant_predicate): Update.
19595 (phi_result_unknown_predicate): Update.
19596 (predicate_for_phi_result): Update.
19597 (array_index_predicate): Update.
19598 (estimate_function_body_sizes): Update.
19599 (estimate_node_size_and_time): Update.
19600 (estimate_ipcp_clone_size_and_time): Update.
19601 (remap_predicate): Rename to ...
19602 (predicate::remap_after_inlining): ... this one.
19603 (remap_hint_predicate): Update.
19604 (inline_merge_summary): Update.
19605 (inline_update_overall_summary): Update.
19606 (estimate_size_after_inlining): Update.
19607 (read_predicate): Rename to ...
19608 (predicate::stream_in): ... this one.
19609 (read_inline_edge_summary): Update.
19610 (write_predicate): Rename to ...
19611 (predicate::stream_out): ... this one.
19612 (write_inline_edge_summary): Update.
19613 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
19614 (clause_t): Turn to uint32_t
19615 (predicate): Turn to class; implement constructor and operators
19617 (size_time_entry): Update.
19618 (inline_summary): Update.
19619 (inline_edge_summary): Update.
19621 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
19623 * fold-const.c (fold_binary_loc): Move transformation...
19624 * match.pd (C - X CMP X): ... here.
19626 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
19628 * config/sparc/sparc.c (sparc_option_override): Set function
19629 alignment for -mcpu=niagara7 to 64 to match the I$ line.
19630 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
19632 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
19634 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
19636 2017-05-18 Marek Polacek <polacek@redhat.com>
19639 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
19640 (pass_ubsan::execute): Call gimple_assign_single_p instead of
19641 gimple_assign_load_p.
19643 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
19645 PR middle-end/80692
19646 * real.c (do_compare): Give decimal_do_compare preference over
19647 comparing just the signs.
19649 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
19651 * doc/md.texi (Canonicalization of Instructions): Describe the
19652 canonical form of instructions that inherently set a condition
19655 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
19657 PR middle-end/80775
19658 * tree-cfg.c: Move deletion of unreachable case statements to after
19659 the merging of consecutive case labels.
19661 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
19663 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
19664 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
19665 restoring of callee-saved registers.
19667 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
19669 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
19670 * config/visium/visium.c (single_set_and_flags): Likewise.
19671 * config/visium/visium.md (Substitutions): Likewise.
19673 2017-05-17 Martin Liska <mliska@suse.cz>
19675 * cfg.c: Introduce dump_flags_t type and
19676 use it instead of int type.
19678 * cfghooks.c: Likewise.
19679 * cfghooks.h (struct cfg_hooks): Likewise.
19680 * cfgrtl.c: Likewise.
19681 * cfgrtl.h: Likewise.
19682 * cgraph.c (cgraph_node::get_body): Likewise.
19683 * coretypes.h: Likewise.
19684 * domwalk.c: Likewise.
19685 * domwalk.h: Likewise.
19686 * dumpfile.c (struct dump_option_value_info): Likewise.
19687 (dump_enable_all): Likewise.
19688 (dump_switch_p_1): Likewise.
19689 (opt_info_switch_p): Likewise.
19690 * dumpfile.h (enum tree_dump_index): Likewise.
19691 (struct dump_file_info): Likewise.
19692 * genemit.c: Likewise.
19693 * generic-match-head.c: Likewise.
19694 * gengtype.c (open_base_files): Likewise.
19695 * gimple-pretty-print.c: Likewise.
19696 * gimple-pretty-print.h: Likewise.
19697 * graph.c (print_graph_cfg): Likewise.
19698 * graphite-scop-detection.c (dot_all_sese): Likewise.
19699 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
19700 * loop-unroll.c (report_unroll): Likewise.
19701 * passes.c (pass_manager::register_one_dump_file): Likewise.
19702 * print-tree.c: Likewise.
19703 * statistics.c: Likewise.
19704 * tree-cfg.c: Likewise.
19705 * tree-cfg.h: Likewise.
19706 * tree-dfa.c: Likewise.
19707 * tree-dfa.h: Likewise.
19708 * tree-dump.c (dump_function): Likewise.
19709 * tree-dump.h (struct dump_info): Likewise.
19710 * tree-pretty-print.c: Likewise.
19711 * tree-pretty-print.h: Likewise.
19712 * tree-ssa-live.c: Likewise.
19713 * tree-ssa-live.h: Likewise.
19714 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
19715 * tree-vect-loop.c: Likewise.
19716 * tree-vect-slp.c: Likewise.
19718 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
19719 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19721 PR tree-optimization/80457
19722 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
19723 of all arguments to a statement as scalar_to_vec operations.
19724 (vectorizable_call): Adjust call to vect_model_simple_cost for
19726 (vectorizable_conversion): Likewise.
19727 (vectorizable_assignment): Likewise.
19728 (vectorizable_shift): Likewise.
19729 (vectorizable_operation): Likewise.
19730 (vectorizable_comparison): Likewise.
19731 (vect_is_simple_cond): Record the def types for operands.
19732 (vectorizable_condition): Likewise, call vect_model_simple_cost.
19733 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
19734 for statement argument count.
19736 2017-05-16 Carl Love <cel@us.ibm.com>
19738 * config/rs6000/rs6000-c: Add support for built-in functions
19739 vector unsigned long long vec_bperm (vector unsigned long long,
19740 vector unsigned char)
19741 vector signed long long vec_mule (vector signed int,
19743 vector unsigned long long vec_mule (vector unsigned int,
19744 vector unsigned int)
19745 vector signed long long vec_mulo (vector signed int,
19747 vector unsigned long long vec_mulo (vector unsigned int,
19748 vector unsigned int)
19749 vector signed char vec_sldw (vector signed char,
19750 vector signed char,
19752 vector unsigned char vec_sldw (vector unsigned char,
19753 vector unsigned char,
19755 vector signed short vec_sldw (vector signed short,
19756 vector signed short,
19758 vector unsigned short vec_sldw (vector unsigned short,
19759 vector unsigned short,
19761 vector signed int vec_sldw (vector signed int,
19764 vector unsigned int vec_sldw (vector unsigned int,
19765 vector unsigned int,
19767 vector signed long long vec_sldw (vector signed long long,
19768 vector signed long long,
19770 vector unsigned long long vec_sldw (vector unsigned long long,
19771 vector unsigned long long,
19773 * config/rs6000/rs6000-c: Add support for built-in functions
19774 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
19775 * config/rs6000/altivec.h: Add defintion for vec_sldw.
19776 * doc/extend.texi: Update the built-in documentation for the
19777 new built-in functions.
19779 2017-05-16 Marek Polacek <polacek@redhat.com>
19783 * tree.c (save_expr): Don't fold the expression.
19785 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
19787 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
19788 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
19789 and (?*y,m). Update insn attributes.
19791 2017-05-16 Martin Liska <mliska@suse.cz>
19793 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
19794 flags argument of print_gimple_stmt, print_gimple_expr,
19795 print_generic_stmt and print_generic_expr.
19796 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
19797 * coretypes.h: Likewise.
19798 * except.c (dump_eh_tree): Likewise.
19799 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
19800 * gimple-pretty-print.h: Likewise.
19801 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
19802 (backprop::push_to_worklist): Likewise.
19803 (backprop::pop_from_worklist): Likewise.
19804 (backprop::process_use): Likewise.
19805 (backprop::intersect_uses): Likewise.
19806 (note_replacement): Likewise.
19807 * gimple-ssa-store-merging.c
19808 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
19809 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
19810 (pass_store_merging::execute): Likewise.
19811 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
19812 (ssa_base_cand_dump_callback): Likewise.
19813 (dump_incr_vec): Likewise.
19814 (replace_refs): Likewise.
19815 (replace_mult_candidate): Likewise.
19816 (create_add_on_incoming_edge): Likewise.
19817 (create_phi_basis): Likewise.
19818 (insert_initializers): Likewise.
19819 (all_phi_incrs_profitable): Likewise.
19820 (introduce_cast_before_cand): Likewise.
19821 (replace_one_candidate): Likewise.
19822 * gimplify.c (gimplify_expr): Likewise.
19823 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
19824 (set_rename): Likewise.
19825 (rename_uses): Likewise.
19826 (copy_loop_phi_nodes): Likewise.
19827 (add_close_phis_to_merge_points): Likewise.
19828 (copy_loop_close_phi_args): Likewise.
19829 (copy_cond_phi_args): Likewise.
19830 (graphite_copy_stmts_from_block): Likewise.
19831 (translate_pending_phi_nodes): Likewise.
19832 * graphite-poly.c (print_pdr): Likewise.
19833 (dump_gbb_cases): Likewise.
19834 (dump_gbb_conditions): Likewise.
19835 (print_scop_params): Likewise.
19836 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
19837 (build_cross_bb_scalars_use): Likewise.
19838 (gather_bbs::before_dom_children): Likewise.
19839 * hsa-dump.c (dump_hsa_immed): Likewise.
19840 * ipa-cp.c (print_ipcp_constant_value): Likewise.
19841 (get_replacement_map): Likewise.
19842 * ipa-inline-analysis.c (dump_condition): Likewise.
19843 (estimate_function_body_sizes): Likewise.
19844 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
19845 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
19846 * ipa-prop.c (ipa_dump_param): Likewise.
19847 (ipa_print_node_jump_functions_for_edge): Likewise.
19848 (ipa_modify_call_arguments): Likewise.
19849 (ipa_modify_expr): Likewise.
19850 (ipa_dump_param_adjustments): Likewise.
19851 (ipa_dump_agg_replacement_values): Likewise.
19852 (ipcp_modif_dom_walker::before_dom_children): Likewise.
19853 * ipa-pure-const.c (check_stmt): Likewise.
19854 (pass_nothrow::execute): Likewise.
19855 * ipa-split.c (execute_split_functions): Likewise.
19856 * omp-offload.c (dump_oacc_loop_part): Likewise.
19857 (dump_oacc_loop): Likewise.
19858 * trans-mem.c (tm_log_emit): Likewise.
19859 (tm_memopt_accumulate_memops): Likewise.
19860 (dump_tm_memopt_set): Likewise.
19861 (dump_tm_memopt_transform): Likewise.
19862 * tree-cfg.c (gimple_verify_flow_info): Likewise.
19863 (print_loop): Likewise.
19864 * tree-chkp-opt.c (chkp_print_addr): Likewise.
19865 (chkp_gather_checks_info): Likewise.
19866 (chkp_get_check_result): Likewise.
19867 (chkp_remove_check_if_pass): Likewise.
19868 (chkp_use_outer_bounds_if_possible): Likewise.
19869 (chkp_reduce_bounds_lifetime): Likewise.
19870 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
19871 (chkp_mark_completed_bounds): Likewise.
19872 (chkp_register_incomplete_bounds): Likewise.
19873 (chkp_mark_invalid_bounds): Likewise.
19874 (chkp_maybe_copy_and_register_bounds): Likewise.
19875 (chkp_build_returned_bound): Likewise.
19876 (chkp_get_bound_for_parm): Likewise.
19877 (chkp_build_bndldx): Likewise.
19878 (chkp_get_bounds_by_definition): Likewise.
19879 (chkp_generate_extern_var_bounds): Likewise.
19880 (chkp_get_bounds_for_decl_addr): Likewise.
19881 * tree-chrec.c (chrec_apply): Likewise.
19882 * tree-data-ref.c (dump_data_reference): Likewise.
19883 (dump_subscript): Likewise.
19884 (dump_data_dependence_relation): Likewise.
19885 (analyze_overlapping_iterations): Likewise.
19886 * tree-inline.c (expand_call_inline): Likewise.
19887 (tree_function_versioning): Likewise.
19888 * tree-into-ssa.c (dump_defs_stack): Likewise.
19889 (dump_currdefs): Likewise.
19890 (dump_names_replaced_by): Likewise.
19891 (dump_update_ssa): Likewise.
19892 (update_ssa): Likewise.
19893 * tree-object-size.c (pass_object_sizes::execute): Likewise.
19894 * tree-parloops.c (build_new_reduction): Likewise.
19895 (try_create_reduction_list): Likewise.
19896 (ref_conflicts_with_region): Likewise.
19897 (oacc_entry_exit_ok_1): Likewise.
19898 (oacc_entry_exit_single_gang): Likewise.
19899 * tree-pretty-print.h: Likewise.
19900 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
19901 (get_scalar_evolution): Likewise.
19902 (add_to_evolution): Likewise.
19903 (get_loop_exit_condition): Likewise.
19904 (analyze_evolution_in_loop): Likewise.
19905 (analyze_initial_condition): Likewise.
19906 (analyze_scalar_evolution): Likewise.
19907 (instantiate_scev): Likewise.
19908 (number_of_latch_executions): Likewise.
19909 (gather_chrec_stats): Likewise.
19910 (final_value_replacement_loop): Likewise.
19911 (scev_const_prop): Likewise.
19912 * tree-sra.c (dump_access): Likewise.
19913 (disqualify_candidate): Likewise.
19914 (create_access): Likewise.
19915 (reject): Likewise.
19916 (maybe_add_sra_candidate): Likewise.
19917 (create_access_replacement): Likewise.
19918 (analyze_access_subtree): Likewise.
19919 (analyze_all_variable_accesses): Likewise.
19920 (sra_modify_assign): Likewise.
19921 (initialize_constant_pool_replacements): Likewise.
19922 (find_param_candidates): Likewise.
19923 (decide_one_param_reduction): Likewise.
19924 (replace_removed_params_ssa_names): Likewise.
19925 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
19926 * tree-ssa-copy.c (dump_copy_of): Likewise.
19927 (copy_prop_visit_cond_stmt): Likewise.
19928 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
19929 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
19930 (record_equivalences_from_stmt): Likewise.
19931 * tree-ssa-dse.c (compute_trims): Likewise.
19932 (delete_dead_call): Likewise.
19933 (delete_dead_assignment): Likewise.
19934 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
19935 (forward_propagate_into_cond): Likewise.
19936 (pass_forwprop::execute): Likewise.
19937 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
19938 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
19940 (move_computations_worker): Likewise.
19941 (execute_sm): Likewise.
19942 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
19943 (remove_exits_and_undefined_stmts): Likewise.
19944 (remove_redundant_iv_tests): Likewise.
19945 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
19946 (adjust_iv_update_pos): Likewise.
19947 * tree-ssa-math-opts.c (bswap_replace): Likewise.
19948 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
19949 (value_replacement): Likewise.
19950 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
19951 * tree-ssa-pre.c (print_pre_expr): Likewise.
19952 (get_representative_for): Likewise.
19953 (create_expression_by_pieces): Likewise.
19954 (insert_into_preds_of_block): Likewise.
19955 (eliminate_insert): Likewise.
19956 (eliminate_dom_walker::before_dom_children): Likewise.
19957 (eliminate): Likewise.
19958 (remove_dead_inserted_code): Likewise.
19959 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
19960 * tree-ssa-reassoc.c (get_rank): Likewise.
19961 (eliminate_duplicate_pair): Likewise.
19962 (eliminate_plus_minus_pair): Likewise.
19963 (eliminate_not_pairs): Likewise.
19964 (undistribute_ops_list): Likewise.
19965 (eliminate_redundant_comparison): Likewise.
19966 (update_range_test): Likewise.
19967 (optimize_range_tests_var_bound): Likewise.
19968 (optimize_vec_cond_expr): Likewise.
19969 (rewrite_expr_tree): Likewise.
19970 (rewrite_expr_tree_parallel): Likewise.
19971 (linearize_expr): Likewise.
19972 (break_up_subtract): Likewise.
19973 (linearize_expr_tree): Likewise.
19974 (attempt_builtin_powi): Likewise.
19975 (attempt_builtin_copysign): Likewise.
19976 (transform_stmt_to_copy): Likewise.
19977 (transform_stmt_to_multiply): Likewise.
19978 (dump_ops_vector): Likewise.
19979 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
19980 (print_scc): Likewise.
19981 (set_ssa_val_to): Likewise.
19982 (visit_reference_op_store): Likewise.
19983 (visit_use): Likewise.
19984 (sccvn_dom_walker::before_dom_children): Likewise.
19985 (run_scc_vn): Likewise.
19986 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
19988 (expr_hash_elt::print): Likewise.
19989 (const_and_copies::pop_to_marker): Likewise.
19990 (const_and_copies::record_const_or_copy_raw): Likewise.
19991 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
19992 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
19993 (dump_predicates): Likewise.
19994 (find_uninit_use): Likewise.
19995 (warn_uninitialized_phi): Likewise.
19996 (pass_late_warn_uninitialized::execute): Likewise.
19997 * tree-ssa.c (verify_vssa): Likewise.
19998 (verify_ssa): Likewise.
19999 (maybe_optimize_var): Likewise.
20000 * tree-vrp.c (dump_value_range): Likewise.
20001 (dump_all_value_ranges): Likewise.
20002 (dump_asserts_for): Likewise.
20003 (register_edge_assert_for_2): Likewise.
20004 (vrp_visit_cond_stmt): Likewise.
20005 (vrp_visit_switch_stmt): Likewise.
20006 (vrp_visit_stmt): Likewise.
20007 (vrp_visit_phi_node): Likewise.
20008 (simplify_cond_using_ranges_1): Likewise.
20009 (fold_predicate_in): Likewise.
20010 (evrp_dom_walker::before_dom_children): Likewise.
20011 (evrp_dom_walker::push_value_range): Likewise.
20012 (evrp_dom_walker::pop_value_range): Likewise.
20013 (execute_early_vrp): Likewise.
20015 2017-05-16 Richard Biener <rguenther@suse.de>
20017 * dwarf2out.c (loc_list_from_tree_1): Do not create
20018 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
20020 2017-05-16 Richard Biener <rguenther@suse.de>
20022 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
20024 (note_variable_value_in_expr): If we resolved the decl ref
20025 do not push to the stack.
20027 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
20029 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
20030 operations in fast-math mode.
20031 (vaddq_f16): Likewise.
20032 (vmul_f16): Likewise.
20033 (vmulq_f16): Likewise.
20034 (vsub_f16): Likewise.
20035 (vsubq_f16): Likewise.
20036 * config/arm/neon.md (add<mode>3): New.
20038 (fma:<VH:mode>3): New. Also remove outdated comment.
20041 2017-05-16 Martin Liska <mliska@suse.cz>
20045 * ipa-devirt.c (warn_types_mismatch): Fix typo.
20046 (odr_types_equivalent_p): Likewise.
20048 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
20050 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
20052 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
20055 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
20056 non-interunit SSE move alternatives with '?'.
20057 (zero-extendsidi peephole2): New peephole to skip intermediate
20058 general register in SSE zero-extend sequence.
20060 2017-05-15 Jeff Law <law@redhat.com>
20062 * reorg.c (relax_delay_slots): Create a new variable to hold
20063 the temporary target rather than clobbering TARGET_LABEL.
20065 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20066 missing argument to extract_bit_field call.
20067 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20069 2017-05-15 Martin Liska <mliska@suse.cz>
20072 * gcc.c (process_command): Do not allow empty argument of -o option.
20074 2017-05-15 Renlin Li <renlin.li@arm.com>
20076 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20077 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20078 * config/aarch64/constraints.md (Usf): Add long call check.
20079 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20080 (call_value): Likewise.
20081 (sibcall): Likewise.
20082 (sibcall_value): Likewise.
20084 (call_value_insn): New.
20085 (sibcall_insn): Update rtx pattern.
20086 (sibcall_value_insn): Likewise.
20087 (call_internal): Remove.
20088 (call_value_internal): Likewise.
20089 (sibcall_internal): Likewise.
20090 (sibcall_value_internal): Likewise.
20091 (call_reg): Likewise.
20092 (call_symbol): Likewise.
20093 (call_value_reg): Likewise.
20094 (call_value_symbol): Likewise.
20096 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20099 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20101 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20103 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20104 compatible with CCGOCmode and with CCZmode.
20106 2017-05-14 Martin Sebor <msebor@redhat.com>
20108 PR middle-end/77671
20109 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20110 (gimple_fold_builtin_snprintf): Same.
20111 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20112 (gimple_fold_builtin_snprintf): Same.
20113 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20114 of character types.
20115 (is_call_safe): New function.
20116 (try_substitute_return_value): Call it.
20117 (try_simplify_call): New function.
20118 (pass_sprintf_length::handle_gimple_call): Call it.
20120 2017-05-14 Martin Sebor <msebor@redhat.com>
20122 PR middle-end/80669
20123 * builtins.c (expand_builtin_stpncpy): Simplify.
20125 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20127 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20128 * config/i386/i386.h
20129 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20130 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20131 (struct machine_function): Add new members call_ms2sysv,
20132 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20133 (struct machine_frame_state): New fields sp_realigned and
20134 sp_realigned_offset.
20135 * config/i386/i386.c
20136 (enum xlogue_stub): New enum.
20137 (enum xlogue_stub_sets): New enum.
20138 (class xlogue_layout): New class.
20139 (struct ix86_frame): New fields stack_realign_allocate_offset,
20140 stack_realign_offset and outlined_save_offset. Modify comments to
20141 detail stack layout when using out-of-line stubs.
20142 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20143 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
20144 -mcall-ms2sysv-xlogues.
20145 (stub_managed_regs): New static variable.
20146 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
20147 registers managed by out-of-line stub.
20148 (disable_call_ms2sysv_xlogues): New function.
20149 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
20150 m->call_ms2sysv when appropriate and compute frame layout for
20152 (sp_valid_at, fp_valid_at): New inline functions.
20153 (choose_basereg): New function.
20154 (choose_baseaddr): Add align parameter, use choose_basereg and modify
20156 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
20157 Use align parameter of choose_baseaddr to generated aligned SSE movs
20159 (pro_epilogue_adjust_stack): Modify to track
20160 machine_frame_state::sp_realigned.
20161 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
20162 (ix86_nsaved_sseregs): Likewise.
20163 (ix86_emit_save_regs): Likewise.
20164 (ix86_emit_save_regs_using_mov): Likewise.
20165 (ix86_emit_save_sse_regs_using_mov): Likewise.
20166 (get_scratch_register_on_entry): Likewise.
20167 (gen_frame_set): New function.
20168 (gen_frame_load): Likewise.
20169 (gen_frame_store): Likewise.
20170 (emit_outlined_ms2sysv_save): Likewise.
20171 (emit_outlined_ms2sysv_restore): Likewise.
20172 (ix86_expand_prologue): Modify stack re-alignment code and call
20173 emit_outlined_ms2sysv_save when appropriate.
20174 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
20175 parameter rtx_insn *insn, which allows the function to be used to only
20176 generate the notes.
20177 (ix86_expand_epilogue): Modify validity checks of frame and stack
20178 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
20179 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
20180 * config/i386/predicates.md
20181 (save_multiple): New predicate.
20182 (restore_multiple): Likewise.
20183 * config/i386/sse.md
20184 (save_multiple<mode>): New pattern.
20185 (save_multiple_realign<mode>): Likewise.
20186 (restore_multiple<mode>): Likewise.
20187 (restore_multiple_and_return<mode>): Likewise.
20188 (restore_multiple_leave_return<mode>): Likewise.
20189 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
20191 2017-05-14 Julia Koval <julia.koval@intel.com>
20193 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
20194 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
20195 (__builtin_ia32_xsetbv): New builtins.
20196 * config/i386/i386.c (ix86_expand_special_args_builtin):
20198 (ix86_expand_builtin): Special expand for new intrinsics.
20199 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
20200 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
20201 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
20203 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20205 * cfganal.c (inverted_post_order_compute): Change argument type
20207 * cfganal.h (inverted_post_order_compute): Adjust prototype.
20208 * df-core.c (rest_of_handle_df_initialize): Adjust.
20209 (rest_of_handle_df_finish): Likewise.
20210 (df_analyze_1): Likewise.
20211 (df_analyze): Likewise.
20212 (loop_inverted_post_order_compute): Change argument to be a vec *.
20213 (df_analyze_loop): Adjust.
20214 (df_get_n_blocks): Likewise.
20215 (df_get_postorder): Likewise.
20216 * df.h (struct df_d): Change field to be a vec.
20217 * lcm.c (compute_laterin): Adjust.
20218 (compute_available): Likewise.
20219 * lra-lives.c (lra_create_live_ranges_1): Likewise.
20220 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
20221 * tree-ssa-pre.c (compute_antic): Likewise.
20223 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20225 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
20226 (depth_first_search::depth_first_search): Change structure init
20227 function to this constructor.
20228 (depth_first_search::add_bb): Rename function to this member.
20229 (depth_first_search::execute): Likewise.
20230 (flow_dfs_compute_reverse_finish): Adjust.
20232 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20234 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
20235 (longest_simple_path): Likewise.
20236 * shrink-wrap.c (spread_components): Likewise.
20237 (disqualify_problematic_components): Likewise.
20238 (emit_common_heads_for_components): Likewise.
20239 (emit_common_tails_for_components): Likewise.
20240 (insert_prologue_epilogue_for_components): Likewise.
20242 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20244 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
20247 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20249 * df-core.c (df_set_blocks): Start using auto_bitmap.
20250 (df_compact_blocks): Likewise.
20251 * df-problems.c (df_rd_confluence_n): Likewise.
20252 * df-scan.c (df_insn_rescan_all): Likewise.
20253 (df_process_deferred_rescans): Likewise.
20254 (df_update_entry_block_defs): Likewise.
20255 (df_update_exit_block_uses): Likewise.
20256 (df_entry_block_bitmap_verify): Likewise.
20257 (df_exit_block_bitmap_verify): Likewise.
20258 (df_scan_verify): Likewise.
20259 * lra-constraints.c (lra_constraints): Likewise.
20260 (undo_optional_reloads): Likewise.
20261 (lra_undo_inheritance): Likewise.
20262 * lra-remat.c (calculate_gen_cands): Likewise.
20263 (do_remat): Likewise.
20264 * lra-spills.c (assign_spill_hard_regs): Likewise.
20265 (spill_pseudos): Likewise.
20266 * tree-ssa-pre.c (bitmap_set_and): Likewise.
20267 (bitmap_set_subtract_values): Likewise.
20269 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20271 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
20272 management with auto_bitmap.
20273 (fix_inter_tick): Likewise.
20274 (fix_recovery_deps): Likewise.
20275 * ira.c (add_store_equivs): Likewise.
20276 (find_moveable_pseudos): Likewise.
20277 (split_live_ranges_for_shrink_wrap): Likewise.
20278 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
20279 (rtx_reuse_manager::seen_def_p): Likewise.
20280 (rtx_reuse_manager::set_seen_def): Likewise.
20281 * print-rtl.h (class rtx_reuse_manager): Likewise.
20283 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20285 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
20287 (migrate_btr_def): Likewise.
20288 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
20289 * df-core.c (loop_post_order_compute): Likewise.
20290 (loop_inverted_post_order_compute): Likewise.
20291 * hsa-common.h: Likewise.
20292 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
20293 * init-regs.c (initialize_uninitialized_regs): Likewise.
20294 * ipa-inline.c (resolve_noninline_speculation): Likewise.
20295 (inline_small_functions): Likewise.
20296 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
20297 * ira.c (combine_and_move_insns): Likewise.
20298 (build_insn_chain): Likewise.
20299 * loop-invariant.c (find_invariants): Likewise.
20300 * lower-subreg.c (propagate_pseudo_copies): Likewise.
20301 * predict.c (tree_predict_by_opcode): Likewise.
20302 (predict_paths_leading_to): Likewise.
20303 (predict_paths_leading_to_edge): Likewise.
20304 (estimate_loops_at_level): Likewise.
20305 (estimate_loops): Likewise.
20306 * shrink-wrap.c (try_shrink_wrapping): Likewise.
20307 (spread_components): Likewise.
20308 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
20309 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
20310 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
20311 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
20312 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
20313 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
20314 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
20315 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
20316 (mark_threaded_blocks): Likewise.
20317 (thread_through_all_blocks): Likewise.
20318 * tree-ssa.c (verify_ssa): Likewise.
20319 (execute_update_addresses_taken): Likewise.
20320 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
20322 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20324 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
20326 (post_order_compute): Likewise.
20327 (inverted_post_order_compute): Likewise.
20328 (pre_and_rev_post_order_compute_fn): Likewise.
20330 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20332 * genrecog.c (int_set::int_set): Explicitly construct our
20333 auto_vec base class.
20334 * vec.h (auto_vec::auto_vec): New constructor.
20336 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20338 * bitmap.h (class auto_bitmap): New constructor taking
20339 bitmap_obstack * argument.
20341 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20343 * bitmap.h (class auto_bitmap): Change type of m_bits to
20344 bitmap_head, and adjust ctor / dtor and member operators.
20346 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
20348 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
20349 when returned register mode doesn't match original mode.
20351 2017-05-12 Jeff Law <law@redhat.com>
20352 Jakub Jelinek <jakub@redhat.com>
20354 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
20355 we look for cc setter after the compare-elim changes.
20356 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
20357 within the vector to match what compare-elim now expects.
20358 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
20359 (xorsi3_flags, one_cmplsi2_flags): Likewise.
20361 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
20362 after the compare-elim changes.
20363 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
20364 the vector to match what compare-elim now expects.
20365 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
20366 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
20367 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
20368 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
20369 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
20371 * config/visium/visium.c (single_set_and_flags): Fix where
20372 we look for cc setter after the compare-elim changes.
20373 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
20374 with the vector to match what compare-elim now expects.
20375 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
20376 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
20377 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
20378 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
20379 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
20380 (neg<mode>2_insn_set_overflow): Likewise.
20382 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
20384 PR middle-end/79794
20385 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
20386 maybe_expand_insn call, set ops[0].target. If still set after call,
20387 set alt_rtl. Add extra arg to recursive calls.
20388 (extract_bit_field): Add alt_rtl argument. Pass to
20390 * expmed.h (extract_bit_field): Fix prototype.
20391 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
20392 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
20393 to extract_bit_field_calls.
20394 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
20395 Pass alt_rtl to extract_bit_field calls.
20396 * calls.c (store_unaligned_arguments_into_psuedos)
20397 load_register_parameters): Pass extra NULL to extract_bit_field calls.
20398 * optabs.c (maybe_legitimize_operand): Clear op->target when call
20400 * optabs.h (struct expand_operand): Add target bitfield.
20402 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20404 * compare-elim.c (try_eliminate_compare): Canonicalize
20405 operation with embedded compare to
20406 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
20407 (set (reg) (operation)].
20409 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
20411 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20414 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
20415 cost of adding a carry flag for ADC instruction.
20416 [case MINUS]: Ignore the cost of subtracting a carry flag
20417 for SBB instruction.
20419 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
20421 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
20423 * config/rs6000/bmiintrin.h: New file.
20424 * config/rs6000/bmi2intrin.h: New file.
20425 * config/rs6000/x86intrin.h: New file.
20427 2017-05-12 Jeff Law <law@redhat.com>
20429 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
20432 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
20434 PR middle-end/80707
20435 * tree-cfg.c: Remove cfg edges of unreachable case statements.
20437 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20439 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20440 early expansion of vector divide builtins.
20441 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
20442 builtins identified as having unsigned arguments.
20444 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20446 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
20447 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
20448 expansion of vector logical operations (and, andc, or, xor,
20451 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20453 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
20454 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
20456 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20458 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20459 early GIMPLE expansion of vector multiplies.
20461 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20463 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
20464 TARGET_HAVE_MOVT conditional.
20465 (movt splitter): Likewise.
20467 2017-05-12 Richard Biener <rguenther@suse.de>
20469 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
20470 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20471 Fold all stmts not inplace.
20473 2017-05-12 Richard Biener <rguenther@suse.de>
20475 PR tree-optimization/80713
20476 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
20477 inserted_exprs bit for not removed stmts.
20479 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
20481 PR middle-end/69921
20482 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
20483 parallelized" attribute for parallelized OpenACC kernels.
20484 * omp-offload.c (execute_oacc_device_lower): Use it.
20486 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
20487 Set "oacc kernels" attribute.
20488 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
20489 parameter. Adjust all users.
20490 (oacc_fn_attrib_kernels_p): Remove function.
20491 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
20492 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
20493 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
20494 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
20495 assert "oacc kernels" attribute is set.
20497 2017-05-11 Carl Love <cel@us.ibm.com>
20499 * config/rs6000/rs6000-c: Add support for built-in functions
20500 vector unsigned char vec_popcnt (vector signed char)
20501 vector unsigned char vec_popcnt (vector unsigned char)
20502 vector unsigned short vec_popcnt (vector signed short)
20503 vector unsigned short vec_popcnt (vector unsigned short)
20504 vector unsigned int vec_popcnt (vector signed int)
20505 vector unsigned int vec_popcnt (vector unsigned int)
20506 vector unsigned long long vec_popcnt (vector signed long long)
20507 vector unsigned long long vec_popcnt (vector unsigned long long)
20508 vector signed long long vec_slo (vector signed long long,
20509 vector signed char)
20510 vector signed long long vec_slo (vector signed long long,
20511 vector unsigned char)
20512 vector unsigned long long vec_slo (vector unsigned long long,
20513 vector signed char)
20514 vector unsigned long long vec_slo (vector unsigned long long,
20515 vector unsigned char)
20516 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
20517 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
20518 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
20519 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
20520 * doc/extend.texi: Update the built-in documentation file for the
20521 new built-in functions.
20523 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20525 * attribs.h (sorted_attr_string): Move machine independent
20526 functions for target clone support from the i386 port to common
20527 code. Rename ix86_function_versions to common_function_versions.
20528 Rename make_name to make_unique_name.
20529 (common_function_versions): Likewise.
20530 (make_unique_name): Likewise.
20531 (make_dispatcher_decl): Likewise.
20532 (is_function_default_version): Likewise.
20533 * attribs.c (attr_strcmp): Likewise.
20534 (sorted_attr_string): Likewise.
20535 (common_function_versions): Likewise.
20536 (make_unique_name): Likewise.
20537 (make_dispatcher_decl): Likewise.
20538 (is_function_default_version): Likewise.
20539 * config/i386/i386.c (attr_strcmp): Likewise.
20540 (sorted_attr_string): Likewise.
20541 (ix86_function_versions): Likewise.
20542 (make_name): Likewise.
20543 (make_dispatcher_decl): Likewise.
20544 (is_function_default_version): Likewise.
20545 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
20547 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20550 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20551 Account for direct move costs for vec_construct of integer
20554 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
20557 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
20558 (UNSPEC_STX_ATOMIC): Ditto.
20559 (loaddi_via_sse): New insn.
20560 (storedi_via_sse): Ditto.
20561 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
20562 Update corresponding peephole2 patterns.
20563 (atomic_storedi_fpu): Ditto.
20565 2017-05-11 Julia Koval <julia.koval@intel.com>
20567 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
20568 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
20570 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
20571 (__builtin_ia32_rsqrt14ss_mask): New builtins.
20572 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
20574 2017-05-11 Nathan Sidwell <nathan@acm.org>
20576 * graphite-poly.c: Include dumpfile.h.
20578 * dumpfle.h (dump_function): Declare here ...
20579 * tree-dump.h (dump_function): ... not here.
20580 * dumpfile.c: #include tree-cfg.h.
20581 (dump_function): Move here from ...
20582 * tree-dump.c (dump_function): ... here.
20583 * gimplify.c: #include splay-tree.h, not tree-dump.h.
20584 * graphite-poly.c: Don't include tree-dump.h.
20585 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
20586 * print-tree.c: Likewise.
20587 * stor-layout.c: Likewise.
20588 * tree-nested.c: Likewise.
20590 * dumpfile.c (dump_start): Use TDF_FLAGS.
20591 (dump_enable_all): Fix TDF_KIND check thinko.
20593 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20595 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20596 array entries to represent two legal parameterizations of the
20597 overloaded __builtin_cmpb function, as represented by the
20598 P6_OV_BUILTIN_CMPB constant.
20599 (altivec_resolve_overloaded_builtin): Add special case handling
20600 for the __builtin_cmpb function, as represented by the
20601 P6_OV_BUILTIN_CMPB constant.
20602 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
20603 (BU_P6_64BIT_2): New macro.
20604 (BU_P6_OVERLOAD_2): New macro
20605 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
20606 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
20607 (CMPB): Add overload support to represent both 32-bit and 64-bit
20608 compare-bytes function.
20609 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
20610 support for TARGET_CMPB.
20611 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
20612 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20613 documentation of the __builtin_cmpb overloaded built-in function.
20615 2017-05-11 Richard Biener <rguenther@suse.de>
20617 PR tree-optimization/80705
20618 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
20619 bases are not vectorizable.
20621 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20623 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
20624 when counting register pressure.
20626 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20628 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
20629 (iv_ca_more_deps): Renamed to ...
20630 (iv_ca_compare_deps): ... this.
20631 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
20633 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20635 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
20637 (determine_group_iv_costs): ... here.
20638 (find_inv_vars_cb): Record inv var if it's not recorded before.
20640 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20642 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
20643 (get_shiftadd_cost): Ditto.
20645 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20647 * tree-ssa-address.c: Include header file.
20648 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
20650 (add_to_parts): Refactor.
20651 (addr_to_parts): New parameter. Update use of move_hint_to_base.
20652 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
20655 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20657 PR tree-optimization/53090
20658 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
20660 (extract_cond_operands): Detect condition with IV on both sides
20661 and return COMP_IV_EXPR_2.
20662 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
20663 (rewrite_use_compare): Simplify by removing call to function
20664 extract_cond_operands.
20666 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20668 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
20669 (extract_cond_operands): Detect condition comparing against non-
20670 invariant bound and return appropriate enum value.
20671 (find_interesting_uses_cond): Update use of extract_cond_operands.
20672 Handle its return value accordingly.
20673 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
20675 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20677 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
20678 nonlinear iv_use computation in loop invariant sensitive way.
20680 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20682 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
20683 (find_iv_candidates): Call relate_compare_use_with_all_cands.
20685 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20687 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
20688 (dump_cand): Support iv_cand.inv_exprs.
20689 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
20691 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
20694 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20696 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
20698 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
20699 as local function. Include necessary header files.
20700 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
20702 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20704 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
20706 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20708 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
20709 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
20710 RSHIFT_EXPR and BIT_NOT_EXPR.
20712 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20714 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
20715 (adjust_setup_cost): New parameter supporting round up adjustment.
20716 (struct address_cost_data): Delete.
20717 (force_expr_to_var_cost): Don't bound cost with spill_cost.
20718 (split_address_cost, ptr_difference_cost): Delete.
20719 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
20720 (struct ainc_cost_data): New struct.
20721 (get_address_cost_ainc): New function.
20722 (get_address_cost, get_computation_cost): Reimplement.
20723 (determine_group_iv_cost_address): Record inv_expr for all uses of
20725 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
20726 (iv_ca_has_deps): Reimplemented to ...
20727 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
20729 (iv_ca_extend): Call iv_ca_more_deps.
20731 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20733 * tree-ssa-address.c (struct mem_address): Move to header file.
20734 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
20735 * tree-ssa-address.h (struct mem_address): Move from C file.
20736 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
20738 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20740 * tree-affine.h (aff_combination_type): New interface.
20741 (aff_combination_zero_p): Remove static.
20742 (aff_combination_const_p): New interface.
20743 (aff_combination_singleton_var_p): New interfaces.
20745 2017-05-11 Richard Biener <rguenther@suse.de>
20747 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20748 Skip unreachable blocks and destinations.
20749 (eliminate): Move stmt removal and fixup ...
20750 (fini_eliminate): ... here. Skip inserted exprs.
20751 (pass_pre::execute): Move fini_pre after fini_eliminate.
20752 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
20753 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
20754 PRE to get rid of dead code that has invalid SSA form and
20755 split critical edges again.
20757 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20759 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20761 2017-05-11 Richard Biener <rguenther@suse.de>
20763 * passes.c (execute_function_todo): Verify loops if they are
20764 said to be up-to-date.
20765 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
20766 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
20768 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
20771 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
20772 handle calling assemble_external ourself.
20775 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
20776 modes with zero size. Enhance comment.
20778 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20780 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
20781 built-ins for vec_xl and vec_xst with short and char pointer
20784 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20786 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
20787 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
20788 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
20789 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
20790 (_mm_maskz_min_round_ss): New intrinsics.
20791 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
20792 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
20793 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
20794 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
20795 (__builtin_ia32_minss_mask_round): New builtins.
20796 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20797 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20798 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
20800 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
20801 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
20803 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
20806 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20808 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
20809 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
20810 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
20811 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
20812 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
20813 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
20814 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
20815 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20816 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20817 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
20818 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
20819 (__builtin_ia32_mulss_mask_round): New builtins.
20820 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20821 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20822 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
20824 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
20825 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20827 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20830 2017-05-10 Julia Koval <julia.koval@intel.com>
20832 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
20833 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
20834 (_mm256_setr_m128i): New intrinsics.
20836 2017-05-10 Julia Koval <julia.koval@intel.com>
20838 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
20839 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
20840 (_mm_maskz_rcp14_ss): New intrinsics.
20841 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
20842 (__builtin_ia32_rcp14ss_mask): New builtins.
20843 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
20845 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
20847 PR tree-optimization/51513
20848 * tree-cfg.c (gimple_seq_unreachable_p): New function.
20849 (assert_unreachable_fallthru_edge_p): Use it.
20850 (group_case_labels_stmt): Likewise.
20851 * tree-cfg.h: Prototype it.
20852 * stmt.c: Include cfghooks.h and tree-cfg.h.
20853 (emit_case_dispatch_table) <gap_label>: New local variable.
20854 Use it to fill dispatch table gaps.
20855 Test for default_label before updating probabilities.
20856 (expand_case) <default_label>: Remove unneeded initialization.
20857 Test for unreachable default case statement and remove its edge.
20858 Set default_label accordingly.
20859 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
20861 2017-05-10 Carl Love <cel@us.ibm.com>
20863 * config/rs6000/rs6000-c: Add support for built-in functions
20864 vector signed char vec_neg (vector signed char)
20865 vector signed short int vec_neg (vector short int)
20866 vector signed int vec_neg (vector signed int)
20867 vector signed long long vec_neg (vector signed long long)
20868 vector float vec_neg (vector float)
20869 vector double vec_neg (vector double)
20870 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
20872 * config/rs6000/altivec.h: Add define for vec_neg
20873 * doc/extend.texi: Update the built-in documentation for the
20874 new built-in functions.
20876 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20878 PR tree-optimization/77644
20879 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
20881 2017-05-10 Nathan Sidwell <nathan@acm.org>
20883 * dumpfile.h (TDI_lang_all): New.
20884 (TDF_KIND): New. Renumber others
20885 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
20887 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
20889 (get_dump_file_name): Adjust suffix generation.
20890 (dump_enable_all): Use TDF_KIND.
20891 * doc/invoke.texi (-fdump-lang-all): Document.
20893 * dumpfile.h: Tabify.
20895 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
20898 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
20899 Move member access before delete.
20901 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
20903 * tree-inline.c (expand_call_inline): Split block at stmt
20906 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
20909 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
20910 are now unused after splitting mov{sf,sd}_hardfloat.
20911 (f32_lr2): Likewise.
20912 (f32_lm): Likewise.
20913 (f32_lm2): Likewise.
20914 (f32_li): Likewise.
20915 (f32_li2): Likewise.
20916 (f32_lv): Likewise.
20917 (f32_sr): Likewise.
20918 (f32_sr2): Likewise.
20919 (f32_sm): Likewise.
20920 (f32_sm2): Likewise.
20921 (f32_si): Likewise.
20922 (f32_si2): Likewise.
20923 (f32_sv): Likewise.
20924 (f32_dm): Likewise.
20925 (f32_vsx): Likewise.
20926 (f32_av): Likewise.
20927 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
20928 For movsf, order stores so the VSX stores occur before the GPR
20929 store which encourages the register allocator to use a traditional
20930 FPR instead of a GPR. For movsd, order the stores so that the GPR
20931 store comes before the VSX stores to allow the power6 to work.
20932 This is due to the power6 not having a 32-bit integer store
20933 instruction from a FPR.
20934 (movsf_hardfloat): Likewise.
20935 (movsd_hardfloat): Likewise.
20937 2017-05-09 Martin Sebor <msebor@redhat.com>
20939 PR translation/80280
20940 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
20943 PR translation/80280
20944 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
20945 data member added in r247778.
20946 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
20948 2017-05-09 Nathan Sidwell <nathan@acm.org>
20950 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
20952 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
20955 2017-05-09 Marek Polacek <polacek@redhat.com>
20957 * doc/invoke.texi: Fix typo.
20959 2017-05-09 Richard Biener <rguenther@suse.de>
20961 * tree-vrp.c (vrp_val_is_max): Adjust comment.
20962 (vrp_val_is_min): Likewise.
20963 (set_value_range_to_value): Likewise.
20964 (set_value_range_to_nonnegative): Likewise.
20965 (gimple_assign_nonzero_p): Likewise.
20966 (gimple_stmt_nonzero_p): Likewise.
20967 (vrp_int_const_binop): Likewise. Remove unreachable case.
20968 (adjust_range_with_scev): Adjust comments.
20969 (compare_range_with_value): Likewise.
20970 (extract_range_from_phi_node): Likewise.
20971 (test_for_singularity): Likewise.
20973 2017-05-09 Richard Biener <rguenther@suse.de>
20975 * tree-vrp.c (get_single_symbol): Add assert that we don't
20976 get overflowed constants as invariant part.
20977 (compare_values_warnv): Add comment before the TREE_NO_WARNING
20978 checks. Use wi::cmp instead of recursing for integer constants.
20979 (compare_values): Just ignore whether we assumed undefined
20980 overflow instead of failing the compare.
20981 (extract_range_for_var_from_comparison_expr): Add comment before the
20982 TREE_NO_WARNING sets.
20983 (test_for_singularity): Likewise.
20984 (extract_range_from_comparison): Do not disable optimization
20985 when we assumed undefined overflow.
20986 (extract_range_basic): Remove init of unused var.
20988 2017-05-09 Richard Biener <rguenther@suse.de>
20990 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
20991 (extract_range_from_multiplicative_op_1): Adjust.
20992 (extract_range_from_binary_expr_1): Use int_const_binop.
20994 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20997 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
20998 rs6000_store_data_bypass_p in seven define_bypass directives and
20999 in several comments.
21000 * config/rs6000/rs6000-protos.h: Add prototype for
21001 rs6000_store_data_bypass_p function.
21002 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
21003 function implements slightly different (rs6000-specific) semantics
21004 than store_data_bypass_p, returning false rather than aborting
21005 with assertion error when arguments do not satisfy the
21006 requirements of store data bypass.
21007 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
21008 rs6000_store_data_bypass_p.
21010 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
21012 * config/xtensa/xtensa-protos.h
21013 (xtensa_initial_elimination_offset): New declaration.
21014 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21015 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
21016 macro definition, add case for FRAME_POINTER_REGNUM when
21017 FRAME_GROWS_DOWNWARD.
21018 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
21019 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
21020 xtensa_initial_elimination_offset.
21022 2017-05-08 Nathan Sidwell <nathan@acm.org>
21024 * doc/invoke.texi: Alphabetize -fdump options.
21026 2017-05-08 Martin Sebor <msebor@redhat.com>
21028 PR translation/80280
21029 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
21031 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
21033 * target.def (compute_frame_layout): New optional target hook.
21034 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
21035 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
21036 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
21038 * reload1.c (verify_initial_elim_offsets): Likewise.
21039 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
21040 (use_simple_return_p): Call arm_compute_frame_layout if needed.
21041 (arm_get_frame_offsets): Split up into this ...
21042 (arm_compute_frame_layout): ... and this function.
21044 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
21046 * config/aarch64/constraints.md (Usa): New constraint.
21047 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
21049 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
21051 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
21052 with_multilib_list after it has been checked.
21054 2017-05-08 Richard Biener <rguenther@suse.de>
21056 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
21057 (bitmap_set_subtract_values): Likewise.
21059 2017-05-08 Richard Biener <rguenther@suse.de>
21061 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
21062 (gimple_assign_nonzero): ... this and remove strict_overflow_p
21064 (gimple_stmt_nonzero_warnv_p): Rename to ...
21065 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21067 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21068 (extract_range_basic): Adjust, do not disable propagation on
21069 strict overflow sensitive simplification.
21070 (vrp_visit_cond_stmt): Likewise.
21072 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21074 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21075 body size unconditionally.
21077 2017-05-07 Jeff Law <law@redhat.com>
21080 2017-05-06 Jeff Law <law@redhat.com>
21081 PR tree-optimization/78496
21082 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21085 PR tree-optimization/78496
21086 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21087 (simplify_stmt_using_ranges): Call it.
21088 (vrp_dom_walker::before_dom_children): Extract equivalences
21089 from an ASSERT_EXPR with an equality comparison against a
21092 2017-05-06 Jeff Law <law@redhat.com>
21094 PR tree-optimization/78496
21095 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21098 PR tree-optimization/78496
21099 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21100 (simplify_stmt_using_ranges): Call it.
21101 (vrp_dom_walker::before_dom_children): Extract equivalences
21102 from an ASSERT_EXPR with an equality comparison against a
21105 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21107 * lra-constraints.c (lra_copy_reg_equiv): New function.
21108 (split_reg): Use it to copy equivalence information from the
21109 original register to the spill register.
21111 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21113 PR rtl-optimization/75964
21114 * simplify-rtx.c (simplify_const_relational_operation): Remove
21115 invalid handling of comparisons of integer ABS.
21117 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21119 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21120 initialize to zero.
21121 (init_regs): Remove declaration.
21122 (function_arg_advance_32): Initialize error_p as boolean variable.
21124 2017-05-05 Nathan Sidwell <nathan@acm.org>
21126 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21127 lines. Use for (;;).
21129 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21131 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21132 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21133 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21134 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21135 VF=2 that require versioning.
21137 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21139 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21142 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21144 * diagnostic.h (diagnostic_override_option_index): Convert from
21145 macro to inline function.
21147 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21149 * diagnostic.c (last_module_changed_p): New function.
21150 (set_last_module): New function.
21151 (diagnostic_report_current_module): Convert macro usage to
21152 the above functions.
21153 * diagnostic.h (diagnostic_context::last_module): Strengthen
21154 from const line_map * to const line_map_ordinary *.
21155 (diagnostic_last_module_changed): Delete macro.
21156 (diagnostic_set_last_module): Delete macro.
21158 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21160 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
21161 with diagnostic_report_diagnostic.
21162 (diagnostic_n_impl_richloc): Likewise.
21163 * diagnostic.h (report_diagnostic): Delete macro.
21164 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
21165 with diagnostic_report_diagnostic.
21166 * substring-locations.c (format_warning_va): Likewise.
21168 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21170 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
21171 save/restor of format_spec. Move option-printing code to...
21172 (print_option_information): ...this new function, and
21173 reimplement by simply printing to the pretty_printer,
21174 rather than appending to the format string.
21176 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21178 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
21179 handling logic into...
21180 (update_effective_level_from_pragmas): ...this new function.
21182 2017-05-04 Andrew Waterman <andrew@sifive.com>
21184 * config/riscv/riscv.opt (mstrict-align): New option.
21185 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
21186 (SLOW_UNALIGNED_ACCESS): Define.
21187 (riscv_slow_unaligned_access): Declare.
21188 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
21190 (riscv_slow_unaligned_access): New variable.
21191 (rocket_tune_info): Set slow_unaligned_access to true.
21192 (optimize_size_tune_info): Set slow_unaligned_access to false.
21193 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
21194 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
21195 (riscv_option_override): Set riscv_slow_unaligned_access.
21196 * doc/invoke.texi: Add -mstrict-align to RISC-V.
21198 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
21200 * config/riscv/riscv.md: Unify indentation.
21202 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21207 * config/rs6000/rs6000.md (u code attribute): Add FIX and
21209 (extendsi<mode>2): Add support for doing sign extension via
21210 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
21211 don't have ISA 3.0 instructions.
21212 (extendsi<mode>2 splitter): Likewise.
21213 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
21214 generate the normal insns since SImode can now go in vector
21215 registers. Disallow the special UNSPECs needed for previous
21216 machines to hide SImode being used. Add new insns
21217 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
21218 (fix_trunc<mode>si2_stfiwx): Likewise.
21219 (fix_trunc<mode>si2_internal): Likewise.
21220 (fixuns_trunc<mode>si2): Likewise.
21221 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21222 (fctiw<u>z_<mode>_smallint): Likewise.
21223 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
21224 of floating point to 32-bit integer from doing a direct move to
21225 the GPR registers to do a store.
21226 (fctiwz_<mode>): Break long line.
21228 2017-05-05 Bin Cheng <bin.cheng@arm.com>
21230 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
21231 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
21232 (addr_list, addr_offset_valid_p): New.
21233 (split_address_groups): Check offset validity with above function.
21234 (gt-tree-ssa-loop-ivopts.h): Include header file.
21236 2017-05-05 Nathan Sidwell <nathan@acm.org>
21238 * config.gcc (arm*-*-*): Add missing 'fi'.
21240 2017-05-05 Steve Ellcey <sellcey@cavium.com>
21242 * doc/invoke.texi (-fopt-info): Explicitly say order of options
21243 included in -fopt-info does not matter.
21244 * doc/optinfo.texi (-fopt-info): Fix description of default
21245 behavour. Explicitly say order of options included in -fopt-info
21248 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21250 * config.gcc: Allow combinations of aprofile and rmprofile values for
21251 --with-multilib-list.
21252 * config/arm/t-multilib: New file.
21253 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
21254 variables. Remove setting of ISA and floating-point ABI in
21255 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
21256 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
21257 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
21258 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
21260 * config/arm/t-rmprofile: Likewise except for the matches changes.
21261 * doc/install.texi (--with-multilib-list): Document the combination of
21262 aprofile and rmprofile values and warn about pitfalls in doing that.
21264 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21266 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
21267 (movdi_aarch64): Likewise.
21269 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21271 PR tree-optimization/80632
21272 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
21274 (build_arrays): Initialize it for virtual phis.
21275 (fix_phi_nodes): Use it for virtual phis.
21277 PR tree-optimization/80558
21278 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
21279 [x, y] op z into [x op, y op z] for op & or | if conditions
21282 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
21283 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21286 * config/arm/arm.md (use_literal_pool): Remove.
21287 (64-bit immediate split): No longer takes cost into consideration
21288 if arm_disable_literal_pool is enabled.
21289 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
21290 used when arm_disable_literal_pool is enabled.
21291 (arm_max_const_double_inline_cost): Remove use of
21292 arm_disable_literal_pool.
21293 (push_minipool_fix): Add assert.
21294 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
21295 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
21296 (no_literal_pool_sf_immediate): New.
21298 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21300 PR tree-optimization/80613
21301 * tree-ssa-dce.c (propagate_necessity): Remove cases for
21302 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
21304 2017-05-05 Richard Biener <rguenther@suse.de>
21306 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
21308 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
21310 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
21311 of this flag from insn conditions due to removal from r247495.
21313 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21315 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
21317 (arm_early_store_addr_dep_ptr): Likewise.
21318 * config/arm/aarch-common-protos.h
21319 (arm_early_load_addr_dep_ptr): Add prototype.
21320 (arm_early_store_addr_dep_ptr): Likewise.
21321 * config/arm/cortex-a53.md: Add new bypasses.
21323 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21325 * tree.c (next_type_uid): Change type to unsigned.
21326 (type_hash_canon): Decrement back next_type_uid if
21327 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
21328 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
21331 2017-05-04 Martin Sebor <msebor@redhat.com>
21333 * builtins.c: Fix a trivial typo in a comment.
21335 PR middle-end/79234
21336 * builtins.c (check_sizes): Adjust to handle reading past the end.
21337 Avoid printing excessive upper bound of ranges. Use %E to print
21338 tree nodes instead of converting them to %wu.
21339 (expand_builtin_memchr): New function.
21340 (compute_dest_size): Rename...
21341 (compute_objsize): ...to this.
21342 (expand_builtin_memcpy): Adjust.
21343 (expand_builtin_mempcpy): Adjust.
21344 (expand_builtin_strcat): Adjust.
21345 (expand_builtin_strcpy): Adjust.
21346 (check_strncat_sizes): Adjust.
21347 (expand_builtin_strncat): Adjust.
21348 (expand_builtin_strncpy): Adjust and simplify.
21349 (expand_builtin_memset): Adjust.
21350 (expand_builtin_bzero): Adjust.
21351 (expand_builtin_memcmp): Adjust.
21352 (expand_builtin): Handle memcmp.
21353 (maybe_emit_chk_warning): Check strncat just once.
21355 2017-05-04 Martin Sebor <msebor@redhat.com>
21357 PR preprocessor/79214
21358 PR middle-end/79222
21359 PR middle-end/79223
21360 * builtins.c (check_sizes): Add inlining context and issue
21361 warnings even when -Wno-system-headers is set.
21362 (check_strncat_sizes): Same.
21363 (expand_builtin_strncat): Same.
21364 (expand_builtin_memmove): New function.
21365 (expand_builtin_stpncpy): Same.
21366 (expand_builtin): Handle memmove and stpncpy.
21368 2017-05-04 Bin Cheng <bin.cheng@arm.com>
21370 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
21371 which is not used any more.
21373 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21375 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
21377 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21379 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
21380 (cortexa53_tunings): Likewise.
21381 (cortexa57_tunings): Likewise.
21382 (cortexa72_tunings): Likewise.
21383 (cortexa73_tunings): Likewise.
21385 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21387 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
21388 Set loop alignment to 8.
21390 2017-05-04 Martin Sebor <msebor@redhat.com>
21392 PR translation/80280
21393 * builtins.c (expand_builtin_object_size): Add missing quoting to
21394 %D and like directives.
21395 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
21396 (hsa_type_for_tree_type): Same.
21397 (verify_function_arguments): Same.
21398 * symtab.c (symbol_table::change_decl_assembler_name): Same.
21399 * varasm.c (get_section): Same.
21402 2017-05-04 Martin Sebor <msebor@redhat.com>
21404 PR translation/80280
21405 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
21407 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21409 * config/aarch64/aarch64.c (generic_addrcost_table):
21410 Change HI/TI mode setting.
21412 2017-05-04 Martin Jambor <mjambor@suse.cz>
21414 PR tree-optimization/80622
21415 * tree-sra.c (comes_initialized_p): New function.
21416 (build_accesses_from_assign): Only set write lazily when
21417 comes_initialized_p is false.
21418 (analyze_access_subtree): Use comes_initialized_p.
21419 (propagate_subaccesses_across_link): Assert !comes_initialized_p
21420 instead of testing for PARM_DECL.
21422 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21424 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
21425 constraint on operand 0 to allow more general addressing modes.
21426 Adjust output template.
21427 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
21429 * config/aarch64/aarch64-protos.h
21430 (aarch64_address_valid_for_prefetch_p): Declare prototype.
21431 * config/aarch64/constraints.md (Dp): New address constraint.
21432 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
21435 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
21437 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
21438 update use of estimate_ipcp_clone_size_and_time.
21439 (estimate_local_effects): Update use of
21440 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
21441 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
21442 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
21443 Return nonspecialized time.
21445 2017-05-04 Richard Biener <rguenther@suse.de>
21447 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
21448 for the last VUSE which def dominates the PHI. Directly call
21450 (get_continuation_for_phi_1): Remove.
21452 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
21454 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
21455 to explain the use of truncating division. Cap the number of
21456 iterations to the maximum given by nb_iterations_upper_bound,
21459 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21461 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
21462 * configure: Regenerate.
21463 * config.in: Regenerate.
21464 * config/i386/driver-mingw32.c: new file.
21465 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
21466 * config.host: Link driver-mingw32.o on MinGW host.
21467 * doc/install.texi: Document new --enable-mingw-wildcard configure
21470 2017-05-04 Marek Polacek <polacek@redhat.com>
21472 PR tree-optimization/80612
21473 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
21475 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21476 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
21478 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
21479 (movt splitter): Likewise.
21480 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
21481 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
21482 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
21483 block for Thumb-1 with MOVT.
21484 (thumb2_legitimate_address_p): Move code block ...
21485 (can_avoid_literal_pool_for_label_p): ... into this new function.
21486 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
21488 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
21489 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
21490 "M-profile targets with the MOVT instruction".
21492 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21494 * config/arm/arm-builtins.c (arm_init_builtins): Rename
21495 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
21496 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
21498 2017-05-04 Martin Liska <mliska@suse.cz>
21500 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
21501 variable cond_code.
21503 2017-05-04 Richard Biener <rguenther@suse.de>
21505 * tree.c (array_at_struct_end_p): Handle arrays at struct
21506 end with flexarrays more conservatively. Refactor and treat
21507 arrays of arrays or aggregates more strict. Fix
21508 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
21509 * tree.c (array_at_struct_end_p): Adjust prototype.
21510 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
21511 * gimple-fold.c (get_range_strlen): Likewise.
21512 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
21514 2017-05-04 Richard Biener <rguenther@suse.de>
21516 PR tree-optimization/31130
21517 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
21519 (supports_overflow_infinity): Likewise.
21520 (is_negative_overflow_infinity): Likewise.
21521 (is_positive_overflow_infinity): Likewise.
21522 (is_overflow_infinity): Likewise.
21523 (stmt_overflow_infinity): Likewise.
21524 (overflow_infinity_range_p): Likewise.
21525 (usable_range_p): Remove as always returning true.
21526 (make_overflow_infinity): Remove.
21527 (negative_overflow_infinity): Likewise.
21528 (positive_overflow_infinity): Likewise.
21529 (avoid_overflow_infinity): Likewise.
21530 (set_value_range): Adjust accordingly.
21531 (set_value_range_to_nonnegative): Likewise, remove now unused
21532 overflow_infinity arg.
21533 (vrp_operand_equal_p): Adjust.
21534 (update_value_range): Likewise.
21535 (range_int_cst_singleton_p): Likewise.
21536 (operand_less_p): Likewise.
21537 (compare_values_warnv): Likewise.
21538 (extract_range_for_var_from_comparison_expr): Likewise.
21539 (vrp_int_const_binop): Likewise.
21540 (zero_nonzero_bits_from_vr): Likewise.
21541 (extract_range_from_multiplicative_op_1): Likewise.
21542 (extract_range_from_binary_expr_1): Likewise.
21543 (extract_range_from_unary_expr): Likewise.
21544 (extract_range_from_comparison): Likewise.
21545 (extract_range_basic): Likewise.
21546 (adjust_range_with_scev): Likewise.
21547 (compare_ranges): Likewise.
21548 (compare_range_with_value): Likewise.
21549 (dump_value_range): Likewise.
21550 (test_for_singularity): Likewise, remove strict_overflow_p parameter
21552 (simplify_cond_using_ranges): Adjust.
21554 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21556 * brig-builtins.def: Added a builtin for class_f64.
21557 * builtin-types.def: Added a builtin type needed by class_f64.
21559 2017-05-03 Jason Merrill <jason@redhat.com>
21561 * timevar.def: Add TV_CONSTEXPR.
21563 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21565 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
21567 2017-05-03 Martin Jambor <mjambor@suse.cz>
21569 * ipa-prop.c (ipa_update_after_lto_read): Removed.
21570 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
21571 * ipa-cp.c (ipcp_propagate_stage): Do not call
21572 ipa_update_after_lto_read.
21573 * ipa-inline.c (ipa_inline): Likewise.
21575 2017-05-03 Martin Jambor <mjambor@suse.cz>
21577 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
21578 tag. Added a default constructor and a destructor.
21579 (ipa_edge_args_sum_t): New class;
21580 (ipa_edge_args_sum): Declare.
21581 (ipa_edge_args_vector): Remove declaration.
21582 (IPA_EDGE_REF): Use ipa_edge_args_sum.
21583 (ipa_free_edge_args_substructures): Remove declaration.
21584 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
21585 (ipa_edge_args_info_available_for_edge_p): Likewise.
21586 * ipa-prop.c (ipa_edge_args_vector): Removed.
21587 (edge_removal_hook_holder): Likewise.
21588 (edge_duplication_hook_holder): Likewise.
21589 (ipa_edge_args_sum): New variable.
21590 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
21591 ipa_edge_args_vector.
21592 (ipa_free_edge_args_substructures): Likewise.
21593 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
21594 ipa_edge_args_vector.
21595 (ipa_edge_removal_hook): Turned into method
21596 ipa_edge_args_sum_t::remove.
21597 (ipa_edge_duplication_hook): Turned into method
21598 ipa_edge_args_sum_t::duplicate.
21599 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
21600 registering edge hooks.
21601 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
21602 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
21603 ipa_edge_args_sum instead of ipa_edge_args_vector.
21604 * ipa-profile.c (ipa_profile): Likewise.
21606 2017-05-03 Martin Jambor <mjambor@suse.cz>
21608 * symbol-summary.h (function_summary): New method exists.
21609 (function_summary::symtab_removal): Deallocate through release.
21610 (call_summary): New class.
21611 (gt_ggc_mx): New overload.
21612 (gt_pch_nx): Likewise.
21613 (gt_pch_nx): Likewise.
21615 2017-05-03 Jeff Law <law@redhat.com>
21617 PR tree-optimization/78496
21618 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
21619 from simplify_cond_using_ranges. Split off code to walk
21620 backwards through casts into ...
21621 (simplify_cond_using_ranges_2): New function.
21622 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
21623 (execute_vrp): After identifying jump threads, call
21624 simplify_cond_using_ranges_2.
21626 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
21629 * ipa-inline.h (inline_summary): Add ctor.
21630 (create_ggc): Do not use ggc_cleared_alloc.
21632 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
21633 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21635 * gcc.c (handle_braces): Support escaping in switch matching
21637 * doc/invoke.texi (Spec Files): Document it.
21638 Remove superfluous @code markup in items.
21640 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21642 * diagnostic-show-locus.c (struct column_range): New struct.
21643 (get_affected_columns): New function.
21644 (get_printed_columns): New function.
21645 (struct correction): New struct.
21646 (correction::ensure_capacity): New function.
21647 (correction::ensure_terminated): New function.
21648 (struct line_corrections): New struct.
21649 (line_corrections::~line_corrections): New dtor.
21650 (line_corrections::add_hint): New function.
21651 (layout::print_trailing_fixits): Reimplement in terms of the new
21653 (selftest::test_overlapped_fixit_printing): New function.
21654 (selftest::diagnostic_show_locus_c_tests): Call it.
21656 2017-05-03 Nathan Sidwell <nathan@acm.org>
21658 Canonicalize canonical type hashing
21659 * tree.h (type_hash_canon_hash): Declare.
21660 * tree.c (type_hash_list, attribute_hash_list): Move into
21661 type_hash_canon_hash.
21662 (build_type_attribute_qual_variant): Break out hash code calc into
21663 type_hash_canon_hash.
21664 (type_hash_canon_hash): New. Generic type hash computation.
21665 (build_range_type_1, build_array_type_1, build_function_type,
21666 build_method_type_directly, build_offset_type, build_complex_type,
21667 make_vector_type): Call it.
21669 2017-05-03 Richard Biener <rguenther@suse.de>
21671 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
21672 When all DRs have unknown misaligned do not always peel
21673 when there is a store but apply the same costing model as if
21674 there were only loads.
21676 2017-05-03 Richard Biener <rguenther@suse.de>
21679 PR tree-optimization/80492
21680 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21681 compare_base_decls returning dont-know properly.
21683 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
21685 * config/arm/iterators.md (CCSI): New mode iterator.
21686 (arch): New mode attribute.
21687 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
21688 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
21689 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
21690 code iterator for success result mode.
21691 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
21692 the corresponding new insn generators.
21694 2017-05-03 Bin Cheng <bin.cheng@arm.com>
21697 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21698 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21700 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
21702 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
21703 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
21704 (DDR_A): Wrap DDR argument in brackets.
21705 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
21706 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
21707 (DDR_REVERSED_P): Likewise.
21709 2017-05-03 Jakub Jelinek <jakub@redhat.com>
21711 PR tree-optimization/79472
21712 * tree-switch-conversion.c (struct switch_conv_info): Add
21713 contiguous_range and default_case_nonstandard fields.
21714 (collect_switch_conv_info): Compute contiguous_range and
21715 default_case_nonstandard fields, don't clear final_bb if
21716 contiguous_range and only the default case doesn't have the required
21718 (check_all_empty_except_final): Set default_case_nonstandard instead
21719 of failing if contiguous_range and the default case doesn't have empty
21721 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
21722 and only the default case doesn't have the required constants. Skip
21724 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
21725 if default_case_nonstandard.
21726 (build_constructors): Build constant 1 just once. Assert that default
21727 values aren't inserted in between cases if contiguous_range. Skip
21729 (build_arrays): Skip virtual phis.
21730 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
21731 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
21732 Handle virtual phis.
21733 (gen_inbound_check): Handle default_case_nonstandard case.
21734 (process_switch): Adjust check_final_bb caller. Call
21735 gather_default_values with the first non-default case instead of
21736 default case if default_case_nonstandard.
21738 2017-05-02 Nathan Sidwell <nathan@acm.org>
21740 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
21741 check. Fix formatting.
21743 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
21745 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
21746 errors when comparing specialized and unspecialized times.
21748 2017-05-02 David Malcolm <dmalcolm@redhat.com>
21750 * diagnostic-show-locus.c
21751 (layout::should_print_annotation_line_p): Make private.
21752 (layout::print_annotation_line): Make private.
21753 (layout::annotation_line_showed_range_p): Make private.
21754 (layout::show_ruler): Make private.
21755 (layout::print_source_line): Make private. Pass in line and
21756 line_width, rather than calling location_get_source_line. Drop
21758 (layout::print_leading_fixits): New method.
21759 (layout::print_any_fixits): Rename to...
21760 (layout::print_trailing_fixits): ...this, and make private.
21761 Don't print newline fixits.
21762 (diagnostic_show_locus): Move logic for printing one row into...
21763 (layout::print_line): ...this new function. Move the
21764 location_get_source_line call and error-handling from
21765 print_source_line to here. Call print_leading_fixits, and rename
21766 print_any_fixits to print_trailing_fixits.
21767 (selftest::test_fixit_insert_containing_newline): Update now that
21768 newlines are partially supported.
21769 (selftest::test_fixit_insert_containing_newline_2): New test.
21770 (selftest::test_fixit_replace_containing_newline): Update comments.
21771 (selftest::diagnostic_show_locus_c_tests): Call the new test.
21772 * edit-context.c (class added_line): New class.
21773 (class edited_line): Describe newline handling in comment.
21774 (edited_line::actually_edited_p): New method.
21775 (edited_line::print_content): Delete redundant decl.
21776 (edited_line::m_predecessors): New field.
21777 (edited_file::print_content): Call edited_line::print_content.
21778 (edited_file::print_diff): Update to support newlines.
21779 (edited_file::print_diff_hunk): Likewise.
21780 (edited_file::print_run_of_changed_lines): New function.
21781 (edited_file::print_diff_line): Convert to...
21782 (print_diff_line): ...this.
21783 (edited_file::get_effective_line_count): New function.
21784 (edited_line::edited_line): Initialize new field m_predecessors.
21785 (edited_line::~edited_line): Clean up m_predecessors.
21786 (edited_line::apply_fixit): Handle newlines.
21787 (edited_line::get_effective_line_count): New function.
21788 (edited_line::print_content): New function.
21789 (edited_line::print_diff_lines): New function.
21790 (selftest::test_applying_fixits_insert_containing_newline): New
21792 (selftest::test_applying_fixits_replace_containing_newline): New
21794 (selftest::insert_line): New function.
21795 (selftest::test_applying_fixits_multiple_lines): Add example of
21797 (selftest::edit_context_c_tests): Call the new tests.
21799 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21801 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
21802 parameter cand. Update dump information.
21803 (get_computation_cost): Update uses.
21805 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21807 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
21808 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
21809 (get_computation_at, rewrite_use_address): Update use of
21810 get_computation_aff.
21812 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21814 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
21815 (get_computation): Delete.
21816 (get_computation_cost): Implement like get_computation_cost_at.
21817 Use get_computation_at.
21818 (get_computation_cost_at): Delete.
21819 (rewrite_use_nonlinear_expr): Use get_computation_at.
21820 (rewrite_use_compare, remove_unused_ivs): Ditto.
21822 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21824 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
21826 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21828 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
21829 (ivopts_global_cost_for_size): Rename parameter and update uses.
21830 (iv_ca_recount_cost): Update uses.
21831 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21832 candidates seperately in n_invs and n_cands.
21833 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
21835 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21837 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
21838 (find_inv_vars_cb): New.
21839 (find_depends): Renamed to ...
21840 (find_inv_vars): ... this.
21841 (add_candidate_1, force_var_cost): Call find_inv_vars.
21842 (split_address_cost, determine_group_iv_cost_cond): Ditto.
21844 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21846 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
21847 inv_vars. Add inv_exprs.
21848 (struct iv_cand): Rename depends_on to inv_vars.
21849 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
21850 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
21851 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
21852 (dump_cand): Dump inv_vars.
21853 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
21854 (record_invariant, find_depends, add_candidate_1): Ditto.
21855 (set_group_iv_cost, force_var_cost): Ditto.
21856 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
21857 (get_computation_cost_at, get_computation_cost): Ditto.
21858 (determine_group_iv_cost_generic): Ditto.
21859 (determine_group_iv_cost_address): Ditto.
21860 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
21861 (determine_group_iv_costs): Ditto.
21862 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
21863 (iv_ca_set_remove_invariants): Renamed to ...
21864 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
21865 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
21866 (iv_ca_set_add_invariants): Renamed to ...
21867 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
21868 (iv_ca_set_cp): Use iv_ca_set_add_invs.
21869 (iv_ca_has_deps): Support inv_vars and inv_exprs.
21870 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
21871 (create_new_ivs): Remove useless dump.
21873 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21875 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
21877 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
21878 (iv_ca_set_no_cp, create_new_iv): Ditto.
21880 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21882 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21884 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21886 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
21887 function tree_check2.
21889 2017-05-02 Martin Liska <mliska@suse.cz>
21891 * doc/gcov.texi: Add missing preposition.
21892 * gcov.c (function_info::function_info): Properly fill up
21893 all member variables.
21895 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21897 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
21899 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21901 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
21903 2017-05-02 Martin Liska <mliska@suse.cz>
21906 * lto-streamer-in.c (lto_read_tree_1): Remove
21907 LTO_STREAMER_DEBUG.
21908 * lto-streamer.c (struct tree_hash_entry): Likewise.
21909 (struct tree_entry_hasher): Likewise.
21910 (tree_entry_hasher::hash): Likewise.
21911 (tree_entry_hasher::equal): Likewise.
21912 (lto_streamer_init): Likewise.
21913 (lto_orig_address_map): Likewise.
21914 (lto_orig_address_get): Likewise.
21915 (lto_orig_address_remove): Likewise.
21916 * lto-streamer.h: Likewise.
21917 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
21918 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
21920 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
21922 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
21923 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
21924 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
21925 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
21926 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
21927 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
21928 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
21929 (mm_maskz_sub_ss): New intrinsics.
21930 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21931 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21932 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
21933 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
21934 (__builtin_ia32_subss_mask_round): New builtins.
21935 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21936 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21937 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
21939 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
21940 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21942 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21945 2017-05-02 Martin Jambor <mjambor@suse.cz>
21947 PR tree-optimization/78687
21948 * tree-sra.c (access): New field parent.
21949 (process_subtree_disqualification): New function.
21950 (disqualify_candidate): Call it.
21951 (build_accesses_from_assign): Reset write flag if creating an
21953 (build_access_subtree): Fill in parent field and also prpagate
21954 down grp_write flag.
21955 (create_artificial_child_access): New parameter set_grp_write, set
21956 grp_write to its value.
21957 (propagate_subaccesses_across_link): Also propagate grp_write flag
21959 (propagate_all_subaccesses): Push the closest parent back to work
21960 queue if add_access_to_work_queue returned true.
21962 2017-05-02 Richard Biener <rguenther@suse.de>
21964 * common.opt (fstrict-overflow): Alias negative to fwrapv.
21965 * doc/invoke.texi (fstrict-overflow): Remove all traces of
21966 -fstrict-overflow documentation.
21967 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
21968 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
21969 flag_strict_overflow.
21970 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
21971 * lto-opts.c (lto_write_options): Do not stream it.
21972 * lto-wrapper.c (merge_and_complain): Do not handle it.
21973 * opts.c (default_options_table): Do not set -fstrict-overflow.
21974 (finish_options): Likewise do not clear it when sanitizing.
21975 * simplify-rtx.c (simplify_const_relational_operation): Do not
21976 test flag_strict_overflow.
21978 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21980 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
21981 using enabled attribute.
21982 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
21983 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
21984 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
21985 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
21986 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
21987 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
21988 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
21989 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
21990 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
21991 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
21993 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21995 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
21997 2017-05-02 Richard Biener <rguenther@suse.de>
21999 PR tree-optimization/80591
22001 2017-04-10 Richard Biener <rguenther@suse.de>
22003 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22006 2017-05-02 Richard Biener <rguenther@suse.de>
22008 PR tree-optimization/80549
22009 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
22010 (cleanup_tree_cfg_noloop): Create forwarders to known loop
22011 headers if they do not have a preheader.
22013 2017-05-02 Martin Liska <mliska@suse.cz>
22016 * common.opt: Fix typo.
22017 * doc/invoke.texi: Likewise.
22019 2017-05-01 Jan Beulich <jbeulich@suse.com>
22021 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
22022 swapping, add (x,x,m,x,n) alternative.
22024 2017-05-01 Nathan Sidwell <nathan@acm.org>
22026 * calls.c (combine_pending_stack_adjustment_and_call): Remove
22027 unnecessary unadjusted_alignment check.
22029 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22032 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
22034 * gimplify.c (gimplify_cilk_detach): New function.
22035 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
22036 * tree-core.h: Document EXPR_CILK_SPAWN.
22037 * tree.h (EXPR_CILK_SPAWN): Define.
22039 2017-05-01 David Malcolm <dmalcolm@redhat.com>
22041 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
22042 to use new fixit_hint representation, using the "replace" logic.
22043 (get_line_span_for_fixit_hint): Likewise.
22044 (layout::print_any_fixits): Likewise.
22045 (selftest::test_one_liner_many_fixits): Rename to...
22046 (selftest::test_one_liner_many_fixits_1): ...this, and update
22047 comment and expected output to reflect that the multiple fix-it
22048 hints are now consolidated into one insertion.
22049 (selftest::test_one_liner_many_fixits_2): New test.
22050 (selftest::test_diagnostic_show_locus_one_liner): Update for
22052 (selftest::test_fixit_consolidation): Update for fix-it API
22054 * diagnostic.c (print_parseable_fixits): Likewise.
22055 * edit-context.c (edited_line::m_line_events): Convert from
22056 auto_vec <line_event *> to auto_vec <line_event>.
22057 (class line_event): Convert from abstract base class to a concrete
22058 class, taking over the role of replace_event.
22059 (class insert_event): Delete.
22060 (class replace_event): Rename to class line_event. Convert to
22062 (edit_context::add_fixits): Reimplement.
22063 (edit_context::apply_insert): Delete.
22064 (edit_context::apply_replace): Rename to...
22065 (edit_context::apply_fixit): ...this. Convert to half-open range.
22066 (edited_file::apply_insert): Delete.
22067 (edited_file::apply_replace): Rename to...
22068 (edited_file::apply_fixit): ...this.
22069 (edited_line::~edited_line): Drop deletion of events.
22070 (edited_line::apply_insert): Delete.
22071 (edited_line::apply_replace): Rename to...
22072 (edited_line::apply_fixit): ...this. Convert to half-open range.
22073 Update for change to type of m_line_events.
22074 * edit-context.h (edit_context::apply_insert): Delete.
22075 (edit_context::apply_replace): Rename to...
22076 (edit_context::apply_fixit): ...this.
22078 2017-05-01 Martin Sebor <msebor@redhat.com>
22080 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22083 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22086 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22087 __get_cpuid_max returns 0.
22088 (__get_cpuid_count): Ditto.
22090 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22092 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22093 replacement expression is another instance of one of its arguments.
22095 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22098 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22099 check for stack push/pop autoinc.
22100 * config/i386/i386.c (ix86_agi_dependent): Return false
22101 if the only reason why modified_in_p returned true is that
22102 addr is SP based and set_insn is a push or pop.
22104 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22106 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22109 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22112 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22113 (account_size_time): Use two predicates - exec_pred and
22115 (evaluate_conditions_for_known_args): Compute both clause and
22117 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22118 (inline_summary_t::duplicate): Update.
22119 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22121 (compute_inline_parameters): Likewise.
22122 (estimate_edge_size_and_time): Update caluclation of time.
22123 (estimate_node_size_and_time): Compute both time and nonspecialized
22125 (estimate_ipcp_clone_size_and_time): Update.
22126 (inline_merge_summary): Update.
22127 (do_estimate_edge_time): Update.
22128 (do_estimate_edge_size): Update.
22129 (do_estimate_edge_hints): Update.
22130 (inline_read_section, inline_write_summary): Stream both new predicates.
22131 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22133 (compute_inlined_call_time): Cleanup.
22134 (big_speedup_p): Update.
22135 (edge_badness): Update.
22136 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22137 (size_time_entry): Replace predicate by exec_predicate and
22138 nonconst_predicate.
22139 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22140 (estimate_edge_time): Return also nonspec_time.
22141 (reset_edge_growth_cache): Update.
22143 2017-04-29 Jakub Jelinek <jakub@redhat.com>
22145 PR rtl-optimization/80491
22146 * ifcvt.c (noce_process_if_block): When looking for x setter
22147 with missing else_bb, don't check only the insn right before
22148 cond_earliest, but look for the last insn that x is modified in
22149 within the same bb.
22151 PR rtl-optimization/80491
22152 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
22154 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
22156 PR tree-optimization/80487
22157 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
22159 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22161 PR tree-optimization/79697
22162 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
22163 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
22164 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
22166 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
22167 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
22169 2017-04-28 Martin Sebor <msebor@redhat.com>
22171 PR tree-optimization/80523
22172 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
22173 (init_target_to_host_charmap, target_to_host, target_strtol10): New
22175 (maybe_warn, format_directive, parse_directive): Use new functions.
22176 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
22178 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
22180 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
22182 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
22184 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
22185 target_header_dir): Set correctly.
22186 * configure: Regenerated.
22187 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22188 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22189 instead of SYSTEM_HEADER_DIR.
22191 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
22193 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
22194 (estimate_local_effects): Likewise.
22195 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
22196 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
22197 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
22198 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
22199 do_estimate_edge_time, estimate_edge_time): Likewise.
22200 * ipa-inline-analysis.c (estimate_node_size_and_time,
22201 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
22202 (estimate_time_after_inlining): Remove.
22204 2017-04-28 Martin Liska <mliska@suse.cz>
22206 * doc/gcov.texi: Enhance documentation of gcov.
22208 2017-04-28 Martin Liska <mliska@suse.cz>
22210 * doc/gcov.texi: Sort options in alphabetic order.
22211 * doc/gcov-dump.texi: Likewise.
22212 * doc/gcov-tool.texi: Likewise.
22213 * gcov.c (print_usage): Likewise.
22214 * gcov-dump.c (print_usage): Likewise.
22215 * gcov-tool.c (print_merge_usage_message): Likewise.
22216 (print_rewrite_usage_message): Likewise.
22217 (print_overlap_usage_message): Likewise.
22219 2017-04-28 Martin Liska <mliska@suse.cz>
22221 PR gcov-profile/53915
22222 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
22224 2017-04-28 Martin Liska <mliska@suse.cz>
22226 PR gcov-profile/79891
22227 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
22228 is marked by compiler as living on a line.
22229 (get_cycles_count): Remove usage of the union.
22230 (output_intermediate_file): Likewise.
22231 (find_source): Fix GNU coding style.
22232 (accumulate_line_counts): Remove old non-all block mode.
22233 (output_lines): Remove usage of the union.
22234 * profile.c (output_location): Include all BBs, even if
22235 belonging to a same line (and file) as a previous BB.
22237 2017-04-28 Martin Liska <mliska@suse.cz>
22239 * gcov.c (process_args): Handle new argument 'w'.
22240 (read_graph_file): Assign ID to BBs.
22241 (output_branch_count): Display BB # if verbose flag is set.
22242 (output_lines): Likewise for arcs.
22243 (print_usage): Add '--verbose' option help.
22244 * doc/gcov.texi: Document --verbose (-w) option.
22246 2017-04-28 Martin Liska <mliska@suse.cz>
22248 * gcov.c (struct block_location_info): New struct.
22249 (process_file): Fill up the new structure.
22250 (read_graph_file): Replace usage of encoding by the newly added
22252 (add_line_counts): Likewise.
22253 (accumulate_line_counts): Remove usage of the union.
22254 (function_info::function_info): New function.
22255 (function_info::~function_info): Likewise.
22256 (process_file): Call delete instead of release_function.
22257 (release_function): Release the function.
22258 (release_structures): Call delete instead of release_function.
22259 (solve_flow_graph): Replace usage of num_blocks.
22260 (find_exception_blocks): Likewise.
22261 (output_lines): Fix GNU coding style.
22263 2017-04-28 Martin Liska <mliska@suse.cz>
22266 * coverage.c (coverage_remove_note_file): New function.
22267 * coverage.h: Declare the function.
22268 * toplev.c (finalize): Clean if an error has been seen.
22270 2017-04-28 Martin Liska <mliska@suse.cz>
22272 PR gcov-profile/80031
22273 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
22274 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
22275 * gcov.c (read_graph_file): Read just number of blocks.
22276 * profile.c (branch_prob): Do not stream 0 flags per a basic
22279 2017-04-28 Martin Liska <mliska@suse.cz>
22281 * gcov-dump.c (tag_*): Add new argument to declarations.
22282 (dump_gcov_file): Likewise.
22283 (tag_blocks): Add and use new argument depth.
22284 (tag_arcs): Likewise.
22285 (tag_lines): Likewise.
22286 (tag_counters): Likewise.
22287 (tag_summary): Likewise.
22288 (dump_working_sets): Use depth to do a proper indentation.
22290 2017-04-28 Jakub Jelinek <jakub@redhat.com>
22293 * cgraph.h (symtab_node::debug_symtab): No longer inline.
22294 * symtab.c (symtab_node::debug_symtab): Move definition here.
22296 2017-04-28 Richard Biener <rguenther@suse.de>
22298 * lto-streamer.h (LTO_major_version): Bump to 7.
22300 2017-04-28 Richard Biener <rguenther@suse.de>
22302 * tree-vrp.c (assert_info): New struct.
22303 (add_assert_info): New helper.
22304 (register_edge_assert_for_2): Refactor to add asserts to a vector
22306 (register_edge_assert_for_1): Likewise.
22307 (register_edge_assert_for): Likewise.
22308 (finish_register_edge_assert_for): New helper actually registering
22309 asserts where live on edge.
22310 (find_conditional_asserts): Adjust.
22311 (find_switch_asserts): Likewise.
22312 (evrp_dom_walker::try_find_new_range): Generalize.
22313 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
22315 2017-04-27 Marek Polacek <polacek@redhat.com>
22318 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
22319 arg10 and arg11 to itype.
22321 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
22323 * doc/extend.texi (Object Size Checking): Improve grammar.
22325 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
22328 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
22329 that the logic for permitting reciprocal estimates matches that
22332 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22335 * tree.c (type_cache_hasher::equal): Only compare
22336 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
22337 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
22338 non-aggregate element types.
22339 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
22340 about the flag on ARRAY_TYPEs in the comment, formatting fix.
22342 2017-04-27 Richard Biener <rguenther@suse.de>
22344 PR middle-end/80533
22345 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
22346 stripping ARRAY_REFs from MEM_EXPR make sure we're not
22347 keeping a reference to a trailing array.
22349 2017-04-27 Richard Biener <rguenther@suse.de>
22351 PR middle-end/80539
22352 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
22353 being in loop-closed SSA form conservatively.
22354 (chrec_fold_multiply_poly_poly): Likewise.
22356 2017-04-27 Tamar Christina <tamar.christina@arm.com>
22358 PR middle-end/79665
22359 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
22360 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
22362 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22365 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
22366 (aarch64_function_arg_alignment): Return unsigned int again, but still
22367 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
22368 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22369 Don't emit -Wpsabi note.
22370 (aarch64_function_arg_boundary): Likewise.
22371 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22374 2017-04-26 Nathan Sidwell <nathan@acm.org>
22376 * tree.h (crc32_unsigned_n): Declare.
22377 (crc32_unsigned, crc32_unsigned): Make inline.
22378 * tree.c (crc32_unsigned_bits): Replace with ...
22379 (crc32_unsigned_n): ... this.
22380 (crc32_unsigned, crc32_byte): Remove.
22381 (crc32_string): Remove unnecessary braces.
22383 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22385 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
22386 * ipa-inline-analysis.c (MAX_TIME): Remove.
22387 (account_size_time): Use sreal for time.
22388 (dump_inline_summary): Update.
22389 (estimate_function_body_sizes): Update.
22390 (estimate_edge_size_and_time): Update.
22391 (estimate_calls_size_and_time): Update.
22392 (estimate_node_size_and_time): Update.
22393 (inline_merge_summary): Update.
22394 (inline_update_overall_summary): Update.
22395 (estimate_time_after_inlining): Update.
22396 (inline_read_section): Update.
22397 (inline_write_summary): Update.
22398 * ipa-inline.c (compute_uninlined_call_time): Update.
22399 (compute_inlined_call_time): Update.
22400 (recursive_inlining): Update.
22401 (inline_small_functions): Update.
22402 (dump_overall_stats): Update.
22403 * ipa-inline.h: Include sreal.h.
22404 (size_time_entry): Turn time to sreal.
22405 (inline_summary): Turn self_time nad time to sreal.
22407 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22409 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
22411 (sreal::stream_out, sreal::stream_in): New.
22412 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
22414 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22416 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
22419 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
22422 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
22424 (dimode_scalar_chain::compute_convert_gain): Ditto.
22425 (dimode_scalar_chain::make_vector_copies): Ditto.
22426 (dimode_scalar_chain::convert_reg): Ditto.
22427 (dimode_scalar_chain::convert_insn): Ditto.
22428 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
22429 (VI248_AVX512BW_1): New mode iterator.
22430 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
22431 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
22434 2017-04-25 Martin Sebor <msebor@redhat.com>
22436 PR tree-optimization/80497
22437 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
22438 constants are representable in HOST_WIDE_INT.
22439 (parse_directive): Ditto.
22441 2017-04-25 Martin Sebor <msebor@redhat.com>
22444 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
22445 (new_zero_array): Adjust signature.
22446 (dom_info::dom_init): Used unsigned rather that size_t.
22447 (dom_info::dom_info): Same.
22449 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22450 Jakub Jelinek <jakub@redhat.com>
22453 * config/arm/arm.c: Include gimple.h.
22454 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22455 returns negative, increment ncrn only if it returned positive.
22456 (arm_needs_doubleword_align): Return int instead of bool,
22457 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
22458 members, but if there is any such non-FIELD_DECL
22459 > PARM_BOUNDARY aligned decl, return -1 instead of false.
22460 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22461 returns negative, increment nregs only if it returned positive.
22462 (arm_setup_incoming_varargs): Likewise.
22463 (arm_function_arg_boundary): Emit -Wpsabi note if
22464 arm_needs_doubleword_align returns negative, return
22465 DOUBLEWORD_ALIGNMENT only if it returned positive.
22467 2017-04-25 Marek Polacek <polacek@redhat.com>
22470 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22471 first argument to type.
22473 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
22476 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
22477 type checks to test for compatibility instead of equality.
22479 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22480 Jakub Jelinek <jakub@redhat.com>
22483 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
22485 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
22486 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
22487 the alignment computation, but return their maximum in warn_alignment.
22488 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22489 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
22491 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
22492 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22495 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22497 * config/arc/simdext.md (dmpyh): Fix typo.
22499 2017-04-25 Richard Biener <rguenther@suse.de>
22501 PR tree-optimization/80492
22502 * alias.c (compare_base_decls): Handle registers with asm
22503 specification conservatively.
22504 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22505 compare_base_decls returning dont-know properly.
22507 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22509 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
22510 (legitimate_offset_address_p): New function.
22511 (arc_legitimate_address_p): Use above function.
22513 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22515 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
22517 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22519 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
22520 ACCH registers whenever they are available.
22522 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22524 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
22525 double regs fix when not used.
22527 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22529 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
22531 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
22532 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
22534 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22536 * config/arc/arc.c (arc_output_addsi): Check for h-register class
22537 when emitting short ADD instructions.
22539 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22541 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
22543 (cmpsi_cc_c_insn): Likewise.
22544 (cbranchsi4_scratch): Compute proper instruction length using
22545 compact_hreg_operand.
22546 * config/arc/predicates.md (compact_hreg_operand): New predicate.
22548 2017-04-25 Richard Biener <rguenther@suse.de>
22550 PR middle-end/80509
22551 * passes.c (pass_manager::pass_manager): Initialize
22552 m_name_to_pass_map.
22554 2017-04-25 Richard Biener <rguenther@suse.de>
22556 PR tree-optimization/79201
22557 * tree-ssa-sink.c (statement_sink_location): Handle calls.
22559 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22562 * config/s390/vector.md: Split MEM->GPR vector moves for
22563 non-s_operand addresses.
22565 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22568 * config/s390/predicates.md (reload_const_wide_int_operand): New
22570 * config/s390/s390.md ("movti"): Remove d/P alternative.
22571 ("movti_bigconst"): New pattern definition.
22573 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22576 * s390-protos.h (s390_expand_cs_hqi): Removed.
22577 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
22578 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
22579 modes as well as CCZ1mode and CCZmode.
22580 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
22581 signature of s390_emit_compare_and_swap.
22582 (s390_expand_cs_hqi): Likewise, make static.
22583 (s390_expand_cs_tdsi): Generate an explicit compare before trying
22584 compare-and-swap, in some cases.
22585 (s390_expand_cs): Wrapper function.
22586 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
22588 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
22589 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
22590 patterns for small and large integers. Forbid symref memory operands.
22591 Move expander to s390.c. Require cc register.
22592 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
22593 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
22594 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
22595 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
22596 symref memory operands. Remove CC mode and call s390_match_ccmode
22598 ("atomic_exchange<mode>"): Allow and implement all integer modes.
22600 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22602 * config/s390/s390.md (define_peephole2): New peephole to help
22603 combining the load-and-test pattern with volatile memory.
22605 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22607 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
22608 with CCZmode for TARGET_Z196.
22610 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22612 PR rtl-optimization/80501
22613 * combine.c (make_compound_operation_int): Set subreg_code to SET
22614 even for AND with mask of the sign bit of mode.
22616 PR rtl-optimization/80500
22617 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
22618 sum's initial value.
22620 2017-04-25 Julian Brown <julian@codesourcery.com>
22621 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22623 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
22625 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
22627 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
22629 2017-04-25 Julian Brown <julian@codesourcery.com>
22630 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22632 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
22633 (thunderx2t99_sha): New Reservation.
22635 2017-04-25 Julian Brown <julian@codesourcery.com>
22636 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22638 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
22639 type for 1-element load.
22641 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
22643 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
22645 2017-04-24 Martin Jambor <mjambor@suse.cz>
22647 PR tree-optimization/80293
22648 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
22649 char arrays not totally scalarizable if it is false.
22650 (analyze_all_variable_accesses): Pass correct value in the new
22651 parameter. Add a statistics counter.
22653 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
22655 PR middle-end/79931
22656 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
22658 2017-04-24 Richard Biener <rguenther@suse.de>
22660 PR tree-optimization/80494
22661 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
22662 out for complex types.
22664 2017-04-24 Richard Biener <rguenther@suse.de>
22666 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
22667 * tree-ssa-sccvn.c (print_scc): Print SCC size.
22668 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
22669 (DFS): Adjust and never fail.
22670 (sccvn_dom_walker::fail): Remove.
22671 (sccvn_dom_walker::before_dom_children): Adjust.
22672 (run_scc_vn): Likewise and never fail.
22673 * tree-ssa-pre.c (pass_pre::execute): Adjust.
22674 (pass_fre::execute): Likewise.
22676 2017-04-24 Richard Biener <rguenther@suse.de>
22678 PR tree-optimization/79725
22679 * tree-ssa-sink.c (statement_sink_location): Return whether
22680 failure reason was zero uses. Move that check later.
22681 (sink_code_in_bb): Deal with zero uses by removing the stmt
22684 2017-04-24 Richard Biener <rguenther@suse.de>
22687 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
22688 pointer-based references.
22690 2017-04-24 Richard Biener <rguenther@suse.de>
22693 * pass_manager.h (pass_manager::operator new): Remove.
22694 (pass_manager::operator delete): Likewise.
22695 * passes.c (pass_manager::operator new): Remove.
22696 (pass_manager::operator delete): Likewise.
22697 (pass_manager::pass_manager): Zero individual pass members.
22699 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
22702 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
22703 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
22704 Check "XEXP (src, 1)" operand here.
22705 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
22706 Check "XEXP (src, 1)" operand here.
22707 (dimode_scalar_chain::make_vector_copies): Detect count register
22708 of a shift instruction. Zero extend count register from QImode
22709 to DImode to satisfy vector shift pattern count operand predicate.
22710 Substitute vector shift count operand with a DImode copy.
22711 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
22714 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
22716 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
22717 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22718 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
22719 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22720 (UNSPEC_NOREX_MEM): Remove definition.
22722 2017-04-21 Richard Biener <rguenther@suse.de>
22724 PR tree-optimization/79547
22725 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22726 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
22727 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
22728 without any constraints.
22730 2017-04-21 Richard Biener <rguenther@suse.de>
22732 PR tree-optimization/78847
22733 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
22735 2017-04-21 Richard Biener <rguenther@suse.de>
22737 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
22738 (build_distinct_type_copy): Likewise.
22739 (build_variant_type_copy): Likewise.
22740 * tree.c (build_qualified_type): Pass down mem-stat info.
22741 (build_distinct_type_copy): Likewise.
22742 (build_variant_type_copy): Likewise.
22744 2017-04-21 Richard Biener <rguenther@suse.de>
22746 PR tree-optimization/80237
22747 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
22749 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
22750 for a simplified result.
22752 2016-04-21 Richard Biener <rguenther@suse.de>
22754 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
22755 sth as strict as a simple_iv but a chrec without symbols and an
22756 operand defined in the loop we are peeling (and not some subloop).
22757 (propagate_constants_for_unrolling): Propagate all constants.
22759 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22762 * config/i386/i386.c (print_reg): Remove assert for disalowed
22763 regno values, call output_operand_lossage instead.
22765 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22768 * config/i386/constraints.md (Yc): New register constraint.
22769 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
22770 Use Yc constraint for alternative 2 of operand 0. Remove
22771 preferred_for_speed attribute.
22773 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22775 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
22776 lastprivate clauses in SIMT case.
22778 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
22780 * doc/invoke.texi (-Wextra-semi): Document new warning option.
22782 2017-04-20 Richard Biener <rguenther@suse.de>
22784 PR tree-optimization/57796
22785 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
22786 as N scalar stores.
22787 (vect_model_load_cost): Cost gathers as N scalar loads.
22789 2017-04-20 Richard Biener <rguenther@suse.de>
22791 * ggc-page.c (ggc_allocated_p): Rename to ...
22792 (safe_lookup_page_table_entry): ... this and return the lookup
22794 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
22796 2017-04-20 Richard Biener <rguenther@suse.de>
22798 PR tree-optimization/80453
22799 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
22800 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
22801 from the conditions.
22802 (vn_phi_eq): Pass them down.
22803 (vn_phi_lookup): Record them.
22804 (vn_phi_insert): Likewise.
22806 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
22808 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
22809 uninitialized variable warning to avoid buffer overrun.
22811 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22814 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
22815 is suppressed for '{ 0 }' in C.
22817 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22819 * BASE-VER: Set to 8.0.0.
22821 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
22823 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
22824 priority .init_array and .fini_array section with SECTION_NOTYPE
22827 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22829 PR middle-end/80423
22830 * tree.h (build_array_type): Add typeless_storage default argument.
22831 * tree.c (type_cache_hasher::equal): Also compare
22832 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
22833 (build_array_type): Add typeless_storage argument, set
22834 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
22836 (build_nonshared_array_type): Adjust build_array_type_1 caller.
22837 (build_array_type): Likewise. Add typeless_storage argument.
22839 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22840 Jakub Jelinek <jakub@redhat.com>
22842 PR tree-optimization/80426
22843 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
22844 operation on symbolic operands, also compute the overflow for the
22845 invariant part when the operation degenerates into a negation.
22847 2017-04-19 Jakub Jelinek <jakub@redhat.com>
22850 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
22851 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
22854 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
22856 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
22859 * config/avr/avr.c (tree.h): Include it.
22860 (cgraph.h): Include it.
22861 (avr_encode_section_info): Don't warn for uninitialized progmem
22862 variable if it's just an alias.
22864 2017-04-19 Richard Biener <rguenther@suse.de>
22867 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
22868 when needed by AutoPGO.
22870 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
22873 * doc/lto.texi: Remove an extra 'that'.
22875 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
22877 PR rtl-optimization/80429
22878 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
22879 are only used in debug insns.
22881 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22882 Vladimir Makarov <vmakarov@redhat.com>
22884 * config/sparc/predicates.md (input_operand): Add comment. Return
22885 true for any memory operand when LRA is in progress.
22886 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
22888 2017-04-18 Jeff Law <law@redhat.com>
22891 * mips.md ({return,simple_return}_internal): Do not overwrite
22894 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22896 PR tree-optimization/80443
22897 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
22898 instead of adding 1, subtract -1 and similarly instead of subtracting
22901 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
22903 PR rtl-optimization/80357
22904 * haifa-sched.c (tmp_bitmap): New variable.
22905 (model_recompute): Handle duplicate use records.
22906 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
22907 (free_global_sched_pressure_data): Free it.
22909 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
22912 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22913 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22914 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22915 instead of SYSTEM_HEADER_DIR.
22917 2017-04-18 Jeff Law <law@redhat.com>
22919 PR middle-end/80422
22920 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
22921 predecessors after walking up the insn chain.
22923 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22926 * dwarf2out.c (modified_type_die): Try harder not to emit internal
22927 sizetype type into debug info.
22929 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22932 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
22933 unneeded test for TARGET_UPPER_REGS_SF.
22934 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
22936 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22939 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
22940 instead of gsi_after_labels.
22942 2017-04-18 Jeff Law <law@redhat.com>
22944 * regcprop.c (maybe_mode_change): Avoid creating copies of the
22948 2017-04-13 Jeff Law <law@redhat.com>
22949 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22950 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22952 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
22955 * config/avr/avr.c (intl.h): Include it.
22956 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
22958 2017-04-18 Martin Liska <mliska@suse.cz>
22960 PR gcov-profile/78783
22961 * gcov-tool.c (gcov_output_files): Validate that destination
22962 file is either removed by the tool or by a user.
22964 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22965 Guy Benyei <guybe@mellanox.com>
22967 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
22968 block, and do not negate it, the stored id is already negative.
22970 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22972 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
22974 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22977 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
22978 masks of options that should be turned off if the VSX vector
22979 options are turned off.
22980 (OTHER_P8_VECTOR_MASKS): Likewise.
22981 (OTHER_VSX_VECTOR_MASKS): Likewise.
22982 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
22983 rs6000_disable_incompatible_switches to validate no type switches
22985 (rs6000_incompatible_switch): New function to disallow turning on
22986 other vector options if -mno-vsx, -mno-power8-vector, or
22987 -mno-power9-vector are specified.
22989 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22991 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
22993 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22995 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
22996 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
22997 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
22998 (ARG_POINTER_CFA_OFFSET): Likewise.
23000 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23002 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
23003 conditions to take advantage of various optimizations.
23005 2017-04-13 Jeff Law <law@redhat.com>
23007 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23008 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23009 (zero_extendsidi2_dext): Likewise.
23011 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23014 * fold-const.c (fold_ternary_loc): Revert
23015 use op0 instead of fold_convert_loc (loc, type, arg0) part of
23018 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
23020 PR rtl-optimization/80343
23021 * lra-remat.c (update_scratch_ops): Assign original hard reg to
23022 new scratch pseudo.
23024 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
23027 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
23028 to ubsan_encode_value.
23030 2017-04-13 Jeff Law <law@redhat.com>
23032 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23033 appearing in DEBUG_INSNs.
23035 2017-04-13 Martin Liska <mliska@suse.cz>
23037 PR gcov-profile/80413
23038 * gcov-io.c (gcov_write_string): Copy to buffer just when
23039 allocated size is greater than zero.
23041 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23044 * dwarf2out.c (decls_for_scope): Ignore declarations of
23045 current_function_decl in BLOCK_NONLOCALIZED_VARS.
23047 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
23050 * ipa-visibility.c (non_local_p): Fix typos.
23051 (localize_node): When localizing symbol in same comdat group,
23052 dissolve the group only when we know external symbols are going
23054 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
23056 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23058 PR tree-optimization/79390
23059 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
23060 order does not result in usable sequence, retry with reversed operand
23066 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23067 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23068 op0 instead of fold_convert_loc (loc, type, arg0).
23070 2017-04-12 Jeff Law <law@redhat.com>
23072 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23073 has a delay slot in the generated code.
23075 * config/cris/cris.md (cris_preferred_reload_class): Return
23076 GENNONACR_REGS rather than GENERAL_REGS.
23078 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23081 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23082 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23083 signedness of the result type.
23085 2017-04-12 Richard Biener <rguenther@suse.de>
23086 Jeff Law <law@redhat.com>
23088 PR tree-optimization/80359
23089 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23090 trim stores to TARGET_MEM_REFs.
23092 2017-04-12 Richard Biener <rguenther@suse.de>
23094 PR tree-optimization/79390
23095 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23096 threading case even more.
23098 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23101 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23102 for quad_address_p for TImode, instead of just not indexed_address.
23104 2017-04-12 Richard Biener <rguenther@suse.de>
23105 Bernd Edlinger <bernd.edlinger@hotmail.de>
23107 PR middle-end/79671
23108 * alias.c (component_uses_parent_alias_set_from): Handle
23109 TYPE_TYPELESS_STORAGE.
23110 (get_alias_set): Likewise.
23111 * tree-core.h (tree_type_common): Add typeless_storage flag.
23112 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23113 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23114 for types containing members with TYPE_TYPELESS_STORAGE.
23115 (place_field): Likewise.
23116 (layout_type): Likewise for ARRAY_TYPE.
23117 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23118 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23119 TYPE_TYPELESS_STORAGE.
23120 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23122 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23125 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23126 first argument to type.
23128 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23132 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23133 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23134 (rs6000_expand_binop_builtin): Likewise.
23135 (rs6000_expand_ternop_builtin): Likewise; also add missing
23136 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23137 vshasigma built-ins.
23138 * doc/extend.texi: Document that vec_xxpermdi's third argument
23139 must be a constant.
23141 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23143 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
23144 Use shift_const cost parameter when calculating gain of STV shifts.
23146 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
23148 PR rtl-optimization/70478
23149 * lra-constraints.c (process_alt_operands): Check memory for
23150 disfavoring memory insn operand.
23152 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23154 PR middle-end/80100
23155 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
23156 left shift in unsigned HOST_WIDE_INT type.
23158 PR rtl-optimization/80385
23159 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
23160 (not (neg X)) into (plus X -1) for complex or non-integral modes.
23163 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
23164 if they have any depend clauses.
23166 2017-04-11 Martin Liska <mliska@suse.cz>
23169 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
23170 * ipa-split.c (split_function): Create a local comdat symbol
23171 if caller is in a comdat group.
23173 2017-04-11 Martin Liska <mliska@suse.cz>
23176 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
23179 2017-04-11 Martin Sebor <msebor@redhat.com>
23181 PR middle-end/80364
23182 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
23183 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
23185 (directive::set_width, directive::set_precision, format_character):
23187 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
23190 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
23193 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
23194 conflict, set target->arch_name instead of target->cpu_name.
23196 2017-04-11 Richard Biener <rguenther@suse.de>
23198 PR tree-optimization/80374
23199 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
23200 build_zero_cst, remove fold_convertible_p check again.
23202 2017-04-11 Martin Liska <mliska@suse.cz>
23205 * ubsan.c (instrument_object_size): Do not instrument register
23208 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23211 * config/i386/i386-builtin-types.def
23212 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
23213 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
23214 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
23215 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
23216 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
23217 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
23218 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
23219 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
23220 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
23221 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
23222 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
23223 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
23224 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
23225 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
23226 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
23227 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
23228 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
23229 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
23230 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
23231 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
23232 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
23233 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
23234 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
23235 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
23236 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
23237 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
23238 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
23239 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
23240 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
23241 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
23242 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
23243 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
23244 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
23245 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
23246 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
23247 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
23248 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
23249 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
23250 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
23251 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
23252 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
23253 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
23254 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
23255 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
23256 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
23257 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
23259 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
23260 flag to second_arg_count, handle 4 argument function type _COUNT
23261 aliases, handle second_arg_count on second argument rather than last.
23263 2017-04-10 Jeff Law <law@redhat.com>
23265 PR tree-optimization/80374
23266 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
23267 record anything if we can not convert integer_zero_node to the
23270 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23273 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23274 Enhance special handling given to the TARGET_P9_MINMAX option in
23275 relation to certain other options.
23277 2017-04-10 Bin Cheng <bin.cheng@arm.com>
23279 PR tree-optimization/80153
23280 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
23281 remove POINTER_PLUS_EXPR's base part directly, rather than through
23284 2017-04-10 Richard Biener <rguenther@suse.de>
23285 Bin Cheng <bin.cheng@arm.com>
23287 PR tree-optimization/80153
23288 * tree-affine.c (aff_combination_to_tree): Get base pointer from
23289 the first element of pointer type aff_tree. Build result expr in
23291 (add_elt_to_tree): Convert to type unconditionally. Remove other
23292 fold_convert calls.
23293 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
23294 (rewrite_use_nonlinear_expr): Check invariant using iv information.
23296 2017-04-10 Richard Biener <rguenther@suse.de>
23298 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23301 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
23303 PR rtl-optimization/70478
23304 * lra-constraints.c (curr_small_class_check): New.
23305 (update_and_check_small_class_inputs): New.
23306 (process_alt_operands): Update curr_small_class_check. Disfavor
23307 alternative insn memory operands. Check available regs for small
23310 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23313 * config/mips/mips.opt (-mvirt): Update description.
23314 * doc/invoke.texi (-mvirt): Likewise.
23316 2017-04-10 Richard Biener <rguenther@suse.de>
23318 PR middle-end/80362
23319 * fold-const.c (fold_binary_loc): Look at unstripped ops when
23320 looking for NEGATE_EXPR in -A / -B to A / B folding.
23322 2017-04-10 Martin Liska <mliska@suse.cz>
23324 PR gcov-profile/80224
23325 * gcov.c (print_usage): Fix usage string.
23326 (get_gcov_intermediate_filename): Remove.
23327 (output_gcov_file): Use both for normal and intermediate format.
23328 (generate_results): Do not initialize special file for
23329 intermediate format.
23331 2017-04-10 Richard Biener <rguenther@suse.de>
23333 PR tree-optimization/80304
23334 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
23337 2017-04-10 Nathan Sidwell <nathan@acm.org>
23340 * config/rs6000/rs6000.c (rs6000_vector_type): New.
23341 (rs6000_init_builtins): Use it.
23343 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23345 * config/arm/arm.md (<mrc>): Add mode to SET source.
23346 (<mrrc>): Likewise.
23348 2017-04-10 Richard Biener <rguenther@suse.de>
23350 PR middle-end/80344
23351 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
23353 2017-04-10 Jakub Jelinek <jakub@redhat.com>
23356 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
23357 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
23358 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
23359 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
23360 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
23361 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
23362 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
23363 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
23364 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
23365 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
23366 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
23367 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
23368 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
23369 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
23370 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
23371 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
23372 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
23373 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
23374 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
23375 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
23376 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
23377 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
23378 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
23380 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
23382 PR rtl-optimization/70478
23383 * lra-constraints.c: Reverse the last patch.
23385 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
23387 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
23388 Add comment for WCHAR_T.
23390 2017-04-08 Martin Liska <mliska@suse.cz>
23393 2017-04-07 Martin Liska <mliska@suse.cz>
23396 * ipa-split.c (split_function): Add function part to a same comdat
23399 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23402 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
23404 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
23406 * rs6000/rs6000.c (vec_load_pendulum): Rename...
23407 (vec_pairing): ...to this.
23408 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
23409 (rs6000_sched_init): Adjust for name change.
23410 (struct rs6000_sched_context): Likewise.
23411 (rs6000_init_sched_context): Likewise.
23412 (rs6000_set_sched_context): Likewise.
23414 2017-04-07 Jakub Jelinek <jakub@redhat.com>
23420 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
23422 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
23423 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
23424 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
23426 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
23428 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
23430 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23432 PR rtl-optimization/70703
23433 * ira-color.c (update_conflict_hard_regno_costs): Use
23434 int64_t instead of HOST_WIDE_INT.
23436 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23438 PR rtl-optimization/70478
23439 * lra-constraints.c (process_alt_operands): Disfavor alternative
23440 insn memory operands.
23442 2017-04-07 Jeff Law <law@redhat.com>
23444 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
23445 CALL and NOTE_INSN_CALL_ARG_LOCATION.
23447 2017-04-07 Martin Liska <mliska@suse.cz>
23450 * config/aarch64/aarch64.c (aarch64_process_target_attr):
23451 Show error message instead of an ICE.
23453 2017-04-07 Martin Liska <mliska@suse.cz>
23456 * ipa-split.c (split_function): Add function part to a same comdat
23459 2017-04-07 Richard Biener <rguenther@suse.de>
23461 PR middle-end/80341
23462 * tree.c (get_unwidened): Also handle ! for_type case for
23464 * convert.c (do_narrow): Split out from ...
23465 (convert_to_integer_1): ... here. Do not pass final truncation
23466 type to get_unwidened for TRUNC_DIV_EXPR.
23468 2017-04-07 Richard Biener <rguenther@suse.de>
23470 * tree-affine.c (wide_int_ext_for_comb): Take type rather
23472 (aff_combination_const): Adjust.
23473 (aff_combination_scale): Likewise.
23474 (aff_combination_add_elt): Likewise.
23475 (aff_combination_add_cst): Likewise.
23476 (aff_combination_convert): Likewise.
23477 (add_elt_to_tree): Likewise. Remove unused argument.
23478 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
23480 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
23482 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
23484 * config/arm/arm.c (arm_default_short_enums): Use
23485 ARM_DEFAULT_SHORT_ENUMS.
23486 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
23488 2017-04-06 Jakub Jelinek <jakub@redhat.com>
23491 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
23492 members with redundant out-of-class redeclaration.
23494 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23497 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
23498 * config/i386/i386.md (*zero_extendsidi2):
23499 Add (?*x,*x) and (?*v,*v) alternatives.
23501 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23504 * config/i386/i386.c (ix86_expand_builtin)
23505 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
23506 mode from insn data. Convert operands to insn operand mode.
23507 Copy operands that don't satisfy insn predicate to a register.
23509 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
23511 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
23514 2017-04-06 Richard Biener <rguenther@suse.de>
23516 PR tree-optimization/80334
23517 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
23518 preserve alignment of accesses.
23520 2017-04-06 Richard Biener <rguenther@suse.de>
23522 PR tree-optimization/80262
23523 * tree-sra.c (build_ref_for_offset): Preserve address-space
23525 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
23526 Drop useless address-space information on MEM_REF offsets.
23528 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
23530 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
23532 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23534 PR rtl-optimization/70703
23535 * ira-color.c (update_conflict_hard_regno_costs): Use
23536 HOST_WIDE_INT instead of long.
23538 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
23541 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
23542 not defined for x86_64 target. Add -mmmx target option when __SSE2__
23544 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
23545 for x86_64 target. Handle -m3dnowa option.
23547 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23549 PR rtl-optimization/70703
23550 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
23551 (update_conflict_hard_regno_costs): Use long instead of unsigned
23552 arithmetic for cost calculation.
23554 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23555 Bernd Edlinger <bernd.edlinger@hotmail.de>
23558 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
23561 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
23564 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
23565 ptr_mode with Pmode throughout.
23566 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
23567 into probe_stack_range and use DImode.
23569 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
23572 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
23573 call_eh_return is true.
23575 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23577 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
23578 Initialize last_match_fntype_index.
23580 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23583 * tree-nvr.c: Include internal-fn.h.
23584 (pass_return_slot::execute): Ignore internal calls without
23587 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23588 Richard Biener <rguenther@suse.de>
23591 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
23592 captures used multiple times, except for the last use.
23593 * generic-match-head.c: Include gimplify.h.
23595 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23597 PR tree-optimization/79390
23598 * target.h (struct noce_if_info): Declare.
23599 * targhooks.h (default_noce_conversion_profitable_p): Declare.
23600 * target.def (noce_conversion_profitable_p): New target hook.
23601 * ifcvt.h (struct noce_if_info): New type, moved from ...
23602 * ifcvt.c (struct noce_if_info): ... here.
23603 (noce_conversion_profitable_p): Renamed to ...
23604 (default_noce_conversion_profitable_p): ... this. No longer
23606 (noce_try_store_flag_constants, noce_try_addcc,
23607 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
23608 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
23609 instead of noce_conversion_profitable_p.
23610 * config/i386/i386.c: Include ifcvt.h.
23611 (ix86_option_override_internal): Don't override
23612 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
23613 (ix86_noce_conversion_profitable_p): New function.
23614 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
23615 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
23616 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
23617 * doc/tm.texi: Regenerated.
23619 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23621 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
23624 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23627 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
23628 instructions for small multiply cores.
23630 2017-04-04 Jeff Law <law@redhat.com>
23632 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
23634 (mips_expand_vec_perm_const): Initialize elements in orig_perm
23635 that are not set by the loop over the elements.
23637 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23640 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
23641 int mode, convert_modes it to mode as unsigned, otherwise use
23642 lowpart_subreg to mode rather than SImode.
23643 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
23644 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
23645 Use DImode instead of SImode for the shift count operand.
23646 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
23649 2017-04-04 Richard Biener <rguenther@suse.de>
23651 PR middle-end/80281
23652 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
23653 arithmetic done for the negate or the plus. Simplify.
23654 (A - (-B) -> A + B): Likewise.
23655 * fold-const.c (split_tree): Make sure to not negate pointers.
23657 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
23659 PR rtl-optimization/60818
23660 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
23661 a compare of comparisons with the thing compared if this results
23662 in a different machine mode.
23664 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
23666 * alias.c (base_alias_check): Fix typo in comment.
23667 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
23668 * cgraphunit.c (symbol_table::compile): Likewise.
23669 * collect2.c (maybe_run_lto_and_relink): Likewise.
23670 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
23671 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
23672 * config/avr/avr.c (avr_map_op_t): Likewise.
23673 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
23674 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
23675 * config/epiphany/epiphany.md (movcc): Likewise.
23676 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
23677 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
23679 * config/mips/mips.c (mips_save_restore_reg): Likewise.
23680 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
23681 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
23682 * config/sh/sh.c (sh_rtx_costs): Likewise.
23683 * fold-const.c (fold_truth_andor): Likewise.
23684 * genautomata.c (collapse_flag): Likewise.
23685 * gengtype.h (struct type::u::s): Likewise.
23686 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
23687 * input.c (FORMAT_AMOUNT): Likewise.
23688 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
23689 (known_aggs_to_agg_replacement_list): Likewise.
23690 * ipa-inline-analysis.c: Likewise.
23691 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
23692 * ipa-polymorphic-call.c
23693 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
23694 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
23695 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
23697 * modulo-sched.c (apply_reg_moves): Likewise.
23698 * omp-expand.c (build_omp_regions_1): Likewise.
23699 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
23700 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
23701 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
23702 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
23703 * value-prof.c: Likewise.
23704 * var-tracking.c (val_reset): Likewise.
23706 2017-04-03 Richard Biener <rguenther@suse.de>
23708 PR tree-optimization/80275
23709 * fold-const.c (split_address_to_core_and_offset): Handle
23712 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
23714 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
23715 descriptors is at least equal to that of functions.
23717 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23719 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
23721 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23724 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
23725 (mov<IMOD4:mode>): New expander.
23726 (*mov<IMOD4:mode>_internal): New insn and split pattern.
23728 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
23730 PR rtl-optimization/79405
23731 * fwprop.c (propagations_left): New variable.
23732 (forward_propagate_into): Decrement it.
23733 (fwprop_init): Initialize it.
23734 (fw_prop): If the variable has reached zero, stop propagating.
23735 (fwprop_addr): Ditto.
23737 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23740 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
23741 a FUNCTION_DECL, pass it as decl instead of origin to
23744 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
23746 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
23749 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
23752 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
23753 TARGET_VSX_SMALL_INTEGER.
23755 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23757 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23758 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
23760 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23762 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
23763 extraction from odd-numbered MSA register.
23765 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23767 PR middle-end/80173
23768 * expmed.c (store_bit_field_1): Don't attempt to create
23769 a word subreg out of hard registers wider than word if they
23770 have HARD_REGNO_NREGS of 1 for their mode.
23772 PR middle-end/80163
23773 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
23774 conversions to integer types wider than word and pointer.
23777 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
23778 (rtx_equal_for_cselib_p): Pass 0 to it.
23779 * cselib.c (cselib_hasher::equal): Likewise.
23780 (rtx_equal_for_cselib_1): Add depth argument. If depth
23781 is 128, don't look up VALUE locs and punt. Increment
23782 depth in recursive calls when walking VALUE locs.
23784 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
23786 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
23787 (make_gcov_file_name): Use the canonical path name for generating
23789 (read_line): Fix handling of files with ascii null bytes.
23791 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
23793 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
23794 to initialise a vector register instead
23795 of using a const_int.
23797 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23799 PR translation/80189
23800 * gimplify.c (omp_default_clause): Use %qs instead of %s in
23801 diagnostic messages.
23803 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
23806 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
23807 (dfp_diex_<mode>): Update mode of operand 1.
23808 * doc/extend.texi (dxex, dxexq): Document change to return type.
23809 (diex, diexq): Document change to argument type.
23811 2017-03-30 Martin Jambor <mjambor@suse.cz>
23814 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
23815 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
23816 it reflects the signature changes performed at the callee side.
23817 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
23818 to cgraph_build_function_type_skip_args.
23819 (build_function_decl_skip_args): Adjust call to the above function.
23821 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23824 * config/i386/sse.md
23825 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
23826 register as dest whenever it is a MEM not rtx_equal_p to the
23827 corresponding dup operand, and when forcing into reg move the
23828 reg into the memory afterwards.
23829 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
23830 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
23831 for the force_reg mode.
23832 (avx512vl_vextractf128<mode>): Use register as dest either
23833 always when a MEM, or when it is a MEM not rtx_equal_p to the
23834 corresponding dup operand, or even not when it is a CONST_VECTOR
23835 depending on the mode and lo vs. hi.
23836 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
23838 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
23839 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
23840 Likewise. Require that operands[2] is even.
23841 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
23842 Remove extraneous parens. Require that operands[2] is a multiple
23844 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
23845 operands[0] is a MEM if <mask_applied>, the predicates/constraints
23846 disallow memory then.
23848 2017-03-30 Richard Biener <rguenther@suse.de>
23850 PR tree-optimization/77498
23851 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
23852 to non-constants over backedges.
23854 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
23856 PR rtl-optimization/80233
23857 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
23858 as last_combined_insn. Do not test for BARRIER_P separately.
23860 2017-03-29 Andreas Schwab <schwab@suse.de>
23863 * calls.c (prepare_call_address): Convert funexp to Pmode before
23864 copying to temp reg.
23866 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23868 PR tree-optimization/80158
23869 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
23870 Handle possible future case of more than one alternate
23872 (replace_rhs_if_not_dup): Likewise.
23873 (replace_one_candidate): Likewise.
23875 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
23877 PR rtl-optimization/80193
23878 * ira.c (ira): Do not check allocation for LRA.
23880 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
23882 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
23883 (nvptx_output_simt_exit): Declare.
23884 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
23885 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
23886 (init_softstack_frame): Move initialization of crtl->is_leaf to...
23887 (nvptx_declare_function_name): ...here. Emit declaration of local
23888 memory space buffer for omp_simt_enter insn.
23889 (nvptx_output_unisimt_switch): New.
23890 (nvptx_output_softstack_switch): New.
23891 (nvptx_output_simt_enter): New.
23892 (nvptx_output_simt_exit): New.
23893 * config/nvptx/nvptx.h (struct machine_function): New fields
23894 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
23895 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
23896 (UNSPECV_SIMT_EXIT): Ditto.
23897 (omp_simt_enter_insn): New insn.
23898 (omp_simt_enter): New expansion.
23899 (omp_simt_exit): New insn.
23900 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
23902 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
23903 (expand_GOMP_SIMT_ENTER_ALLOC): New.
23904 (expand_GOMP_SIMT_EXIT): New.
23905 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
23906 (GOMP_SIMT_ENTER_ALLOC): Ditto.
23907 (GOMP_SIMT_EXIT): Ditto.
23908 * target-insns.def (omp_simt_enter): New insn.
23909 (omp_simt_exit): Ditto.
23910 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
23912 (lower_rec_simd_input_clauses): Implement SIMT privatization.
23913 (lower_rec_input_clauses): Likewise.
23914 (lower_lastprivate_clauses): Handle SIMT privatization.
23916 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
23917 (ompdevlow_adjust_simt_enter): New.
23918 (find_simtpriv_var_op): New.
23919 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
23920 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
23922 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
23923 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
23924 (copy_decl_for_dup_finish): Ditto.
23926 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
23928 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
23931 * config/i386/i386.c (ix86_option_override_internal): Always
23932 allow -mpreferred-stack-boundary=3 for 64-bit targets.
23934 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23936 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
23938 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23940 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
23941 mark new edge's irreducible flag accordign to it.
23942 (vect_do_peeling): Check loop preheader edge's irreducible flag
23943 and pass it to function slpeel_add_loop_guard.
23945 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
23947 PR tree-optimization/80218
23948 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23949 Update block frequencies and counts.
23951 2017-03-28 Richard Biener <rguenther@suse.de>
23953 PR tree-optimization/78644
23954 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
23955 of a simplification result we may not use it at all.
23957 2017-03-28 Richard Biener <rguenther@suse.de>
23960 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
23961 without arguments, generate default definition of a SSA name.
23963 2017-03-28 Richard Biener <rguenther@suse.de>
23965 PR middle-end/80222
23966 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
23967 TYPE_REF_CAN_ALIAS_ALL references.
23968 * fold-const.c (fold_indirect_ref_1): Likewise.
23970 2017-03-28 Martin Liska <mliska@suse.cz>
23973 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
23974 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
23976 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23977 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
23979 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
23980 (EXTRA_SPECS): Define.
23981 (SUBTARGET_EXTRA_SPECS): Likewise.
23982 (SUBTARGET_CPP_SPEC): Likewise.
23983 * config/arc/elf.h (EXTRA_SPECS): Renamed to
23984 SUBTARGET_EXTRA_SPECS.
23985 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
23987 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23989 * config/arc/simdext.md (vst64_insn): Update pattern.
23990 (vld32wh_insn): Likewise.
23991 (vld32wl_insn): Likewise.
23992 (vld64_insn): Likewise.
23993 (vld32_insn): Likewise.
23995 2017-03-28 Marek Polacek <polacek@redhat.com>
23998 * fold-const.c (fold_comparison): Use protected_set_expr_location
23999 instead of SET_EXPR_LOCATION.
24001 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
24003 * tree.c (add_expr): Avoid name lookup warning.
24005 2017-03-27 Jeff Law <law@redhat.com>
24007 PR tree-optimization/80216
24008 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
24009 function name. Limit recursion depth.
24010 (record_temporary_equivalences): Corresponding changes.
24012 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
24014 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
24017 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24020 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
24022 * cfgcleanup.c (reg_note_cfa_p): New array.
24023 (insns_have_identical_cfa_notes): New function.
24024 (old_insns_match_p): Don't cross-jump in between /f
24025 and non-/f instructions. If both i1 and i2 are frame related,
24026 verify all CFA notes, their order and content.
24028 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24031 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
24032 HImode and SImode with zero extend to DImode to one insn.
24033 (bswap<mode>2_extenddi): Likewise.
24034 (bswapsi2_extenddi): Likewise.
24035 (bswaphi2_extendsi): Likewise.
24036 (bswaphi2): Combine bswap HImode and SImode into one insn.
24037 Separate memory insns from swapping register.
24038 (bswapsi2): Likewise.
24039 (bswap<mode>2): Likewise.
24040 (bswaphi2_internal): Delete, no longer used.
24041 (bswapsi2_internal): Likewise.
24042 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
24043 store, and gpr<-gpr swap insns.
24044 (bswap<mode>2_store): Likewise.
24045 (bswaphi2_reg): Register only splitter, combine with the splitter.
24046 (bswaphi2 splitter): Likewise.
24047 (bswapsi2_reg): Likewise.
24048 (bswapsi2 splitter): Likewise.
24049 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
24050 the insns into load, store, and register/register insns.
24051 (bswapdi2_ldbrx): Likewise.
24052 (bswapdi2_load): Likewise.
24053 (bswapdi2_store): Likewise.
24054 (bswapdi2_reg): Likewise.
24056 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
24058 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
24059 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
24061 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
24064 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24066 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24067 special handling for target option conflicts between dform
24068 options (-mpower9-dform, -mpower9-dform-vector,
24069 -mpower9-dform-scalar) and -mno-direct-move.
24071 2017-03-27 Richard Biener <rguenther@suse.de>
24073 PR tree-optimization/80181
24074 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24076 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24078 * config/arc/predicates.md (move_double_src_operand): Replace the
24079 call to move_double_src_operand with a call to address_operand.
24081 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24083 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24084 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24085 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24087 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24089 * config/arc/predicates.md (long_immediate_loadstore_operand):
24090 Consider scaled addresses cases.
24092 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24094 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24095 restored when in interrupt.
24096 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24097 doesn't have delay slot.
24099 2017-03-27 Richard Biener <rguenther@suse.de>
24102 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24103 inlined thunk clones.
24105 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24108 * asan.c (instrument_derefs): Copy over last operand from
24109 original COMPONENT_REF to the new COMPONENT_REF with
24110 DECL_BIT_FIELD_REPRESENTATIVE.
24111 * ubsan.c (instrument_object_size): Likewise.
24113 2017-03-27 Richard Biener <rguenther@suse.de>
24115 PR tree-optimization/80170
24116 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24117 sure DR/SCEV didnt fold in constants we do not see when looking
24118 at the reference base alignment.
24120 2017-03-27 Richard Biener <rguenther@suse.de>
24122 PR middle-end/80171
24123 * gimple-fold.c (fold_ctor_reference): Properly guard against
24124 NULL return value from canonicalize_constructor_val.
24126 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24129 * config/i386/i386.c (ix86_expand_builtin)
24130 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24131 flags reg setting and flags reg using instructions.
24132 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24133 clobbering instructions to zero extend op2.
24135 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24137 * doc/install.texi (Configuration) <--with-aix-soname>:
24138 Update link to AIX ld.
24140 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24142 PR rtl-optimization/80160
24143 PR rtl-optimization/80159
24144 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
24145 reg_alternate_class into account.
24147 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
24150 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
24151 to consider in curr_insn_transform.
24153 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24155 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
24156 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
24157 and emit_mode_inner.
24159 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24161 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
24162 argument to the overloaded builtin variants. Use the new flag to
24163 deprecate certain builtin variants.
24164 * config/s390/s390-builtin-types.def: Add new builtin types.
24165 * config/s390/s390-builtins.h: Support new flags field for
24166 overloaded builtins.
24167 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
24168 (s390_macro_to_expand): Enable vector float data type.
24169 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24170 builtins by incrementing the __VEC__ version number.
24171 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
24173 (s390_resolve_overloaded_builtin): Emit error messages depending
24174 on the builtin flags.
24175 * config/s390/s390.c (s390_expand_builtin): Support additional
24176 flags argument. Change error message to match the messages
24177 emitted in s390-c.c.
24178 * config/s390/s390.md: New UNSPEC_* constants.
24179 (op_type): Add new instruction types.
24180 * config/s390/vecintrin.h: Add new builtins and test data class
24182 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
24183 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
24184 (VEC_INEXACT, VEC_NOINEXACT): New constants.
24185 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
24186 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
24187 ("vec_mergel<mode>"): V_HW -> VEC_HW.
24189 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
24190 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
24191 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
24192 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
24194 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
24195 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
24196 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
24197 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
24199 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
24200 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
24201 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
24202 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
24203 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
24204 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
24205 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
24207 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
24208 ("vec_scatter_element<V_HW_4:mode>_DI")
24209 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
24210 ("vec_fpint<mode>", "vflls")
24211 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
24212 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
24213 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
24214 ("*vec_cmphe<mode>_cc"): ... these.
24216 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
24217 mode constant instead of magic value.
24219 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24221 * config/s390/s390.c (s390_expand_vec_compare): Support other
24222 vector floating point modes than just V2DF.
24223 (s390_expand_vcond): Likewise.
24224 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
24225 (s390_cannot_change_mode_class): Prevent mode changes between TF
24226 and V1TF in vector registers.
24227 * config/s390/s390.md (DF, SF): New mode attributes.
24228 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
24229 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
24230 SFmode support for VRs.
24231 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
24233 (VFT, VF_HW): New mode iterators.
24234 (vw, sdx): New mode attributes.
24235 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
24236 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
24237 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
24238 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
24239 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
24240 also the new vector floating point modes. Renaming to ...
24242 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
24243 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
24244 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
24245 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
24246 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
24247 ("vec_unordered<mode>"): ... these.
24249 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
24250 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
24251 ("*vec_extendv2df"): New insn definitions.
24253 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24255 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
24256 ("mulditi3_2", "*muldi3_sign"): New patterns.
24257 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
24258 rename the pattern definition.
24260 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24262 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
24264 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
24266 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24268 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
24269 instruction if possible.
24270 * config/s390/vector.md (vec_halfnumelts): New mode
24272 ("*vec_vllezlf<mode>"): New pattern.
24274 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24276 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
24277 ("popcountv4si2", "popcountv2di2"): Rename to ...
24278 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
24279 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
24281 ("popcount<mode>2_vxe"): New pattern.
24283 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24285 * common/config/s390/s390-common.c (processor_flags_table): Add
24287 * config.gcc: Add arch12.
24288 * config/s390/driver-native.c (s390_host_detect_local_cpu):
24289 Default to arch12 for unknown CPU model numbers.
24290 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
24291 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
24292 PROCESSOR_max sanity check.
24293 * config/s390/s390-opts.h (enum processor_type): Add
24295 * config/s390/s390.c (processor_table): Add arch12.
24296 (s390_expand_builtin): Add check for B_VXE flag.
24297 (s390_issue_rate): Add PROCESSOR_ARCH12.
24298 (s390_get_sched_attrmask): Likewise.
24299 (s390_get_unit_mask): Likewise.
24300 (s390_sched_score): Enable z13 scheduling for arch12.
24301 (s390_sched_reorder): Likewise.
24302 (s390_sched_variable_issue): Likewise.
24303 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
24305 (s390_tune_attr): Use z13 scheduling also for arch12.
24306 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
24307 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
24308 (TARGET_VXE_P): New macros.
24309 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
24310 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
24311 * config/s390/s390.opt: Add arch12 as processor_type.
24313 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24315 * config/s390/s390.md
24316 ("fixuns_truncdddi2", "fixuns_trunctddi2")
24317 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
24318 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
24320 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
24321 Rename expanders to ...
24323 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
24324 ("fixuns_truncdddi2_emu"): ... these.
24326 ("fixuns_trunc<mode>si2_emu"): New expander.
24328 ("*fixuns_truncdfdi2_z13"): Rename to ...
24329 ("*fixuns_truncdfdi2_vx"): ... this.
24331 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24333 * config/s390/2964.md: Remove the single element vector compare
24334 instructions which are no longer used.
24335 * config/s390/s390.c (s390_select_ccmode): Remove handling of
24337 (s390_canonicalize_comparison): Remove handling of DFmode
24339 (s390_expand_vec_compare_scalar): Remove function.
24340 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
24341 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
24343 ("*cmp<mode>_ccs"): Add wfcdb instruction.
24345 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24347 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
24349 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
24350 will anyway by matched by mov<mode>_64dfp.
24352 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24354 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
24355 vlef/vstef. Add missing operand to vleif.
24357 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24359 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
24360 pair for all vector types with 64 bit elements.
24361 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
24362 * config/s390/vector.md (V_HW_64): ... here.
24363 (V_128_NOSINGLE): New mode iterator.
24364 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
24365 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
24366 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
24367 ("*vec_load_pairv2di"): Change to ...
24368 ("*vec_load_pair<mode>"): ... this one.
24370 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24372 * config/s390/constraints.md: Add comments.
24373 (jKK): Reject element sizes > 8 bytes.
24374 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
24376 * config/s390/s390.md: Add the s_operand checks formerly in
24377 s390_split_ok_p to various splitters where they are still
24379 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
24380 for 128 bit vectors. Plus two splitters.
24382 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24384 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
24387 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24390 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
24391 error if the boundary argument is not constant.
24393 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24395 PR rtl-optimization/80112
24396 * loop-doloop.c (doloop_condition_get): Don't check condition
24397 if cmp isn't SET with IF_THEN_ELSE src.
24399 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24401 PR tree-optimization/80158
24402 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
24403 replacing a candidate statement, also replace it for the
24404 candidate's alternate interpretation.
24405 (replace_rhs_if_not_dup): Likewise.
24406 (replace_one_candidate): Likewise.
24408 2017-03-24 Richard Biener <rguenther@suse.de>
24410 PR tree-optimization/80167
24411 * graphite-isl-ast-to-gimple.c
24412 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
24414 (translate_isl_ast_to_gimple::get_rename): Likewise.
24416 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24418 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
24419 handling of certain combinations of target options, including the
24420 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
24421 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
24423 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24426 * config/arm/arm.md (*load_multiple): Add reload_completed to
24427 matching condition.
24429 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24430 Richard Biener <rguenth@suse.de>
24432 PR tree-optimization/79908
24433 PR tree-optimization/80136
24434 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24435 been cast away, gimplify_and_add suffices.
24437 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
24439 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
24441 2017-03-23 Richard Biener <rguenther@suse.de>
24443 PR tree-optimization/80032
24444 * gimplify.c (gimple_push_cleanup): Forced unconditional
24445 cleanups still have to go to the conditional_cleanups
24448 2017-03-22 Jakub Jelinek <jakub@redhat.com>
24450 PR tree-optimization/80072
24451 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
24453 (next_operand_entry_id): Change type to unsigned int.
24454 (sort_by_operand_rank): Make sure to return the right return value
24455 even if unsigned fields are bigger than INT_MAX.
24456 (struct oecount): Change cnt and id type to unsigned int.
24457 (oecount_hasher::equal): Formatting fix.
24458 (oecount_cmp): Make sure to return the right return value
24459 even if unsigned fields are bigger than INT_MAX.
24460 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
24463 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
24464 TREE_READONLY on result if writing it more than once.
24467 * doc/invoke.texi (-fsanitize=thread): Document that with
24468 -fnon-call-exceptions atomics are not able to throw
24472 * tsan.c: Include tree-eh.h.
24473 (instrument_builtin_call): Call maybe_clean_eh_stmt or
24474 maybe_clean_or_replace_eh_stmt where needed.
24475 (instrument_memory_accesses): Add cfg_changed argument.
24476 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
24477 if it returned true.
24478 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
24480 PR rtl-optimization/63191
24481 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
24482 wrapper function, moved the whole old content into ...
24483 (ix86_delegitimize_address_1): ... this. New inline function.
24484 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
24485 true as last argument instead of ix86_delegitimize_address.
24487 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24489 * config/aarch64/aarch64.c (generic_branch_cost): Copy
24490 cortexa57_branch_cost.
24492 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24494 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
24496 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24499 * doc/md.texi (Constraints): Document wA constraint.
24500 * config/rs6000/constraints.md (wA): New.
24501 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
24502 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
24503 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
24504 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
24506 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
24509 * gimplify.c (is_oacc_declared): New function.
24510 (oacc_default_clause): Use it to set default flags for acc declared
24511 variables inside parallel regions.
24512 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
24513 declared variables.
24514 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
24515 declare attribute to any decl as necessary.
24517 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24520 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
24521 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
24522 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
24523 (arm_arch_lpae): This.
24524 * config/arm/arm.c (arm_arch7ve): Rename into ...
24525 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
24526 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
24529 2017-03-22 Martin Liska <mliska@suse.cz>
24532 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
24533 error message instead of an ICE.
24535 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24537 * doc/extend.texi (6.11 Additional Floating Types): Revise.
24539 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
24541 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24543 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24546 2017-03-21 Martin Sebor <msebor@redhat.com>
24548 * doc/extend.texi: Use "cannot" instead of "can't."
24549 * doc/hostconfig.texi: Same.
24550 * doc/install.texi: Same.
24551 * doc/invoke.texi: Same.
24552 * doc/loop.texi: Same.
24553 * doc/md.texi: Same.
24554 * doc/objc.texi: Same.
24555 * doc/rtl.texi: Same.
24556 * doc/tm.texi: Same.
24557 * doc/tm.texi.in: Same.
24558 * doc/trouble.texi: Same.
24560 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
24563 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
24564 (collect_checksum_attributes): Set it.
24565 (die_checksum_ordered): Use it.
24567 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24569 PR tree-optimization/79908
24570 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
24571 change: For a VA_ARG whose LHS has been cast away, use
24572 force_gimple_operand to construct the side effects.
24574 2017-03-21 David Malcolm <dmalcolm@redhat.com>
24576 PR translation/80001
24577 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
24578 more amenable to translation.
24579 (oacc_loop_auto_partitions): Likewise.
24581 2017-03-21 Marek Polacek <polacek@redhat.com>
24582 Martin Sebor <msebor@redhat.com>
24584 PR tree-optimization/80109
24585 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
24586 on INTEGRAL_TYPE_P.
24588 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24589 Segher Boessenkool <segher@kernel.crashing.org>
24592 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
24593 check reg_used_between_p between insn and one of succ or succ2
24594 depending on if succ is artificial insn not inserted into insn
24597 2017-03-21 Martin Liska <mliska@suse.cz>
24599 PR gcov-profile/80081
24600 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
24601 * doc/gcc.texi: Include gcov-dump stuff.
24602 * doc/gcov-dump.texi: New file.
24604 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
24606 PR rtl-optimization/79150
24607 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
24608 conditional jump, if the jump is the last insn of the loop.
24610 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24611 Richard Biener <rguenth@suse.de>
24613 PR tree-optimization/79908
24614 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24615 been cast away, use force_gimple_operand to construct the side
24618 2017-03-21 Martin Liska <mliska@suse.cz>
24620 PR libfortran/79956
24621 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
24624 2017-03-21 Brad Spengler <spender@grsecurity.net>
24627 * plugin.c (htab_hash_plugin): New function.
24628 (add_new_plugin): Use it and adjust.
24629 (parse_plugin_arg_opt): Adjust.
24630 (init_one_plugin): Likewise.
24632 2017-03-21 Richard Biener <rguenther@suse.de>
24634 PR tree-optimization/80032
24635 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
24636 if set force the cleanup to happen unconditionally.
24637 (gimplify_target_expr): Push inserted clobbers with force_uncond
24638 to avoid them being removed by control-dependent DCE.
24640 2017-03-21 Richard Biener <rguenther@suse.de>
24642 PR tree-optimization/80122
24643 * tree-inline.c (copy_bb): Do not expans va-arg packs or
24644 va_arg_pack_len when the inlined call stmt requires pack
24646 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
24648 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24651 * tsan.c (instrument_builtin_call): If the memory model argument
24652 is not a constant, assume it is valid.
24655 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
24658 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24660 PR rtl-optimization/79910
24661 * combine.c (can_combine_p): Do not allow combining an I0 or I1
24662 if its dest is used by an insn before I2 (other than the combined
24663 insns themselves, which are properly handled already).
24665 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24668 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24670 * combine.c (record_used_regs): New static function.
24671 (try_combine): Handle situations where there is an additional
24672 instruction between I2 and I3 which needs to have a LOG_LINK
24676 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24678 * combine.c (try_combine): Delete redundant i1 test. Call
24679 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24681 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24684 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
24685 alternatives 13/14.
24687 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24689 PR tree-optimization/80054
24690 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
24691 the optimization if a PHI or any of its arguments is not dominated
24692 by the candidate's basis. Use gphi* rather than gimple* as
24694 (replace_profitable_candidates): Clean up a gimple* variable that
24695 should be a gphi* variable.
24697 2017-03-20 Martin Sebor <msebor@redhat.com>
24700 * doc/extend.texi (attribute constructor): Document present limitation.
24702 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
24705 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
24706 __POWER9_VECTOR__ #ifdef control, change template definition to
24707 use Power9-specific built-in function.
24708 (vec_any_eq): Likewise.
24709 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
24710 to control outcomes from this test.
24711 (vector_ae_<mode>p): For VEC_F modes, likewise.
24713 2017-03-20 Ian Lance Taylor <iant@google.com>
24715 * config/i386/i386.c (ix86_function_regparm): Save an extra
24716 register for -fsplit-stack with DECL_STATIC_CHAIN.
24718 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24721 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
24722 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
24724 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24726 * config/riscv/riscv.c (riscv_print_operand): Use "fence
24728 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
24731 2017-03-20 Marek Polacek <polacek@redhat.com>
24734 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
24736 2017-03-20 Richard Biener <rguenther@suse.de>
24738 PR tree-optimization/80113
24739 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
24740 allocate extra SSA name for PHI def.
24741 (add_close_phis_to_outer_loops): Likewise.
24742 (add_close_phis_to_merge_points): Likewise.
24743 (copy_loop_close_phi_args): Likewise.
24744 (copy_cond_phi_nodes): Likewise.
24746 2017-03-20 Martin Liska <mliska@suse.cz>
24748 PR middle-end/79753
24749 * tree-chkp.c (chkp_build_returned_bound): Do not build
24750 returned bounds for a LHS that's not a BOUNDED_P type.
24752 2017-03-20 Martin Liska <mliska@suse.cz>
24756 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
24757 COMPLEX_CST and VECTOR_CST.
24759 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24762 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
24763 target operand. A new splitter adds the clobber statement in case
24764 the target operand is dead anyway.
24766 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
24768 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
24769 to age-old versions of binutils and glibc.
24771 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
24773 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
24775 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24777 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
24779 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24781 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
24782 requirement for binutils 2.13.
24784 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24786 * combine.c (try_combine): Delete redundant i1 test. Call
24787 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24789 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
24791 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
24792 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
24794 <riscv64-*-elf>: Re-arrange section
24795 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
24796 <riscv32-*-linux>: Likewise.
24797 <riscv64-*-elf>: Likewise
24798 <riscv64-*-linux>: Likewise.
24800 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
24803 * aarch64.opt(verbose-cost-dump): Fix typo.
24805 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
24808 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
24809 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
24811 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24813 * reload.c (find_reloads): When reloading a nonoffsettable address,
24814 use RELOAD_OTHER for it and its address reloads.
24816 PR rtl-optimization/79910
24817 * combine.c (record_used_regs): New static function.
24818 (try_combine): Handle situations where there is an additional
24819 instruction between I2 and I3 which needs to have a LOG_LINK
24822 2017-03-17 Jeff Law <law@redhat.com>
24824 PR tree-optimization/71437
24825 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
24826 conditional in the hash table first.
24827 (vrp_dom_walker::before_dom_children): Extract condition from
24828 ASSERT_EXPR. Record condition, its inverion and any implied
24829 conditions as well.
24831 2017-03-17 Marek Polacek <polacek@redhat.com>
24832 Markus Trippelsdorf <markus@trippelsdorf.de>
24834 PR tree-optimization/80079
24835 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
24838 2017-03-17 Richard Biener <rguenther@suse.de>
24840 PR middle-end/80075
24841 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
24842 Properly verify the LHS before the RHS possibly claims to be
24844 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
24847 2017-03-17 Martin Jambor <mjambor@suse.cz>
24849 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
24850 (List of -O2 options): Likewise.
24851 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
24854 2017-03-17 Tom de Vries <tom@codesourcery.com>
24856 * gcov-dump.c (print_usage): Print bug_report_url.
24858 2017-03-17 Richard Biener <rguenther@suse.de>
24860 PR middle-end/80050
24861 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
24862 (parser::peek): Likewise.
24864 2017-03-17 Richard Biener <rguenther@suse.de>
24866 PR tree-optimization/80048
24867 * sese.c (free_sese_info): Properly release rename_map and
24868 copied_bb_map elements.
24870 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
24872 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
24873 Add linked-list forward and backlinks. Insert on
24874 construction, remove on destruction.
24875 (class pass_store_merging): Add m_stores_head field.
24876 (pass_store_merging::terminate_and_process_all_chains):
24877 Iterate over m_stores_head list.
24878 (pass_store_merging::terminate_all_aliasing_chains):
24880 (pass_store_merging::execute): Check for debug stmts first.
24881 Push new chains onto the m_stores_head stack.
24883 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24886 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
24887 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
24888 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
24890 2017-03-16 Jeff Law <law@redhat.com>
24892 PR tree-optimization/71437
24893 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
24894 member function. Implementation moved into after_dom_children
24895 member function and into the threader's thread_outgoing_edges
24897 (dom_opt_dom_walker::after_dom_children): Simplify by moving
24898 some code into new thread_outgoing_edges.
24899 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
24900 definition. Simplify marker handling (do it here). Assume we always
24901 have the available expression and the const/copies tables.
24902 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
24904 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
24905 * tree-vrp.c (equiv_stack): No longer file scoped.
24906 (vrp_dom_walker): New class.
24907 (vrp_dom_walker::before_dom_children): New member function.
24908 (vrp_dom_walker::after_dom_children): Likewise.
24909 (identify_jump_threads): Setup domwalker. Use it rather than
24910 walking edges in a random order by hand. Simplify setup/finalization.
24911 (finalize_jump_threads): Remove.
24912 (vrp_finalize): Do not call identify_jump_threads here.
24913 (execute_vrp): Do it here instead and call thread_through_all_blocks
24916 PR tree-optimization/71437
24917 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
24919 (simplify_stmt_for_jump_threading): Add basic_block argument. All
24921 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
24922 (dom_opt_dom_walker::thread_across_edge): Remove
24923 handle_dominating_asserts argument. All callers changed.
24924 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
24925 changes. Remove calls to lhs_of_dominating_assert. Other
24926 uses of handle_dominating_asserts turn into unconditional code
24927 (simplify_control_stmt_condition_1): Likewise.
24928 (simplify_control_stmt_condition): Likewise.
24929 (thread_through_normal_block, thread_across_edge): Likewise.
24930 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
24931 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
24932 object if it is not an SSA_NAME.
24933 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
24934 before calling into the VRP specific simplifiers.
24935 (identify_jump_threads): Remove handle_dominating_asserts
24938 2017-03-16 Jakub Jelinek <jakub@redhat.com>
24941 * tree-diagnostic.c (default_tree_printer): No longer static.
24942 * tree-diagnostic.h (default_tree_printer): New prototype.
24944 2017-03-16 Tamar Christina <tamar.christina@arm.com>
24946 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
24947 Change ins into fmov.
24949 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24951 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
24952 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
24953 Use h_con constraint for operand 1.
24954 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
24955 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
24957 2017-03-15 Jeff Law <law@redhat.com>
24959 PR tree-optimization/71437
24960 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
24961 (record_temporary_equivalences): Use it.
24963 PR tree-optimization/71437
24964 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
24965 tree-ssa-scopedtables.
24966 (lookup_avail_expr, build_and_record_new_cond): Likewise.
24967 (record_conditions, record_cond, vuse_eq): Likewise.
24968 (record_edge_info): Adjust to API tweak of record_conditions.
24969 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
24970 (record_temporary_equivalences, optimize_stmt): Likewise.
24971 (eliminate_redundant_computations): Likewise.
24972 (record_equivalences_from_stmt): Likewise.
24973 * tree-ssa-scopedtables.c: Include options.h and params.h.
24974 (vuse_eq): New function, moved from tree-ssa-dom.c
24975 (build_and_record_new_cond): Likewise.
24976 (record_conditions): Likewise. Accept vector of conditions rather
24977 than edge_equivalence structure for first argument.
24978 for the first argument.
24979 (avail_exprs_stack::lookup_avail_expr): New member function, moved
24980 from tree-ssa-dom.c.
24981 (avail_exprs_stack::record_cond): Likewise.
24982 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
24983 from tree-ssa-dom.c.
24984 (avail_exprs_stack): Add new member functions lookup_avail_expr
24986 (record_conditions): Declare.
24988 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
24991 * lra-constraints.c (process_alt_operands): Increase reject for
24992 reloading an input/output operand.
24994 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
24997 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
24998 insns to convert from signed/unsigned char/short to IEEE 128-bit
25000 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
25002 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
25005 * config/i386/i386.c (ix86_vector_duplicate_value): Create
25006 subreg of inner mode for values already in registers.
25008 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
25010 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
25011 iteration reg is used after the loop.
25013 2017-03-14 Martin Sebor <msebor@redhat.com>
25015 PR tree-optimization/79800
25016 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
25017 precision in negative-positive range.
25018 (format_floating): Call non-const overload with adjusted precision.
25020 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25023 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
25026 2017-03-14 Martin Sebor <msebor@redhat.com>
25028 PR middle-end/80020
25029 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
25030 * builtins.def (aligned_alloc): Use it.
25033 * Makefile.in (GTFILES): Add calls.c.
25034 * calls.c: Include "gt-calls.h".
25036 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
25038 PR rtl-optimization/79728
25039 * regs.h (struct target_regs): New field
25040 x_contains_allocatable_regs_of_mode.
25041 (contains_allocatable_regs_of_mode): New macro.
25042 * reginfo.c (init_reg_sets_1): Initialize it, and change
25043 contains_reg_of_mode so it includes global regs as well.
25044 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
25045 rather than contains_regs_of_mode.
25047 2017-03-14 Martin Liska <mliska@suse.cz>
25049 * doc/invoke.texi: Document options that can't be combined with
25050 -fcheck-pointer-bounds.
25052 2017-03-14 Martin Liska <mliska@suse.cz>
25054 PR middle-end/79831
25055 * doc/invoke.texi (-Wchkp): Document the option.
25057 2017-03-14 Martin Liska <mliska@suse.cz>
25059 * Makefile.in: Install gcov-dump.
25061 2017-03-14 Martin Liska <mliska@suse.cz>
25063 * multiple_target.c (expand_target_clones): Bail out for
25064 an invalid attribute.
25066 2017-03-14 Richard Biener <rguenther@suse.de>
25068 * alias.c (struct alias_set_entry): Pack properly.
25069 * cfgloop.h (struct loop): Likewise.
25070 * cse.c (struct set): Likewise.
25071 * ipa-utils.c (struct searchc_env): Likewise.
25072 * loop-invariant.c (struct invariant): Likewise.
25073 * lra-remat.c (struct cand): Likewise.
25074 * recog.c (struct change_t): Likewise.
25075 * rtl.h (struct address_info): Likewise.
25076 * symbol-summary.h (function_summary): Likewise.
25077 * tree-loop-distribution.c (struct partition): Likewise.
25078 * tree-object-size.c (struct object_size_info): Likewise.
25079 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25080 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25081 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25082 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25083 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25084 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25085 (struct _stmt_vec_info): Likewise.
25087 2017-03-14 Martin Liska <mliska@suse.cz>
25090 * multiple_target.c (create_dispatcher_calls): Check that
25091 a target can create a function dispatcher.
25093 2017-03-14 Martin Liska <mliska@suse.cz>
25096 * multiple_target.c (expand_target_clones): Drop local.local
25097 flag for default implementation.
25099 2017-03-14 Richard Biener <rguenther@suse.de>
25101 PR tree-optimization/80030
25102 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25104 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25106 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25107 gcc_fallthrough() instead of __attribute__((fallthrough));
25109 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25111 * doc/gcc.texi: Remove "up" link to (DIR).
25112 * doc/gccint.texi: Ditto.
25114 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25116 * doc/install.texi (Specific) <avr>: Remove reference to
25119 2017-03-13 Jeff Law <law@redhat.com>
25121 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25122 attribute rather than comments.
25124 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25125 match_scratch operand is highest.
25127 2017-03-13 Martin Liska <mliska@suse.cz>
25129 PR middle-end/78339
25130 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25131 is a CHKP clone, use original declaration.
25133 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25135 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25136 (arc_conditional_register_usage): Use a different allocation order
25137 when optimizing for size.
25138 * common/config/arc/arc-common.c (arc_option_optimization_table):
25139 Section anchors default on when optimizing for size.
25141 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25143 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
25145 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25147 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
25148 * config/arc/arc.md (cpu_facility): Add cd variant.
25149 (*movqi_insn): Add code density variant.
25150 (*movhi_insn): Likewise.
25151 (*movqi_insn): Likewise.
25152 (*addsi3_mixed): Likewise.
25153 (subsi3_insn): Likewise.
25155 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25157 * config/arc/arc.md (movsi_cond_exec): Update constraint.
25159 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25161 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
25162 expressions with MINUS and UNARY ops.
25164 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25167 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
25169 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
25170 between vec_select and vector argument.
25171 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
25172 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
25173 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
25174 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
25175 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
25176 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
25178 2017-03-13 Richard Biener <rguenther@suse.de>
25181 * params.def (vect-max-peeling-for-alignment): Fix typo.
25183 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25185 * doc/install.texi (Specific) <mips-*-*>: Remove description of
25186 issue that only occurred with binutils below 2.18.
25188 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25190 * doc/install.texi (Specific) <cris-axis-elf>: No longer
25191 refer to binutils 2.11/2.12 minimum.
25193 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25195 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
25196 ftp.kernel.org and simplify binutils requirement.
25198 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
25200 * doc/invoke.texi (Warning Options): Fix spelling of link-time
25202 (Optimize Options): Ditto. Also remove redundancy.
25204 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25206 PR translation/79848
25207 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
25209 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
25210 to G_ to avoid double translation.
25212 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25214 PR translation/79923
25215 * auto-profile.c (get_combined_location): Convert leading
25216 character of diagnostics to lower case and remove trailing period.
25217 (read_profile): Likewise for various diagnostics.
25218 * config/arm/arm.c (arm_option_override): Remove trailing period
25219 from various diagnostics.
25220 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
25221 (msp430_expand_delay_cycles): Likewise.
25223 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25226 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
25227 full command-line argument, rather than just "str".
25228 (aarch64_validate_march): Likewise.
25229 (aarch64_validate_mtune): Likewise.
25231 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
25233 PR rtl-optimization/78911
25234 * lra-assigns.c (must_not_spill_p): New function.
25235 (spill_for): Use it.
25237 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25239 PR tree-optimization/79981
25240 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
25241 ATOMIC_COMPARE_EXCHANGE ifn result.
25242 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
25243 IFN_ATOMIC_COMPARE_EXCHANGE.
25245 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25248 * opts.c (parse_sanitizer_options): Add missing question mark to
25249 "did you mean" message.
25251 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25253 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
25255 (VMULEUH_UNS): Likewise.
25256 (VMULOUB_UNS): Likewise.
25257 (VMULOUH_UNS): Likewise.
25258 * config/rs6000/rs6000.c (builtin_function_type): Remove
25259 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
25261 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25264 * read-rtl-function.c (function_reader::read_rtx_operand): Update
25265 x with result of extra_parsing_for_operand_code_0.
25266 (function_reader::extra_parsing_for_operand_code_0): Convert
25267 return type from void to rtx, returning x. When reading
25268 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
25269 larger size containing struct block_symbol.
25271 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
25273 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
25274 -mfloat128-hardware without -m64.
25276 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
25279 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
25280 entries to the case statement that marks unsigned arguments to
25281 overloaded functions.
25283 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25285 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
25286 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
25288 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
25291 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
25292 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
25294 2017-03-10 Martin Liska <mliska@suse.cz>
25298 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
25299 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
25302 2017-03-10 Olivier Hainque <hainque@adacore.com>
25304 * tree-switch-conversion (array_value_type): Start by resetting
25305 candidate type to it's main variant.
25307 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25309 PR rtl-optimization/79909
25310 * combine.c (try_combine): Use simplify_replace_rtx on individual
25311 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
25312 of the whole CALL_INSN_FUNCTION_USAGE.
25314 PR tree-optimization/79972
25315 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
25316 get_range_info on SSA_NAMEs. Formatting fixes.
25318 2017-03-10 Richard Biener <rguenther@suse.de>
25319 Jakub Jelinek <jakub@redhat.com>
25321 PR tree-optimization/77975
25322 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
25323 edge to be constant.
25324 (get_val_for): For constant x return it. Formatting fix.
25325 (loop_niter_by_eval): Avoid pointless looping if the next iteration
25326 would use the same bases as the current one.
25328 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25330 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
25331 instead of vec_select for V1TImode.
25332 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
25334 (VSX_LE_128): Add V1TI to this mode iterator.
25335 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
25336 (*vsx_le_perm_store_<mode>): Likewise.
25337 (pre-reload splitter for VSX stores): Likewise.
25338 (post-reload splitter for VSX stores): Likewise.
25339 (*vsx_xxpermdi2_le_<mode>): Likewise.
25340 (*vsx_lxvd2x2_le_<mode>): Likewise.
25341 (*vsx_stxvd2x2_le_<mode>): Likewise.
25343 2017-03-09 Michael Eager <eager@eagercon.com>
25345 Correct failures with --enable-checking=yes,rtl.
25347 * config/microblaze/microblaze.c (microblaze_expand_shift):
25348 Replace GET_CODE test with CONST_INT_P and INTVAL test with
25349 test for const0_rtx.
25350 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
25351 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
25353 2017-03-09 Richard Biener <rguenther@suse.de>
25355 PR tree-optimization/79977
25356 * graphite-scop-detection.c (scop_detection::merge_sese):
25357 Handle the case of extra exits to blocks dominating the entry.
25359 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
25361 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25364 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
25366 PR rtl-optimization/79949
25367 * lra-constraints.c (process_alt_operands): Check memory when
25368 trying to predict a cycle. Print about the overall increase.
25370 2017-03-09 Richard Biener <rguenther@suse.de>
25372 PR middle-end/79971
25373 * gimple-expr.c (useless_type_conversion_p): Preserve
25374 TYPE_SATURATING for fixed-point types.
25376 2017-03-09 Richard Biener <rguenther@suse.de>
25379 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
25380 alignment of BLKmode params.
25382 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25385 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
25386 (VALL_NO_V2Q): Likewise.
25388 * config/aarch64/aarch64-simd.md
25389 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
25391 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
25392 VALL_NO_V2Q mode iterator.
25393 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
25395 2017-03-09 Martin Liska <mliska@suse.cz>
25397 PR tree-optimization/79631
25398 * tree-chkp-opt.c (chkp_is_constant_addr): Call
25399 tree_int_cst_sign_bit just for INTEGER constants.
25401 2017-03-09 Martin Liska <mliska@suse.cz>
25405 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
25408 2017-03-09 Marek Polacek <polacek@redhat.com>
25411 * tree.c (inchash::add_expr): Handle TREE_VEC.
25413 2017-03-09 Martin Liska <mliska@suse.cz>
25416 (chkp_narrow_size_and_offset): New function.
25417 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
25418 (void chkp_parse_bit_field_ref): New function.
25419 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
25420 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
25422 2017-03-09 Martin Liska <mliska@suse.cz>
25425 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
25426 (chkp_find_bounds_1): Remove gcc_unreachable.
25428 2017-03-09 Jakub Jelinek <jakub@redhat.com>
25431 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
25432 BUILT_IN_SYNC*, determine the access type from the size suffix and
25433 always build a MEM_REF with that type. Handle forgotten
25434 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
25437 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
25438 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
25439 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
25440 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
25441 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
25442 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
25443 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
25444 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
25445 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
25446 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
25447 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
25448 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
25449 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
25450 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
25451 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
25452 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
25453 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
25454 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
25455 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
25456 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
25457 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
25458 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
25459 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
25460 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
25461 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
25462 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
25463 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
25464 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
25465 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
25466 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
25467 definitions outside of __OPTIMIZE__ guarded section.
25470 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
25471 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
25472 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
25473 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
25476 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25478 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
25479 ("vfenez<mode>"): Add missing constraints.
25481 2017-03-08 Martin Sebor <msebor@redhat.com>
25484 * config/nds32/nds32.c (nds32_option_override):
25485 Fix misspelled diagnostic.
25487 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25490 * gimplify.c (gimplify_omp_for): Replace index var in outer
25491 taskloop statement with an artificial variable and add
25492 OMP_CLAUSE_PRIVATE clause for it.
25494 2017-03-08 Richard Biener <rguenther@suse.de>
25496 PR tree-optimization/79955
25497 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
25498 for accesses that are completely outside of the variable.
25500 2017-03-08 Andrew Haley <aph@redhat.com>
25502 PR tree-optimization/79943
25503 * tree-ssa-loop-split.c (compute_new_first_bound): When
25504 calculating the new upper bound, (END-BEG) should be added, not
25507 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25509 * config/avr/avr.md (setmemhi): Make sure match_dup
25510 operand number comes before match_scratch.
25512 2017-03-08 Richard Biener <rguenther@suse.de>
25514 PR tree-optimization/79920
25515 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
25516 with ncopies == 1 to ...
25517 (vect_transform_slp_perm_load): ... here. Properly compute
25518 all element loads by iterating VF times over the group. Do
25519 not handle ncopies (computed in a broken way) in
25520 vect_create_mask_and_perm.
25522 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25525 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
25526 is a uniform vector, use uniform_vector_p return value instead of
25527 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
25529 2017-03-07 Marek Polacek <polacek@redhat.com>
25531 PR middle-end/79809
25532 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
25533 (alloca_call_type): Likewise.
25535 2017-03-07 Martin Liska <mliska@suse.cz>
25537 * gcov.c (process_args): Put comment to correct location.
25539 2017-03-07 Martin Liska <mliska@suse.cz>
25541 PR middle-end/68270
25542 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
25543 Use array_at_struct_end_p instead of DECL_CHAIN (field).
25544 (chkp_narrow_bounds_for_field): Likewise.
25545 (chkp_parse_array_and_component_ref): Pass one more argument to
25548 2017-03-07 Richard Biener <rguenther@suse.de>
25550 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
25553 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25555 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
25556 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
25558 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25561 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
25562 to end of description.
25563 (PARAM_MAX_STORES_TO_MERGE): Likewise.
25565 2017-03-07 Jakub Jelinek <jakub@redhat.com>
25567 PR rtl-optimization/79901
25568 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
25570 (*avx512f_<code><mode>3<mask_name>): ... this.
25571 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
25572 iterator instead of VI8_AVX2_AVX512BW.
25574 PR rtl-optimization/79901
25575 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
25576 min/max expander, expand it using expand_vec_cond_expr.
25579 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
25582 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25585 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
25586 to void * for PCH reasons.
25587 * dwarf2out.c (output_loc_operands, output_die): Cast
25588 v.val_vec.array to unsigned char *.
25590 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
25593 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
25596 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
25598 PR rtl-optimization/79571
25599 * lra-constraints.c (process_alt_operands): Calculate static
25600 reject and subtract it from overall when only addresses will be
25603 2017-03-06 Julia Koval <julia.koval@intel.com>
25606 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
25607 incoming stack boundary to 128 for 64-bit targets.
25609 2017-03-06 Richard Biener <rguenther@suse.de>
25611 PR tree-optimization/79894
25612 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
25613 to NULL after folding it.
25615 2017-03-06 Richard Biener <rguenther@suse.de>
25617 PR tree-optimization/79824
25618 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
25619 check disabling peeling for gaps.
25621 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
25623 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25624 attributes): Document gettimeofday.
25626 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25628 * config/s390/s390.c (s390_option_override_internal): Set
25629 PARAM_MIN_VECT_LOOP_BOUND
25631 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25633 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
25634 * config/s390/s390.md: Likewise.
25636 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25639 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
25640 (<avx2_avx512>_perm<mode>): Rename to ...
25641 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
25643 (<avx512>_perm<mode>_mask): Rename to ...
25644 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
25646 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
25647 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
25648 instead of VI8F_256_512.
25649 (avx512f_perm<mode>): New define_expand.
25650 (avx512f_perm<mode>_mask): Likewise.
25651 (avx512f_perm<mode>_1<mask_name>): New define_insn.
25652 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
25654 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25656 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
25657 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
25659 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
25661 2017-03-06 Martin Liska <mliska@suse.cz>
25664 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
25665 when having a SSA NAME w/o VAR_DECL assigned to it.
25667 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25669 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
25670 msa_dpsub_<su>_d): Fix MODE for vec_select.
25672 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25674 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
25676 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
25678 2017-03-06 Richard Biener <rguenther@suse.de>
25680 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
25681 * plugin.c (register_plugin_info): Likewise.
25682 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
25684 2017-03-05 Jakub Jelinek <jakub@redhat.com>
25686 * config/i386/sse.md (sse_storehps, sse_storelps,
25687 avx_<castmode><avxsizesuffix>_<castmode>,
25688 avx512f_<castmode><avxsizesuffix>_<castmode>,
25689 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
25690 in condition that at least one operand is not a MEM.
25692 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25694 PR middle-end/79805
25695 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
25696 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
25698 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
25699 gimple_call_nothrow_p flag based on whether original builtin can throw.
25700 If it can, emit following stmts on the fallthrough edge.
25701 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
25702 don't create new bb if inserting just debug stmts on the edge, try to
25703 insert them on the fallthru bb or just reset debug stmts.
25705 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
25708 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
25709 restore recog_data (including the operand rtxes inside it) around
25710 the call to get_insn_template.
25712 2017-03-03 Martin Sebor <msebor@redhat.com>
25714 PR tree-optimization/79699
25715 * context.c (context::~context): Free MPFR caches to avoid
25716 a memory leak on program exit.
25718 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25720 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
25721 Use wide_int::ulow () instead of .elt (0).
25723 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25725 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
25726 (*pushxf): Limit oF constraint to 32bit targets and add oC
25727 constraint for 64bit targets.
25728 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
25729 (*pushdf): Change rmF constraint to rmC.
25731 2017-03-03 Martin Liska <mliska@suse.cz>
25733 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
25734 Remove unused variable.
25736 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25739 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
25740 is a memory operand, increase num_memory.
25741 (ix86_expand_args_builtin): Likewise.
25743 2017-03-03 Jan Hubicka <jh@suse.cz>
25746 * ipa-devirt.c (maybe_record_node): Properly handle
25747 __cxa_pure_virtual visibility.
25749 2017-03-03 Martin Liska <mliska@suse.cz>
25751 PR tree-optimization/79803
25752 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
25754 (pass_loop_prefetch::execute): Disabled optimization if an
25755 assumption about L1 cache size is not met.
25757 2017-03-03 Martin Liska <mliska@suse.cz>
25759 PR rtl-optimization/79574
25760 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
25761 (hash_scan_set): Likewise.
25762 (dump_hash_table): Likewise.
25763 (hoist_code): Likewise.
25765 2017-03-03 Richard Biener <rguenther@suse.de>
25767 * fixed-value.c (fixed_from_string): Restore use of elt (1)
25768 in place of uhigh ().
25769 (fixed_convert_from_real): Likewise.
25771 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25774 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
25776 2017-03-03 Richard Biener <rguenther@suse.de>
25778 PR middle-end/79818
25779 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
25780 TYPE_OVERFLOW_UNDEFINED check.
25782 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25784 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
25786 (vector_ae_<mode>_p): Likewise.
25787 (vector_nez_<mode>_p): Likewise.
25788 (vector_ne_v2di_p): Likewise.
25789 (vector_ae_v2di_p): Likewise.
25790 (vector_ne_<mode>_p): Likewise.
25791 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
25793 (vsx_tsqrt<mode>2_fe): Likewise.
25795 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
25798 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
25800 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25802 PR rtl-optimization/79780
25803 * cprop.c (one_cprop_pass): When second and further conditional trap
25804 in a single basic block is turned into an unconditional trap, turn it
25805 into a deleted note to avoid RTL verification failures.
25807 2017-03-02 Richard Biener <rguenther@suse.de>
25809 * fold-const.c (const_binop): Use ulow () instead of elt (0).
25811 2017-03-02 Richard Biener <rguenther@suse.de>
25813 PR tree-optimization/79345
25815 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
25816 param and abort the walk, returning -1 if it is hit.
25817 (walk_aliased_vdefs): Take a limit param and pass it on.
25818 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
25819 defaulting to 0 and return a signed int.
25820 * tree-ssa-uninit.c (struct check_defs_data): New struct.
25821 (check_defs): New helper.
25822 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
25823 about uninitialized memory.
25824 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
25825 bogus uninitialized warning.
25826 (fixed_convert_from_real): Likewise.
25828 2017-03-02 Bin Cheng <bin.cheng@arm.com>
25830 PR tree-optimization/66768
25831 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
25832 iv_use if base object can't be determined.
25834 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25836 PR tree-optimization/79345
25837 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
25838 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
25839 (get_pattern_stats): Initialize it.
25840 * genemit.c (gen_expand): Verify match_scratch numbers come after
25841 match_operand/match_dup numbers.
25842 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
25843 match_scratch numbers.
25844 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
25846 * config/s390/s390.md (trunctdsd2): Likewise.
25848 2017-03-02 Richard Biener <rguenther@suse.de>
25850 * wide-int.h (wide_int_storage::operator=): Implement in terms
25853 2017-03-02 Richard Biener <rguenther@suse.de>
25855 PR tree-optimization/79777
25856 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
25857 the to insert expression to sth existing.
25859 2017-03-01 Martin Sebor <msebor@redhat.com>
25861 PR middle-end/79692
25862 * gimple-ssa-sprintf.c
25863 (directive::known_width_and_precision): New function.
25864 (format_integer): Use it.
25865 (get_mpfr_format_length): Consider the full range of precision
25866 when computing %g output with the # flag. Set the likely byte
25867 count to 3 rather than 1 when precision is indeterminate.
25868 (format_floating): Correct the lower bound of precision.
25870 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25872 * doc/invoke.texi: Document default code model for 64-bit Linux.
25874 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25877 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
25878 udiv rather than div since input pattern is unsigned.
25880 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
25882 * config/i386/i386.c (print_reg): Warn for values of
25883 unsupported size in integer register.
25885 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
25888 * config/rs6000/predicates.md (current_file_function_operand): Do
25889 not allow self calls to be local if the function is replaceable.
25891 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
25894 * config/rs6000/altivec.h (vec_ctz and others): Change the
25895 preprocessor macro that controls conditional compilation from
25896 _ARCH_PWR9 to __POWER9_VECTOR__.
25897 (vec_all_ne): Change parameterization of __altivec_scalar_pred
25898 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
25899 control (instead of _ARCH_PWR9 control) so that template
25900 definition uses power9-specific function.
25901 (vec_any_eq): Likewise.
25902 (vec_all_ne): Change macro definition to use a power9-specific
25903 expansion under #ifdef __POWER9_VECTOR__ control (instead of
25904 _ARCH_PWR9 control).
25905 (vec_any_eq) Likewise.
25906 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
25907 expansion for CMPNEF to remove support for xvcmpnesp instruction.
25908 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
25909 support for xvcmpnedp instruction.
25910 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
25911 macro expansion so that Power9 implementation of vec_all_ne does
25912 not use the AltiVec predicate framework.
25913 (VCMPNEH_P): Likewise.
25914 (VCMPNEW_P): Likewise.
25915 (VCMPNED_P): Likewise.
25916 (VCMPNEFP_P): Likewise.
25917 (VCMPNEDP_P): Likewise.
25918 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
25919 implementation of vec_any_eq to not use AltiVec predicate
25921 (VCMPAEH_P): Likewise.
25922 (VCMPAEW_P): Likewise.
25923 (VCMPAED_P): Likewise.
25924 (VCMPAEFP_P): Likewise.
25925 (VCMPAEDP_P): Likewise.
25926 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
25927 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
25928 not use the AltiVec predicate framework.
25929 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
25930 of vec_any_eq to not use AltiVec predicate framework.
25931 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25932 support for predefined __POWER9_VECTOR__ macro to indicate that
25933 Power9 instruction selection is enabled.
25934 (altivec_overloaded_builtins): Remove extraneous
25935 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
25936 function argument types RS6000_BTI_bool_V16QI and
25937 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
25938 entry for overloaded function argument types RS6000_BTI_bool_V4SI
25939 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
25940 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
25941 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
25942 Power9 for implementations of vec_cmpne. Change the signature for
25943 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
25944 (representing vec_all_ne) to remove the previously described first
25945 argument of type RS6000_BTI_INTSI, as this was an artifact of
25946 reliance on the AltiVec predicate framework, which is no longer
25947 used in the implementation of these functions. Add
25948 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
25949 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
25950 since, unlike the AltiVec predicate framework implementation, we
25951 do not share function descriptors between vec_alle and vec_anyeq.
25952 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
25953 set of modes that receive special treatment even when
25954 TARGET_P9_VECTOR is true. The special treatment emits code that
25955 does not depend on Power9 instructions.
25956 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
25957 define_expand to not rely on AltiVec predicate framework.
25958 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25960 (vector_ne_v2di_p): Change this define_expand to not rely on
25961 AltiVec predicate framework.
25962 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
25964 (vector_ne_<mode>_p): Change this define_expand to not rely on
25965 AltiVec predicate framework.
25966 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25968 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
25969 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
25970 define_insn pattern.
25971 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
25972 define_insn pattern because the xvcmpne<VSs>. instruction is not
25974 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
25975 instruction is not supported.
25977 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25979 * config/nvptx/nvptx.c: Include intl.h.
25981 2017-03-01 Martin Jambor <mjambor@suse.cz>
25984 * ipa-prop.h (ipa_bits): Removed field known.
25985 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
25986 to pointers. Adjusted their comments to warn about their sharing.
25987 (ipcp_transformation_summary): Change bits to a vector of pointers.
25988 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
25989 (ipa_get_ipa_bits_for_value): Declare.
25990 * tree-vrp.h (value_range): Mark as GTY((for_user)).
25991 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
25992 (ipa_bits_hash_table): Likewise.
25993 (ipa_vr_ggc_hash_traits): Likewise.
25994 (ipa_vr_hash_table): Likewise.
25995 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
25996 being pointers and vr_known being removed.
25997 (ipa_set_jf_unknown): Likewise.
25998 (ipa_get_ipa_bits_for_value): New function.
25999 (ipa_set_jfunc_bits): Likewise.
26000 (ipa_get_value_range): New overloaded functions.
26001 (ipa_set_jfunc_vr): Likewise.
26002 (ipa_compute_jump_functions_for_edge): Use the above functions to
26003 construct bits and vr parts of jump functions.
26004 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
26005 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26007 (ipcp_grow_transformations_if_necessary): Also allocate
26008 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26010 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
26011 them. Fix too long lines.
26012 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
26013 vr_known being removed.
26014 (ipa_read_jump_function): Use new setter functions to construct bits
26015 and vr parts of jump functions or set them to NULL.
26016 (write_ipcp_transformation_info): Adjust for bits being pointers.
26017 (read_ipcp_transformation_info): Likewise.
26018 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
26020 Include gt-ipa-prop.h.
26021 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
26023 (ipcp_store_bits_results): Likewise.
26024 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
26025 Do not write to existing jump functions but use a temporary instead.
26027 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26030 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
26031 attempt to use its first operand as BIT_FIELD_REF base.
26033 2017-03-01 Richard Biener <rguenther@suse.de>
26035 PR middle-end/79721
26036 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
26037 interpolating formula in wrapping arithmetic.
26038 (chrec_apply): Convert chrec_evaluate return value to wanted type.
26040 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26042 PR tree-optimization/79734
26043 * tree-vect-generic.c (expand_vector_condition): Optimize
26044 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
26045 Handle VEC_COND_EXPR where comparison has different inner width from
26046 type's inner width.
26048 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
26050 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
26051 markup, and similar issues. Remove @opindex entries for things
26052 that aren't options. Add missing -mmpy-option entries.
26054 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26056 PR tree-optimization/79737
26057 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
26058 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
26059 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
26060 instead of byte_size. Formatting fix.
26061 (shift_bytes_in_array_right): Formatting fix.
26063 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26066 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26067 condition on optimize for the leaf function test.
26069 2017-02-28 Martin Liska <mliska@suse.cz>
26072 * read-rtl-function.c (function_reader::handle_unknown_directive):
26073 Bail out when one uses -flto.
26075 2017-02-28 Martin Liska <mliska@suse.cz>
26077 * common.opt: Replace space with tabular for options of <number>
26079 * config/i386/i386.opt: Show <number> value for
26080 -mlarge-data-threshold.
26081 * opts.c (print_filtered_help): Do not display number in hexadecimal
26084 2017-02-28 Martin Liska <mliska@suse.cz>
26086 * common.opt: Fix --help=option -Q for options which are of
26089 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26091 * config/i386/i386.c (print_reg): Error out for values
26092 of 8-bit size in invalid integer register.
26094 2017-02-28 Martin Sebor <msebor@redhat.com>
26096 PR tree-optimization/79691
26097 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26099 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26102 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26103 gcc_unreachable with output_operand_lossage.
26105 2017-02-28 Richard Biener <rguenther@suse.de>
26107 PR tree-optimization/79740
26108 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26110 (visit_nary_op): Insert the nary into the hashtable if we
26111 pattern-matched sth.
26112 * tree-ssa-pre.c (eliminate_insert): Robustify.
26114 2017-02-28 Richard Biener <rguenther@suse.de>
26116 PR middle-end/79731
26117 * fold-const.c (decode_field_reference): Reject out-of-bound
26120 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26122 * config/i386/i386.c: Include intl.h.
26123 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26124 instead of just cond ? "..." : "...".
26125 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26126 * coverage.c (read_counts_file): Likewise.
26127 * omp-offload.c: Include intl.h.
26128 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26129 of just cond ? "..." : "...".
26130 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26131 of just cond ? "..." : "...".
26133 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26136 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26138 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26139 'tune for' CPU name.
26140 * config/arm/arm-cpu-data.h: Regenerated.
26142 2017-02-28 Richard Biener <rguenther@suse.de>
26144 PR tree-optimization/79732
26145 * tree-inline.c (expand_call_inline): Do not shadow var.
26147 2017-02-28 Richard Biener <rguenther@suse.de>
26149 PR tree-optimization/79723
26150 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
26151 address-space properly.
26153 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
26155 * doc/optinfo.texi (Optimization groups): Fix option used for
26157 * doc/invoke.texi (-fopt-info): Document "omp".
26158 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
26159 (OPTGROUP_ALL): Add OPTGROUP_OMP.
26160 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
26161 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
26162 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
26164 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
26166 * dumpfile.c (optgroup_options): Instead of "openmp", associate
26167 OPTGROUP_OMP with "omp".
26169 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
26172 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
26173 for arithmetic shift of unsigned V2DI.
26175 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
26177 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
26178 arc/linux.h headers.
26179 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
26180 (LINK_SPEC): Likewise.
26181 (ARC_TLS_EXTRA_START_SPEC): Likewise.
26182 (EXTRA_SPECS): Likewise.
26183 (STARTFILE_SPEC): Likewise.
26184 (ENDFILE_SPEC): Likewise.
26185 (LIB_SPEC): Likewise.
26186 (TARGET_SDATA_DEFAULT): Likewise.
26187 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
26188 (MULTILIB_DEFAULTS): Likewise.
26189 (DWARF2_UNWIND_INFO): Likewise.
26190 * config/arc/big.h: New file.
26191 * config/arc/elf.h: Likewise.
26192 * config/arc/linux.h: Likewise.
26193 * config/arc/t-uClibc: Remove.
26195 2017-02-27 Bin Cheng <bin.cheng@arm.com>
26197 PR tree-optimization/77536
26198 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
26199 (tree_transform_and_unroll_loop): Use above function to compute the
26200 estimated niter of unrolled loop and use it when scaling profile.
26201 Also use count info rather than frequency if it's non-zero.
26202 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
26203 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
26204 (vect_transform_loop): Call above function.
26206 2017-02-27 Richard Biener <rguenther@suse.de>
26208 PR tree-optimization/45397
26209 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
26210 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
26211 (visit_nary_op): Add pattern matching for CSEing sign-changed
26212 or truncated operations with wider ones.
26214 2017-02-27 Richard Biener <rguenther@suse.de>
26216 PR tree-optimization/79690
26217 * tree-vect-stmts.c (vectorizable_store): Use vector type
26218 built from the DR with address-space.
26220 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
26222 * doc/invoke.texi (Optimize Options): Refine the description
26223 of asan-use-after-return.
26225 2017-02-25 Alan Modra <amodra@gmail.com>
26227 PR rtl-optimization/79584
26228 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
26229 base, not ad->base_term, the reg within base. Remove assertion
26230 that ad->base == ad->base_term. Replace gen_int_mode using
26231 bogus mode with const0_rtx.
26233 2017-02-25 Jakub Jelinek <jakub@redhat.com>
26235 PR middle-end/79396
26236 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
26237 FMA_EXPR like tcc_binary or tcc_unary.
26239 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
26242 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
26244 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
26245 (output_loc_operands): Handle DW_OP_call_ref and
26246 DW_OP_GNU_variable_value.
26247 (struct variable_value_struct): New type.
26248 (struct variable_value_hasher): Likewise.
26249 (variable_value_hash): New variable.
26250 (string_types): Remove.
26251 (copy_loc_descr): New function.
26252 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
26253 (prepend_loc_descr_to_each): New function.
26254 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
26255 instead of add_loc_descr_to_each if the first argument is single
26256 location list and the second has multiple.
26257 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
26258 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
26259 when looking for variable value which doesn't have other location info.
26260 (loc_list_from_tree): Formatting fix.
26261 (gen_array_type_die): Simplify DW_AT_string_length handling.
26262 (adjust_string_types): Remove.
26263 (gen_subprogram_die): Don't call adjust_string_types nor test/set
26264 string_types. Call resolve_variable_values.
26265 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
26266 (resolve_addr_in_expr): Likewise. Add A argument.
26267 (copy_deref_exprloc): Remove deref argument. Adjust for the
26268 original expression being DW_OP_GNU_variable_value with optionally
26269 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
26270 optionally after it.
26271 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
26272 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
26273 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
26274 (variable_value_hasher::hash, variable_value_hasher::equal): New
26276 (resolve_variable_value_in_expr, resolve_variable_value,
26277 resolve_variable_values, note_variable_value_in_expr,
26278 note_variable_value): New functions.
26279 (dwarf2out_early_finish): Call note_variable_value on all toplevel
26282 2017-02-24 Jakub Jelinek <jakub@redhat.com>
26285 * opts.h (handle_generated_option): Add GENERATED_P argument.
26286 * opts-common.c (handle_option): Adjust function comment.
26287 (handle_generated_option): Add GENERATED_P argument, pass it to
26289 (control_warning_option): Pass false to handle_generated_option
26291 * opts.c (maybe_default_option): Pass true to handle_generated_option
26293 * optc-gen.awk: Likewise.
26295 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26297 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
26298 a REG, look at the REG it is a SUBREG of.
26299 (splitter for cmpeqsi_t): Ditto.
26301 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26303 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
26304 the special USEs with the pattern of the insn, not the insn itself.
26306 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
26309 * doc/invoke.texi: Document -mload-store-pairs.
26311 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26312 Sandra Loosemore <sandra@codesourcery.com>
26314 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
26315 argument isn't a CONST_INT.
26316 (nios2_alternate_compare_const): Assert op is a CONST_INT.
26317 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
26318 (nios2_validate_compare): Bypass alternate compare logic if *op2
26319 is not a CONST_INT.
26320 (ldstwm_operation_p): Return false if first_base is not a REG or
26321 if first_offset is not a CONST_INT.
26323 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26325 * config/cris/cris.md: Use correct operand in a define_peephole2.
26327 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26329 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
26331 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26333 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
26334 this_insn if it is an INSN or JUMP_INSN.
26335 (force_offsettable): Look at base, not at addr.
26336 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
26337 on things that aren't necessarily CONST_INTs.
26339 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
26341 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
26342 -mfpmath=sse is the default also for x86-32 targets with SSE2
26343 instruction set when @option{-ffast-math} is enabled
26345 2017-02-24 Jeff Law <law@redhat.com>
26347 PR rtl-optimizatoin/79286
26348 * ira.c (update_equiv_regs): Drop may_trap_p exception to
26351 2017-02-24 Richard Biener <rguenther@suse.de>
26353 PR tree-optimization/79389
26354 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
26357 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
26359 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
26360 function comment to reflect reality.
26361 (loop_exits_before_overflow): Fix typo in function description.
26363 2017-02-24 Richard Biener <rguenther@suse.de>
26365 PR tree-optimization/79389
26366 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
26367 properly that a threading opportunity exists. Detect conditional
26368 copy/constant propagation opportunities.
26370 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
26372 * config/visium/visium.md (type): Add trap.
26373 (b): New mode attribute.
26374 (*btst): Rename into...
26375 (*btst<mode>): ...this and adjust.
26376 (*cbranchsi4_btst_insn): Rename into...
26377 (*cbranch<mode>4_btst_insn): ...this and adjust.
26378 (trap): New define_insn.
26380 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26382 PR tree-optimization/79389
26383 * ifcvt.c (struct noce_if_info): Add rev_cond field.
26384 (noce_reversed_cond_code): New function.
26385 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
26386 reversed_comparison_code. Formatting fix.
26387 (noce_try_store_flag): Test rev_cond != NULL in addition to
26388 reversed_comparison_code.
26389 (noce_try_store_flag_constants): Likewise.
26390 (noce_try_store_flag_mask): Likewise.
26391 (noce_try_addcc): Use rev_cond if non-NULL instead of
26392 reversed_comparison_code.
26393 (noce_try_cmove_arith): Likewise. Formatting fixes.
26394 (noce_try_minmax, noce_try_abs): Clear rev_cond.
26395 (noce_find_if_block): Initialize rev_cond.
26396 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
26397 instead of false as last argument never attempt to reverse it
26400 2017-02-23 Bin Cheng <bin.cheng@arm.com>
26402 PR tree-optimization/79663
26403 * tree-predcom.c (combine_chains): Process refs in reverse order
26404 only for ZERO length chains, and add explaining comment.
26406 2017-02-23 Jeff Law <law@redhat.com>
26408 PR tree-optimization/79578
26409 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
26410 in call to operand_equal_p.
26412 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
26415 * config/i386/cpuid.h: Fix another undefined behavior.
26417 2017-02-23 Richard Biener <rguenther@suse.de>
26419 PR tree-optimization/79683
26420 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
26421 vector types for data-refs.
26423 2017-02-23 Martin Liska <mliska@suse.cz>
26425 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
26427 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26429 PR middle-end/79665
26430 * internal-fn.c (get_range_pos_neg): Moved to ...
26431 * tree.c (get_range_pos_neg): ... here. No longer static.
26432 * tree.h (get_range_pos_neg): New prototype.
26433 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
26434 are known to be in between 0 and signed maximum inclusive, try to
26435 expand both unsigned and signed divmod and use the cheaper one from
26438 2017-02-22 Jeff Law <law@redhat.com>
26440 PR tree-optimization/79578
26441 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
26442 to compare base operands.
26444 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
26447 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
26448 gpc_reg_operand instead of fpr_reg_operand.
26450 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
26452 * config/mips/mips.c (mips_return_in_memory): Force FP
26453 vector types to be returned in memory for o32 ABI.
26455 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26457 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
26458 instead of DW_TAG_member for static data member declarations and don't
26459 set no_linkage_name for static inline data members.
26460 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
26463 2017-02-22 Martin Liska <mliska@suse.cz>
26465 * doc/invoke.texi: Replace inequality signs with square brackets
26468 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26470 PR tree-optimization/68644
26471 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
26473 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
26476 * lra-constraints.c (simplify_operand_subreg): Handle
26477 WORD_REGISTER_OPERATIONS targets.
26479 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26482 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
26483 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
26484 elimination by swapping fld*.
26486 2017-02-22 Richard Biener <rguenther@suse.de>
26488 PR tree-optimization/79673
26489 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
26490 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
26491 irrelevant address-space qualifiers and avoiding a
26492 ADDR_SPACE_CONVERT_EXPR from fold_convert.
26494 2017-02-22 Richard Biener <rguenther@suse.de>
26496 PR tree-optimization/79666
26497 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
26498 to not symbolically negate if that may introduce undefined
26501 2017-02-22 Martin Liska <mliska@suse.cz>
26504 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
26505 * data-streamer-out.c (streamer_write_gcov_count_stream):
26507 * value-prof.c (stream_out_histogram_value): Make assert more
26508 precise based on type of counter.
26510 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
26513 * config/i386/i386.md (standard_x87sse_constant_load splitter):
26514 Use nonimmediate_operand instead of memory_operand for operand 1.
26515 (float-extend standard_x87sse_constant_load splitter): Ditto.
26517 2017-02-21 Jeff Law <law@redhat.com>
26519 PR tree-optimization/79621
26520 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
26521 blocks with edges to themselves.
26523 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26526 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
26527 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
26528 Use gimple_call_builtin_p.
26531 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
26532 on temporarily removed DEBUG_INSNs.
26534 PR tree-optimization/79649
26535 * tree-loop-distribution.c (classify_partition): Give up on
26536 non-generic address space loads/stores.
26538 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
26540 * doc/loop.texi (Loop manipulation): Remove nonexistent
26541 tree_ssa_loop_version from the documentation.
26542 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
26544 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26547 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
26548 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
26549 * config/rs6000/rs6000.c: Include except.h.
26550 (rs6000_expand_split_stack_prologue): Call
26551 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
26553 2017-02-21 Martin Jambor <mjambor@suse.cz>
26556 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
26557 have been analyzed.
26559 2017-02-21 Martin Jambor <mjambor@suse.cz>
26561 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
26562 for backward compatibility only.
26563 * doc/invoke.texi (Option Summary): Remove all references to
26564 -fipa-cp-alignment.
26566 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
26570 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26572 * lra-constraints.c (curr_insn_transform): Handle
26573 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26575 2017-02-21 Martin Liska <mliska@suse.cz>
26577 * config/i386/i386.opt: Replace -masm-dialect with -masm.
26579 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
26581 PR translation/79638
26582 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
26584 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
26587 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
26588 (arm_function_ok_for_sibcall): Return false for an indirect call by
26589 descriptor if all the argument registers are used.
26590 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
26591 alignment of the function.
26593 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26595 PR tree-optimization/61441
26596 * simplify-rtx.c (simplify_const_unary_operation): For
26597 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
26598 the sNaN unmodified.
26600 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26602 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26603 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26604 instead of SYSTEM_HEADER_DIR.
26606 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
26607 Martin Liška <mliska@suse.cz>
26609 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
26610 Fix typos and grammar, use active voice, and clarify.
26612 2017-02-20 Marek Polacek <polacek@redhat.com>
26614 PR middle-end/79537
26615 * gimplify.c (gimplify_expr): Handle unused *&&L;.
26618 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
26620 2017-02-20 Jakub Jelinek <jakub@redhat.com>
26623 * config/i386/i386.c (ix86_expand_builtin): Handle
26624 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
26625 ix86_builtins_isa[fcode].isa as a requirement of those
26626 flags and any other flag in the bitmask.
26627 (ix86_init_mmx_sse_builtins): Use 0 instead of
26628 ~OPTION_MASK_ISA_64BIT as mask.
26629 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
26630 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
26631 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
26632 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
26634 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26637 * lra-constraints.c (split_reg): Check requested split mode
26638 is supported by the register.
26640 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26642 * lra-constraints.c (simplify_operand_subreg): Remove early
26645 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26648 * lra-constraints.c (curr_insn_transform): Tighten condition
26649 for converting SUBREG reloads from OP_OUT to OP_INOUT.
26651 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26654 * lra-constraints.c (curr_insn_transform): Handle
26655 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26657 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
26660 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
26662 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
26664 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
26667 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
26670 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26672 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
26675 2017-02-18 Jakub Jelinek <jakub@redhat.com>
26677 * final.c (last_columnnum, override_columnnum): New variables.
26678 (final_start_function): Set last_columnnum, pass it to begin_prologue
26679 hook and pass 0 to dwarf2out_begin_prologue.
26680 (final_scan_insn): Update override_columnnum. Pass last_columnnum
26681 to source_line debug hook.
26682 (notice_source_line): Compute last_columnnum and for debug_column_info
26683 return true on column changes.
26684 * debug.h (struct gcc_debug_hooks): Add column argument to
26685 source_line and begin_prologue hooks.
26686 (debug_nothing_int_charstar_int_bool): Remove prototype.
26687 (debug_nothing_int_int_charstar,
26688 debug_nothing_int_int_charstar_int_bool): New prototypes.
26689 (dwarf2out_begin_prologue): Add column argument.
26690 * debug.c (do_nothing_debug_hooks): Adjust source_line and
26691 begin_prologue hooks.
26692 (debug_nothing_int_charstar_int_bool): Remove.
26693 (debug_nothing_int_int_charstar,
26694 debug_nothing_int_int_charstar_int_bool): New functions.
26695 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
26696 through to dwarf2out_source_line.
26697 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
26698 (dwarf2out_source_line): Add column argument, emit it if requested.
26699 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
26701 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26702 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26703 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
26704 through to dwarf2out_begin_prologue.
26705 (vmsdbgout_source_line): Add column argument, pass it through to
26706 dwarf2out_source_line.
26707 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
26708 dbxout_source_line caller.
26709 (dbxout_source_line): Add column argument.
26711 * common.opt (gno-column-info, gcolumn-info): New options.
26712 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
26713 (check_die): Also test for multiple DW_AT_decl_column attributes.
26714 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
26715 DW_AT_decl_column if requested.
26716 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
26718 (gen_variable_die): Likewise.
26719 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
26720 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
26723 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
26724 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
26725 (ix86_handle_option): Handle OPT_m3dnowa.
26726 * doc/invoke.texi (-m3dnowa): Document.
26727 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
26728 -m3dnowa instead of -m3dnow -march=athlon.
26731 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
26732 instead of gcc_assert for K, r and R code checks. Formatting fixes.
26734 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26737 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
26738 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
26739 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
26740 generator for vsx_xxpermdi_<mode>_be.
26741 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
26742 force big-endian semantics.
26743 (vsx_xxpermdi_<mode>_be): New define_expand with same
26744 implementation as previous version of vsx_xxpermdi_<mode>.
26746 2017-02-17 Jakub Jelinek <jakub@redhat.com>
26748 PR tree-optimization/79327
26749 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
26750 variable, its initialization and use.
26752 2017-02-17 Julia Koval <julia.koval@intel.com>
26754 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
26755 (OPTION_MASK_ISA_PKU_UNSET): New.
26756 (ix86_handle_option): Handle -mrdpid.
26757 * config/i386/cpuid.h (bit_RDPID): New.
26758 * config/i386/driver-i386.c (host_detect_local_cpu):
26759 Detect RDPID feature.
26760 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
26761 * config/i386/i386-c.c (ix86_target_macros_internal):
26763 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
26764 (ix86_valid_target_attribute_inner_p): Add "rdpid".
26765 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26766 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
26767 * config/i386/i386.md (define_insn "rdpid"): New.
26768 * config/i386/i386.opt Add -mrdpid.
26769 * config/i386/immintrin.h (_rdpid_u32): New.
26771 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
26773 PR rtl-optimization/79541
26774 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
26775 instead of transforming it into USE.
26777 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
26779 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
26780 If HONOR_SNANS (SFmode) force the input to a register.
26781 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
26782 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
26783 an frsp or similar insn.
26785 2017-02-17 Martin Liska <mliska@suse.cz>
26787 PR rtl-optimization/79577
26788 * params.def (selsched-max-sched-times): Increase minimum to 1.
26790 2017-02-17 Martin Liska <mliska@suse.cz>
26792 PR rtl-optimization/79574
26793 * gcse.c (want_to_gcse_p): Prevent integer overflow.
26795 2017-02-17 Martin Liska <mliska@suse.cz>
26797 PR tree-optimization/79529
26798 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
26799 ssa_defined_default_def_p to handle cases which are implicitly
26801 * tree-ssa.c (ssa_defined_default_def_p): New function.
26802 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
26803 which are implicitly defined.
26804 * tree-ssa.h (ssa_defined_default_def_p): Declare.
26806 2017-02-17 Richard Biener <rguenther@suse.de>
26808 PR middle-end/79576
26809 * params.def (max-ssa-name-query-depth): Limit to 10.
26811 2017-02-17 Richard Biener <rguenther@suse.de>
26813 PR tree-optimization/79552
26814 * tree-ssa-structalias.c (visit_loadstore): Properly verify
26817 2017-02-17 Richard Biener <rguenther@suse.de>
26820 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
26822 2017-02-17 Marek Polacek <polacek@redhat.com>
26824 PR middle-end/79536
26825 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
26826 (fold_negate_expr): New wrapper.
26828 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
26830 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
26831 Correct terminology and de-emphasize pre-standard behavior.
26833 2017-02-16 Alan Modra <amodra@gmail.com>
26835 PR rtl-optimization/79286
26836 * ira.c (def_dominates_uses): New function.
26837 (update_equiv_regs): Don't create an equivalence for insns that
26838 may trap where the register def does not dominate the use.
26840 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
26842 PR rtl-optimization/78127
26843 * lra.c (lra): Call lra_eliminate before finish the loop after
26846 2017-02-16 Richard Biener <rguenther@suse.de>
26848 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
26850 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
26851 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
26852 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
26853 (isl_val_int_from_wi): New function.
26854 (extract_affine_gmp): Rename to ...
26855 (extract_affine_wi): ... this, take a widest_int.
26856 (extract_affine_int): Just wrap extract_affine_wi.
26857 (add_param_constraints): Use isl_val_int_from_wi.
26858 (add_loop_constraints): Likewise, and extract_affine_wi.
26860 2017-02-15 Jeff Law <law@redhat.com>
26862 PR middle-end/79521
26863 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
26864 ira_init_register_move_cost_if_necessary.
26866 2017-02-15 Martin Sebor <msebor@redhat.com>
26868 PR middle-end/32003
26869 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
26870 removed in a prior commit.
26872 2017-02-15 Bin Cheng <bin.cheng@arm.com>
26874 PR tree-optimization/79347
26875 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
26876 counters during peeling.
26878 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
26880 * Makefile.in (site.exp): Remove "set ISLVER".
26882 2017-02-15 Jakub Jelinek <jakub@redhat.com>
26885 * real.c (real_from_integer): Call real_convert even for decimal.
26887 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
26890 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
26892 2017-02-14 Andrew Pinski <apinski@cavium.com>
26894 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
26895 cores and change the partno/implementer to be correct.
26896 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
26897 the 'B" as the implementer.
26898 * config/aarch64/aarch64-tune.md: Regenerate.
26900 2017-02-14 Carl Love <cel@us.ibm.com>
26902 * config/rs6000/rs6000.c: Add case statement entry to make the
26903 xvcvuxdsp built-in argument unsigned.
26904 * config/rs6000/vsx.md: Fix the source and return operand types so they
26905 match the instruction definitions from the ISA document. Fix typo
26906 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
26909 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
26912 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
26913 member early_clobber_alts.
26914 * lra-lives.c (reg_early_clobber_p): New.
26915 (process_bb_lives): Use it.
26916 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
26917 (debug_operand_data): Initialize early_clobber_alts.
26918 (setup_operand_alternative): Set up early_clobber_alts.
26919 (collect_non_operand_hard_regs): Ditto. Pass early clobber
26920 alternatives to new_insn_reg.
26921 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
26923 (lra_update_insn_regno_info): Pass the new arg.
26925 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26927 PR middle-end/79505
26928 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
26929 (new_oacc_loop_raw): Don't clear already cleared fields.
26932 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
26933 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
26934 _mm512_prefetch_i64gather_ps): New inline functions and macros.
26936 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
26939 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
26941 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
26944 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
26945 the extra instruction to the right place to store 128-bit constant
26948 2017-02-14 Martin Sebor <msebor@redhat.com>
26950 PR middle-end/79448
26951 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
26952 warning for strings of unknown length.
26954 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26956 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
26958 2017-02-14 Jeff Law <law@redhat.com>
26961 * ira-costs.c (scan_one_insn): Initialize register move costs
26962 for pseudos seen in USE/CLOBBER insns.
26964 PR tree-optimization/79095
26965 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
26966 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
26967 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
26968 if the operands are known to be not equal, then the resulting range
26970 (intersect_ranges): If the new range is ~[0,0] and the old range is
26971 wide, then prefer ~[0,0].
26972 * tree-vrp.c (overflow_comparison_p_1): New function.
26973 (overflow_comparison_p): New function.
26974 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
26975 if NAME is used in an overflow test.
26976 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
26977 overflow check that can be expressed as an equality test, then adjust
26978 ops to be that equality test.
26980 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26982 * config/s390/s390-builtin-types.def: Remove flags argument.
26983 * config/s390/s390.c (s390_init_builtins): Likewise.
26985 2017-02-14 Martin Liska <mliska@suse.cz>
26987 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
26988 vector. Fix trailing white spaces.
26990 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
26992 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
26995 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26997 PR rtl-optimization/68664
26998 * config/arm/arm.c (arm_sched_can_speculate_insn):
26999 New function. Declare prototype.
27000 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27002 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27004 PR rtl-optimization/68664
27005 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
27007 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27009 2017-02-14 Amit Pawar <amit.pawar@amd.com>
27011 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
27012 max skip bytes for function, loop and jump.
27014 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27016 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
27017 ABS_EXPR for gimple dump.
27019 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27022 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
27024 PR tree-optimization/79408
27025 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
27026 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
27027 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
27028 also if rhs1 is INTEGER_CST.
27030 2017-02-14 Richard Biener <rguenther@suse.de>
27032 PR middle-end/79432
27033 * tree-into-ssa.c (insert_phi_nodes): When the function can
27034 have abnormal edges rewrite SSA names with broken use-def
27035 dominance out of SSA and register them for PHI insertion.
27037 2017-02-13 Martin Sebor <msebor@redhat.com>
27039 PR middle-end/79496
27040 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
27041 clearing info.nowrite flag when snprintf size argument is a range.
27043 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27045 * cprop.c (cprop_jump): Add missing space in string literal.
27046 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
27047 (get_constraint_for_component_ref): Likewise.
27048 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
27049 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
27050 * lra-constraints.c (process_alt_operands): Likewise.
27051 * ipa-inline.c (inline_small_functions): Likewise.
27052 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
27053 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27054 * trans-mem.c (diagnose_tm_1_op): Likewise.
27055 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
27056 (grid_parallel_clauses_gridifiable): Likewise.
27058 * config/nvptx/mkoffload.c (process): Add space in between
27061 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
27062 "MOD4_SSE_REGS" and "ALL_REGS".
27064 * spellcheck.c (test_data): Add , in between "foo" and "food".
27066 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27069 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27070 boundary crossing check and subsequent code generation agree.
27072 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27074 * config/aarch64/aarch64.c (has_memory_op): Delete.
27075 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27078 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27080 PR rtl-optimization/79388
27081 PR rtl-optimization/79450
27082 * combine.c (distribute_notes): When removing TEM_INSN for which
27083 corresponding dest has last value recorded, invalidate that last
27086 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27088 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27089 of explicit '@'. Add missing assembly comment marker on branch costs
27092 2017-02-13 Nathan Sidwell <nathan@acm.org>
27094 * gengtype-lex.l (<in_struct>): Add '/'.
27096 2017-02-13 Martin Liska <mliska@suse.cz>
27099 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27101 2017-02-13 Richard Biener <rguenther@suse.de>
27103 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27105 * configure: Re-generate.
27106 * config.in: Likewise.
27107 * graphite-dependences.c: Simplify as if
27108 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27109 * graphite-isl-ast-to-gimple.c: Likewise.
27110 * graphite-optimize-isl.c: Likewise.
27111 * graphite-poly.c: Likewise.
27112 * graphite-sese-to-poly.c: Likewise.
27113 * graphite.h: Likewise.
27114 * toplev.c: Include isl/version.h and use isl_version () for
27115 printing the ISL version.
27116 * doc/install.texi: Update ISL requirement.
27118 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27120 * doc/standards.texi (Standards): Update reference to
27123 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27125 * doc/extend.texi (Named Address Spaces): sourceware.org now
27127 * doc/install.texi (Binaries): Ditto.
27130 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27132 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27133 standard terminology "stringize"/"stringizing" throughout.
27134 * doc/cppinternals.texi: Likewise.
27136 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27138 * doc/extend.texi: Fix some spelling mistakes and typos.
27139 * doc/invoke.texi: Likewise.
27141 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27144 * params.def (inline-min-speedup) Change from 10 to 8.
27146 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27148 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
27151 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27154 * ipa-inline-analysis.c (get_minimal_bb): New function.
27155 (record_modified): Use it.
27156 (remap_edge_change_prob): Handle also ancestor functions.
27158 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
27160 * doc/contrib.texi (Contributors): Remove broken link into
27161 the Mauve CVS repository.
27163 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27165 PR middle-end/79454
27166 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
27167 result computation whenever lhs doesn't have vector mode, not
27168 just when it has BLKmode.
27170 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27172 * doc/makefile.texi (profiledbootstrap): Refer to the
27173 installation instructions only in textual form.
27175 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27178 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
27180 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27182 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
27183 (Specific): Update mingw-w64 reference.
27185 (Specific): Remove broken link to Renesas RX processor.
27187 2017-02-10 Richard Biener <rguenther@suse.de>
27189 * toplev.c (process_options): Do not mention obsolete graphite
27190 options when printing sorry message about missing graphite support.
27191 Mention -floop-nest-optimize.
27193 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
27195 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
27196 (vtst_p16): Likewise.
27197 (vtstq_p8): Likewise.
27198 (vtstq_p16): Likewise.
27200 (vtstq_p64): Likewise.
27201 * config/arm/arm_neon.h (vgetq_lane_p64): New.
27202 (vset_lane_p64): New.
27203 (vsetq_lane_p64): New.
27205 2017-02-10 Jakub Jelinek <jakub@redhat.com>
27207 PR tree-optimization/79411
27208 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
27209 stmt operands are SSA_NAMEs used in abnormal phis.
27210 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
27213 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27216 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
27219 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27221 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
27223 2017-02-09 Jakub Jelinek <jakub@redhat.com>
27225 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
27229 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
27230 not arbitrary TREE_CONSTANT.
27233 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
27234 "omp declare target link" attribute unless is_global_var.
27235 * omp-offload.c (find_link_var_op): Likewise.
27237 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
27238 Chung-Lin Tang <cltang@codesourcery.com>
27240 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
27242 (gimplify_adjust_omp_clauses): Don't delete TILE.
27243 (gimplify_omp_for): Deal with TILE.
27244 * internal-fn.c (expand_GOACC_TILE): New function.
27245 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
27247 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
27248 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
27250 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
27251 avoid DIV for outermost collapse var.
27252 (expand_oacc_for): Insert tile element loop as needed. Adjust.
27253 Remove out of date comments, fix whitespace.
27254 * omp-general.c (omp_extract_for_data): Deal with tiling.
27255 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
27256 adjust OLF_DIM_BASE value.
27257 (struct omp_for_data): Add tiling field.
27258 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
27259 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
27260 for auto loops. Remove default auto determining, moved to
27261 oacc_loop_fixed_partitions.
27262 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
27263 stmts, add e_mask field.
27264 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
27265 (oacc_thread_numbers): Use oacc_dim_call.
27266 (oacc_xform_tile): New.
27267 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
27268 (finish_oacc_loop): Adjust for ifns vector.
27269 (oacc_loop_discover_walk): Append loop abstraction sites to list,
27270 add case for GOACC_TILE fns.
27271 (oacc_loop_xform_loop): Delete.
27272 (oacc_loop_process): Iterate over call list directly, and add
27273 handling for GOACC_TILE fns.
27274 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
27276 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
27277 vector partitioning to outer loops. Assign 2 partitions to loops
27278 when available. Add TILE handling.
27279 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
27280 (execite_oacc_device_lower): Process GOACC_TILE fns,
27281 ignore unknown specs.
27282 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
27283 * tree.c (omp_clause_num_ops): Adjust TILE ops.
27284 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
27286 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
27288 * configure.ac (ACX_BUGURL): Update.
27289 * configure: Regenerate.
27291 2017-02-09 Richard Biener <rguenther@suse.de>
27293 PR tree-optimization/69823
27294 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27295 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
27297 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27299 * config/arc/arc-c.def: Add __NPS400__ definition.
27300 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
27301 (TARGET_NPS400): Define.
27303 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27305 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
27307 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
27308 pointer, arch_info.
27309 (arc_cpu_types): Fill the arch_info field with a pointer into the
27310 arc_arch_types table.
27311 (arc_selected_cpu): Declare.
27312 * config/arc/arc.c (arc_selected_cpu): Make global.
27313 (arc_selected_arch): Delete.
27314 (arc_base_cpu): Delete.
27315 (arc_override_options): Remove references to deleted variables,
27316 update access to arch information.
27317 (ARC_OPT): Update access to arch information.
27318 (ARC_OPTX): Likewise.
27319 * config/arc/arc.h (arc_base_cpu): Remove declaration.
27320 (TARGET_ARC600): Update access to arch information.
27321 (TARGET_ARC601): Likewise.
27322 (TARGET_ARC700): Likewise.
27323 (TARGET_EM): Likewise.
27324 (TARGET_HS): Likewise.
27325 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
27328 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
27331 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
27332 condition/operands for integer GE/LE/GEU/LEU operations.
27334 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
27336 PR translation/79397
27337 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
27340 2017-02-08 Martin Jambor <mjambor@suse.cz>
27343 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
27344 whether allocation happened.
27345 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
27346 nothing was allocated.
27348 2017-02-08 Jakub Jelinek <jakub@redhat.com>
27350 PR tree-optimization/79408
27351 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
27352 constant, but SSA_NAME with a known integer range, use the minimum
27353 of that range instead of op1 to determine if modulo can be replaced
27354 with its first operand.
27356 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27358 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
27360 2017-02-08 Richard Biener <rguenther@suse.de>
27362 PR tree-optimization/71824
27363 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27364 Check all loops contained in the merged region.
27366 2017-02-07 Andrew Pinski <apinski@cavium.com>
27368 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
27370 2017-02-07 Andrew Pinski <apinski@cavium.com>
27372 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
27373 (thunderxt88): Likewise.
27374 (thunderxt81): Disable LSE and change v8.1 to v8.
27375 (thunderxt83): Likewise.
27377 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27378 Richard Biener <rguenther@suse.de>
27380 PR middle-end/79399
27381 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
27382 type from int to size_t.
27383 * ira-costs.c (struct_costs_size): Change type from int to size_t.
27385 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27387 PR rtl-optimization/79386
27388 * cprop.c (bypass_conditional_jumps): Initialize
27389 bypass_last_basic_block already before splitting bbs after
27390 unconditional traps...
27391 (bypass_conditional_jumps): ... rather than here.
27394 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
27395 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
27396 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
27397 fix -masm=intel patterns.
27399 2017-02-07 Richard Biener <rguenther@suse.de>
27401 PR tree-optimization/79256
27402 PR middle-end/79278
27403 * builtins.c (get_object_alignment_2): Use min_align_of_type
27404 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
27405 and ADJUST_FIELD_ALIGN.
27407 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
27409 * doc/tm.texi: Regenerate.
27410 * stor-layout.c (layout_decl): Adjust.
27411 (update_alignment_for_field): Likewise.
27412 (place_field): Likewise.
27413 (min_align_of_type): Likewise.
27414 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
27415 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
27416 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
27417 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
27418 * config/frv/frv.c (frv_adjust_field_align): Likewise.
27419 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
27420 * config/i386/i386.c (x86_field_alignment): Likewise.
27421 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
27422 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
27423 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
27424 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
27425 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
27426 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
27430 2017-01-30 Richard Biener <rguenther@suse.de>
27432 PR tree-optimization/79256
27433 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27434 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27437 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
27439 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
27440 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
27441 builtins to SImode and emit a zero-extend, if necessary.
27443 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27445 * docs/invoke.texi (RISC-V Options): Alphabetize.
27447 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27449 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
27452 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27454 * config/riscv/riscv.c: New file.
27455 * common/config/riscv/riscv-common.c: Likewise.
27456 * config.gcc: Likewise.
27457 * config/riscv/constraints.md: Likewise.
27458 * config/riscv/elf.h: Likewise.
27459 * config/riscv/generic.md: Likewise.
27460 * config/riscv/linux.h: Likewise.
27461 * config/riscv/multilib-generator: Likewise.
27462 * config/riscv/peephole.md: Likewise.
27463 * config/riscv/pic.md: Likewise.
27464 * config/riscv/predicates.md: Likewise.
27465 * config/riscv/riscv-builtins.c: Likewise.
27466 * config/riscv/riscv-c.c: Likewise.
27467 * config/riscv/riscv-ftypes.def: Likewise.
27468 * config/riscv/riscv-modes.def: Likewise.
27469 * config/riscv/riscv-opts.h: Likewise.
27470 * config/riscv/riscv-protos.h: Likewise.
27471 * config/riscv/riscv.h: Likewise.
27472 * config/riscv/riscv.md: Likewise.
27473 * config/riscv/riscv.opt: Likewise.
27474 * config/riscv/sync.md: Likewise.
27475 * config/riscv/t-elf-multilib: Likewise.
27476 * config/riscv/t-linux: Likewise.
27477 * config/riscv/t-linux-multilib: Likewise.
27478 * config/riscv/t-riscv: Likewise.
27479 * configure.ac: Likewise.
27480 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
27481 Waterman as RISC-V maintainers.
27482 * doc/install.texi: Add RISC-V entries.
27483 * doc/invoke.texi: Add RISC-V options section.
27484 * doc/md.texi: Add RISC-V constraints section.
27485 * configure: Regenerated.
27487 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27490 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
27491 false values to be constant vectors with all 0 or all 1 bits set.
27492 (vcondu<mode><mode>): Likewise.
27493 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
27495 (fpmask_comparison_operator): Update comment.
27496 (vecint_comparison_operator): New predicate.
27497 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
27498 vector conditionals when the true and false values are constant
27499 vectors with all 0 bits or all 1 bits set.
27501 2017-02-06 Martin Sebor <msebor@redhat.com>
27503 PR tree-optimization/79376
27504 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
27506 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
27508 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
27509 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
27510 to simplify split condition.
27512 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27514 * omp-expand.c (oxpand_omp_atomic_fetch_op,
27515 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
27518 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
27520 PR rtl-optimization/68664
27521 * target.def (can_speculate_insn): New hook.
27522 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
27523 * doc/tm.texi: Regenerate.
27524 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
27525 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
27526 (rs6000_sched_can_speculate_insn): New function.
27528 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27530 PR tree-optimization/79284
27531 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
27532 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
27533 vectorizable_mask_load_store, vectorizable_operation,
27534 vect_is_simple_cond, get_same_sized_vectype): Use it instead
27535 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
27536 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
27537 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
27538 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
27539 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27540 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
27541 is_gimple_assign (stmt). Replace another such test with
27542 is_gimple_assign (stmt).
27544 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27547 * config/avr/avr.c (rtl-iter.h): Include it.
27548 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
27549 (avr_legitimate_combined_insn): ...and implementation.
27551 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
27553 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
27554 * config/s390/s390.c (s390_const_operand_ok)
27555 (s390_canonicalize_comparison, s390_extract_part)
27556 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
27557 (s390_contiguous_bitmask_p, s390_rtx_costs)
27558 (legitimize_pic_address): Likewise.
27559 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
27560 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
27561 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
27562 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
27563 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
27565 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27567 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
27568 REGNO($0) == REGNO($1).
27570 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27572 * config/s390/linux.h(SIZE_TYPE): Add comment.
27574 2017-02-06 Julian Brown <julian@codesourcery.com>
27575 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27576 Virendra Pathak <virendra.pathak@broadcom.com>
27578 * config/aarch64/aarch64-cores.def: Change the scheduler
27580 * config/aarch64/aarch64.md: Include thunderx2t99.md.
27581 * config/aarch64/thunderx2t99.md: New file.
27583 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27585 * doc/standards.texi (Go Language): Update link to language
27588 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27590 * tree-eh.c (lower_resx): Sanitize profile.
27591 (cleanup_empty_eh_move_lp): Likewise.
27593 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27596 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
27598 * cfgloopmanip.h (loop_version): Update prototype.
27599 * modulo-sched.c (sms_schedule): Update call of loop_version.
27600 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
27601 * tree-parloops.c (gen_parallel_loop): Likewise.
27602 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
27603 * tree-ssa-loop-split.c (split_loop): Likewise.
27604 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
27605 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
27607 2017-02-05 Martin Liska <mliska@suse.cz>
27610 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
27612 (print_operand_address): Initialize a struct to zero.
27614 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27616 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
27617 garbage collector only in textual form.
27619 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27621 * doc/extend.texi (x86 specific memory model extensions for
27622 transactional memory): Simplify a phrase.
27624 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
27627 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
27628 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
27629 (atomic_storedi_1): Likewise.
27631 2017-02-04 Jakub Jelinek <jakub@redhat.com>
27633 PR tree-optimization/79338
27634 * tree-parloops.c (gather_scalar_reductions): Don't call
27635 vect_analyze_loop_form for loop->inner before destroying loop's
27638 2017-02-03 Martin Sebor <msebor@redhat.com>
27640 PR tree-optimization/79327
27641 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
27642 when precision has resulted in leading zeros.
27643 (format_integer): Adjust the likely counter to assume an unknown
27644 argument that may be zero is non-zero.
27646 2017-02-03 Jason Merrill <jason@redhat.com>
27649 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
27650 avoid copying non-taken branch.
27652 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27654 PR tree-optimization/79340
27655 * tree-vect-loop.c (vectorizable_reduction): Release
27656 vec_defs elements after safe_splicing them into other vectors.
27659 PR tree-optimization/79327
27660 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
27661 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
27663 (format_integer): Use wide_int_to_tree instead of build_int_cst
27664 + to_?hwi. If argmin is NULL, just set argmin and argmax to
27665 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
27666 of shortest and longest sequence.
27668 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
27670 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
27671 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
27673 2017-02-03 Walter Lee <walt@tilera.com>
27676 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
27677 after initial stackframe link reg save.
27678 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
27680 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27683 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
27684 wu for stxssp alternative.
27686 2017-02-03 Martin Sebor <msebor@redhat.com>
27688 PR tree-optimization/79352
27689 * gimple-fold.c (get_range_strlen): Add argument.
27690 (get_range_strlen): Change return type to bool.
27691 (get_maxval_strlen): Pass in a dummy argument.
27692 * gimple-fold.h (get_range_strlen): Change return type to bool.
27693 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
27694 * tree.h (array_at_struct_end_p): Add argument.
27695 * tree.c (array_at_struct_end_p): Handle it.
27697 2017-02-03 Martin Liska <mliska@suse.cz>
27700 * multiple_target.c (create_dispatcher_calls): Redirect edge
27701 from a caller of a dispatcher.
27702 (expand_target_clones): Make the clones local.
27703 (ipa_target_clone): Do both target clones and resolvers.
27704 (ipa_dispatcher_calls): Remove the pass.
27705 (pass_dispatcher_calls::gate): Likewise.
27706 (make_pass_dispatcher_calls): Likewise.
27707 * passes.def (pass_target_clone): Put as very first IPA early
27710 2017-02-03 Martin Liska <mliska@suse.cz>
27712 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
27713 in case of a function with ifunc attribute.
27715 2017-02-03 Martin Liska <mliska@suse.cz>
27717 * cgraph.c (cgraph_node::dump): Dump function version info.
27718 * symtab.c (symtab_node::dump_base): Add missing new line.
27720 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27722 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
27723 (ifcombine_ifandif): Use it.
27725 2017-02-03 Martin Liska <mliska@suse.cz>
27727 * doc/invoke.texi: Document default value for
27728 use-after-scope-direct-emission-threshold.
27730 2017-02-03 Martin Liska <mliska@suse.cz>
27732 PR tree-optimization/79339
27733 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
27734 (format_floating): Likewise.
27736 2017-02-03 Martin Liska <mliska@suse.cz>
27739 * ipa-prop.c (ipa_node_params_t::insert): Remove current
27741 (ipa_node_params_t::remove): Likewise.
27742 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
27743 initialization from removed ipa_node_params_t::insert.
27744 (ipa_node_params::~ipa_node_params): Move from removed
27745 ipa_node_params_t::release.
27746 * symbol-summary.h (symbol_summary::m_released): New member.
27747 Do not release a summary twice. Do not allow to call finalizer
27748 for types of a summary that live in GGC memory.
27750 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27752 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
27755 2017-02-02 Martin Sebor <msebor@redhat.com>
27757 PR middle-end/79275
27758 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
27759 (format_string): Tighten up the range of output for non-constant
27760 strings and correct the expected range for wide non-constant strings.
27762 2017-02-02 Martin Sebor <msebor@redhat.com>
27764 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
27766 PR middle-end/32003
27767 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
27769 (-fdump-tree-@var): Add to index and document how to come up
27770 with pass-specific option and dump file names.
27771 (-fdump-passes): Clarify where to look for output.
27773 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27775 PR middle-end/77445
27776 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
27777 statistics of the analyzed path; allow threading for speed when
27778 any of BBs along the path are optimized for speed.
27780 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
27782 PR middle-end/78468
27783 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
27784 settings of the virtual registers.
27787 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
27789 * explow.c (get_dynamic_stack_size): Take known alignment of stack
27790 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
27793 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27795 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
27796 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
27798 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27800 * config/s390/s390.md: Add missing comments with the expanded
27802 * config/s390/vector.md: Likewise.
27803 * config/s390/vx-builtins.md: Likewise.
27805 2017-02-02 Jakub Jelinek <jakub@redhat.com>
27808 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
27809 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
27810 conditions on a single line.
27812 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27814 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27815 __S390_VX__ to __VX__.
27817 2017-02-01 Andrew Pinski <apinski@cavium.com>
27819 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
27820 stmt_info to record_stmt_cost.
27821 (vect_get_known_peeling_cost): Pass stmt_info if known to
27823 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
27824 cpu_vector_cost field into
27825 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
27826 field into vec_int_stmt_cost and vec_fp_stmt_cost.
27827 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
27828 splitting of scalar_stmt_cost and vec_stmt_cost.
27829 (thunderx_vector_cost): Likewise.
27830 (cortexa57_vector_cost): LIkewise.
27831 (exynosm1_vector_cost): Likewise.
27832 (xgene1_vector_cost): Likewise.
27833 (thunderx2t99_vector_cost): Improve after the splitting of the two
27835 (aarch64_builtin_vectorization_cost): Update for the splitting of
27836 scalar_stmt_cost and vec_stmt_cost.
27838 2017-02-01 Torvald Riegel <triegel@redhat.com>
27839 Richard Henderson <rth@redhat.com>
27841 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
27842 conditional on existance of a fast atomic load.
27843 * optabs-query.c (can_atomic_load_p): New function.
27844 * optabs-query.h (can_atomic_load_p): Declare it.
27845 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
27846 no fast atomic load is available for the particular size of access.
27847 (expand_atomic_compare_and_swap): Likewise.
27848 (expand_atomic_load): Likewise.
27849 (expand_atomic_store): Likewise.
27850 (expand_atomic_fetch_op): Likewise.
27851 * testsuite/lib/target-supports.exp
27852 (check_effective_target_sync_int_128): Remove x86 because it provides
27853 no fast atomic load.
27854 (check_effective_target_sync_int_128_runtime): Likewise.
27856 2017-02-01 Richard Biener <rguenther@suse.de>
27858 * graphite.c: Include tree-vectorizer.h for find_loop_location.
27859 (graphite_transform_loops): Provide opt-info for optimized nests.
27860 * tree-parloop.c (parallelize_loops): Provide opt-info for
27861 parallelized loops.
27863 2017-02-01 Richard Biener <rguenther@suse.de>
27865 PR middle-end/79315
27866 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
27867 was not set before.
27869 2017-02-01 Richard Biener <rguenther@suse.de>
27871 PR tree-optimization/71824
27872 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27873 Verify the loops are valid in the merged SESE region.
27874 (scop_detection::can_represent_loop_1): Check analyzing the
27875 evolution of the number of iterations in the region succeeds.
27877 2017-01-31 Ian Lance Taylor <iant@golang.org>
27879 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
27880 REG_ARGS_SIZE note to 32-bit push insns and call insn.
27882 2017-01-31 David Malcolm <dmalcolm@redhat.com>
27884 PR preprocessor/79210
27885 * input.c (get_substring_ranges_for_loc): Replace line_width
27886 assertion with error-handling.
27888 2017-01-31 Richard Biener <rguenther@suse.de>
27890 PR tree-optimization/77318
27891 * graphite-sese-to-poly.c (extract_affine): Fix assert.
27892 (create_pw_aff_from_tree): Take loop parameter.
27893 (add_condition_to_pbb): Pass loop of the condition to
27894 create_pw_aff_from_tree.
27896 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27898 * config/s390/s390.c (s390_asan_shadow_offset): New function.
27899 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
27901 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
27905 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
27907 (convert_int_to_float128): Likewise.
27908 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
27909 (convert_int_to_float128): Likewise.
27910 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
27911 (UNSPEC_IEEE128_CONVERT): Likewise.
27912 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
27913 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
27914 Use local variables for IBM extended format.
27915 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
27916 (fix_trunc<mode>si2_fprs): Likewise.
27917 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
27918 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
27919 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
27920 to know that we can now have integers of all sizes in vector
27922 (fix<uns>_<mode>di2_hw): Likewise.
27923 (float<uns>_<mode>si2_hw): Likewise.
27924 (fix_<mode>si2_hw): Likewise.
27925 (fixuns_<mode>si2_hw): Likewise.
27926 (float<uns>_<mode>di2_hw): Likewise.
27927 (float_<mode>di2_hw): Likewise.
27928 (float_<mode>si2_hw): Likewise.
27929 (floatuns_<mode>di2_hw): Likewise.
27930 (floatuns_<mode>si2_hw): Likewise.
27931 (xscvqp<su>wz_<mode>): Delete, no longer used.
27932 (xscvqp<su>dz_<mode>): Likewise.
27933 (xscv<su>dqp_<mode>): Likewise.
27934 (ieee128_mfvsrd_64bit): Likewise.
27935 (ieee128_mfvsrd_32bit): Likewise.
27936 (ieee128_mfvsrwz): Likewise.
27937 (ieee128_mtvsrw): Likewise.
27938 (ieee128_mtvsrd_64bit): Likewise.
27939 (ieee128_mtvsrd_32bit): Likewise.
27941 2017-01-31 Martin Liska <mliska@suse.cz>
27944 * ipa-prop.c (ipa_free_all_node_params): Call release method
27945 instead of ~sumbol_summary to not to trigger double times
27948 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
27950 PR tree-optimization/71691
27951 * bitmap.h (class auto_bitmap): New.
27952 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
27953 is_maybe_undefined instead of ssa_undefined_value_p.
27955 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27957 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27958 __S390_ARCH_LEVEL__ to __ARCH__.
27960 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27962 PR tree-optimization/79267
27963 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
27964 if should_remove_lhs_p is true.
27966 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
27969 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
27970 (add_alignment_attribute): New.
27971 (base_type_die): Add alignment attribute.
27972 (subrange_type_die): Likewise.
27973 (modified_type_die): Likewise.
27974 (gen_array_type_die): Likewise.
27975 (gen_descr_array_type_die: Likewise.
27976 (gen_enumeration_type_die): Likewise.
27977 (gen_subprogram_die): Likewise.
27978 (gen_variable_die): Likewise.
27979 (gen_field_die): Likewise.
27980 (gen_ptr_to_mbr_type_die): Likewise.
27981 (gen_struct_or_union_type_die): Likewise.
27982 (gen_subroutine_type_die): Likewise.
27983 (gen_typedef_die): Likewise.
27984 (base_type_cmp): Compare alignment attribute.
27986 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27989 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
27990 (setb_unsigned) New pattern for setb with CCUNS.
27991 * config/rs6000/rs6000.c (expand_block_compare): Use a different
27992 subfc./subfe sequence to avoid overflow problems. Generate a
27993 shorter sequence with cmpld/setb for power9.
27994 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
27995 for generating subfc. instruction.
27996 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
27997 now uses this instruction.
27999 2017-01-30 Ian Lance Taylor <iant@google.com>
28002 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
28003 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
28005 2017-01-30 Martin Sebor <msebor@redhat.com>
28007 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
28008 Move constant to the right of a relational operator.
28009 (get_mpfr_format_length, format_character, format_string): Ditto.
28010 (should_warn_p, maybe_warn): Same.
28012 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
28014 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
28017 * asan.c (get_translation_unit_decl): Remove function.
28018 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
28020 2017-01-30 Martin Liska <mliska@suse.cz>
28022 PR gcov-profile/79259
28023 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
28024 -fprofile-generate.
28026 2017-01-30 Martin Liska <mliska@suse.cz>
28029 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
28030 Initialize variables with NULL value.
28032 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
28035 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
28037 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
28039 2017-01-30 Richard Biener <rguenther@suse.de>
28041 PR tree-optimization/79276
28042 * tree-vrp.c (process_assert_insertions): Properly adjust common
28043 when removing a duplicate.
28045 * gcc.dg/torture/pr79276.c: New testcase.
28047 2017-01-30 Richard Biener <rguenther@suse.de>
28049 PR tree-optimization/79256
28050 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28051 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28053 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
28055 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
28058 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
28059 ("*r<noxa>sbg_<mode>_sll_bitmask")
28060 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
28061 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
28062 Use contiguous_bitmask_nowrap_operand.
28064 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28067 * config/rs6000/altivec.h (vec_xl): Revise #define.
28068 (vec_xst): Likewise.
28070 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28072 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28074 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28076 PR rtl-optimization/79194
28077 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28078 traps before call to bypass_conditional_jumps.
28080 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28082 PR tree-optimization/71374
28083 * lra-constraints.c (check_conflict_input_operands): New.
28084 (match_reload): Use it.
28086 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28089 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28090 account to calculate conflict_set.
28092 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28094 PR rtl-optimization/78559
28095 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28096 other_insn in combine.
28098 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28100 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28101 uint16_type_node for BT_UINT16.
28103 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28105 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28106 "RTL Tests" to menu.
28107 (GIMPLE Tests): New node.
28108 (RTL Tests): New node.
28110 2017-01-27 Richard Biener <rguenther@suse.de>
28112 PR tree-optimization/79245
28113 * tree-loop-distribution.c (distribute_loop): Apply cost
28114 modeling also to detected patterns.
28116 2017-01-27 Richard Biener <rguenther@suse.de>
28118 PR tree-optimization/71433
28119 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28120 (compare_assert_loc): New function.
28121 (process_assert_insertions): Sort and optimize assert locations
28122 to remove duplicates and push down identical assertions on
28123 edges to their destination block.
28125 2017-01-27 Richard Biener <rguenther@suse.de>
28127 PR tree-optimization/79244
28128 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28129 out SSA names even if abnormal.
28131 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28133 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28134 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28135 instead of MPFR_RNDN.
28137 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28140 * arm.c (arm_option_override): Don't call build_target_option_node
28141 until after doing all option overrides.
28142 (arm_valid_target_attribute_tree): Likewise.
28144 2017-01-27 Martin Liska <mliska@suse.cz>
28146 * doc/invoke.texi (-fprofile-arcs): Document profiling support
28147 for {cd}tors and C++ {cd}tors.
28149 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
28151 * config/s390/s390.md ("*setmem_long_and")
28152 ("*setmem_long_and_31z"): Use zero_extend instead of and.
28154 2017-01-26 Martin Sebor <msebor@redhat.com>
28156 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
28159 2017-01-26 Martin Sebor <msebor@redhat.com>
28161 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
28162 HAVE_DFmode before using XFmode or DFmode.
28163 (parse_directive): Avoid using the z length modifier to avoid
28164 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
28166 PR middle-end/78703
28167 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
28168 to accept adjustment as an array.
28169 (get_int_range): New function.
28170 (struct directive): Make width and prec arrays.
28171 (directive::set_width, directive::set_precision): Call get_int_range.
28172 (format_integer, format_floating): Handle width and precision ranges.
28173 (format_string, parse_directive): Same.
28175 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28178 * dwarf2out.c (generate_skeleton_bottom_up): For children with
28179 comdat_type_p set, just clone them, but keep the children in the
28183 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
28184 which have direct callers with -fvar-tracking-assignments enabled
28186 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
28187 inside of type units.
28189 2017-01-26 Martin Sebor <msebor@redhat.com>
28191 PR middle-end/78703
28192 * gimple-ssa-sprintf.c (struct result_range): Add likely and
28194 (struct format_result): Replace number_chars, number_chars_min,
28195 and number_chars_max with a single member of struct result_range.
28197 (format_result::operator+=): Adjust.
28198 (struct fmtresult): Remove bounded. Handle likely and unlikely
28200 (fmtresult::adjust_for_width_or_precision): New function.
28201 (fmtresult:type_max_digits): New function.
28202 (bytes_remaining): Handle likely and unlikely counters.
28203 (min_bytes_remaining): Remove.
28204 (format_percent): Simplify.
28205 (format_integer, format_floating): Set likely and unlikely counters.
28206 (get_string_length, format_character, format_string): Same.
28207 (format_plain, should_warn_p): New function.
28208 (maybe_warn): Call should_warn_p. Update diagnostic messages
28209 and handle those for all directives, including plain strings.
28210 (format_directive): Handle likely and unlikely counters.
28211 Remove unnecessary quoting from diagnostics. Add an informational
28213 (add_bytes): Remove.
28214 (pass_sprintf_length::compute_format_length): Simplify.
28215 (try_substitute_return_value): Handle likely and unlikely counters.
28217 2017-01-26 Carl Love <cel@us.ibm.com>
28219 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
28220 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
28222 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
28225 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
28226 endianess for subregs into account.
28227 * lra-constraints.c (lra_constraints): Do risky transformations
28228 always on the first iteration.
28229 * lra-lives.c (check_pseudos_live_through_calls): Add arg
28230 last_call_used_reg_set.
28231 (process_bb_lives): Define and use last_call_used_reg_set.
28232 * lra.c (lra): Always continue after lra_constraints on the first
28235 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
28237 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
28239 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
28241 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28243 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
28244 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
28245 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
28246 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
28247 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
28248 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
28249 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
28250 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
28251 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
28253 2017-01-26 Marek Polacek <polacek@redhat.com>
28256 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
28257 for the third operand.
28259 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28261 PR middle-end/79236
28262 * omp-low.c (struct omp_context): Add simt_stmt field.
28263 (scan_omp_for): Return omp_context *.
28264 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
28265 context to the _simt_ SIMD stmt.
28266 (lower_omp_for): For combined SIMD with sibling _simt_
28267 SIMD, make sure to use the same decls in _looptemp_
28268 clauses as in the sibling.
28270 2017-01-26 David Sherwood <david.sherwood@arm.com>
28272 PR middle-end/79212
28273 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
28276 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28279 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
28280 emit fld b; fld a; if possible.
28282 * brig-builtins.def: Update copyright years.
28283 * config/arm/arm_acle_builtins.def: Update copyright years.
28285 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
28288 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
28289 constraint instead of o for the stxsd instruction.
28291 2017-01-25 Carl Love <cel@us.ibm.com>
28293 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
28294 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
28296 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
28298 * doc/invoke.texi (C++ Dialect Options): Fix typo.
28300 2017-01-25 Richard Biener <rguenther@suse.de>
28302 PR tree-optimization/69264
28303 * target.def (vector_alignment_reachable): Improve documentation.
28304 * doc/tm.texi: Regenerate.
28305 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
28307 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
28308 earlier changes with respect to TYPE_USER_ALIGN.
28309 (vector_alignment_reachable_p): Likewise. Improve dumping.
28311 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28314 * config/arm/arm.md (xordi3): Force constant operand into a register
28317 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28319 * doc/invoke.texi (-fstore-merging): Correct default optimization
28320 levels at which it is enabled.
28321 (-O): Move -fstore-merging from list to...
28324 2017-01-25 Richard Biener <rguenther@suse.de>
28327 * omp-expand.c: Include debug.h.
28328 (expand_omp_taskreg): Make sure to generate early debug before
28329 outlining anything from a function.
28330 (expand_omp_target): Likewise.
28331 (grid_expand_target_grid_body): Likewise.
28333 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
28336 * asan.c (get_translation_unit_decl): New function.
28337 (asan_add_global): Extract modules file name from globals
28338 TRANSLATION_UNIT_DECL name.
28340 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
28343 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
28344 for long calls with APCS frame and VFP.
28346 2017-01-24 David Malcolm <dmalcolm@redhat.com>
28348 * cfg.c (original_copy_tables_initialized_p): New function.
28349 * cfg.h (original_copy_tables_initialized_p): New decl.
28350 * cfgrtl.c (relink_block_chain): Guard the call to
28351 free_original_copy_tables with a call to
28352 original_copy_tables_initialized_p.
28353 * cgraph.h (symtab_node::native_rtl_p): New decl.
28354 * cgraphunit.c (symtab_node::native_rtl_p): New function.
28355 (symtab_node::needed_p): Don't assert for early assembly output
28356 for __RTL functions.
28357 (cgraph_node::finalize_function): Set "force_output" for __RTL
28359 (cgraph_node::analyze): Bail out early for __RTL functions.
28360 (analyze_functions): Update assertion to support __RTL functions.
28361 (cgraph_node::expand): Bail out early for __RTL functions.
28362 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
28364 * function.h (struct function): Update comment for field
28366 * gimple-expr.c: Include "tree-pass.h".
28367 (gimple_has_body_p): Return false for __RTL functions.
28368 * Makefile.in (OBJS): Add run-rtl-passes.o.
28369 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
28371 (gcc::pass_manager::get_clean_slate): New accessor.
28372 * passes.c: Include "insn-addr.h".
28373 (should_skip_pass_p): Add logging. Update logic for running
28374 "expand" to be compatible with both __GIMPLE and __RTL. Guard
28375 property-provider override so it is only done for gimple passes.
28377 (skip_pass): New function.
28378 (execute_one_pass): Call skip_pass when skipping passes.
28379 * read-md.c (md_reader::read_char): Support filtering
28380 the input to a subset of line numbers.
28381 (md_reader::md_reader): Initialize fields
28382 m_first_line and m_last_line.
28383 (md_reader::read_file_fragment): New function.
28384 * read-md.h (md_reader::read_file_fragment): New decl.
28385 (md_reader::m_first_line): New field.
28386 (md_reader::m_last_line): New field.
28387 * read-rtl-function.c (function_reader::create_function): Only
28388 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
28389 curr_properties. Set DECL_INITIAL to a dummy block.
28390 (read_rtl_function_body_from_file_range): New function.
28391 * read-rtl-function.h (read_rtl_function_body_from_file_range):
28393 * run-rtl-passes.c: New file.
28394 * run-rtl-passes.h: New file.
28396 2017-01-24 Jeff Law <law@redhat.com>
28398 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
28401 2017-01-24 Bin Cheng <bin.cheng@arm.com>
28403 PR tree-optimization/79159
28404 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
28405 (record_nonwrapping_iv): Improve boundary using above function if no
28406 value range information.
28408 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
28409 Martin Jambor <mjambor@suse.cz>
28411 * brig-builtins.def: New file.
28412 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
28413 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
28414 (DEF_HSAIL_SAT_BUILTIN): Likewise.
28415 (DEF_HSAIL_INTR_BUILTIN): Likewise.
28416 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
28417 * builtin-types.def (BT_INT8): New.
28418 (BT_INT16): Likewise.
28419 (BT_UINT8): Likewise.
28420 (BT_UINT16): Likewise.
28421 (BT_FN_ULONG): Likewise.
28422 (BT_FN_UINT_INT): Likewise.
28423 (BT_FN_UINT_ULONG): Likewise.
28424 (BT_FN_UINT_LONG): Likewise.
28425 (BT_FN_UINT_PTR): Likewise.
28426 (BT_FN_ULONG_PTR): Likewise.
28427 (BT_FN_INT8_FLOAT): Likewise.
28428 (BT_FN_INT16_FLOAT): Likewise.
28429 (BT_FN_UINT32_FLOAT): Likewise.
28430 (BT_FN_UINT16_FLOAT): Likewise.
28431 (BT_FN_UINT8_FLOAT): Likewise.
28432 (BT_FN_UINT64_FLOAT): Likewise.
28433 (BT_FN_UINT16_UINT32): Likewise.
28434 (BT_FN_UINT32_UINT16): Likewise.
28435 (BT_FN_UINT16_UINT16_UINT16): Likewise.
28436 (BT_FN_INT_PTR_INT): Likewise.
28437 (BT_FN_UINT_PTR_UINT): Likewise.
28438 (BT_FN_LONG_PTR_LONG): Likewise.
28439 (BT_FN_ULONG_PTR_ULONG): Likewise.
28440 (BT_FN_VOID_UINT64_UINT64): Likewise.
28441 (BT_FN_UINT8_UINT8_UINT8): Likewise.
28442 (BT_FN_INT8_INT8_INT8): Likewise.
28443 (BT_FN_INT16_INT16_INT16): Likewise.
28444 (BT_FN_INT_INT_INT): Likewise.
28445 (BT_FN_UINT_FLOAT_UINT): Likewise.
28446 (BT_FN_FLOAT_UINT_UINT): Likewise.
28447 (BT_FN_ULONG_UINT_UINT): Likewise.
28448 (BT_FN_ULONG_UINT_PTR): Likewise.
28449 (BT_FN_ULONG_ULONG_ULONG): Likewise.
28450 (BT_FN_UINT_UINT_UINT): Likewise.
28451 (BT_FN_VOID_UINT_PTR): Likewise.
28452 (BT_FN_UINT_UINT_PTR: Likewise.
28453 (BT_FN_UINT32_UINT64_PTR): Likewise.
28454 (BT_FN_INT_INT_UINT_UINT): Likewise.
28455 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
28456 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
28457 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
28458 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
28459 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
28460 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
28461 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
28462 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
28463 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
28464 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
28465 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
28466 * doc/frontends.texi: List BRIG FE.
28467 * doc/install.texi (Testing): Add BRIG tesring requirements.
28468 * doc/invoke.texi (Overall Options): Mention BRIG.
28469 * doc/standards.texi (Standards): Doucment BRIG HSA version.
28471 2017-01-24 Richard Biener <rguenther@suse.de>
28473 PR translation/79208
28474 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
28476 2017-01-24 Martin Jambor <mjambor@suse.cz>
28479 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
28480 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
28481 and known_contexts.
28483 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
28485 PR middle-end/79123
28486 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
28487 casts from signed to unsigned really don't have a range.
28489 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
28491 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
28492 GMP_RNDx for compatiblity.
28494 2017-01-24 Martin Liska <mliska@suse.cz>
28497 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
28498 that would prevent us to call alloca with -1 as argument.
28500 2017-01-24 Jakub Jelinek <jakub@redhat.com>
28502 * dwarf2out.c (output_compilation_unit_header, output_file_names):
28503 Avoid -Wformat-security warning.
28505 2017-01-23 Andrew Pinski <apinski@cavium.com>
28507 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
28510 2017-01-23 Martin Sebor <msebor@redhat.com>
28512 PR middle-end/78703
28513 * gimple-ssa-sprintf.c (warn_level): New global.
28514 (format_integer): Use it here and throughout the rest of the file.
28515 Use the same switch to compute sign as base.
28516 (maybe_warn): New function.
28517 (format_directive): Factor out warnings into maybe_warn.
28518 Add debugging output. Use warn_level.
28519 (add_bytes): Use warn_level.
28520 (pass_sprintf_length::compute_format_length): Add debugging output.
28521 (try_substitute_return_value): Same.
28522 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
28524 PR middle-end/78703
28525 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
28526 (struct fmtresult, format_integer, format_floating): Adjust.
28527 (fmtresult::fmtresult): Set max correctly in two argument ctor.
28528 (get_string_length, format_string,format_directive): Same.
28529 (pass_sprintf_length::compute_format_length): Same.
28530 (try_substitute_return_value): Simplify slightly.
28532 PR middle-end/78703
28533 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
28534 (fmtresult::operator+=): Outlined.
28535 (struct fmtresult): Add ctors.
28536 (struct conversion_spec): Rename...
28537 (struct directive): ...to this. Add and remove data members.
28538 (directive::set_width, directive::set_precision): New functions.
28539 (format_percent): Use fmtresult ctor.
28540 (get_width_and_precision): Remove.
28541 (format_integer): Make naming changes. Avoid computing width and
28543 (format_floating): Same. Adjust indentation.
28544 (format_character, format_none): New functions.
28545 (format_string): Moved character handling to format_character.
28546 (format_directive): Remove arguments, change return type.
28547 (parse_directive): New function.
28548 (pass_sprintf_length::compute_format_length): Move directive
28549 parsing to parse_directive.
28551 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28553 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
28554 (assign_assembler_name_if_needed): ... this.
28555 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
28556 (assign_assembler_name_if_needed): ... this.
28557 (free_lang_data_in_cgraph): Adjust callers.
28558 * cgraphunit.c (cgraph_node::analyze): Likewise.
28559 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
28562 2017-01-23 Richard Biener <rguenther@suse.de>
28564 PR tree-optimization/79088
28565 PR tree-optimization/79188
28566 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
28567 resetting loop bounds after last path deletion. Reset loop
28568 bounds of the target loop, make code match the comments.
28569 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
28570 Make sure loops need no fixups.
28572 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28574 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
28575 exponent support with double type for first argument.
28576 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
28577 type returned by __builtin_vec_extract_sig,
28578 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
28579 functions from "vector int" to "vector unsigned int" or from
28580 "vector long long int" to "vector unsigned long long int".
28581 Changed type returned by __builtin_vec_extract_exp,
28582 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
28583 functions from "vector int" to "vector unsigned int" or from
28584 "vector long long int" to "vector unsigned long long int".
28585 Changed return type of __builtin_vec_test_data_class,
28586 __builtin_vec_test_data_class_sp, and
28587 __builtin_vec_test_data_class_dp from "vector int" to
28588 "vector bool int" or from "vector long long int" to "vector bool
28589 long long int" and changed second argument type from "unsigned
28590 int" to "int". Added new overloaded function forms "vector float
28591 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
28592 "vector float __builtin_vec_insert_exp_sp (vector float, vector
28593 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
28594 double, vector unsigned long long int)" and "vector double
28595 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
28596 long int)". Changed return type of
28597 __builtin_scalar_test_data_class and
28598 __builtin_scalar_test_data_class_sp and
28599 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
28600 int" and changed second argument from "unsigned int" to "int".
28601 Changed type returned by __builtin_scalar_test_neg,
28602 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
28603 from "int" to "bool int". Added new overloaded function form
28604 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
28605 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
28606 exponent double-precision with floating point first argument.
28607 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
28608 documentation of scalar_test_data_class, scalar_test_neg,
28609 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
28610 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
28611 vec_test_data_class built-in functions to reflect refinements in
28612 their type signatures.
28614 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
28616 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
28618 (aarch64_elf_asm_destructor): Likewise.
28620 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
28622 PR rtl-optimization/78634
28623 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
28624 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
28625 * ifcvt.c (noce_try_cmove): Add missing cost check.
28627 PR rtl-optimization/71724
28628 * combine.c (if_then_else_cond): Look for situations where it is
28629 beneficial to undo the work of one of the recursive calls.
28631 2017-01-23 Bin Cheng <bin.cheng@arm.com>
28633 PR tree-optimization/70754
28634 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
28635 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
28636 combined stmt before it if not NULL.
28637 (combine_chains): Process refs reversely and compute dominance point
28640 2017-01-23 Martin Liska <mliska@suse.cz>
28642 PR tree-optimization/79196
28643 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
28644 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
28646 (strlen_optimize_stmt): Call the renamed function.
28648 2017-01-23 Michael Matz <matz@suse.de>
28650 PR tree-optimization/78384
28651 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
28653 2017-01-23 Richard Biener <rguenther@suse.de>
28655 PR tree-optimization/79186
28656 * tree-vrp.c (register_new_assert_for): Make sure we've seen
28657 both incoming edges before moving an assert.
28659 2017-01-23 Martin Jambor <mjambor@suse.cz>
28661 * ipa-prop.c (load_from_param_1): Removed.
28662 (load_from_unmodified_param): Bits from load_from_param_1 put back
28664 (load_from_param): Removed.
28665 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
28666 with stmt. Reverted back to use of load_from_unmodified_param.
28668 2017-01-23 Martin Jambor <mjambor@suse.cz>
28671 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
28672 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
28673 field a pointer to garbage collected vector, mark lattices and
28674 ipcp_orig_node with GTY((skip)).
28675 (ipa_get_param_count): Adjust to descriptors being a pointer.
28676 (ipa_get_param): Likewise.
28677 (ipa_get_type): Likewise.
28678 (ipa_get_param_move_cost): Likewise.
28679 (ipa_set_param_used): Likewise.
28680 (ipa_get_controlled_uses): Likewise.
28681 (ipa_set_controlled_uses): Likewise.
28682 (ipa_is_param_used): Likewise.
28683 (ipa_node_params_t): Move into garbage collector. New methods insert
28685 (ipa_node_params_sum): Annotate wth GTY(()).
28686 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
28688 (ipa_load_from_parm_agg): Adjust declaration.
28689 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
28690 * ipa-profile.c (ipa_profile): Likewise.
28691 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
28692 (ipa_populate_param_decls): Make descriptors parameter garbage
28694 (ipa_dump_param): Adjust to descriptors being a pointer.
28695 (ipa_alloc_node_params): Likewise.
28696 (ipa_initialize_node_params): Likewise.
28697 (load_from_param_1): Make descriptors parameter garbage collected.
28698 (load_from_unmodified_param): Likewise.
28699 (load_from_param): Likewise.
28700 (ipa_load_from_parm_agg): Likewise.
28701 (ipa_node_params::~ipa_node_params): Removed.
28702 (ipa_free_all_node_params): Remove call to delete operator.
28703 (ipa_node_params_t::insert): New.
28704 (ipa_node_params_t::remove): Likewise.
28705 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
28706 copy known_csts and known_contexts vectors.
28707 (ipa_read_node_info): Adjust to descriptors being a pointer.
28708 (ipcp_modif_dom_walker): Make m_descriptors field garbage
28710 (ipcp_transform_function): Make descriptors variable garbage
28713 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
28715 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
28716 * config/i386/avx512dqintrin.h: Ditto.
28717 * config/i386/avx512fintrin.h: Ditto.
28718 * config/i386/i386.c: Handle new builtins.
28719 * config/i386/i386-builtin.def: Add new builtins.
28720 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
28721 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
28723 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28724 Martin Liska <mliska@suse.cz>
28726 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
28727 * asan.c (asan_expand_poison_ifn): Support stores and use
28728 appropriate ASAN report function.
28729 * internal-fn.c (expand_ASAN_POISON_USE): New function.
28730 * internal-fn.def (ASAN_POISON_USE): Declare.
28731 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
28732 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
28733 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
28734 ASAN_POISON calls w/o LHS.
28735 * tree-ssa.c (execute_update_addresses_taken): Create clobber
28736 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
28737 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
28738 * gimplify.c (asan_poison_variables): Add attribute
28739 use_after_scope_memory to variables that really needs to live
28741 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
28742 having the attribute.
28744 2017-01-23 Martin Liska <mliska@suse.cz>
28746 * asan.c (create_asan_shadow_var): New function.
28747 (asan_expand_poison_ifn): Likewise.
28748 * asan.h (asan_expand_poison_ifn): New declaration.
28749 * internal-fn.c (expand_ASAN_POISON): Likewise.
28750 * internal-fn.def (ASAN_POISON): New builtin.
28751 * sanopt.c (pass_sanopt::execute): Expand
28752 asan_expand_poison_ifn.
28753 * tree-inline.c (copy_decl_for_dup_finish): Make function
28755 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
28756 * tree-ssa.c (is_asan_mark_p): New function.
28757 (execute_update_addresses_taken): Rewrite local variables
28758 (identified just by use-after-scope as addressable) into SSA.
28760 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28762 * doc/install.texi (Specific): opensource.apple.com uses https
28763 now. Remove trailing slash.
28765 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28767 * README.Portability: Remove note on an Irix compatibility issue.
28769 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
28771 * gcov.c (INCLUDE_ALGORITHM): Define.
28772 (INCLUDE_VECTOR): Define.
28773 No longer include <vector> and <algorithm> directly.
28775 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
28777 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
28779 * doc/invoke.texi (Code Gen Options): Ditto.
28781 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
28784 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
28786 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
28788 rtl-optimization/79125
28789 * cprop.c (local_cprop_pass): Handle cases where we make an
28790 unconditional trap.
28792 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
28796 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
28797 read from, for big endian.
28799 2017-01-20 Jiong Wang <jiong.wang@arm.com>
28801 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
28802 register pauth builtins for LP64 only.
28804 2017-01-20 Marek Polacek <polacek@redhat.com>
28807 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
28810 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28812 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
28814 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
28815 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
28816 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
28818 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28821 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
28822 in big-endian mode when they are not a single duplicated value.
28824 2017-01-20 Richard Biener <rguenther@suse.de>
28826 * BASE-VER: Bump to 7.0.1.
28828 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28830 * omp-low.c (omplow_simd_context): New struct. Use it...
28831 (lower_rec_simd_input_clauses): ...here and...
28832 (lower_rec_input_clauses): ...here to hold common data. Adjust all
28833 references to idx, lane, max_vf, is_simt.
28835 2017-01-20 Graham Markall <graham.markall@embecosm.com>
28837 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
28840 2017-01-20 Martin Jambor <mjambor@suse.cz>
28842 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
28843 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
28845 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
28846 (GTFILES): Rename hsa.c to hsa-common.c.
28847 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
28848 * hsa-dump.c: Likewise.
28849 * hsa-gen.c: Likewise.
28850 * hsa-regalloc.c: Likewise.
28851 * ipa-hsa.c: Likewise.
28852 * omp-expand.c: Likewise.
28853 * omp-low.c: Likewise.
28854 * toplev.c: Likewise.
28856 2017-01-20 Marek Polacek <polacek@redhat.com>
28859 * doc/invoke.texi: Document -Wduplicated-branches.
28860 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
28861 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
28862 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
28863 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
28864 return 0 only when not OEP_LEXICOGRAPHIC.
28865 (fold_build_cleanup_point_expr): Use the expression
28866 location when building CLEANUP_POINT_EXPR.
28867 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
28868 * tree.c (add_expr): Handle error_mark_node.
28870 2017-01-20 Martin Liska <mliska@suse.cz>
28873 * tree-profile.c (init_ic_make_global_vars): Do not call
28875 (gimple_init_gcov_profiler): Likewise.
28877 2017-01-20 Martin Liska <mliska@suse.cz>
28880 * cgraph.h (maybe_create_reference): Remove argument and
28882 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
28884 * ipa-cp.c (create_specialized_node): Likewise.
28885 * symtab.c (symtab_node::maybe_create_reference): Handle
28886 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
28888 2017-01-20 Martin Liska <mliska@suse.cz>
28890 * read-rtl-function.c (function_reader::create_function): Use
28891 build_decl instread of build_decl_stat.
28893 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
28895 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
28896 * config/i386/avx512dqintrin.h: Ditto.
28897 * config/i386/avx512fintrin.h: Ditto.
28898 * config/i386/i386-builtin-types.def: Add new types.
28899 * config/i386/i386.c: Handle new types.
28900 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
28901 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
28902 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
28903 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
28904 (__builtin_ia32_kshiftridi): New.
28905 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
28907 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
28911 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
28912 define to rs6000_init_stack_protect_guard.
28913 (rs6000_init_stack_protect_guard): New function.
28915 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28916 Yunqiang Su <yunqiang.su@imgtec.com>
28918 * config.gcc (supported_defaults): Add madd4.
28919 (with_madd4): Add validation.
28920 (all_defaults): Add madd4.
28921 * config/mips/mips.opt (mmadd4): New option.
28922 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28924 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28926 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
28927 (ISA_HAS_FUSED_MADD4): Likewise.
28928 * doc/invoke.texi (-mmadd4): Document the new option.
28929 * doc/install.texi (--with-madd4): Document the new option.
28931 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28933 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
28934 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
28935 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
28936 (aarch64_init_pauth_hint_builtins): New.
28937 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
28938 (aarch64_expand_builtin): Expand new builtins.
28940 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28942 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
28943 * combine-stack-adj.c (no_unhandled_cfa): Handle
28944 REG_CFA_TOGGLE_RA_MANGLE.
28945 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
28946 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
28947 info for return address signing.
28948 (aarch64_expand_epilogue): Likewise.
28950 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28952 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
28953 * config/aarch64/aarch64-protos.h
28954 (aarch64_return_address_signing_enabled): New declaration.
28955 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
28957 (aarch64_expand_prologue): Sign return address before it's pushed onto
28959 (aarch64_expand_epilogue): Authenticate return address fetched from
28961 (aarch64_override_options): Sanity check for ILP32 and ISA level.
28962 (aarch64_attributes): New function attributes for "sign-return-address".
28963 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
28964 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
28965 ("*do_return"): Generate combined instructions according to key index.
28966 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
28967 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
28969 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
28970 * config/aarch64/aarch64.opt (msign-return-address=): New.
28971 * doc/extend.texi (AArch64 Function Attributes): Documents
28972 "sign-return-address=".
28973 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
28975 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28977 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
28978 overall option summary.
28980 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28982 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
28983 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
28984 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
28985 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
28987 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
28989 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
28990 -mpower9-minmax by default for -mcpu=power9.
28991 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
28992 128-bit floating point.
28994 2017-01-20 Alan Modra <amodra@gmail.com>
28996 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
28997 optimizing for size.
28999 2017-01-20 Alan Modra <amodra@gmail.com>
29002 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
29003 for strcmp and strncmp from corresponding builtin decl.
29005 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29007 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
29008 instead of i386/rtems-64.h.
29009 * config/i386/rtems-64.h: Remove.
29011 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29015 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
29017 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
29019 2017-01-19 Tamar Christina <tamar.christina@arm.com>
29021 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
29022 Change int to HOST_WIDE_INT.
29023 * config/aarch64/aarch64-protos.h
29024 (aarch64_simd_gen_const_vector_dup): Likewise.
29025 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
29027 2017-01-19 David Malcolm <dmalcolm@redhat.com>
29029 * langhooks-def.h (lhd_type_for_size): New decl.
29030 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
29031 * langhooks.c (lhd_type_for_size): New function, taken from
29034 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
29036 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
29037 define_bypass for CR latency.
29038 (power9-cracked-alu): Update bypass latency and remove power9-branch.
29039 (power9-alu2): Add define_bypass for CR latency.
29041 (power9-mul): Update insn latency.
29042 (power9-mul-compare): Update insn latency, bypass latency and remove
29045 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29047 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
29049 * config/aarch64/aarch64.md
29050 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
29051 aarch64_nopcrelative_literal_loads.
29052 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
29054 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
29056 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
29057 TARGET_LOONGSON_3A.
29058 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
29060 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
29063 * config.gcc (supported_defaults): Add lxc1-sxc1.
29064 (with_lxc1_sxc1): Add validation.
29065 (all_defaults): Add lxc1-sxc1.
29066 * config/mips/mips.opt (mlxc1-sxc1): New option.
29067 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29069 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29070 __mips_no_lxc1_sxc1.
29071 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29072 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29073 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29075 2017-01-19 Richard Biener <rguenther@suse.de>
29077 PR tree-optimization/72488
29078 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29079 sure to restore SSA info.
29080 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29082 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29084 PR rtl-optimization/79121
29085 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29086 of the inner type when shifting an extended value.
29088 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29091 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29092 interposable aliases.
29094 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29097 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29098 Use the evmergelohi instruction.
29099 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29100 (mov_sitf_e500_subreg8_2_be): Likewise.
29101 (mov_sitf_e500_subreg12_2_le): Likewise.
29102 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29103 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29104 (mov_sitf_e500_subreg8_2_le): Likewise.
29105 (mov_sitf_e500_subreg12_2_be): Likewise.
29107 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29109 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29110 attribute from vecsimple to vecperm.
29111 (altivec_vbpermq2): Likewise.
29113 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29116 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29118 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29119 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29120 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29121 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29122 case where N arg is SIZE_MAX.
29123 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29124 (cmpstrsi): Add pattern.
29126 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29128 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29129 __builtin_vec_revb builtins.
29130 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29131 built-in functions to support generation of the ISA 3.0 XXBR<x>
29132 vector byte reverse instructions.
29133 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29134 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29135 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29136 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29137 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29138 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29139 (P9V_BUILTIN_VEC_REVB): Likewise.
29140 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29141 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29142 (p9_xxbrq_v16qi): Likewise.
29143 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
29144 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
29145 (p9_xxbrh_v8hi): Likewise.
29146 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
29147 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
29148 vec_revb built-in functions.
29150 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
29152 PR rtl-optimization/78952
29153 * config/i386/i386.md (any_extract): New code iterator.
29154 (*insvqi_2): Use any_extract for source operand.
29155 (*insvqi_3): Use any_shiftrt for source operand.
29157 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29159 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
29161 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
29163 2017-01-18 Matthias Klose <doko@ubuntu.com>
29165 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
29167 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29169 * config/rs6000/altivec.h (vec_bperm): Change #define.
29170 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
29171 (altivec_vbpermq2): New define_insn.
29172 (altivec_vbpermd): Likewise.
29173 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
29174 function interface.
29175 (VBPERMD): Likewise.
29176 (VBPERM): New polymorphic function interface.
29177 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
29178 Add entries for P9V_BUILTIN_VEC_VBPERM.
29179 * doc/extend.texi: Add interfaces for vec_bperm.
29181 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29183 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
29184 first letter of error messages.
29185 (s390_resolve_overloaded_builtin): Likewise.
29186 * config/s390/s390.c (s390_expand_builtin): Likewise.
29187 (s390_invalid_arg_for_unprototyped_fn): Likewise.
29188 (s390_valid_target_attribute_inner_p): Likewise.
29189 * config/s390/s390.md ("tabort"): Likewise.
29191 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
29193 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
29194 (ISA_AVOID_DIV_HILO): New macro.
29195 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
29196 (ISA_HAS_DDIV): Likewise.
29198 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29200 * doc/invoke.texi (fabi-version): Correct number of occurrences.
29202 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29204 * doc/invoke.texi (fabi-version): Spelling fix.
29206 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29209 * doc/invoke.texi (fabi-version): Mention mangling fix for
29212 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29215 * doc/invoke.texi (fabi-version): Document discriminator mangling.
29217 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
29220 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
29221 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
29223 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
29225 (stack_protect_test): Ditto.
29226 * config/rs6000/rs6000.opt (mstack-protector-guard=,
29227 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
29229 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
29230 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
29231 -mstack-protector-guard-offset=.
29232 (RS/6000 and PowerPC Options): Ditto.
29234 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29236 * config/i386/i386.h (MASK_CLASS_P): New define.
29237 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
29238 there are no registers from different register sets also when
29239 mask registers are used. Update function comment.
29240 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
29241 to (*k/*r) and (*k/*km) alternatives.
29243 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29245 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
29246 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
29247 (EH_RETURN_HANDLER_RTX): New define.
29248 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
29249 Force frame pointer in EH return functions.
29250 (aarch64_expand_epilogue): Add barrier for eh_return.
29251 (aarch64_final_eh_return_addr): Remove.
29252 (aarch64_eh_return_handler_rtx): New function.
29253 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
29255 (aarch64_eh_return_handler_rtx): New prototype.
29257 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29259 * config/rs6000/altivec.h (vec_rlmi): New #define.
29260 (vec_vrlnm): Likewise.
29261 (vec_rlnm): Likewise.
29262 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
29263 (UNSPEC_VRLNM): Likewise.
29264 (VIlong): New mode iterator.
29265 (altivec_vrl<VI_char>mi): New define_insn.
29266 (altivec_vrl<VI_char>nm): Likewise.
29267 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
29269 (VRLDNM): Likewise.
29270 (RLNM): New polymorphic function entry.
29271 (VRLWMI): New monomorphic function entry.
29272 (VRLDMI): Likewise.
29273 (RLMI): New polymorphic function entry.
29274 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
29275 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
29276 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
29279 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29282 * dwarf2out.c (field_byte_offset): Restore the
29283 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
29284 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
29285 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
29288 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29291 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
29293 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29296 * dwarf2out.c (add_data_member_location_attribute): For constant
29297 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
29298 instead of DW_AT_data_member_location, DW_AT_bit_offset and
29299 DW_AT_byte_size attributes.
29301 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29303 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
29304 after forcing to constant memory when the code model is medium.
29306 2017-01-17 Julia Koval <julia.koval@intel.com>
29309 * config/i386/avx512fintrin.h
29310 (_mm512_i32gather_ps): Change __addr type to void const*.
29311 (_mm512_mask_i32gather_ps): Ditto.
29312 (_mm512_i32gather_pd): Ditto.
29313 (_mm512_mask_i32gather_pd): Ditto.
29314 (_mm512_i64gather_ps): Ditto.
29315 (_mm512_mask_i64gather_ps): Ditto.
29316 (_mm512_i64gather_pd): Ditto.
29317 (_mm512_mask_i64gather_pd): Ditto.
29318 (_mm512_i32gather_epi32): Ditto.
29319 (_mm512_mask_i32gather_epi32): Ditto.
29320 (_mm512_i32gather_epi64): Ditto.
29321 (_mm512_mask_i32gather_epi64): Ditto.
29322 (_mm512_i64gather_epi32): Ditto.
29323 (_mm512_mask_i64gather_epi32): Ditto.
29324 (_mm512_i64gather_epi64): Ditto.
29325 (_mm512_mask_i64gather_epi64): Ditto.
29326 (_mm512_i32scatter_ps): Change __addr type to void*.
29327 (_mm512_mask_i32scatter_ps): Ditto.
29328 (_mm512_i32scatter_pd): Ditto.
29329 (_mm512_mask_i32scatter_pd): Ditto.
29330 (_mm512_i64scatter_ps): Ditto.
29331 (_mm512_mask_i64scatter_ps): Ditto.
29332 (_mm512_i64scatter_pd): Ditto.
29333 (_mm512_mask_i64scatter_pd): Ditto.
29334 (_mm512_i32scatter_epi32): Ditto.
29335 (_mm512_mask_i32scatter_epi32): Ditto.
29336 (_mm512_i32scatter_epi64): Ditto.
29337 (_mm512_mask_i32scatter_epi64): Ditto.
29338 (_mm512_i64scatter_epi32): Ditto.
29339 (_mm512_mask_i64scatter_epi32): Ditto.
29340 (_mm512_i64scatter_epi64): Ditto.
29341 (_mm512_mask_i64scatter_epi64): Ditto.
29342 * config/i386/avx512pfintrin.h
29343 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
29344 (_mm512_mask_prefetch_i32gather_ps): Ditto.
29345 (_mm512_mask_prefetch_i64gather_pd): Ditto.
29346 (_mm512_mask_prefetch_i64gather_ps): Ditto.
29347 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
29348 (_mm512_prefetch_i32scatter_ps): Ditto.
29349 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
29350 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
29351 (_mm512_prefetch_i64scatter_pd): Ditto.
29352 (_mm512_prefetch_i64scatter_ps): Ditto.
29353 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
29354 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
29355 * config/i386/avx512vlintrin.h
29356 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
29357 (_mm_mmask_i32gather_ps): Ditto.
29358 (_mm256_mmask_i32gather_pd): Ditto.
29359 (_mm_mmask_i32gather_pd): Ditto.
29360 (_mm256_mmask_i64gather_ps): Ditto.
29361 (_mm_mmask_i64gather_ps): Ditto.
29362 (_mm256_mmask_i64gather_pd): Ditto.
29363 (_mm_mmask_i64gather_pd): Ditto.
29364 (_mm256_mmask_i32gather_epi32): Ditto.
29365 (_mm_mmask_i32gather_epi32): Ditto.
29366 (_mm256_mmask_i32gather_epi64): Ditto.
29367 (_mm_mmask_i32gather_epi64): Ditto.
29368 (_mm256_mmask_i64gather_epi32): Ditto.
29369 (_mm_mmask_i64gather_epi32): Ditto.
29370 (_mm256_mmask_i64gather_epi64): Ditto.
29371 (_mm_mmask_i64gather_epi64): Ditto.
29372 (_mm256_i32scatter_ps): Change __addr type to void*.
29373 (_mm256_mask_i32scatter_ps): Ditto.
29374 (_mm_i32scatter_ps): Ditto.
29375 (_mm_mask_i32scatter_ps): Ditto.
29376 (_mm256_i32scatter_pd): Ditto.
29377 (_mm256_mask_i32scatter_pd): Ditto.
29378 (_mm_i32scatter_pd): Ditto.
29379 (_mm_mask_i32scatter_pd): Ditto.
29380 (_mm256_i64scatter_ps): Ditto.
29381 (_mm256_mask_i64scatter_ps): Ditto.
29382 (_mm_i64scatter_ps): Ditto.
29383 (_mm_mask_i64scatter_ps): Ditto.
29384 (_mm256_i64scatter_pd): Ditto.
29385 (_mm256_mask_i64scatter_pd): Ditto.
29386 (_mm_i64scatter_pd): Ditto.
29387 (_mm_mask_i64scatter_pd): Ditto.
29388 (_mm256_i32scatter_epi32): Ditto.
29389 (_mm256_mask_i32scatter_epi32): Ditto.
29390 (_mm_i32scatter_epi32): Ditto.
29391 (_mm_mask_i32scatter_epi32): Ditto.
29392 (_mm256_i32scatter_epi64): Ditto.
29393 (_mm256_mask_i32scatter_epi64): Ditto.
29394 (_mm_i32scatter_epi64): Ditto.
29395 (_mm_mask_i32scatter_epi64): Ditto.
29396 (_mm256_i64scatter_epi32): Ditto.
29397 (_mm256_mask_i64scatter_epi32): Ditto.
29398 (_mm_i64scatter_epi32): Ditto.
29399 (_mm_mask_i64scatter_epi32): Ditto.
29400 (_mm256_i64scatter_epi64): Ditto.
29401 (_mm256_mask_i64scatter_epi64): Ditto.
29402 (_mm_i64scatter_epi64): Ditto.
29403 (_mm_mask_i64scatter_epi64): Ditto.
29404 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
29405 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
29406 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
29407 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
29408 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
29409 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
29410 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
29411 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
29412 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
29413 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
29414 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
29415 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
29416 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
29417 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
29418 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
29419 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
29420 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
29421 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
29422 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
29423 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
29424 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
29425 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
29426 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
29427 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
29428 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
29429 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
29430 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
29431 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
29432 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
29433 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
29434 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
29435 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
29436 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
29437 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
29438 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
29439 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
29440 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
29441 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
29442 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
29443 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
29444 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
29445 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
29446 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
29447 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
29448 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
29449 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
29450 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
29451 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
29452 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
29453 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
29454 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
29455 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
29456 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
29457 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
29458 definitions accordingly.
29460 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
29461 Kuan-Lin Chen <kuanlinchentw@gmail.com>
29464 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
29467 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
29470 * ira-conflicts.c (ira_build_conflicts): Update total conflict
29471 hard regs for inner regno.
29473 2017-01-17 Martin Liska <mliska@suse.cz>
29476 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
29477 assumption and add comment.
29479 2017-01-17 Nathan Sidwell <nathan@acm.org>
29481 * ipa-visibility.c (localize_node): New function, broken out of ...
29482 (function_and_variable_visibility): ... here. Call it.
29484 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29486 PR middle-end/77445
29487 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
29488 correctly set frequency of oudgoing edge.
29489 (duplicate_thread_path): Fix profile updating.
29491 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29494 * configure.ac: Add GCC_BASE_VER.
29495 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
29496 version from BASE-VER file.
29497 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
29498 (gcc.o): Depend on $(BASEVER).
29499 * common.opt (dumpfullversion): New option.
29500 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
29501 * doc/invoke.texi: Document -dumpfullversion.
29502 * doc/install.texi: Document --with-gcc-major-version-only.
29503 * configure: Regenerated.
29505 2017-01-17 Richard Biener <rguenther@suse.de>
29507 PR tree-optimization/71433
29508 * tree-vrp.c (register_new_assert_for): Merge same asserts
29509 on all incoming edges.
29510 (process_assert_insertions_for): Handle insertions at the
29513 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
29515 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
29516 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
29518 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
29521 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
29524 2017-01-17 Alan Modra <amodra@gmail.com>
29527 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
29528 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
29529 symbolic stack limit when pic.
29531 2017-01-16 Martin Sebor <msebor@redhat.com>
29533 PR tree-optimization/78608
29534 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
29536 2017-01-16 Jeff Law <law@redhat.com>
29539 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
29540 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
29541 for several include directories that may be relative to sysroot.
29542 * config/i386/x-mingw32 (gplus_includedir): Define.
29543 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
29544 (native_system_includedir): Likewise.
29545 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
29546 override if TARGET_SYSTEM_ROOT is defined.
29547 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
29549 PR tree-optimization/79090
29550 PR tree-optimization/33562
29551 PR tree-optimization/61912
29552 PR tree-optimization/77485
29553 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
29554 and computed trims into the dump file.
29556 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29558 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
29560 2017-01-16 Jakub Jelinek <jakub@redhat.com>
29563 * gimplify.c (gimplify_init_constructor): If want_value and
29564 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
29568 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
29569 sequence. Formatting fixes.
29570 (doloop_optimize): Formatting fixes.
29573 * gcc.c (debug_level_greater_than_spec_func): New function.
29574 (static_spec_functions): Add debug-level-gt spec function.
29575 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
29577 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29578 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29579 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
29580 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
29581 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
29582 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
29584 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
29586 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
29587 QImode fixups to general and mask registers only.
29589 2017-01-16 Carl Love <cel@us.ibm.com>
29591 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
29592 for built-in functions
29593 vector signed char vec_nabs (vector signed char)
29594 vector signed short vec_nabs (vector signed short)
29595 vector signed int vec_nabs (vector signed int)
29596 vector signed long long vec_nabs (vector signed long long)
29597 vector float vec_nabs (vector float)
29598 vector double vec_nabs (vector double)
29599 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
29601 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
29602 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
29603 * doc/extend.texi: Update the documentation file for the new built-in
29606 2017-01-16 Martin Sebor <msebor@redhat.com>
29608 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
29611 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29613 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
29614 UNSPEC_VSX__XXSPLTD to require special splat handling.
29616 2017-01-16 David Malcolm <dmalcolm@redhat.com>
29619 * system.h: Poison strndup.
29621 2017-01-16 Alan Modra <amodra@gmail.com>
29624 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
29627 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
29629 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
29631 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
29633 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
29634 call recog here. Assert that INSN_CODE (insn) is non-negative.
29636 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29639 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
29641 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
29642 in the currently scheduled RTL fragment.
29644 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29646 PR rtl-optimization/78751
29647 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
29650 2017-01-14 Jeff Law <law@redhat.com>
29652 PR tree-optimization/79090
29653 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
29654 variable length stores.
29655 (compute_trims): Delete dead assignment to *trim_tail.
29656 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
29659 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
29661 PR rtl-optimization/78626
29662 PR rtl-optimization/78727
29663 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
29664 of a block, and split such blocks after everything else is finished.
29666 2017-01-14 Alan Modra <amodra@gmail.com>
29669 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
29670 target legitimate_combined_insn.
29671 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
29672 (rs6000_legitimate_combined_insn): New function.
29673 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
29675 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
29676 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
29677 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
29679 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
29681 * doc/frontends.texi (G++ and GCC): Remove references to Java.
29683 2017-01-13 Jeff Law <law@redhat.com>
29685 PR tree-optimization/33562
29686 PR tree-optimization/61912
29687 PR tree-optimization/77485
29688 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
29690 (delete_dead_assignment): Likewise.
29691 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
29692 statement to delete_dead_call and delete_dead_assignment.
29694 2017-01-13 David Malcolm <dmalcolm@redhat.com>
29697 * substring-locations.c (format_warning_va): Strengthen case 1 so
29698 that both endpoints of the substring must be within the format
29699 range for just the substring to be printed.
29701 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
29703 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
29704 * config/i386/i386.c (ix86_target_string): Add missing options
29705 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
29706 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
29707 flags_other and ix86_target_other to flags2_other. Display unknown
29709 (ix86_valid_target_attribute_inner_p): Add missing options and
29710 reorder options by implied ISAs, as in ix86_target_string.
29712 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29714 * hash-table.h (hash_table::too_empty_p): New function.
29715 (hash_table::expand): Use it.
29716 (hash_table::traverse): Likewise.
29717 (hash_table::empty_slot): Use sizeof (value_type) instead of
29718 sizeof (PTR) to convert bytes to elements. Shrink the table
29719 if the current size is excessive for the current number of
29722 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29724 * ira-costs.c (record_reg_classes): Break from the inner loop
29725 early once alt_fail is known to be true. Update outer loop
29726 handling accordingly.
29728 2017-01-13 Jeff Law <law@redhat.com>
29730 * tree-ssa-dse.c (decrement_count): New function.
29731 (increment_start_addr, maybe_trim_memstar_call): Likewise.
29732 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
29733 when we know the partially dead statement is a mem* function.
29735 PR tree-optimization/61912
29736 PR tree-optimization/77485
29737 * tree-ssa-dse.c: Include expr.h.
29738 (maybe_trim_constructor_store): New function.
29739 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
29741 PR tree-optimization/33562
29742 PR tree-optimization/61912
29743 PR tree-optimization/77485
29744 * doc/invoke.texi: Document new dse-max-object-size param.
29745 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
29746 * tree-ssa-dse.c: Include params.h.
29747 (dse_store_status): New enum.
29748 (initialize_ao_ref_for_dse): New, partially extracted from
29750 (valid_ao_ref_for_dse, normalize_ref): New.
29751 (setup_live_bytes_from_ref, compute_trims): Likewise.
29752 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
29753 (maybe_trim_partially_dead_store): Likewise.
29754 (maybe_trim_complex_store): Likewise.
29755 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
29756 Track what bytes live from the original store. Return tri-state
29757 for dead, partially dead or live.
29758 (dse_dom_walker): Add constructor, destructor and new private members.
29759 (delete_dead_call, delete_dead_assignment): New extracted from
29761 (dse_optimize_stmt): Make a member of dse_dom_walker.
29762 Use initialize_ao_ref_for_dse.
29764 PR tree-optimization/33562
29765 PR tree-optimization/61912
29766 PR tree-optimization/77485
29767 * sbitmap.h (bitmap_count_bits): Prototype.
29768 (bitmap_clear_range, bitmap_set_range): Likewise.
29769 * sbitmap.c (bitmap_clear_range): New function.
29770 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
29772 2017-01-13 Martin Liska <mliska@suse.cz>
29775 * function.c (set_cfun): Add new argument force.
29776 * function.h (set_cfun): Likewise.
29777 * ipa-inline-transform.c (inline_call): Use the function when
29778 strict alising from is dropped for function we inline to.
29780 2017-01-13 Richard Biener <rguenther@suse.de>
29782 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
29783 for dumping GIMPLE INTEGER_CSTs.
29785 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29787 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
29788 to 201112L since C++17.
29790 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
29793 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
29794 if -fsanitize=kernel-address is present.
29796 2017-01-13 Richard Biener <rguenther@suse.de>
29798 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
29799 as _Literal ( type ) number in case usual suffixes do not
29800 preserve all information.
29802 2017-01-13 Richard Biener <rguenther@suse.de>
29804 PR tree-optimization/77283
29805 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
29806 and ssa-iterators.h.
29807 (is_feasible_trace): Implement a cost model based on joiner
29810 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
29813 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
29814 char or short to __float128/_Float128 directly.
29816 2017-01-12 Martin Sebor <msebor@redhat.com>
29818 to -Wformat-overflow.
29819 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
29820 (min_bytes_remaining): Same.
29821 (get_string_length): Same.
29822 (format_string): Same.
29823 (format_directive): Same.
29825 (pass_sprintf_length::handle_gimple_call): Same.
29827 2017-01-12 Jakub Jelinek <jakub@redhat.com>
29829 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
29830 info.nowrite calls with no lhs that can't throw. Return bool
29831 whether gsi_remove has been called or not.
29832 (pass_sprintf_length::handle_gimple_call): Return bool whether
29833 try_substitute_return_value called gsi_remove. Formatting fix.
29834 (pass_sprintf_length::execute): Don't use gsi_remove if
29835 handle_gimple_call returned true.
29838 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
29839 be removed due to side-effects, don't remove following barrier nor
29840 turn the successor edge into fallthru edge.
29842 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29845 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
29846 element-reversing loads and stores as not swappable.
29848 2017-01-12 Nathan Sidwell <nathan@acm.org>
29849 Nicolai Stange <nicstange@gmail.com>
29851 * combine.c (try_combine): Don't ignore result of overlap checking
29852 loop. Combine overlap & asm check into single loop.
29854 2017-01-12 Richard Biener <rguenther@suse.de>
29856 * tree-pretty-print.c (dump_generic_node): Provide -gimple
29857 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
29859 2017-01-12 Richard Biener <rguenther@suse.de>
29861 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
29862 and TS_TARGET_OPTION directly derive from TS_BASE.
29863 * tree-core.h (tree_optimization_option): Derive from tree_base.
29864 (tree_target_option): Likewise.
29866 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
29868 * config/i386/i386.c (memory_address_length): Increase len
29869 only when rip_relative_addr_p returns false.
29871 2017-01-11 Julia Koval <julia.koval@intel.com>
29873 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
29874 (OPTION_MASK_ISA_SGX_SET): New.
29875 (ix86_handle_option): Handle OPT_msgx.
29876 * config.gcc: Added sgxintrin.h.
29877 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
29878 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
29879 * config/i386/i386.c (ix86_target_string): Add -msgx.
29881 (ix86_option_override_internal): Handle new options.
29882 (ix86_valid_target_attribute_inner_p): Add sgx.
29883 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
29884 * config/i386/i386.opt: Add msgx.
29885 * config/i386/sgxintrin.h: New file.
29886 * config/i386/x86intrin.h: Add sgxintrin.h.
29888 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29891 * fold-const.c (maybe_nonzero_address): Return 1 for function
29893 (tree_single_nonzero_warnv_p): Don't handle function local objects
29897 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
29900 2017-01-11 David Malcolm <dmalcolm@redhat.com>
29903 * opts.c: Include "spellcheck.h"
29904 (struct string_fragment): New struct.
29905 (struct edit_distance_traits<const string_fragment &>): New
29907 (get_closest_sanitizer_option): New function.
29908 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
29910 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29912 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
29914 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
29915 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
29916 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
29917 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
29918 for initial die_offset if dwarf_split_debug_info.
29919 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
29920 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
29922 (output_skeleton_debug_sections): Formatting fix. Use
29923 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
29924 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
29926 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
29928 * config/arm/cortex-a53.md: Add bypasses for
29929 cortex_a53_r2f_cvt.
29930 (cortex_a53_r2f): Only use for transfers.
29931 (cortex_a53_f2r): Likewise.
29932 (cortex_a53_r2f_cvt): Add reservation for conversions.
29933 (cortex_a53_f2r_cvt): Likewise.
29935 2017-01-11 Tamar Christina <tamar.christina@arm.com>
29937 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
29938 to all inlined functions, change static to extern.
29940 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
29943 * config/arm/arm.c (legitimize_pic_address): Handle reference to
29945 (arm_assemble_integer): Likewise.
29947 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
29949 * config.gcc: Use new awk script to check CPU, FPU and architecture
29950 parameters for --with-... options.
29951 * config/arm/parsecpu.awk: New file
29952 * config/arm/arm-cpus.in: New file.
29953 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
29955 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
29957 * config/arm/t-arm: Update dependency rules.
29958 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
29959 of processing .def files.
29960 * config/arm/genopt.sh: Deleted.
29961 * config/arm/gentune.sh: Deleted.
29962 * config/arm/arm-cores.def: Deleted.
29963 * config/arm/arm-arches.def: Deleted.
29964 * config/arm/arm-fpus.def: Deleted.
29965 * config/arm/arm-tune.md: Regenerated.
29966 * config/arm/arm-tables.opt: Regenerated.
29967 * config/arm/arm-cpu.h: New generated file.
29968 * config/arm/arm-cpu-data.h: New generated file.
29969 * config/arm/arm-cpu-cdata.h: New generated file.
29971 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
29974 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
29976 (input_varpool_node): Unpack dynamically_initialized bit.
29978 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
29980 PR rtl-optimization/79032
29981 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
29982 the alignment of the adjusted memory reference against that of MODE,
29983 instead of the alignment of the original memory reference.
29985 2017-01-11 Martin Jambor <mjambor@suse.cz>
29987 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
29989 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
29990 decorated functions.
29992 2017-01-11 Richard Biener <rguenther@suse.de>
29994 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
29995 set range/nonnull info for PHI results. Do not set it on
29996 stmts marked for removal.
29998 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
30000 * expr.c (store_field): In the bitfield case, fetch the return value
30001 from the registers before applying a single big-endian adjustment.
30002 Always do a final load for a BLKmode value not larger than a word.
30004 2017-01-10 David Malcolm <dmalcolm@redhat.com>
30007 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30008 that we correctly handle column numbers greater than
30009 LINE_MAP_MAX_COLUMN_NUMBER.
30011 2017-01-10 Martin Sebor <msebor@redhat.com>
30013 PR middle-end/78245
30014 * gimple-ssa-sprintf.c (get_destination_size): Call
30015 {init,fini}object_sizes.
30016 * tree-object-size.c (addr_object_size): Adjust.
30017 (pass_through_call): Adjust.
30018 (pass_object_sizes::execute): Adjust.
30019 * tree-object-size.h (fini_object_sizes): Declare.
30021 2017-01-10 Martin Sebor <msebor@redhat.com>
30023 PR tree-optimization/78775
30024 * builtins.c (get_size_range): Move...
30025 * calls.c: ...to here.
30026 (alloc_max_size): Accept zero argument.
30027 (operand_signed_p): Remove.
30028 (maybe_warn_alloc_args_overflow): Call get_size_range.
30029 * calls.h (get_size_range): Declare.
30031 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
30033 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
30034 from TI's devices.csv file as of September 2016.
30035 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
30037 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
30039 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
30040 * doc/invoke.texi: Likewise.
30041 * doc/md.texi: Likewise.
30042 * doc/objc.texi: Likewise.
30044 2017-01-10 Joshua Conner <joshconner@google.com>
30046 * config/arm/fuchsia-elf.h: New file.
30047 * config/fuchsia.h: New file.
30048 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
30049 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
30051 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
30053 2016-01-10 Richard Biener <rguenther@suse.de>
30055 PR tree-optimization/79034
30056 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30057 Propagate out degenerate PHIs in the joiner.
30059 2017-01-10 Martin Liska <mliska@suse.cz>
30061 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
30062 (sort_congruence_classes_by_decl_uid): Likewise.
30063 (sort_congruence_class_groups_by_decl_uid): Likewise.
30064 (sem_item_optimizer::merge_classes): Sort class, groups in these
30065 classes and members in the groups by DECL_UID of declarations.
30066 This would make merge operations stable.
30068 2017-01-10 Martin Liska <mliska@suse.cz>
30070 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30071 usage of m_classes_vec.
30072 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30073 (sem_item_optimizer::get_group_by_hash): Likewise.
30074 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30075 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30076 (sem_item_optimizer::verify_classes): Likewise.
30077 (sem_item_optimizer::process_cong_reduction): Likewise.
30078 (sem_item_optimizer::dump_cong_classes): Likewise.
30079 (sem_item_optimizer::merge_classes): Likewise.
30080 * ipa-icf.h (congruence_class_hash): Rename from
30081 congruence_class_group_hash. Remove declaration of m_classes_vec.
30083 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30085 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30086 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30087 * config.gcc: Add avx512vpopcntdqintrin.h.
30088 * config/i386/avx512vpopcntdqintrin.h: New.
30089 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30090 * config/i386/i386-builtin-types.def: Add new types.
30091 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30092 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30093 __builtin_ia32_vpopcountq_v8di_mask): New.
30094 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30095 __AVX512VPOPCNTDQ__.
30096 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30097 (PTA_AVX512VPOPCNTDQ): Define.
30098 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30099 TARGET_AVX512VPOPCNTDQ_P): Define.
30100 * config/i386/i386.opt: Add mavx512vpopcntdq.
30101 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30102 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30104 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30106 PR middle-end/77484
30107 * predict.def (PRED_CALL): Set to 67.
30109 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30111 * expr.c (store_field): In the bitfield case, if the value comes from
30112 a function call and is of an aggregate type returned in registers, do
30113 not modify the field mode; extract the value in all cases if the mode
30114 is BLKmode and the size is not larger than a word.
30116 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30119 * config/i386/cpuid.h: Fix undefined behavior.
30121 2017-01-04 Jeff Law <law@redhat.com>
30123 PR tree-optimization/79007
30124 PR tree-optimization/67955
30125 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30126 conservative for pt.null when flag_non_call_exceptions is on.
30128 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30130 PR translation/79019
30131 PR translation/79020
30132 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30133 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30134 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30136 * config/avr/avr.opt (maccumulate-args): Likewise.
30137 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30138 * common.opt (freport-bug): Likewise.
30139 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30140 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30141 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30142 translatable string.
30143 * config/i386/i386.c (function_value_32): Likewise.
30144 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
30145 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
30147 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
30148 * common/config/msp430/msp430-common.c (msp430_handle_option):
30150 * symtab.c (symtab_node::verify_base): Likewise.
30151 * opts.c (set_debug_level): Likewise.
30152 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
30153 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
30154 missing whitespace to translatable strings.
30155 * config/avr/avr.md (bswapsi2): Fix typo in comment.
30156 * config/sh/superh.h: Likewise.
30157 * config/i386/xopintrin.h: Likewise.
30158 * config/i386/znver1.md: Likewise.
30159 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
30160 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
30161 * double-int.h (struct double_int): Likewise.
30162 * double-int.c (div_and_round_double): Likewise.
30163 * wide-int.cc: Likewise.
30164 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
30165 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
30166 * cfgcleanup.c (crossjumps_occured): Renamed to ...
30167 (crossjumps_occurred): ... this.
30168 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
30171 PR tree-optimization/78899
30172 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
30173 returning bool return struct loop *, NULL for failure and the new
30175 (versionable_outer_loop_p): Don't version outer loop if it has
30176 dont_vectorized bit set.
30177 (tree_if_conversion): When versioning outer loop, ensure
30178 tree_if_conversion is performed also on the inner loop of the
30179 non-vectorizable outer loop copy.
30180 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
30181 LOOP_VECTORIZED in inner loop of the scalar outer loop and
30182 prevent vectorization of it.
30183 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
30184 the outer loop vectorization of the non-scalar version is attempted
30185 before vectorization of the inner loop in scalar version. If
30186 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
30187 vectorization of its inner loop.
30188 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
30189 has 2 inner loops, rename also on edges from bb whose single pred
30190 is outer_loop->header. Fix typo in function comment.
30192 2017-01-09 Martin Sebor <msebor@redhat.com>
30195 * asan.c (asan_emit_stack_protection): Increase local buffer size
30196 to avoid snprintf truncation warning.
30198 2017-01-09 Andrew Pinski <apinski@cavium.com>
30200 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
30201 to reference thunderx2t99 for the tuning structure
30202 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
30204 (thunderx2t99_extra_costs): This.
30205 * config/aarch64/aarch64-tune.md: Regenerate.
30206 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
30207 (vulcan_addrcost_table): This.
30208 (vulcan_regmove_cost): Rename to ...
30209 (thunderx2t99_regmove_cost): This.
30210 (vulcan_vector_cost): Rename to ...
30211 (thunderx2t99_vector_cost): this.
30212 (vulcan_branch_cost): Rename to ...
30213 (thunderx2t99_branch_cost): This.
30214 (vulcan_tunings): Rename to ...
30215 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
30216 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
30218 2017-01-09 Martin Jambor <mjambor@suse.cz>
30222 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
30223 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
30224 (propagate_vr_accross_jump_function): Use the above function for all
30225 value range computations for pass-through jump functions and type
30226 converasion from explicit value range values.
30227 (ipcp_propagate_stage): Do not attempt to deduce types of formal
30228 parameters from TYPE_ARG_TYPES.
30229 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
30230 (ipa_write_node_info): Stream type of the actual argument.
30231 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
30233 2017-01-09 Martin Liska <mliska@suse.cz>
30236 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
30237 (lookup_compiler): Do not show error message with have_E.
30239 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30241 PR tree-optimization/78938
30242 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
30243 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
30244 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
30245 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
30248 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30250 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
30253 2017-01-09 Richard Biener <rguenther@suse.de>
30255 PR tree-optimization/78997
30256 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
30257 name condition properly.
30259 2017-01-09 Richard Biener <rguenther@suse.de>
30262 * dwarf2out.c (is_cxx): New overload with context.
30263 (is_naming_typedef_decl): Use it.
30265 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30267 * invoke.texi (Option Summary): Correct spacing in option lists
30268 and add line breaks to fix over-long lines.
30270 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30272 PR middle-end/17660
30274 * extend.texi (Common Variable Attributes): Add xref to GCC
30275 Internals manual to explain mode attribute keywords.
30277 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30280 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
30281 and Preprocessor Options.
30282 (Options for Linking): Document -pthread here....
30283 (RS/6000 and PowerPC Options): ...not here.
30284 (Solaris 2 Options): ...or here.
30285 * doc/cppopts.texi: Document -pthread.
30287 2017-01-08 Martin Sebor <msebor@redhat.com>
30289 PR middle-end/77708
30290 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
30291 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
30292 New member functions.
30293 (format_directive): Used them.
30295 (pass_sprintf_length::handle_gimple_call): Same.
30296 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
30297 to avoid truncation for any argument.
30298 (extract_affine_mul): Same.
30299 * tree.c (get_file_function_name): Same.
30301 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30303 PR middle-end/77484
30304 * predict.def (PRED_INDIR_CALL): Set to 86.
30306 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30308 PR preprocessor/54124
30309 * doc/cppopts.texi: Reformat -d subtable to list the full name
30310 of the options. Add cross-reference to the docs for the general
30311 compiler -d options.
30312 * doc/invoke.texi (Developer Options): Add cross-reference to the
30313 preprocessor-specific -d option documentation.
30315 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30317 PR preprocessor/13498
30318 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
30319 redudant material, and reflect new command-line options.
30320 (System Headers): Likewise.
30322 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30324 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
30325 -isystem, and -idirafter. Copy-edit.
30326 * doc/cppopts.texi: Copy-edit. Remove contradiction about
30327 default for -ftrack-macro-expansion. Delete obsolete and
30328 badly-formatted implementation details about -fdebug-cpp output.
30329 * doc/cppwarnopts.texi: Copy-edit.
30331 2017-01-07 David Malcolm <dmalcolm@redhat.com>
30334 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30335 that the transition from a max line width >= 1<<10 to narrower
30336 lines works correctly.
30338 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
30340 * doc/options.texi (PerFunction): New.
30341 * opt-functions.awk (switch_flags): Map both Optimization and
30342 PerFunction to CL_OPTIMIZATION.
30343 * opth-gen.awk: Test for PerFunction flag along with
30345 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
30346 it only when the latter is present. Skip those that don't in
30347 the hash function generator.
30348 * common.opt (fvar-tracking): Mark as PerFunction instead of
30350 (fvar-tracking-assignments): Likewise.
30351 (fvar-tracking-assignments-toggle): Likewise.
30352 (fvar-tracking-uninit): Likewise.
30354 2017-01-07 Jakub Jelinek <jakub@redhat.com>
30356 PR translation/79018
30357 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
30360 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
30363 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
30364 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
30365 TARGET_LONG_JUMP_TABLE_OFFSETS.
30366 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
30367 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
30368 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
30369 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
30370 * config/m68k/m68k.md (tablejump expander): Likewise.
30371 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
30372 TARGET_LONG_JUMP_TABLE_OFFSETS.
30373 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
30374 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
30376 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
30377 David Holsgrove <david.holsgrove@xilinx.com>
30379 * common/config/microblaze/microblaze-common.c
30380 (TARGET_EXCEPT_UNWIND_INFO): Remove.
30381 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
30383 * config/microblaze/microblaze.c (microblaze_must_save_register)
30384 (microblaze_expand_epilogue, microblaze_return_addr): Handle
30386 (microblaze_eh_return): New function.
30387 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
30388 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
30389 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
30390 * config/microblaze/microblaze.md (eh_return): New pattern.
30392 2017-01-06 Jakub Jelinek <jakub@redhat.com>
30394 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
30395 GCC_DIAGNOSTIC_STRINGIFY): Define.
30397 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
30399 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30401 * config/arm/arm.md (<mcrr>): New.
30403 * config/arm/arm.c (arm_arch5te): New.
30404 (arm_option_override): Set arm_arch5te.
30405 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
30407 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
30408 (arm_mcrr_qualifiers): ... this. New.
30409 (MRRC_QUALIFIERS): Define to...
30410 (arm_mrrc_qualifiers): ... this. New.
30411 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
30413 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
30414 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
30415 (MRRCI, mrrc, MRRC): New.
30416 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
30417 VUNSPEC_MRRC2): New.
30419 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30421 * config/arm/arm.md (<mcr>): New.
30423 * config/arm/arm.c (arm_coproc_builtin_available): Add
30424 support for mcr, mrc, mcr2 and mrc2.
30425 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
30426 (arm_mcr_qualifiers): ... this. New.
30427 (MRC_QUALIFIERS): Define to ...
30428 (arm_mrc_qualifiers): ... this. New.
30429 (MCR_QUALIFIERS): Define to ...
30430 (arm_mcr_qualifiers): ... this. New.
30431 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
30433 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
30434 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
30435 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
30436 VUNSPEC_MRC2): New.
30438 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30440 * config/arm/arm.md (*ldc): New.
30444 * config/arm/arm.c (arm_coproc_builtin_available): Add
30445 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
30446 (arm_coproc_ldc_stc_legitimate_address): New.
30447 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
30448 'qualifier_const_pointer'.
30449 (LDC_QUALIFIERS): Define to...
30450 (arm_ldc_qualifiers): ... this. New.
30451 (STC_QUALIFIERS): Define to...
30452 (arm_stc_qualifiers): ... this. New.
30453 * config/arm/arm-protos.h
30454 (arm_coproc_ldc_stc_legitimate_address): New.
30455 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
30456 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
30457 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
30458 stc2, stcl, stc2l): New.
30459 * config/arm/constraints.md (Uz): New.
30460 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
30461 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
30462 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
30463 VUNSPEC_STC2L): New.
30465 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30467 * config/arm/arm.md (<cdp>): New.
30468 * config/arm/arm.c (neon_const_bounds): Rename this ...
30469 (arm_const_bounds): ... this.
30470 (arm_coproc_builtin_available): New.
30471 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
30472 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
30473 (CDP_QUALIFIERS): Define to...
30474 (arm_cdp_qualifiers): ... this. New.
30476 (arm_expand_builtin_args): Add case for 6 arguments.
30477 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
30478 (arm_const_bounds): ... this.
30479 (arm_coproc_builtin_available): New.
30480 * config/arm/arm_acle.h (__arm_cdp): New.
30482 * config/arm/arm_acle_builtins.def (cdp): New.
30484 * config/arm/iterators.md (CDPI,CDP,cdp): New.
30485 * config/arm/neon.md: Rename all 'neon_const_bounds' to
30486 'arm_const_bounds'.
30487 * config/arm/types.md (coproc): New.
30488 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
30489 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
30490 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
30491 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
30493 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30495 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
30496 (UBINOP_QUALIFIERS): New.
30498 (acle_builtin_data): New. Change comment.
30499 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
30500 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
30501 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
30502 arm_acle_builtins.def.
30503 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
30504 (arm_init_acle_builtins): New.
30505 (CRC32_BUILTIN): Remove.
30506 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
30507 crc32cb, crc32ch and crc32cw.
30508 (arm_init_crc32_builtins): Remove.
30509 (arm_init_builtins): Use arm_init_acle_builtins rather
30510 than arm_init_crc32_builtins.
30511 (arm_expand_acle_builtin): New.
30512 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
30513 * config/arm/arm_acle_builtins.def: New.
30515 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30517 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
30518 (arm_builtin_datum): ... this.
30519 (arm_init_neon_builtin): Rename to ...
30520 (arm_init_builtin): ... this. Add a new parameters PREFIX
30521 and USE_SIG_IN_NAME.
30522 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
30523 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
30524 'arm_builtin_datum'.
30525 (arm_init_vfp_builtins): Likewise.
30526 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
30527 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
30528 (arm_expand_neon_args): Rename to ...
30529 (arm_expand_builtin_args): ... this. Rename builtin_arg
30530 enum values and differentiate between ARG_BUILTIN_MEMORY
30531 and ARG_BUILTIN_NEON_MEMORY.
30532 (arm_expand_neon_builtin_1): Rename to ...
30533 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
30534 values, arm_expand_builtin_args and add bool parameter NEON.
30535 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
30536 (arm_expand_vfp_builtin): Likewise.
30537 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
30539 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30541 PR middle-end/77484
30542 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
30543 * predict.c (tree_estimate_probability_bb): Reverse direction of
30544 polymorphic call predictor.
30546 2017-01-06 David Malcolm <dmalcolm@redhat.com>
30548 * passes.c (execute_one_pass): Split out pass-skipping logic into...
30549 (determine_pass_name_match): ...this new function and...
30550 (should_skip_pass_p): ...this new function.
30552 2017-01-06 Nathan Sidwell <nathan@acm.org>
30554 * ipa-visibility.c (function_and_variable_visibility): Reformat
30555 comments and long lines. Remove extrneous if.
30556 * symtab.c (symtab_node::make_decl_local): Fix code format.
30557 (symtab_node::set_section_for_node): Fix comment typo.
30559 2017-01-06 Martin Liska <mliska@suse.cz>
30562 * lra-constraints.c: Rename invariant to lra_invariant.
30563 * predict.c (set_even_probabilities): Initialize e to NULL.
30565 2017-01-05 Martin Sebor <msebor@redhat.com>
30567 PR tree-optimization/78910
30568 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
30569 (format_integer): Correct off-by-one error in the handling
30570 of precision with negative numbers in signed conversions..
30572 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
30574 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
30576 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30578 PR tree-optimization/71016
30579 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
30580 factor_out_conditional_conversion. Formatting fix.
30581 (factor_out_conditional_conversion): Add cond_stmt argument.
30582 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
30583 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
30586 2017-01-05 David Malcolm <dmalcolm@redhat.com>
30588 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
30589 read-rtl-function.o, and selftest-rtl.o.
30590 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
30591 (selftest::aarch64_test_loading_full_dump): New function.
30592 (selftest::aarch64_run_selftests): New function.
30593 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
30594 selftest::aarch64_run_selftests.
30595 * config/i386/i386.c
30596 (selftest::ix86_test_loading_dump_fragment_1): New function.
30597 (selftest::ix86_test_loading_call_insn): New function.
30598 (selftest::ix86_test_loading_full_dump): New function.
30599 (selftest::ix86_test_loading_unspec): New function.
30600 (selftest::ix86_run_selftests): Call the new functions.
30601 * emit-rtl.c (maybe_set_max_label_num): New function.
30602 * emit-rtl.h (maybe_set_max_label_num): New decl.
30603 * function.c (instantiate_decls): Guard call to
30604 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
30605 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
30607 * gensupport.c (gen_reader::gen_reader): Pass "false"
30608 for new "compact" param of rtx_reader.
30609 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
30610 rather than an empty string for NULL strings.
30611 * read-md.c: Potentially include config.h rather than bconfig.h.
30612 Wrap include of errors.h with #ifdef GENERATOR_FILE.
30613 (have_error): New global, copied from errors.c.
30614 (md_reader::read_name): Rename to...
30615 (md_reader::read_name_1): ...this, adding "out_loc" param,
30616 and converting "missing name or number" to returning false, rather
30618 (md_reader::read_name): Reimplement in terms of read_name_1.
30619 (md_reader::read_name_or_nil): New function.
30620 (md_reader::read_string): Handle "(nil)" by returning NULL.
30621 (md_reader::md_reader): Add new param "compact".
30622 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
30623 (md_reader::read_file): New method.
30624 * read-md.h (md_reader::md_reader): Add new param "compact".
30625 (md_reader::read_file): New method.
30626 (md_reader::is_compact): New accessor.
30627 (md_reader::read_name): Convert return type from void to file_location.
30628 (md_reader::read_name_or_nil): New decl.
30629 (md_reader::read_name_1): New decl.
30630 (md_reader::m_compact): New field.
30631 (noop_reader::noop_reader): Pass "false" for new "compact" param
30633 (rtx_reader::rtx_reader): Add new "compact" param.
30634 (rtx_reader::read_rtx_operand): Make virtual and convert return
30635 type from void to rtx.
30636 (rtx_reader::read_until): New decl.
30637 (rtx_reader::handle_any_trailing_information): New virtual function.
30638 (rtx_reader::postprocess): New virtual function.
30639 (rtx_reader::finalize_string): New virtual function.
30640 (rtx_reader::m_in_call_function_usage): New field.
30641 (rtx_reader::m_reuse_rtx_by_id): New field.
30642 * read-rtl-function.c: New file.
30643 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
30644 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
30645 (selftest::verify_three_block_rtl_cfg): New decl.
30646 * read-rtl-function.h: New file.
30647 * read-rtl.c: Potentially include config.h rather than bconfig.h.
30648 For host, include function.h, memmodel.h, and emit-rtl.h.
30649 (one_time_initialization): New function.
30650 (struct compact_insn_name): New struct.
30651 (compact_insn_names): New array.
30652 (find_code): Handle insn codes in compact dumps.
30653 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
30654 (bind_subst_iter_and_attr): Likewise.
30655 (add_condition_to_string): Likewise.
30656 (add_condition_to_rtx): Likewise.
30657 (apply_attribute_uses): Likewise.
30658 (add_current_iterators): Likewise.
30659 (apply_iterators): Likewise.
30660 (initialize_iterators): Guard usage of apply_subst_iterator with
30661 #ifdef GENERATOR_FILE.
30662 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
30663 (md_reader::read_mapping): Likewise.
30664 (add_define_attr_for_define_subst): Likewise.
30665 (add_define_subst_attr): Likewise.
30666 (read_subst_mapping): Likewise.
30667 (check_code_iterator): Likewise.
30668 (rtx_reader::read_rtx): Likewise. Move one-time initialization
30670 (one_time_initialization): New function.
30671 (rtx_reader::read_until): New method.
30672 (read_flags): New function.
30673 (parse_reg_note_name): New function.
30674 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
30675 Handle reuse_rtx ids.
30676 Wrap iterator lookup within #ifdef GENERATOR_FILE.
30677 Add parsing support for RTL dumps, mirroring the special-cases in
30678 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
30679 values, and calling handle_any_trailing_information.
30680 (rtx_reader::read_rtx_operand): Convert return type from void
30681 to rtx, returning return_rtx. Handle case 'e'. Call
30682 finalize_string on XSTR and XTMPL fields.
30683 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
30684 "(nil)" values were omitted. Call the postprocess vfunc on the
30686 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
30687 class ctor. Initialize m_in_call_function_usage. Call
30688 one_time_initialization.
30689 * rtl-tests.c (selftest::test_uncond_jump): Call
30690 set_new_first_and_last_insn.
30691 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
30692 * selftest-rtl.c: New file.
30693 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
30694 (selftest::get_insn_by_uid): New decl.
30695 * selftest-run-tests.c (selftest::run_tests): Call
30696 read_rtl_function_c_tests.
30697 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
30698 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
30701 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
30703 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
30704 operands in a special way. Assert that pos+len <= mode precision.
30706 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30708 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
30709 3 argument Alias with unlimited for the negative form.
30710 (fno-vect-cost-model): Removed.
30712 2017-01-05 Martin Liska <mliska@suse.cz>
30714 * hsa-gen.c (gen_hsa_divmod): New function.
30715 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
30717 2017-01-05 Martin Liska <mliska@suse.cz>
30720 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
30723 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30725 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
30726 small constant length operands.
30728 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30730 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
30731 between loop iterations.
30733 2017-01-05 Martin Liska <mliska@suse.cz>
30736 * gimplify.c (gimplify_decl_expr): Compare to
30737 asan_poisoned_variables instread of checking flags.
30738 (gimplify_target_expr): Likewise.
30739 (gimplify_expr): Likewise.
30740 (gimplify_function_tree): Conditionally initialize
30741 asan_poisoned_variables.
30743 2017-01-04 Jeff Law <law@redhat.com>
30745 PR tree-optimizatin/78812
30746 * rtl.h (contains_mem_rtx_p): Prototype.
30747 * ifcvt.c (containts_mem_rtx_p): Move from here to...
30748 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
30749 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
30750 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
30751 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
30753 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30755 * input.c (assert_char_at_range): Default-initialize actual_range.
30757 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30759 * df-scan.c (df_ref_create_structure): Make regno unsigned,
30760 to match the caller.
30762 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30764 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
30765 insns after final jump in test to emit dummy move.
30767 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30769 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
30770 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
30772 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30774 * multiple_target.c (create_dispatcher_calls): Init e_next.
30775 * tree-ssa-loop-split.c (split_loop): Init border.
30776 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
30779 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
30784 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
30785 (altivec_register_operand): Do not return true if the operand
30786 contains a SUBREG mixing SImode and SFmode.
30787 (vsx_register_operand): Likewise.
30788 (vsx_reg_sfsubreg_ok): New predicate.
30789 (vfloat_operand): Do not return true if the operand contains a
30790 SUBREG mixing SImode and SFmode.
30791 (vint_operand): Likewise.
30792 (vlogical_operand): Likewise.
30793 (gpc_reg_operand): Likewise.
30794 (int_reg_operand): Likewise.
30795 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
30796 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
30797 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
30799 (rs6000_emit_move_si_sf_subreg): New helper function.
30800 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
30801 fixup SUBREGs involving SImode and SFmode.
30802 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
30803 numbers for the new peephole2 optimization.
30804 (peephole2 for SFmode unions): New peephole2 to optimize cases in
30805 the GLIBC math library that do AND/IOR/XOR operations on single
30806 precision floating point.
30807 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
30808 target macros to say whether we need to avoid SUBREGs mixing
30810 (TARGET_ALLOW_SF_SUBREG): Likewise.
30811 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
30812 (UNSPEC_SI_FROM_SF): Likewise.
30813 (iorxor): Change spacing.
30814 (and_ior_xor): New iterator for AND, IOR, and XOR.
30815 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
30816 (movdi_from_sf_zero_ext): Likewise.
30817 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
30818 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
30819 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
30820 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
30821 (fms<mode>4): Likewise.
30822 (fnma<mode>4): Likewise.
30823 (fnms<mode>4): Likewise.
30824 (nfma<mode>4): Likewise.
30825 (nfms<mode>4): Likewise.
30827 2017-01-04 Marek Polacek <polacek@redhat.com>
30830 * doc/invoke.texi: Document -Wpointer-compare.
30832 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30834 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
30837 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
30838 descriptions for -gdwarf-5 and emit them as uleb128 instead of
30841 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
30844 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
30845 documentation of the powerpc_popcntb_ok attribute.
30846 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30847 code to issue warning messages if a requested CPU configuration is
30848 not supported by the binary (assembler and loader) toolchain.
30849 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
30850 made to define a built-in function that has been disabled.
30851 (paired_init_builtins): Add assertion to prevent ICE if attempt is
30852 made to define a built-in function that has been disabled.
30853 (altivec_init_builtins): Add comment explaining why definition
30854 of the DST built-in functions is not preceded by an assertion
30855 check. Add assertions to prevent ICE if attempts are made to
30856 define an altivec predicate or an abs* built-in function that has
30858 (htm_init_builtins): Add comment explaining why definition of the
30859 htm built-in functions is not preceded by an assertion check.
30861 2017-01-04 Jeff Law <law@redhat.com>
30863 PR tree-optimizatin/67955
30864 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
30865 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
30866 the points-to solution does not include pt_null. Use DECL_PT_UID
30869 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
30871 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
30872 Use gen_int_mode instead of gen_lopwart for const_int operands.
30874 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30876 PR tree-optimization/71563
30877 * match.pd: Simplify X << Y into X if Y is known to be 0 or
30878 out of range value - has low bits known to be zero.
30880 2017-01-04 Alan Modra <amodra@gmail.com>
30882 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
30883 * configure: Regenerate.
30884 * config.in: Regenerate.
30886 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30889 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
30890 a substring of the message, but strcmp with the whole message. Ifdef
30891 ENABLE_NLS, translate the message first using dgettext.
30893 2017-01-03 Jeff Law <law@redhat.com>
30895 PR tree-optimizatin/78856
30896 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
30897 (mark_threaded_blocks): Remove code to truncate thread paths that
30898 cross multiple loop headers. Instead invalidate the cached loop
30899 iteration information and handle case of a thread path walking
30900 into an irreducible region.
30902 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
30905 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
30906 assertions. Add support for doing the signbit if the IEEE 128-bit
30907 floating point value is in a GPR.
30908 * config/rs6000/rs6000.md (Fsignbit): Delete.
30909 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
30910 Update the length attribute if the value is in a GPR.
30911 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
30912 the sign or zero extension instruction, since the value is always 0/1.
30913 (signbit<mode>2_dm2): Delete using <Fsignbit>.
30916 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
30917 extracting SImode to a GPR register so that we can generate a
30918 store, limit the vector to be in a traditional Altivec register
30919 for the vextuwrx instruction.
30921 2017-01-03 Ian Lance Taylor <iant@google.com>
30923 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
30925 2017-01-03 Martin Sebor <msebor@redhat.com>
30927 PR tree-optimization/78696
30928 * gimple-ssa-sprintf.c (format_floating): Correct handling of
30929 precision. Use MPFR for %f for greater fidelity. Correct handling
30931 (pass_sprintf_length::compute_format_length): Set width and precision
30932 specified by asrerisk to void_node for vararg functions.
30933 (try_substitute_return_value): Adjust dump output.
30935 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
30937 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
30939 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
30941 * doc/invoke.texi (SPARC options): Document -mlra as the default.
30942 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
30943 -mlra/-mno-lra was passed to the compiler.
30945 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
30947 PR rtl-optimization/65618
30948 * emit-rtl.c (try_split): Move initialization of "before" and
30949 "after" to just before the call to emit_insn_after_setloc.
30951 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
30953 * doc/md.texi (Standard Names): Remove reference to Java frontend.
30955 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
30957 * dwarf2out.c (gen_enumeration_type_die): When
30958 -gno-strict-dwarf, add a DW_AT_encoding attribute.
30960 2017-01-03 Jakub Jelinek <jakub@redhat.com>
30962 PR tree-optimization/78965
30963 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
30964 Change first argument from const call_info & to call_info &. For %n
30965 set info.nowrite to false.
30967 PR middle-end/78901
30968 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
30969 possibly throwing calls.
30971 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
30972 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
30973 and fns handling, rather than in a separate case SSA_NAME.
30975 2017-01-02 Jeff Law <law@redhat.com>
30977 * config/darwin-driver.c (darwin_driver_init): Const-correctness
30978 fixes for first_period and second_period variables.
30980 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
30983 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
30984 (*insvqi_1): New insn pattern.
30985 (*insvqi_1_mem_rex64): Ditto.
30986 (*insvqi_2): Ditto.
30987 (*insvqi_3): Rename from *insvqi.
30989 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
30991 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
30993 * doc/cfg.texi (Edges): Remove reference to Java.
30994 (Maintaining the CFG): Ditto.
30996 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30998 PR middle-end/77674
30999 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
31000 transparent aliases.
31002 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31004 PR middle-end/77484
31005 * predict.def (PRED_CALL): Update hitrate.
31006 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
31007 * predict.c (tree_estimate_probability_bb): Split CALL predictor
31008 into direct/indirect/polymorphic variants.
31010 2017-01-01 Jakub Jelinek <jakub@redhat.com>
31012 Update copyright years.
31014 * gcc.c (process_command): Update copyright notice dates.
31015 * gcov-dump.c (print_version): Ditto.
31016 * gcov.c (print_version): Ditto.
31017 * gcov-tool.c (print_version): Ditto.
31018 * gengtype.c (create_file): Ditto.
31019 * doc/cpp.texi: Bump @copying's copyright year.
31020 * doc/cppinternals.texi: Ditto.
31021 * doc/gcc.texi: Ditto.
31022 * doc/gccint.texi: Ditto.
31023 * doc/gcov.texi: Ditto.
31024 * doc/install.texi: Ditto.
31025 * doc/invoke.texi: Ditto.
31027 Copyright (C) 2017 Free Software Foundation, Inc.
31029 Copying and distribution of this file, with or without modification,
31030 are permitted in any medium without royalty provided the copyright
31031 notice and this notice are preserved.