1 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
5 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7 * config/i386/i386.md (isa): Remove fma_avx512f.
8 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
9 <avx512>_fmadd_<mode>_mask3<round_name>,
10 <avx512>_fmsub_<mode>_mask<round_name>,
11 <avx512>_fmsub_<mode>_mask3<round_name>,
12 <avx512>_fnmadd_<mode>_mask<round_name>,
13 <avx512>_fnmadd_<mode>_mask3<round_name>,
14 <avx512>_fnmsub_<mode>_mask<round_name>,
15 <avx512>_fnmsub_<mode>_mask3<round_name>,
16 <avx512>_fmaddsub_<mode>_mask<round_name>,
17 <avx512>_fmaddsub_<mode>_mask3<round_name>,
18 <avx512>_fmsubadd_<mode>_mask<round_name>,
19 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
20 (*vec_widen_umult_even_v16si<mask_name>,
21 *vec_widen_smult_even_v16si<mask_name>): Likewise.
22 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
24 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
26 * extend.texi: Add 'nocf_check' documentation.
27 * gimple.texi: Add second parameter to
28 gimple_build_call_from_tree.
29 * invoke.texi: Add -fcf-protection documentation.
30 * rtl.texi: Add REG_CALL_NOTRACK documenation.
32 2017-10-20 Richard Biener <rguenther@suse.de>
34 PR tree-optimization/82473
35 * tree-vect-loop.c (vectorizable_reduction): Properly get at
36 the largest input type.
38 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
40 * c-attribs.c (handle_nocf_check_attribute): New function.
41 (c_common_attribute_table): Add 'nocf_check' handling.
42 * gimple-parser.c: Add second argument NULL to
43 gimple_build_call_from_tree.
44 * attrib.c (comp_type_attributes): Check nocf_check attribute.
45 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
47 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
48 * common.opt: Add fcf-protection flag.
49 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
50 * flag-types.h: Add enum cf_protection_level.
51 * gimple.c (gimple_build_call_from_tree): Add second parameter.
52 Add 'nocf_check' attribute propagation to gimple call.
53 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
54 (gimple_build_call_from_tree): Update prototype.
55 (gimple_call_nocf_check_p): New function.
56 (gimple_call_set_nocf_check): Likewise.
57 * gimplify.c: Add second argument to gimple_build_call_from_tree.
58 * ipa-icf.c: Add nocf_check attribute in statement hash.
59 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
60 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
61 * toplev.c (process_options): Add flag_cf_protection handling.
63 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
65 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
67 2017-10-20 Richard Biener <rguenther@suse.de>
69 PR tree-optimization/82603
70 * tree-if-conv.c (predicate_mem_writes): Make sure to only
71 remove false predicated stores.
73 2017-10-20 Richard Biener <rguenther@suse.de>
75 * graphite-isl-ast-to-gimple.c
76 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
77 Remove return value and simplify, dump copied stmt after lhs
79 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
80 Reduce dump verbosity.
81 (gsi_insert_earliest): Likewise.
82 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
83 * graphite.c (print_global_statistics): Adjust dumping.
84 (print_graphite_scop_statistics): Likewise.
85 (print_graphite_statistics): Do not dump loops here.
86 (graphite_transform_loops): But here.
88 2017-10-20 Nicolas Roche <roche@adacore.com>
90 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
91 * configure: Regenerate.
93 2017-10-20 Jakub Jelinek <jakub@redhat.com>
96 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
97 functions when optimizing replace GIMPLE_RETURN stmts with
98 calls to __builtin_unreachable ().
101 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
102 for -fsanitize=thread link of executables.
103 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
107 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
109 (<shift_insn><mode>3<mask_name>): Change the last of the 3
110 define_insns for logical vector shifts to use VI248_AVX512BW
111 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
112 condition, useless isa and prefix attributes. Change the first
113 2 of these define_insns to ...
114 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
115 define_insn for avx512vl.
116 (<shift_insn><mode>3): ... and this, new define_insn without
117 masking for non-avx512vl.
120 * config/i386/sse.md (*andnot<mode>3,
121 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
122 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
123 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
124 not applied use empty suffix even for TARGET_AVX512VL.
125 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
126 is applied, supply evex,evex or evex,evex,evex instead of just
129 2017-10-20 Julia Koval <julia.koval@intel.com>
131 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
132 (OPTION_MASK_ISA_GFNI_UNSET): New.
133 (ix86_handle_option): Handle OPT_mgfni.
134 * config/i386/cpuid.h (bit_GFNI): New.
135 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
136 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
137 * config/i386/i386.c (ix86_target_string): Add -mgfni.
138 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
139 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
140 * config/i386/i386.opt: Add mgfni.
142 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
144 * config/msp430/msp430.c (msp430_option_override): Disable
145 -fdelete-null-pointer-checks.
146 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
148 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
150 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
151 of x87 and SSE instructions.
153 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
155 * asan.c (create_cond_insert_point): Do not update edge count.
156 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
157 (afdo_propagate_circuit): Likewise.
158 (afdo_calculate_branch_prob): Likewise.
159 (afdo_annotate_cfg): Likewise.
160 * basic-block.h (struct edge_def): Remove count.
161 (edge_def::count): New accessor.
162 * bb-reorder.c (rotate_loop): Update.
163 (find_traces_1_round): Update.
164 (connect_traces): Update.
165 (sanitize_hot_paths): Update.
166 * cfg.c (unchecked_make_edge): Update.
167 (make_single_succ_edge): Update.
168 (check_bb_profile): Update.
169 (dump_edge_info): Update.
170 (update_bb_profile_for_threading): Update.
171 (scale_bbs_frequencies_int): Update.
172 (scale_bbs_frequencies_gcov_type): Update.
173 (scale_bbs_frequencies_profile_count): Update.
174 (scale_bbs_frequencies): Update.
175 * cfganal.c (connect_infinite_loops_to_exit): Update.
176 * cfgbuild.c (compute_outgoing_frequencies): Update.
177 (find_many_sub_basic_blocks): Update.
178 * cfgcleanup.c (try_forward_edges): Update.
179 (try_crossjump_to_edge): Update
180 * cfgexpand.c (expand_gimple_cond): Update
181 (expand_gimple_tailcall): Update
182 (construct_exit_block): Update
183 * cfghooks.c (verify_flow_info): Update
184 (redirect_edge_succ_nodup): Update
186 (make_forwarder_block): Update
187 (duplicate_block): Update
188 (account_profile_record): Update
189 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
190 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
191 * cfgloopmanip.c (scale_loop_profile): Update.
193 (lv_adjust_loop_entry_edge): Update.
194 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
195 (force_nonfallthru_and_redirect): Update.
196 (purge_dead_edges): Update.
197 (rtl_flow_call_edges_add): Update.
198 * cgraphunit.c (init_lowered_empty_function): Update.
199 (cgraph_node::expand_thunk): Update.
200 * gimple-pretty-print.c (dump_probability): Update.
201 (dump_edge_probability): Update.
202 * gimple-ssa-isolate-paths.c (isolate_path): Update.
203 * haifa-sched.c (sched_create_recovery_edges): Update.
204 * hsa-gen.c (convert_switch_statements): Update.
205 * ifcvt.c (dead_or_predicable): Update.
206 * ipa-inline-transform.c (inline_transform): Update.
207 * ipa-split.c (split_function): Update.
208 * ipa-utils.c (ipa_merge_profiles): Update.
209 * loop-doloop.c (add_test): Update.
210 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
211 * lto-streamer-in.c (input_cfg): Update.
212 (input_function): Update.
213 * lto-streamer-out.c (output_cfg): Update.
214 * modulo-sched.c (sms_schedule): Update.
215 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
216 * predict.c (maybe_hot_edge_p): Update.
217 (unlikely_executed_edge_p): Update.
218 (probably_never_executed_edge_p): Update.
219 (dump_prediction): Update.
220 (drop_profile): Update.
221 (propagate_unlikely_bbs_forward): Update.
222 (determine_unlikely_bbs): Update.
223 (force_edge_cold): Update.
224 * profile.c (compute_branch_probabilities): Update.
225 * reg-stack.c (better_edge): Update.
226 * shrink-wrap.c (handle_simple_exit): Update.
227 * tracer.c (better_p): Update.
228 * trans-mem.c (expand_transaction): Update.
229 (split_bb_make_tm_edge): Update.
230 * tree-call-cdce.c: Update.
231 * tree-cfg.c (gimple_find_sub_bbs): Update.
232 (gimple_split_edge): Update.
233 (gimple_duplicate_sese_region): Update.
234 (gimple_duplicate_sese_tail): Update.
235 (gimple_flow_call_edges_add): Update.
236 (insert_cond_bb): Update.
237 (execute_fixup_cfg): Update.
238 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
239 * tree-complex.c (expand_complex_div_wide): Update.
240 * tree-eh.c (lower_resx): Update.
241 (unsplit_eh): Update.
242 (cleanup_empty_eh_move_lp): Update.
243 * tree-inline.c (copy_edges_for_bb): Update.
244 (freqs_to_counts): Update.
245 (copy_cfg_body): Update.
246 * tree-ssa-dce.c (remove_dead_stmt): Update.
247 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
248 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
249 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
250 (unloop_loops): Update.
251 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
252 * tree-ssa-loop-split.c (connect_loops): Update.
253 (split_loop): Update.
254 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
255 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
256 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
257 * tree-ssa-reassoc.c (branch_fixup): Update.
258 * tree-ssa-tail-merge.c (replace_block_by): Update.
259 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
260 (compute_path_counts): Update.
261 (update_profile): Update.
262 (recompute_probabilities): Update.
263 (update_joiner_offpath_counts): Update.
264 (estimated_freqs_path): Update.
265 (freqs_to_counts_path): Update.
266 (clear_counts_path): Update.
267 (ssa_fix_duplicate_block_edges): Update.
268 (duplicate_thread_path): Update.
269 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
270 (case_bit_test_cmp): Update.
271 (collect_switch_conv_info): Update.
272 (gen_inbound_check): Update.
273 (do_jump_if_equal): Update.
274 (emit_cmp_and_jump_insns): Update.
275 * tree-tailcall.c (decrease_profile): Update.
276 (eliminate_tail_call): Update.
277 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
278 (vect_do_peeling): Update.
279 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
280 * ubsan.c (ubsan_expand_null_ifn): Update.
281 (ubsan_expand_ptr_ifn): Update.
282 * value-prof.c (gimple_divmod_fixed_value): Update.
283 (gimple_mod_pow2): Update.
284 (gimple_mod_subtract): Update.
286 (gimple_stringop_fixed_value): Update.
288 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
291 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
293 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
295 PR rtl-optimization/82395
296 * ira-color.c (allocno_priority_compare_func): Fix comparison step
297 based on non_spilled_static_chain_regno_p.
299 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
301 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
302 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
303 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
305 2017-10-19 Martin Sebor <msebor@redhat.com>
307 PR tree-optimization/82596
308 * tree.c (array_at_struct_end_p): Handle STRING_CST.
310 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
312 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
313 (get_mem_refs_of_builtin_call): Likewise.
314 * builtins.c (expand_builtin_apply): Adjust call to
315 allocate_dynamic_stack_space.
316 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
317 the third argument to allocate_dynamic_stack_space, otherwise -1.
318 (expand_builtin): Deal with all alloca variants.
319 (is_inexpensive_builtin): Likewise.
320 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
321 * calls.c (special_function_p): Deal with all alloca variants.
322 (initialize_argument_information): Adjust call to
323 allocate_dynamic_stack_space.
324 (expand_call): Likewise.
325 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
326 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
327 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
328 use it for the stack usage computation.
329 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
330 * function.c (gimplify_parameters): Call build_alloca_call_expr.
331 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
332 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
333 (in_loop_p): Remove first argument and useless check.
334 (pass_walloca::execute): Remove useless test and adjust call to above.
335 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
336 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
337 (gimplify_call_expr): Deal with all alloca variants.
338 * hsa-gen.c (gen_hsa_alloca): Likewise.
339 (gen_hsa_insns_for_call): Likewise.
340 * ipa-pure-const.c (special_builtin_state): Likewise.
341 * tree-chkp.c (chkp_build_returned_bound): Likewise.
342 * tree-object-size.c (alloc_object_size): Likewise.
343 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
344 (call_may_clobber_ref_p_1): Likewise.
345 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
346 (ccp_fold_stmt): Likewise.
347 (optimize_stack_restore): Likewise.
348 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
349 (mark_all_reaching_defs_necessary_1): Likewise.
350 (propagate_necessity): Likewise.
351 (eliminate_unnecessary_stmts): Likewise.
352 * tree.c (build_common_builtin_nodes): Build
353 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
354 (build_alloca_call_expr): New function.
355 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
356 (CASE_BUILT_IN_ALLOCA): Likewise.
357 (build_alloca_call_expr): Declare.
358 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
360 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
363 * dwarf2out.c (new_die_raw): New static inline function.
364 (new_die): Use it to create the DIE.
365 (add_AT_external_die_ref): Likewise.
366 (clone_die): Likewise.
367 (clone_as_declaration): Likewise.
368 (dwarf2out_vms_debug_main_pointer): Likewise.
369 (base_type_die): Likewise. Remove early return for corner cases.
370 Do not call add_pubtype on the DIE here.
371 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
372 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
373 typedefs for base types with DW_AT_endianity. Make sure a DIE with
374 native order exists for base types, attach the DIE manually and call
375 add_pubtype on it. Do not equate a reverse order DIE to the type.
377 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
379 * config/arm/arm.c (align_ok_ldrd_strd): New function.
380 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
382 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
384 2017-10-19 Jakub Jelinek <jakub@redhat.com>
386 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
387 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
388 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
389 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
390 * opts.c (sanitizer_opts): Add builtin.
391 * ubsan.c (instrument_builtin): New function.
392 (pass_ubsan::execute): Call it.
393 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
394 * doc/invoke.texi: Document -fsanitize=builtin.
396 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
397 builtins, store max (log2 (align), 0) into uchar field instead of
398 align into uptr field.
399 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
400 store uchar 0 field instead of uptr 0 field.
401 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
402 instead of passing one address of struct with 2 locations pass
403 two addresses of structs with 1 location each.
404 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
405 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
406 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
407 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
408 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
409 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
410 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
411 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
413 2017-10-19 Martin Liska <mliska@suse.cz>
416 * file-find.c (remove_prefix): Remove.
417 * file-find.h (remove_prefix): Likewise.
418 * gcc-ar.c: Remove smartness of lookup.
420 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
422 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
423 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
424 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
426 2017-10-19 Jakub Jelinek <jakub@redhat.com>
429 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
430 (setcc + and to xor + setcc): New peephole2.
432 2017-10-19 Tom de Vries <tom@codesourcery.com>
434 * doc/sourcebuild.texi (Test Directives, Variants of
435 dg-require-support): Add dg-require-stack-size.
437 2017-10-19 Martin Liska <mliska@suse.cz>
440 * gimplify.c (gimplify_decl_expr): Do not instrument variables
441 that have a large alignment.
442 (gimplify_target_expr): Likewise.
444 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
446 PR rtl-optimization/82602
447 * ira.c (rtx_moveable_p): Return false for volatile asm.
449 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
452 * config/i386/i386-modes.def (CCGZ): New CC mode.
453 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
454 * config/i386/predicates.md (ix86_comparison_operator):
456 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
457 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
458 with double-word subtraction.
459 (put_condition_code): Handle CCGZmode.
461 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
463 * wide-int.cc (debug (const wide_int &)): New.
464 (debug (const wide_int *)): New.
465 (debug (const widest_int &)): New.
466 (debug (const widest_int *)): New.
468 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
471 * lra-constraints.c (curr_insn_transform): Use non-input operand
472 instead of output one for matched reload.
474 2017-10-18 Bin Cheng <bin.cheng@arm.com>
476 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
477 (tree-ssa-loop-ivopts.h): New header file.
478 (struct builtin_info): New fields.
479 (classify_builtin_1): Compute and record base and offset parts for
480 memset builtin partition by calling strip_offset.
481 (offset_cmp, fuse_memset_builtins): New functions.
482 (finalize_partitions): Fuse adjacent memset partitions by calling
484 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
485 Expose the interface.
486 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
488 2017-10-18 Bin Cheng <bin.cheng@arm.com>
490 PR tree-optimization/82574
491 * tree-loop-distribution.c (find_single_drs): New parameter. Check
492 that data reference must be executed exactly once per iteration
493 against the outermost loop in nest.
494 (classify_partition): Update call to above function.
496 2017-10-18 Richard Biener <rguenther@suse.de>
498 PR tree-optimization/82591
499 * graphite.c (graphite_transform_loops): Move code gen message
501 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
502 Here. Handle scop_to_isl_ast failing.
503 (scop_to_isl_ast): Limit the number of ISL operations.
505 2017-10-18 Richard Biener <rguenther@suse.de>
507 * graphite-isl-ast-to-gimple.c
508 (translate_isl_ast_to_gimple::set_rename): Simplify.
509 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
510 (graphite_copy_stmts_from_block): ... here.
511 (copy_bb_and_scalar_dependences): Simplify.
512 (add_parameters_to_ivs_params): Canonicalize.
513 (generate_entry_out_of_ssa_copies): Simplify.
514 * graphite-sese-to-poly.c (extract_affine_name): Simplify
515 by passing in ISL dimension.
516 (parameter_index_in_region_1): Rename to ...
517 (parameter_index_in_region): ... this.
518 (extract_affine): Adjust assert, pass down parameter index.
519 (add_param_constraints): Use range-info when available.
520 (build_scop_context): Adjust.
521 * sese.c (new_sese_info): Adjust.
522 (free_sese_info): Likewise.
523 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
524 Remove unused typedefs.
525 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
527 2017-10-18 Martin Liska <mliska@suse.cz>
529 * combine.c (simplify_compare_const): Add gcc_fallthrough.
531 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
533 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
534 (s390_sched_init): Do not reset s390_sched_state if we entered the
535 current basic block via a fallthru edge and all others are unlikely.
537 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
539 * config/s390/s390.c (NUM_SIDES): New variable.
540 (LONGRUNNING_THRESHOLD): New variable.
541 (LATENCY_FACTOR): New variable.
542 (s390_sched_score): Decrease score for long-running instructions on
544 (s390_sched_variable_issue): Perform bookkeeping for long-running
547 2017-10-18 Richard Biener <rguenther@suse.de>
549 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
550 Simplify with removal of the parameter rename map.
551 (set_rename): Likewise.
552 (should_copy_to_new_region): Likewise.
553 (graphite_copy_stmts_from_block): Likewise.
554 (copy_bb_and_scalar_dependences): Remove initialization of
555 unused copied_bb_map.
557 (copy_internal_parameters): Likewise.
558 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
559 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
561 (parameter_index_in_region_1): Rename to ...
562 (assign_parameter_index_in_region): ... this. Assert we have
563 a parameter we handle.
564 (scan_tree_for_params): Adjust.
565 * sese.h (parameter_rename_map_t): Remove.
566 (struct sese_info_t): Remove unused parameter_rename_map and
567 copied_bb_map members.
568 * sese.c (new_sese_info): Adjust.
569 (free_sese_info): Likewise.
571 2017-10-18 Martin Liska <mliska@suse.cz>
574 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
577 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
579 * doc/invoke.texi (ffunction-sections and fdata-sections):
582 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
584 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
585 the use statement can throw internally.
587 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
589 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
590 any RTX present on the RHS of a SET.
591 * compare-elim.c (try_eliminate_compare): Restore comment.
593 2017-10-17 Jakub Jelinek <jakub@redhat.com>
595 * langhooks.h (struct lang_hooks): Document that tree_size langhook
596 may be also called on tcc_type nodes.
597 * langhooks.c (lhd_tree_size): Likewise.
599 2017-10-17 David Malcolm <dmalcolm@redhat.com>
601 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
602 format_warning_at_substring.
603 (maybe_warn): Convert source_range * param to a location_t. Pass
604 UNKNOWN_LOCATION rather than NULL to fmtwarn.
605 (format_directive): Remove code to extract source_ranges and
606 source_range * in favor of just a location_t.
607 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
609 * substring-locations.c (format_warning_va): Convert
610 source_range * param to a location_t.
611 (format_warning_at_substring): Likewise.
612 * substring-locations.h (format_warning_va): Likewise.
613 (format_warning_at_substring): Likewise.
615 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
617 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
619 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
620 and scatter/gather ops.
622 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
623 vec_gather_load and vec_scatter_store.
624 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
625 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
626 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
627 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
628 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
629 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
631 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
633 * reg-stack.c (compare_for_stack_reg): Add bool argument.
634 Detect FTST instruction and handle its register pops. Only pop
635 second operand if can_pop_second_op is true.
636 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
637 set can_pop_second_op to false in the compare_for_stack_reg call.
639 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
640 output_fp_compare for stack register operands.
641 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
642 instructions here. Do not emit stack register pops here. Assert
643 that FCOMPP pops next to top stack register. Rewrite function.
645 2017-10-17 Nathan Sidwell <nathan@acm.org>
648 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
649 use DECL_ASSEMBLER_NAME_RAW.
652 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
655 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
656 Wilco Dijkstra <wilco.dijkstra@arm.com>
658 * builtins.c (expand_builtin_update_setjmp_buf): Add a
659 converstion to Pmode from the buf_addr.
661 2017-10-17 Richard Biener <rguenther@suse.de>
663 * graphite-dependences.c (scop_get_reads_and_writes): Change
664 output parameters to references.
666 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
668 PR 71026/tree-optimization
669 * fold-const.c (distribute_real_division): Removed.
670 (fold_binary_loc): Remove calls to distribute_real_divison.
672 2017-10-17 Richard Biener <rguenther@suse.de>
674 * graphite-scop-detection.c
675 (scop_detection::stmt_has_simple_data_refs_p): Always use
676 the full nest as region.
677 (try_generate_gimple_bb): Likewise.
678 * sese.c (scalar_evolution_in_region): Simplify now that
679 SCEV can handle instantiation in regions.
680 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
681 in the non-loop part of a function if requested.
683 2017-10-17 Richard Biener <rguenther@suse.de>
685 PR tree-optimization/82563
686 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
688 (graphite_regenerate_ast_isl): Call it.
689 * graphite-scop-detection.c (build_scops): Remove entry edge split.
691 2017-10-17 Jakub Jelinek <jakub@redhat.com>
693 PR tree-optimization/82549
694 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
695 Formatting fixes. Instead of calling make_bit_field_ref with negative
698 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
700 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
701 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
702 _mm_maskz_reduce_ss): New.
703 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
704 __builtin_ia32_reducess_mask): Ditto..
705 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
706 * config/i386/sse.md (reduces<mode>): Renamed to ...
707 (reduces<mode><mask_scalar_name>): ... this.
708 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
710 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
711 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
713 2017-10-16 David Malcolm <dmalcolm@redhat.com>
715 * Makefile.in (OBJS): Add unique-ptr-tests.o.
716 * selftest-run-tests.c (selftest::run_tests): Call
717 selftest::unique_ptr_tests_cc_tests.
718 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
719 * unique-ptr-tests.cc: New file.
721 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
724 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
726 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
727 (make_hard_regno_born, make_hard_regno_dead): Update
728 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
730 2017-10-16 Jeff Law <law@redhat.com>
732 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
734 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
736 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
738 2017-10-16 Olivier Hainque <hainque@adacore.com>
740 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
741 with_cpu if we were configured for an e500v2 target cpu name.
743 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
745 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
746 * doc/invoke.texi: Document +nodsp as a valid extension for
749 2017-10-16 Martin Liska <mliska@suse.cz>
751 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
752 (test_set_range): Likewise.
753 (test_range_functions): Rename to ...
754 (test_bit_in_range): ... this.
755 (sbitmap_c_tests): Add new test.
757 2017-10-16 Tamar Christina <tamar.christina@arm.com>
759 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
760 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
761 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
763 2017-10-16 Tamar Christina <tamar.christina@arm.com>
765 * config/aarch64/aarch64-builtins.c
766 (aarch64_types_quadopu_lane_qualifiers): New.
767 (TYPES_QUADOPU_LANE): New.
768 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
769 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
770 (aarch64_<sur>dot_laneq<vsi2qi>): New.
771 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
772 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
773 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
774 (Vdottype, DOTPROD): New.
775 (sur): Add SDOT and UDOT.
777 2017-10-16 Tamar Christina <tamar.christina@arm.com>
779 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
780 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
781 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
782 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
783 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
784 (cortex-a75.cortex-a55): Likewise.
785 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
787 2017-10-16 Tamar Christina <tamar.christina@arm.com>
789 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
790 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
791 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
792 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
793 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
794 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
795 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
796 * config/arm/types.md (neon_dot, neon_dot_q): New.
797 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
799 2017-10-16 Tamar Christina <tamar.christina@arm.com>
801 * config/arm/arm.h (TARGET_DOTPROD): New.
802 * config/arm/arm.c (arm_arch_dotprod): New.
803 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
804 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
805 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
806 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
807 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
808 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
809 * doc/invoke.texi (armv8.2-a): Document dotprod
811 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
813 * i386.c (ix86_vec_cost): New function.
814 (ix86_rtx_costs): Handle vector operations better.
815 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
816 * x86-tune-costs.h: Add new costs to all tables.
818 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
820 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
822 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
823 divsd, sqrtss and sqrtsd
824 * x86-tune-costs.h: Add new entries to all costs.
825 (znver1_cost): Fix to match real instruction latencies.
827 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
828 Michael Collison <michael.collison@arm.com>
830 * compare-elim.c: Include emit-rtl.h.
831 (can_merge_compare_into_arith): New function.
832 (try_validate_parallel): Likewise.
833 (try_merge_compare): Likewise.
834 (try_eliminate_compare): Call the above when no previous clobber
836 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
839 2017-10-14 Jakub Jelinek <jakub@redhat.com>
843 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
844 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
848 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
849 to be any operand_equal_p operands. For & (B - 1) require
850 B to be power of 2. Recognize
851 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
853 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
856 * optabs.c (expand_memory_blockage): Fix call of
857 targetm.have_memory_blockage.
859 2017-10-14 Jakub Jelinek <jakub@redhat.com>
862 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
863 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
864 objects to extra_objs instead of overwriting it.
866 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
868 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
869 Use any_fp_register_operand as operand[3] predicate. Simplify
870 equality test for operands[2] and operands[4] memory location.
871 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
872 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
873 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
874 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
875 any_fp_register_operand as operand[1] predicate. Simplify
876 equality test for operands[0] and operands[3] memory location.
877 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
878 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
879 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
881 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
883 * target-insns.def: Add memory_blockage.
884 * optabs.c (expand_memory_blockage): New function.
885 (expand_asm_memory_barrier): Rename ...
886 (expand_asm_memory_blockage): ... to this.
887 (expand_mem_thread_fence): Call expand_memory_blockage
888 instead of expand_asm_memory_barrier.
889 (expand_mem_singnal_fence): Ditto.
890 (expand_atomic_load): Ditto.
891 (expand_atomic_store): Ditto.
892 * doc/md.texi (Standard Pattern Names For Generation):
893 Document memory_blockage instruction pattern.
895 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
897 * config/rl78/rl78.c (rl78_emit_libcall): New function.
898 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
899 * config/rl78/rl78.md: New define_expand "adddi3".
901 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
903 * cfghooks.c (verify_flow_info): Disable check that all probabilities
906 2017-10-13 Jeff Law <law@redhat.com>
908 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
910 2017-10-13 Jakub Jelinek <jakub@redhat.com>
913 * internal-fn.c (expand_mul_overflow): If both operands have
914 the same highpart of -1 or 0 and the topmost bit of lowpart
915 is different, overflow is if res <= 0 rather than res < 0.
917 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
919 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
920 TARGET_P9_VECTOR code for unaligned_load case.
922 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
924 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
926 2017-10-13 Nathan Sidwell <nathan@acm.org>
928 * tree-core.h (tree_contains_struct): Make bool.
929 * tree.c (tree_contains_struct): Likewise.
930 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
931 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
932 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
933 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
934 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
936 2017-10-13 Richard Biener <rguenther@suse.de>
938 * graphite-isl-ast-to-gimple.c
939 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
940 parameters and dominance check.
941 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
942 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
943 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
944 Do not update SSA form here or do intermediate IL verification.
945 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
946 (graphite_initialize): Remove check on the number of loops in
947 the function and inline into graphite_transform_loops.
948 (graphite_finalize): Inline into graphite_transform_loops.
949 (graphite_transform_loops): Perform SSA update and IL verification
951 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
953 2017-10-13 Richard Biener <rguenther@suse.de>
955 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
956 graphite_expression_type_precision): Avoid global constructor
958 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
959 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
960 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
961 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
962 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
963 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
965 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
968 * config/i386/i386.h (ix86_red_zone_size): New.
969 * config/i386/i386.md (push peephole2s): Replace
970 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
972 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
973 Alan Hayward <alan.hayward@arm.com>
974 David Sherwood <david.sherwood@arm.com>
976 * combine.c (can_change_dest_mode): Reject changes in
977 REGMODE_NATURAL_SIZE.
979 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
980 Alan Hayward <alan.hayward@arm.com>
981 David Sherwood <david.sherwood@arm.com>
983 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
984 (expand_debug_source_expr): Likewise.
985 * combine.c (combine_simplify_rtx): Likewise.
986 * cse.c (fold_rtx): Likewise.
987 * fwprop.c (canonicalize_address): Likewise.
988 * targhooks.c (default_shift_truncation_mask): Likewise.
990 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
991 Alan Hayward <alan.hayward@arm.com>
992 David Sherwood <david.sherwood@arm.com>
994 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
995 (widened_mode): Likewise.
996 (expand_unop): Likewise.
997 * ree.c (transform_ifelse): Likewise.
998 (merge_def_and_ext): Likewise.
999 (combine_reaching_defs): Likewise.
1000 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1002 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1003 Alan Hayward <alan.hayward@arm.com>
1004 David Sherwood <david.sherwood@arm.com>
1006 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
1007 * combine.c (gen_lowpart_for_combine): Likewise.
1008 * dwarf2out.c (rtl_for_decl_location): Likewise.
1009 * final.c (alter_subreg): Likewise.
1010 * rtlhooks.c (gen_lowpart_general): Likewise.
1011 (gen_lowpart_if_possible): Likewise.
1013 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1014 Alan Hayward <alan.hayward@arm.com>
1015 David Sherwood <david.sherwood@arm.com>
1017 * calls.c (expand_call): Use subreg_lowpart_offset.
1018 * cse.c (cse_insn): Likewise.
1019 * regcprop.c (copy_value): Likewise.
1020 (copyprop_hardreg_forward_1): Likewise.
1022 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1025 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
1026 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
1027 =Q constraints to +Q and into insn condition add check
1028 that operands[0] and operands[1] are equal.
1029 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
1030 =Q constraints to +Q and into insn condition add check
1031 that operands[0] is equal to either operands[1] or operands[2].
1034 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
1035 instead of handling MINUS_EXPR twice (once for each argument),
1036 canonicalize operand order and handle just once, use rtype where
1037 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
1040 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
1041 any values of __C while still being pattern recognizable as a simple
1044 2017-10-13 Richard Biener <rguenther@suse.de>
1046 PR tree-optimization/82451
1048 2017-10-02 Richard Biener <rguenther@suse.de>
1050 PR tree-optimization/82355
1051 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1052 a mapping for the enclosing loop but avoid generating one for
1054 (copy_bb_and_scalar_dependences): Remove premature codegen
1055 error on PHIs in blocks duplicated into multiple places.
1056 * graphite-scop-detection.c
1057 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1058 in the region use it as loop and nest to analyze the DR in.
1059 (try_generate_gimple_bb): Likewise.
1060 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1061 (add_loop_constraints): For blocks in a loop not in the region
1062 create a dimension with a single iteration.
1063 * sese.h (gbb_loop_at_index): Remove assert.
1065 * cfgloop.c (loop_preheader_edge): For the loop tree root
1066 return the single successor of the entry block.
1067 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1068 Reset the SCEV hashtable and niters.
1069 * graphite-scop-detection.c
1070 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
1071 assert that we only have POLYNOMIAL_CHREC that vary in loops
1072 contained in the region.
1073 (scop_detection::graphite_can_represent_expr): Adjust.
1074 (scop_detection::stmt_has_simple_data_refs_p): For loops
1075 not in the region set loop to NULL. The nest is now the
1076 entry edge to the region.
1077 (try_generate_gimple_bb): Likewise.
1078 * sese.c (scalar_evolution_in_region): Adjust for
1079 instantiate_scev change.
1080 * tree-data-ref.h (graphite_find_data_references_in_stmt):
1081 Make nest parameter the edge into the region.
1082 (create_data_ref): Likewise.
1083 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
1084 entry edge into a region and adjust instantiate_scev calls.
1085 (create_data_ref): Likewise.
1086 (graphite_find_data_references_in_stmt): Likewise.
1087 (find_data_references_in_stmt): Pass the loop preheader edge
1088 from the nest argument.
1089 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
1090 parameter the edge into the region.
1091 (instantiate_parameters): Use the loop preheader edge as entry.
1092 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
1094 (get_instantiated_value_entry): Make instantiate_below parameter
1095 the edge into the region.
1096 (instantiate_scev_name): Likewise. Adjust dominance checks,
1097 when we cannot use loop-based instantiation instantiate by
1098 walking use-def chains.
1099 (instantiate_scev_poly): Adjust.
1100 (instantiate_scev_binary): Likewise.
1101 (instantiate_scev_convert): Likewise.
1102 (instantiate_scev_not): Likewise.
1103 (instantiate_array_ref): Remove.
1104 (instantiate_scev_3): Likewise.
1105 (instantiate_scev_2): Likewise.
1106 (instantiate_scev_1): Likewise.
1107 (instantiate_scev_r): Do not blindly handle N-operand trees.
1108 Do not instantiate array-refs. Handle all constants and invariants.
1109 (instantiate_scev): Make instantiate_below parameter
1110 the edge into the region.
1111 (resolve_mixers): Use the loop preheader edge for the region
1112 parameter to instantiate_scev_r.
1113 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1115 2017-10-13 Richard Biener <rguenther@suse.de>
1117 PR tree-optimization/82525
1118 * graphite-isl-ast-to-gimple.c
1119 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1121 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1122 Fail code generation when we cannot represent the isl integer.
1123 (binary_op_to_tree): Elide modulo operations that are no-ops
1124 in the type we code generate. Remove now superfluous code
1127 2017-10-13 Richard Biener <rguenther@suse.de>
1129 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1130 (scop_detection::harmful_loop_in_region): Remove premature
1131 IV type restriction.
1132 (scop_detection::graphite_can_represent_scev): We can handle
1133 pointer IVs just fine.
1135 2017-10-13 Alan Modra <amodra@gmail.com>
1137 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1138 "Clobbers and Scratch Registers". Add paragraph on
1139 alternative to clobbers for scratch registers and OpenBLAS
1142 2017-10-13 Alan Modra <amodra@gmail.com>
1144 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1145 example of a memory input for a string of known length. Move
1146 commentary out of table. Add a number of new examples
1147 covering array memory inputs.
1149 2017-10-12 Martin Liska <mliska@suse.cz>
1151 PR tree-optimization/82493
1152 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1153 (test_range_functions): New function.
1154 (sbitmap_c_tests): Likewise.
1155 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1156 * selftest.h (sbitmap_c_tests): New function.
1158 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1160 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1162 * config/rs6000/amo.h: Fix spacing issue.
1164 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1167 * config/i386/i386.md (*ashl<mode>3_mask_1,
1168 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1169 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1172 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1174 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
1175 (profile_probability): Set max_probability
1176 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
1178 * profile-count.c (profile_probability::differs_from_p): Do not
1179 rely on max_probaiblity == 10000
1181 2017-10-12 Jeff Law <law@redhat.com>
1183 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
1186 2017-10-12 Martin Sebor <msebor@redhat.com>
1190 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
1191 (handle_alias_pairs): Call it.
1192 * common.opt (-Wattribute-alias): New option.
1193 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
1194 * doc/invoke.texi (-Wattribute-alias): Document.
1196 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
1199 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1201 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1203 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1204 (make_hard_regno_born, make_hard_regno_dead): Update
1205 bb_killed_pseudos and bb_gen_pseudos.
1207 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1209 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
1211 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
1213 * config/alpha/alpha.c (alpha_split_conditional_move):
1214 Use std::swap instead of manually swapping.
1215 (alpha_stdarg_optimize_hook): Ditto.
1216 (alpha_canonicalize_comparison): Ditto.
1218 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1220 * tree-loop-distribution.c (struct builtin_info): New struct.
1221 (struct partition): Refactor fields into struct builtin_info.
1222 (partition_free): Free struct builtin_info.
1223 (build_size_arg_loc, build_addr_arg_loc): Delete.
1224 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
1225 information from struct builtin_info.
1226 (find_single_drs): New function refactored from classify_partition.
1227 Also moved builtin validity checks to this function.
1228 (compute_access_range, alloc_builtin): New functions.
1229 (classify_builtin_st, classify_builtin_ldst): New functions.
1230 (classify_partition): Refactor code into functions find_single_drs,
1231 classify_builtin_st and classify_builtin_ldst.
1232 (distribute_loop): Don't do runtime alias check when distributing
1234 (find_seed_stmts_for_distribution): New function.
1235 (pass_loop_distribution::execute): Refactor code finding seed
1236 stmts into above function. Support distribution for the innermost
1237 two-level loop nest. Adjust dump information.
1239 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1241 * tree-loop-distribution.c: Adjust the general comment.
1242 (NUM_PARTITION_THRESHOLD): New macro.
1243 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
1244 (classify_partition): Skip builtin pattern of loop nest's inner loop.
1245 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
1246 in call to build_partition_graph.
1247 (finalize_partitions): New parameter. Make loop distribution more
1248 conservative by fusing more partitions.
1249 (distribute_loop): Don't do runtime alias check in case of loop nest
1251 (find_seed_stmts_for_distribution): New function.
1252 (prepare_perfect_loop_nest): New function.
1253 (pass_loop_distribution::execute): Refactor code finding seed stmts
1254 and loop nest into above functions. Support loop nest distribution.
1255 Adjust dump information accordingly.
1257 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1259 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
1260 and set PTYPE_SEQUENTIAL for merged partition.
1262 2017-10-12 Richard Biener <rguenther@suse.de>
1264 PR tree-optimization/69728
1266 2017-09-19 Richard Biener <rguenther@suse.de>
1268 PR tree-optimization/69728
1269 * graphite-sese-to-poly.c (schedule_error): New global.
1270 (add_loop_schedule): Handle empty domain by failing the
1272 (build_original_schedule): Handle schedule_error.
1274 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
1275 domain by returning an unchanged schedule.
1277 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1279 * genrecog.c (validate_pattern): For VEC_SELECT verify that
1280 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
1282 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
1284 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
1287 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1290 * expr.c (store_field): Don't optimize away bitsize == 0 store
1291 from CALL_EXPR with addressable return type.
1293 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1295 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
1296 * config/rs6000/rs6000.md (sel): Delete mode attribute.
1297 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
1298 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
1299 TARGET_ISEL instead of TARGET_ISEL<sel>.
1301 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
1303 * config/rs6000/rs6000.c
1304 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
1306 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1308 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
1310 (reg_or_cint_operand): Fix comment.
1311 (reg_or_zero_operand): New predicate.
1312 * config/rs6000/rs6000-protos.h (output_isel): Delete.
1313 * config/rs6000/rs6000.c (output_isel): Delete.
1314 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
1315 instead of reg_or_cint_operand. Output instruction directly (not via
1317 (isel_unsigned_<mode>): Ditto.
1318 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
1319 gpc_reg_operand. Add an instruction alternative for this. Output
1320 instruction directly.
1321 (*isel_reversed_unsigned_<mode>): Ditto.
1323 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1325 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
1326 (TARGET_CANONICALIZE_COMPARISON): Define.
1328 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
1331 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1332 Check whether the dest is REG before adding REG_EQUIV note.
1334 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1337 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1339 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1340 (make_hard_regno_born, make_hard_regno_dead): Update
1341 bb_killed_pseudos and bb_gen_pseudos.
1343 2017-10-11 Nathan Sidwell <nathan@acm.org>
1345 * incpath.h (enum incpath_kind): Name enum, prefix values.
1346 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
1347 * incpath.c (heads, tails): Use INC_MAX.
1348 (add_env_var_paths, add_standard_paths): Use incpath_kind.
1349 (merge_include_chains, split_quote_chain,
1350 register_include_chains): Update incpath_kind names.
1351 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
1352 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
1354 (add_framework_path, darwin_register_objc_includes): Likewise.
1355 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
1357 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1359 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1360 Do not use float_operator operator predicate.
1361 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1362 * config/i386/predicates.md (float_operator): Remove predicate.
1364 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1366 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
1367 (*jccxf_i387): Ditto.
1368 (*jcc<mode>_i387): Ditto.
1369 (*jccu<mode>_i387): Ditto.
1370 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
1371 (*jcc_*_i387 splitters): Remove.
1372 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
1373 * config/i386/i386.c (ix86_split_fp_branch): Remove.
1374 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
1377 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1379 * profile-count.h (slow_safe_scale_64bit): New function.
1380 (safe_scale_64bit): New inline.
1381 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
1382 * profile-count.c: Include wide-int.h
1383 (slow_safe_scale_64bit): New.
1385 2017-10-11 Nathan Sidwell <nathan@acm.org>
1387 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
1388 HAS_DECL_ASSEMBLER_NAME_P.
1389 * gimple-expr.c (gimple_decl_printable_name: Check
1390 HAS_DECL_ASSEMBLER_NAME_P too.
1391 * ipa-utils.h (type_in_anonymous_namespace_p): Check
1392 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
1393 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
1394 * passes.c (rest_of_decl_compilation): Check
1395 HAS_DECL_ASSEMBLER_NAME_P too.
1396 * recog.c (verify_changes): Likewise.
1397 * tree-pretty-print.c (dump_decl_name): Likewise.
1398 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
1400 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
1401 (SET_DECL_ASSEMBLER_NAME): Use it.
1402 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
1403 (COPY_DECL_ASSEMBLER_NAME): Likewise.
1404 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
1406 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1408 * config.gcc (i386, x86_64): Add extra objects.
1409 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
1410 (ix86_min_insn_size): Declare.
1411 (ix86_issue_rate): Declare.
1412 (ix86_adjust_cost): Declare.
1413 (ia32_multipass_dfa_lookahead): Declare.
1414 (ix86_macro_fusion_p): Declare.
1415 (ix86_macro_fusion_pair_p): Declare.
1416 (ix86_bd_has_dispatch): Declare.
1417 (ix86_bd_do_dispatch): Declare.
1418 (ix86_core2i7_init_hooks): Declare.
1419 (ix86_atom_sched_reorder): Declare.
1420 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
1421 (COSTS_N_BYTES): Move to x86-tune-costs.h.
1422 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
1423 (rip_relative_addr_p): Rename to ...
1424 (ix86_rip_relative_addr_p): ... this one; export.
1425 (memory_address_length): Update.
1426 (ix86_issue_rate): Move to x86-tune-sched.c.
1427 (ix86_flags_dependent): Move to x86-tune-sched.c.
1428 (ix86_agi_dependent): Move to x86-tune-sched.c.
1429 (exact_dependency_1): Move to x86-tune-sched.c.
1430 (exact_store_load_dependency): Move to x86-tune-sched.c.
1431 (ix86_adjust_cost): Move to x86-tune-sched.c.
1432 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
1433 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
1434 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
1435 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
1436 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
1437 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
1438 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
1439 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
1440 (min_insn_size): Rename to ...
1441 (ix86_min_insn_size): ... this one; export.
1442 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
1443 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
1444 (core2i7_first_cycle_multipass_backtrack): Move to
1445 x86-tune-sched-core.c.
1446 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
1447 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
1448 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
1449 (ix86_avoid_jump_mispredicts): Update.
1450 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
1451 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
1452 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
1453 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
1454 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
1455 (MAX_INSN): Move to ix86-tune-sched-bd.c.
1456 (MAX_IMM): Move to ix86-tune-sched-bd.c.
1457 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
1458 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
1459 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
1460 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
1461 (MAX_STORE): Move to ix86-tune-sched-bd.c.
1462 (BIG): Move to ix86-tune-sched-bd.c.
1463 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
1464 (enum insn_path): Move to ix86-tune-sched-bd.c.
1465 (get_mem_group): Move to ix86-tune-sched-bd.c.
1466 (is_cmp): Move to ix86-tune-sched-bd.c.
1467 (dispatch_violation): Move to ix86-tune-sched-bd.c.
1468 (is_branch): Move to ix86-tune-sched-bd.c.
1469 (is_prefetch): Move to ix86-tune-sched-bd.c.
1470 (init_window): Move to ix86-tune-sched-bd.c.
1471 (allocate_window): Move to ix86-tune-sched-bd.c.
1472 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
1473 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
1474 (process_end_window): Move to ix86-tune-sched-bd.c.
1475 (allocate_next_window): Move to ix86-tune-sched-bd.c.
1476 (find_constant): Move to ix86-tune-sched-bd.c.
1477 (get_num_immediates): Move to ix86-tune-sched-bd.c.
1478 (has_immediate): Move to ix86-tune-sched-bd.c.
1479 (get_insn_path): Move to ix86-tune-sched-bd.c.
1480 (get_insn_group): Move to ix86-tune-sched-bd.c.
1481 (count_num_restricted): Move to ix86-tune-sched-bd.c.
1482 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
1483 (add_insn_window): Move to ix86-tune-sched-bd.c.
1484 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
1485 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
1486 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
1487 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
1488 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
1489 (do_dispatch): Move to ix86-tune-sched-bd.c.
1490 (has_dispatch): Move to ix86-tune-sched-bd.c.
1491 * config/i386/t-i386: Add new object files.
1492 * config/i386/x86-tune-costs.h: New file.
1493 * config/i386/x86-tune-sched-atom.c: New file.
1494 * config/i386/x86-tune-sched-bd.c: New file.
1495 * config/i386/x86-tune-sched-core.c: New file.
1496 * config/i386/x86-tune-sched.c: New file.
1498 2017-10-11 Liu Hao <lh_mouse@126.com>
1500 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
1501 the generic version below instead.
1502 (should_colorize): Recognize Windows consoles as terminals
1504 * pretty-print.c [__MINGW32__] (write_all): New function.
1505 [__MINGW32__] (find_esc_head): Likewise.
1506 [__MINGW32__] (find_esc_terminator): Likewise.
1507 [__MINGW32__] (eat_esc_sequence): Likewise.
1508 [__MINGW32__] (mingw_ansi_fputs): New function that handles
1510 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
1513 2017-10-11 Richard Biener <rguenther@suse.de>
1515 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
1516 Properly call analyze_scalar_evolution with the loop of the stmt.
1518 2017-10-11 Richard Biener <rguenther@suse.de>
1520 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
1521 * tree-core.h (tree_base): Add chrec_var union member.
1522 * tree.h (CHREC_VAR): Remove.
1523 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
1524 * tree-chrec.h (build_polynomial_chrec): Adjust.
1525 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
1526 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
1528 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
1530 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
1531 * match.pd: ... here.
1532 ((T) X == (T) Y): Relax condition.
1534 2017-10-11 Bin Cheng <bin.cheng@arm.com>
1536 PR tree-optimization/82472
1537 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
1539 (break_alias_scc_partitions): Update postorder number.
1541 2017-10-11 Martin Liska <mliska@suse.cz>
1544 * opts.c (parse_no_sanitize_attribute): Do not use error_value
1546 * opts.h (parse_no_sanitize_attribute): Remove last argument.
1548 2017-10-11 Martin Liska <mliska@suse.cz>
1550 * print-rtl.c (print_insn): Move declaration of idbuf
1551 to same scope as name.
1553 2017-10-11 Martin Liska <mliska@suse.cz>
1558 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1561 2017-10-11 Martin Liska <mliska@suse.cz>
1564 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1567 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
1569 * config.gcc (default_gnu_indirect_function): Default to yes for
1570 arm*-*-linux* with glibc.
1572 2017-10-11 Richard Biener <rguenther@suse.de>
1574 * tree-scalar-evolution.c (get_scalar_evolution): Handle
1575 default-defs and types we do not want to analyze.
1576 (interpret_loop_phi): Replace unreachable code with an assert.
1577 (compute_scalar_evolution_in_loop): Remove and inline ...
1578 (analyze_scalar_evolution_1): ... here, replacing condition with
1579 what makes the intent clearer. Remove handling of cases
1580 get_scalar_evolution now handles.
1582 2017-10-10 Jim Wilson <wilson@tuliptree.org>
1584 PR rtl-optimization/81434
1585 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
1586 comment for main loop. In sched_group_found if, also add checks for
1587 pass and min_cost_group.
1589 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
1591 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
1592 (rs6000_insn_cost): New function.
1593 * config/rs6000/rs6000.md (cost): New attribute.
1595 2017-10-10 Jakub Jelinek <jakub@redhat.com>
1596 H.J. Lu <hongjiu.lu@intel.com>
1600 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
1601 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
1602 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
1603 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
1604 to OPTION_MASK_ISA_AVX512VL - builtins that have both
1605 OPTION_MASK_ISA_MMX and some other bit set require both
1606 mmx and the ISAs without the mmx bit.
1607 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
1608 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
1609 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
1610 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
1611 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
1612 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
1613 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
1614 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
1615 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
1616 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
1617 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
1618 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
1619 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
1620 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
1621 Add OPTION_MASK_ISA_MMX.
1623 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
1625 * config.gcc (armv7*-*-freebsd*): New target.
1626 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
1628 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
1630 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
1631 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
1634 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
1636 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
1637 a template parameter.
1638 (WIDE_INT_REF_FOR): Update accordingly.
1639 * tree.h (wi::int_traits <const_tree>): Delete.
1640 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
1641 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
1642 (wi::tree_to_wide_ref): New typedef.
1643 (wi::to_wide): New function.
1644 * calls.c (get_size_range): Use wi::to_wide when operating on
1646 * cgraph.c (cgraph_node::create_thunk): Likewise.
1647 * config/i386/i386.c (ix86_data_alignment): Likewise.
1648 (ix86_local_alignment): Likewise.
1649 * dbxout.c (stabstr_O): Likewise.
1650 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
1651 * expr.c (const_vector_from_tree): Likewise.
1652 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
1653 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
1654 (fold_negate_expr_1, int_const_binop_1, const_binop)
1655 (fold_convert_const_int_from_real, optimize_bit_field_compare)
1656 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
1657 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
1658 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
1659 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
1660 (fold_not_const, round_up_loc): Likewise.
1661 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
1662 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
1663 (alloca_call_type): Likewise.
1664 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
1665 * godump.c (go_output_typedef): Likewise.
1666 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1667 * internal-fn.c (get_min_precision): Likewise.
1668 * ipa-cp.c (ipcp_store_vr_results): Likewise.
1669 * ipa-polymorphic-call.c
1670 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1671 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
1672 (ipa_modify_call_arguments): Likewise.
1673 * match.pd: Likewise.
1674 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
1675 * print-tree.c (print_node_brief, print_node): Likewise.
1676 * stmt.c (expand_case): Likewise.
1677 * stor-layout.c (layout_type): Likewise.
1678 * tree-affine.c (tree_to_aff_combination): Likewise.
1679 * tree-cfg.c (group_case_labels_stmt): Likewise.
1680 * tree-data-ref.c (dr_analyze_indices): Likewise.
1681 (prune_runtime_alias_test_list): Likewise.
1682 * tree-dump.c (dequeue_and_dump): Likewise.
1683 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
1684 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
1685 * tree-pretty-print.c (dump_generic_node): Likewise.
1686 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
1687 (simple_iv_with_niters): Likewise.
1688 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1689 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
1690 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
1691 * tree-ssa-loop-niter.c (split_to_var_and_offset)
1692 (refine_value_range_using_guard, number_of_iterations_ne_max)
1693 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
1694 (get_cst_init_from_scev, record_nonwrapping_iv)
1695 (scev_var_range_cant_overflow): Likewise.
1696 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
1697 * tree-ssa-pre.c (compute_avail): Likewise.
1698 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1699 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
1700 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1701 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
1702 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
1703 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
1704 (dump_case_nodes, try_switch_expansion): Likewise.
1705 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
1706 (vect_do_peeling): Likewise.
1707 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
1708 * tree-vect-stmts.c (vectorizable_load): Likewise.
1709 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
1710 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
1711 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
1712 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
1713 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
1714 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
1715 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
1716 (evrp_dom_walker::before_dom_children): Likewise.
1717 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
1718 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
1719 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
1720 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
1721 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
1722 (get_range_pos_neg): Likewise.
1723 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
1724 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
1725 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
1726 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
1727 * config/avr/avr.c (avr_fold_builtin): Likewise.
1728 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
1729 * config/msp430/msp430.c (msp430_attr): Likewise.
1730 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
1731 * config/powerpcspe/powerpcspe-c.c
1732 (altivec_resolve_overloaded_builtin): Likewise.
1733 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
1734 (rs6000_expand_ternop_builtin): Likewise.
1735 * config/rs6000/rs6000-c.c
1736 (altivec_resolve_overloaded_builtin): Likewise.
1737 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
1738 (rs6000_expand_ternop_builtin): Likewise.
1739 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
1741 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1743 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
1744 when copying loop nest with only one inner loop.
1746 2017-10-10 Richard Biener <rguenther@suse.de>
1748 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
1749 blocks if SCEV is active.
1750 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
1752 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
1753 (scev_initialize): Assert we are not yet initialized.
1755 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1757 * tree-loop-distribution.c (generate_loops_for_partition): Remove
1758 inner loop's exit stmt by making it always exit the loop, otherwise
1759 we would generate an infinite empty loop.
1761 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1763 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
1764 renaming variables in new preheader if it's deleted.
1766 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1768 * tree-loop-distribution.c (struct partition): Remove unused field
1769 loops of the structure.
1770 (partition_alloc, partition_free): Ditto.
1771 (build_rdg_partition_for_vertex): Ditto.
1773 2017-10-09 Jeff Law <law@redhat.com>
1775 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
1776 return type to match prototype and documentation.
1778 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1780 * config/rs6000/rs6000.c (processor_costs): Move to ...
1781 * config/rs6000/rs6000.h: ... here.
1782 (rs6000_cost): Declare.
1784 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
1786 * except.c (setjmp_fn): New global variable.
1787 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
1788 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
1789 if DONT_USE_BUILTIN_SETJMP is defined.
1791 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1793 * target.def (insn_cost): New hook.
1794 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
1795 * doc/tm.texi: Regenerate.
1796 * rtlanal.c (insn_cost): Use the new hook.
1798 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1800 * combine.c (combine_validate_cost): Compute the new insn_cost,
1801 not just pattern_cost.
1802 (try_combine): Adjust comment.
1804 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1806 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
1808 * combine.c (uid_insn_cost): Adjust comment.
1809 (combine_validate_cost): Adjust comment. Use pattern_cost instead
1811 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
1812 * dse.c (find_shift_sequence): Ditto.
1813 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
1814 (bb_valid_for_noce_process_p): Use pattern_cost.
1815 * rtl.h (insn_rtx_cost): Delete.
1816 (pattern_cost): New prototype.
1817 (insn_cost): New prototype.
1818 * rtlanal.c (insn_rtx_cost): Rename to...
1819 (pattern_cost): ... this.
1822 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
1824 * config/i386/i386.md (*jcc_2): Remove insn pattern.
1825 (*jcc<mode>_0_r_i387): Ditto.
1826 (*jccxf_r_i387): Ditto.
1827 (*jcc<mode>_r_i387): Ditto.
1828 (*jccu<mode>_r_i387): Ditto.
1829 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
1830 (*jcc): Rename from *jcc_1.
1832 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1834 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
1835 deferred rescans after the lvx/stvx recombination pre-pass.
1837 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1839 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
1840 memory operation instruction support.
1841 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
1842 (rs6000-ibm-aix[789]*): Likewise.
1843 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
1844 Document new functions.
1846 2017-10-09 Richard Biener <rguenther@suse.de>
1848 PR tree-optimization/82397
1849 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
1850 equality only for semantically equal trees.
1852 2017-10-09 Richard Biener <rguenther@suse.de>
1854 PR tree-optimization/82449
1855 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
1856 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
1857 allow constant addresses.
1858 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
1861 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1863 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
1866 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1869 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
1872 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1875 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
1877 2017-10-09 Jakub Jelinek <jakub@redhat.com>
1880 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
1881 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
1883 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
1885 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
1886 (WI_BINARY_PREDICATE_RESULT): Likewise.
1887 (wi::binary_traits::operator_result): New type.
1888 (wi::binary_traits::predicate_result): Likewise.
1889 (generic_wide_int::operator~, unary generic_wide_int::operator-)
1890 (generic_wide_int::operator==, generic_wide_int::operator!=)
1891 (generic_wide_int::operator&, generic_wide_int::and_not)
1892 (generic_wide_int::operator|, generic_wide_int::or_not)
1893 (generic_wide_int::operator^, generic_wide_int::operator+
1894 (binary generic_wide_int::operator-, generic_wide_int::operator*):
1896 (operator~, unary operator-, operator==, operator!=, operator&)
1897 (operator|, operator^, operator+, binary operator-, operator*): New
1899 * expr.c (get_inner_reference): Use wi::bit_and_not.
1900 * fold-const.c (fold_binary_loc): Likewise.
1901 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
1902 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
1903 (bit_value_binop): Likewise.
1904 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
1905 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
1906 (extract_range_from_binary_expr_1): Likewise.
1907 (masked_increment): Likewise.
1908 (simplify_bit_ops_using_ranges): Likewise.
1910 2017-10-09 Martin Jambor <mjambor@suse.cz>
1913 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
1914 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
1915 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
1917 (hsa_fixup_mov_insn_type): New function.
1918 (hsa_op_with_type::get_in_type): Use it.
1919 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
1920 immediates in an assert.
1921 (hsa_op_with_type::extend_int_to_32bit): New method.
1922 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
1923 types. Convert to dest type if necessary.
1924 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
1925 (reg_for_gimple_ssa): Pass false as min32int to
1926 hsa_type_for_scalar_tree_type.
1927 (gen_hsa_addr): Fixup type when creating addresable temporary.
1928 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
1929 (gen_hsa_unary_operation): Extend operands and convert to dest type if
1930 necessary. Call hsa_fixup_mov_insn_type.
1931 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
1932 extend operands and convert to dest type if necessary.
1933 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
1934 to dest type if necessary.
1935 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
1936 if conversion nt necessary and size matches.
1937 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
1938 to dest type if necessary.
1939 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
1940 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
1942 (gen_hsa_clrsb): Likewise.
1943 (gen_hsa_ffs): Likewise.
1944 (gen_hsa_divmod): Extend operands and convert to dest type if
1946 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
1948 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1950 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
1951 Remove empty default arguments. Use a brace block as output
1953 (conditional return): Ditto.
1955 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
1956 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
1958 (group_ending_nop): Ditto.
1959 (doloop_end): Ditto.
1960 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
1961 (splitters for those): Ditto.
1963 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1965 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
1966 a conditional jump (and the compare for it) so that pc_rtx is the
1968 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
1969 for the deleted and renamed ctr<mode>_internal[234] patterns.
1970 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
1971 Delete second conditional return pattern.
1972 (ctr<mode>_internal2): Delete this second bdnz pattern.
1973 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
1974 (ctr<mode>_internal4): Delete this second bdz pattern.
1976 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1978 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
1979 (always_initialized_rtx_for_ssa_name_p): New predicate.
1980 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
1981 (finish_out_of_ssa): Free new field of SA.
1982 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
1983 * tree-ssa-coalesce.c: Include tree-ssa.h.
1984 (get_parm_default_def_partitions): Remove extern keyword.
1985 (get_undefined_value_partitions): New function.
1986 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
1987 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
1989 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
1991 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1993 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
1995 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1997 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
1998 for targets that preffer 128bit.
2000 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2002 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
2004 2017-10-08 Olivier Hainque <hainque@adacore.com>
2006 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
2007 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
2009 (thumb_set_return_address): Likewise.
2011 2017-10-08 Olivier Hainque <hainque@adacore.com>
2013 * common/config/arm/arm-common.c (arm_except_unwind_info):
2014 Handle DWARF2_UNWIND_INFO.
2016 2017-10-07 Michael Collison <michael.collison@arm.com>
2018 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
2021 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
2023 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
2024 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
2025 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
2026 defined, force the creation of a new block for a dispatch label.
2028 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
2030 * invoke.texi (Wsuggest-attribute=cold): Document.
2031 * common.opt (Wsuggest-attribute=cold): New
2032 * ipa-pure-const.c (warn_function_cold): New function.
2033 * predict.c (compute_function_frequency): Use it.
2034 * predict.h (warn_function_cold): Declare.
2036 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
2038 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
2041 2017-10-06 Martin Liska <mliska@suse.cz>
2043 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
2044 keyword for member functions.
2045 (struct sanopt_tree_couple): New struct.
2046 (struct sanopt_tree_couple_hash): New function.
2047 (struct sanopt_ctx): Add new hash_map.
2048 (has_dominating_ubsan_ptr_check): New function.
2049 (record_ubsan_ptr_check_stmt): Likewise.
2050 (maybe_optimize_ubsan_ptr_ifn): Likewise.
2051 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
2052 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
2054 2017-10-06 Sudakshina Das <sudi.das@arm.com>
2057 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
2058 aarch64_simd_valid_immediate on CONST_VECTORs.
2059 (aarch64_reg_or_bic_imm): Likewise.
2061 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
2063 PR rtl-optimization/82396
2064 * haifa-sched.c (ready_sort_real): Disable qsort checking.
2066 2017-10-06 Sebastian Pop <sebpop@gmail.com>
2068 * graphite-dependences.c (scop_get_reads): Move code to...
2069 (scop_get_must_writes): Move code to...
2070 (scop_get_may_writes): Move code to...
2071 (scop_get_reads_and_writes): ... here.
2072 (scop_get_dependences): Call scop_get_reads_and_writes.
2074 2017-10-06 Jakub Jelinek <jakub@redhat.com>
2076 PR tree-optimization/82434
2077 * fold-const.h (can_native_encode_type_p,
2078 can_native_encode_string_p): Remove.
2079 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
2080 don't encode anything, just return what would be otherwise returned.
2081 (native_encode_fixed, native_encode_complex, native_encode_vector):
2083 (native_encode_string): Likewise. Inline by hand
2084 can_native_encode_string_p.
2085 (can_native_encode_type_p): Remove.
2086 (can_native_encode_string_p): Remove.
2087 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
2088 STRING_CSTs using can_native_encode_string_p, test all
2089 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
2090 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
2091 argument from native_encode_expr.
2092 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
2093 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
2096 2017-10-06 Richard Biener <rguenther@suse.de>
2098 PR tree-optimization/82397
2099 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2100 operand_equal_p but rely on data_ref_compare_tree for detecting
2102 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2103 to match up with dr_group_sort_cmp.
2105 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2108 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2110 * config/s390/s390-builtin-types.def: Regenerate.
2112 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2115 * config/s390/s390-builtin-types.def: Regenerate.
2116 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2117 Change flag from B_VXE to B_VX.
2118 (s390_vec_min_dbl): Remove B_VXE flag.
2120 2017-10-06 Richard Biener <rguenther@suse.de>
2122 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2123 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2124 translate_isl_ast_to_gimple::is_valid_rename,
2125 translate_isl_ast_to_gimple::get_rename,
2126 translate_isl_ast_to_gimple::get_def_bb_for_const,
2127 translate_isl_ast_to_gimple::get_new_name,
2128 translate_isl_ast_to_gimple::collect_all_ssa_names,
2129 translate_isl_ast_to_gimple::copy_loop_phi_args,
2130 translate_isl_ast_to_gimple::collect_all_ssa_names,
2131 translate_isl_ast_to_gimple::copy_loop_phi_args,
2132 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2133 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2134 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2135 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2136 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2137 translate_isl_ast_to_gimple::copy_cond_phi_args,
2138 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2139 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2140 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2141 translate_isl_ast_to_gimple::rename_uses,
2142 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2143 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2144 (set_rename_for_each_def): Likewise.
2145 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2146 here. Handle rewriting SCEV analyzable uses here.
2147 (copy_bb_and_scalar_dependences): Generate code for PHI
2149 (graphite_regenerate_ast_isl): Adjust.
2150 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2151 (add_write, add_read): New functions.
2152 (build_cross_bb_scalars_def): Use it and simplify.
2153 (build_cross_bb_scalars_use): Likewise.
2154 (graphite_find_cross_bb_scalar_vars): Inline into...
2155 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2156 simulating out-of-SSA. Compute liveout and add dependencies.
2157 (build_scops): Force an empty entry block.
2158 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2160 (sese_build_liveouts): Declare.
2161 (sese_trivially_empty_bb_p): Likewise.
2162 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2163 compute liveout and debug_liveout.
2164 (sese_bad_liveouts_use): Remove.
2165 (sese_reset_debug_liveouts_bb): Likewise.
2166 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2167 (sese_build_liveouts): Build liveout and debug_liveout and store
2169 (new_sese_info): Adjust.
2170 (free_sese_info): Likewise.
2171 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2172 do not build liveout here.
2173 (move_sese_in_condition): Adjust region entry.
2174 (scev_analyzable_p): Match up with chrec_apply requirements.
2175 (sese_trivially_empty_bb_p): New.
2176 * tree-into-ssa.c (get_reaching_def): Properly support generating
2177 default-defs for incremental rewrite of anonymous names.
2179 2017-10-06 Richard Biener <rguenther@suse.de>
2181 * graphite-sese-to-poly.c (extract_affine): For casts increasing
2182 precision do not perform modulo reduction.
2184 2017-10-06 Richard Biener <rguenther@suse.de>
2186 PR tree-optimization/82436
2187 * tree-vect-slp.c (vect_supported_load_permutation_p): More
2188 conservatively choose the vectorization factor when checking
2189 whether we can perform the required load permutation.
2190 (vect_transform_slp_perm_load): Assert when we may not fail.
2192 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2194 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
2195 message for incompatible -msdata=* and -mcall-* options.
2197 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2199 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
2200 rate for post-reload scheduling.
2202 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2204 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
2206 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2208 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
2209 to improve monte carlo in scimark.
2211 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2213 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
2214 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
2215 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
2216 pentium4_cost, nocona_cost): Set reassociation width to 1.
2217 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
2218 width to 2 for fp operations and 1 otherwise.
2219 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
2221 (atom_cost): Set reassociation width to 2.
2222 (slm_cost, generic_cost): Set fp reassociation width
2223 to 2 and 1 otherwise.
2224 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
2225 (core_cost): Set fp reassociation width to 4 and vector to 2.
2226 (ix86_reassociation_width): Rewrite using cost table; special case
2227 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
2228 and TARGET_AVX128_OPTIMAL.
2229 * config/i386/i386.h (processor_costs): Add
2230 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
2231 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
2232 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
2233 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
2234 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
2235 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
2237 2017-10-05 Nathan Sidwell <nathan@acm.org>
2239 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
2241 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2243 * config/arm/arm.c (arm_test_fpu_data): New.
2244 (arm_run_selftests): Call arm_test_fpu_data.
2246 2017-10-04 Nathan Sidwell <nathan@acm.org>
2248 * toplev.c (toplev::main): Remove excess parens on pretty_printer
2250 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
2252 2017-10-04 Sudakshina Das <sudi.das@arm.com>
2254 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
2255 check type for aarch64_simd_valid_immediate.
2256 (aarch64_output_simd_mov_immediate): Update prototype.
2257 (aarch64_simd_valid_immediate): Update prototype.
2258 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
2259 support for ORR-immediate.
2260 (and<mode>3): modified pattern to add support for BIC-immediate.
2261 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
2262 now checks for valid immediate for BIC and ORR based on new enum
2264 (aarch64_output_simd_mov_immediate): Function now used to output
2265 BIC/ORR imm as well based on new enum argument.
2266 * config/aarch64/constraints.md (Do): New vector immediate constraint.
2268 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
2269 (aarch64_reg_or_bic_imm): Likewise.
2271 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2273 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
2274 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
2276 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2280 PR rtl-optimization/82396
2281 * haifa-sched.c (autopref_multipass_init): Simplify
2283 (autopref_rank_data): Simplify sort order.
2284 * sched-int.h (autopref_multipass_data_): Remove
2285 multi_mem_insn_p, min_offset and max_offset.
2287 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2289 * doc/sourcebuild.texi: Document vect_peeling_profitable.
2291 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2293 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
2296 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2298 * doc/sourcebuild.texi: Document vect_long_mult.
2300 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
2302 PR tree-optimization/82413
2303 * fold-const.c (build_range_check): Use widest_int when comparing
2304 the maximum ETYPE value with HIGH.
2306 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2308 PR rtl-optimization/82396
2309 * haifa-sched.c (autopref_multipass_init): Simplify
2311 (autopref_rank_data): Simplify sort order.
2312 * sched-int.h (autopref_multipass_data_): Remove
2313 multi_mem_insn_p, min_offset and max_offset.
2315 2017-10-04 Jakub Jelinek <jakub@redhat.com>
2317 PR tree-optimization/82381
2318 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
2319 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
2322 PR tree-optimization/82374
2323 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
2324 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
2325 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
2326 current_function_decl to the new decl.
2328 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
2330 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
2331 helper macro for IEEE float128 hardware built-in functions.
2332 (SQRTF128_ODD): Add built-in functions with the round-to-odd
2334 (TRUNCF128_ODD): Likewise.
2335 (ADDF128_ODD): Likewise.
2336 (SUBF128_ODD): Likewise.
2337 (MULF128_ODD): Likewise.
2338 (DIVF128_ODD): Likewise.
2339 (FMAF128_ODD): Likewise.
2340 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
2341 UNSPEC_TRUNC_ROUND_TO_ODD.
2342 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
2343 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
2344 floating point round to odd instructions.
2345 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
2346 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
2347 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
2348 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
2349 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
2350 (trunc<mode>sf2_hw): Change the truncate with round to odd
2351 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
2352 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
2353 to odd hardware instructions.
2354 (sub<mode>3_odd): Likewise.
2355 (mul<mode>3_odd): Likewise.
2356 (div<mode>3_odd): Likewise.
2357 (sqrt<mode>2_odd): Likewise.
2358 (fma<mode>4_odd): Likewise.
2359 (fms<mode>4_odd): Likewise.
2360 (nfma<mode>4_odd): Likewise.
2361 (nfms<mode>4_odd): Likewise.
2362 (trunc<mode>df2_odd): Change the truncate with round to odd
2363 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
2365 * doc/extend.texi (PowerPC built-in functions): Update documentation
2366 for existing IEEE float128-bit built-in functions. Add built-in
2367 functions that generate the IEEE 128-bit floating point round to
2370 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
2372 PR rtl-optimization/77729
2373 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
2374 to (X&(C1&~C2))|C2 transformations.
2376 2017-10-03 Martin Jambor <mjambor@suse.cz>
2378 PR tree-optimization/82363
2379 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
2380 mismatch, mark lacc written regardless of racc.
2382 2017-10-03 Jakub Jelinek <jakub@redhat.com>
2384 PR tree-optimization/82381
2385 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
2386 stmt_to_insert nor wheather SSA_NAMEs are default defs.
2387 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
2388 fallthrough into reassoc_stmt_dominates_stmt_p.
2391 * combine.c (combine_instructions): Don't combine in unreachable
2394 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2397 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
2398 function to not use the have_cpu variable. Do not set cpu_index,
2399 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
2401 (rs6000_valid_attribute_p): Remove duplicate initializations of
2402 old_optimize and func_optimize.
2403 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
2404 (rs6000_activate_target_options): Make global.
2405 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
2408 2017-10-02 Jakub Jelinek <jakub@redhat.com>
2410 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
2411 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
2412 Set *poffset to 0 and *psize and *pmax_size to -1 if
2413 *poffset + *psize overflows in HOST_WIDE_INT.
2415 PR tree-optimization/82387
2416 PR tree-optimization/82388
2417 PR tree-optimization/82389
2418 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
2419 instead of live_bytes non-NULL.
2421 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
2424 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
2427 2017-10-02 Richard Biener <rguenther@suse.de>
2429 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
2430 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
2431 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
2433 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2435 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
2436 requested precision matches the type's.
2437 * calls.c (alloc_max_size): Calculate the new candidate size as
2438 a widest_int and use wi::to_widest when comparing it with the
2439 current candidate size.
2440 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
2441 zero rather than integer_zero_node.
2442 * match.pd: Check for a no-op conversion before using wi::add
2443 rather than after. Use tree_to_uhwi when summing small shift
2444 counts into an unsigned int.
2446 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2447 Alan Hayward <alan.hayward@arm.com>
2448 David Sherwood <david.sherwood@arm.com>
2451 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
2452 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
2453 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
2454 POINTER_AND_FP_REGS.
2456 2017-10-02 Richard Biener <rguenther@suse.de>
2458 PR tree-optimization/82355
2459 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2460 a mapping for the enclosing loop but avoid generating one for
2462 (copy_bb_and_scalar_dependences): Remove premature codegen
2463 error on PHIs in blocks duplicated into multiple places.
2464 * graphite-scop-detection.c
2465 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2466 in the region use it as loop and nest to analyze the DR in.
2467 (try_generate_gimple_bb): Likewise.
2468 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2469 (add_loop_constraints): For blocks in a loop not in the region
2470 create a dimension with a single iteration.
2471 * sese.h (gbb_loop_at_index): Remove assert.
2473 2017-10-01 Kevin Buettner <kevinb@redhat.com>
2475 * omp-expand.c (adjust_context_scope): New function.
2476 (expand_parallel_call): Call adjust_context_scope.
2478 2017-10-01 Jeff Law <law@redhat.com>
2480 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
2481 dom_opt_dom_walker class with direct access to private members.
2482 Add comments. Call test_for_singularity.
2483 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
2484 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
2485 m_dummy_cond anymore.
2486 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
2488 (pass_dominator:execute): Build the dummy_cond here and pass it
2489 to the dom_opt_dom_walker ctor.
2490 (test_for_singularity): New function.
2492 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
2493 Maya Rashish <coypu@sdf.org>
2495 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
2496 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
2497 (alpha*-*-netbsd*) Use nbsd_tm_file.
2498 (arm*-*-netbsdelf*) Likewise.
2499 (i[34567]86-*-netbsdelf*) Likewise.
2500 (x86_64-*-netbsd*) Likewise.
2501 (mips*-*-netbsd*) Likewise.
2502 (powerpc-*-netbsd*) Likewise.
2503 (sh*-*-netbsd*) Likewise.
2504 (sparc-*-netbsdelf*) Likewise.
2505 (sparc64-*-netbsd*) Likewise.
2506 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
2508 (vax-*-netbsdelf*) Likewise.
2509 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
2510 (UINT_FAST8_TYPE) Likewise.
2511 (INT_FAST16_TYPE) Check CHAR_FAST16.
2512 (UINT_FAST16_TYPE) Likewise.
2514 2017-09-30 Jakub Jelinek <jakub@redhat.com>
2517 * config/i386/i386.md
2518 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
2519 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
2520 *divmodsi4_zext_2): New define_insn_and_split.
2521 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
2522 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
2523 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
2524 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
2525 New define_insn_and_split.
2526 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
2527 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
2528 operands[1] having DImode when mode is SImode.
2530 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
2531 always SImode for DIV and MOD in REG_EQUAL notes.
2533 2017-09-29 Yury Gribov <tetra2005@gmail.com>
2536 * match.pd: Fix handling of NaNs in pattern.
2538 2017-09-29 Jeff Law <law@redhat.com>
2540 * sbitmap.c (bitmap_bit_in_range_p): New function.
2541 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
2542 * tree-ssa-dse.c (live_bytes_read): New function.
2543 (dse_classify_store): Ignore reads of dead bytes.
2545 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
2546 typos and whitespace errors.
2547 * config/i386/predicates.md (address_no_seg_operand): Likewise.
2548 * config/s390/s390.c (s390_emit_prologue): Likewise.
2550 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2553 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
2554 with a symbol for LRA.
2556 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2558 PR rtl-optimization/82338
2559 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
2561 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
2563 * genmodes.c (calc_wider_mode): Suppress qsort macro.
2564 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
2565 (qsort_chk): Declare.
2566 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
2567 (qsort_chk): New function.
2569 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2571 PR tree-optimization/82337
2572 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
2573 phi definition if the PHI result appears in an abnormal PHI.
2574 (find_basis_for_base_expr): Don't record a basis if the LHS of the
2575 basis appears in an abnormal PHI.
2577 2017-09-29 Richard Biener <rguenther@suse.de>
2579 * graphite-isl-ast-to-gimple.c
2580 (translate_isl_ast_to_gimple::set_codegen_error): New function.
2581 (binary_op_to_tree): Use it.
2582 (get_rename_from_scev): Likewise.
2583 (copy_loop_phi_nodes): Likewise.
2584 (copy_bb_and_scalar_dependences): Likewise.
2585 (translate_pending_phi_nodes): Likewise.
2587 2017-09-29 Jakub Jelinek <jakub@redhat.com>
2590 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
2591 for movabsq $(i32 << shift), r64.
2593 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
2595 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
2596 index when encoding %esp as %rsp to avoid 0x67 prefix.
2598 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
2600 * config/i386/i386.md (*movsf_internal, *movdf_internal):
2601 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
2603 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
2605 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
2606 Extensions with more than 16 double VFP registers.
2607 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
2608 to_clear_mask and all code related to it. Replace the remaining
2609 entry by a sbitmap and adapt code accordingly.
2611 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
2613 * brig-builtins.def: Change pure attributes to const.
2615 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2617 * config.gcc (default_gnu_indirect_function): Default to yes for
2618 sparc*-*-linux* with glibc.
2620 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2622 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
2623 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
2624 when creating .init_array and .fini_array sections with priority
2627 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
2630 * config/aarch64/aarch64.c
2631 (aarch64_builtin_support_vector_misalignment): Always return false
2632 when misalignment is unknown.
2634 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
2636 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
2637 this function to return false if the definition used by the swap
2638 instruction is artificial, or if the memory address from which the
2639 constant value is loaded is not represented by a base address held
2640 in a register or if the base address register is a frame or stack
2641 pointer. Additionally, return false if the base address of the
2642 loaded constant is a SYMBOL_REF but is not considered to be a
2644 (replace_swapped_load_constant): New function.
2645 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
2646 loaded constant vector with a load of a swapped constant vector.
2648 2017-09-27 Carl Love <cel@us.ibm.com>
2650 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
2651 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
2652 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
2655 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
2657 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
2658 first, always call autopref_rank_data otherwise.
2660 2017-09-27 Richard Biener <rguenther@suse.de>
2662 * graphite-scop-detection.c (find_scop_parameters): Move
2663 loop bound handling ...
2664 (gather_bbs::before_dom_children): ... here, avoiding the need
2665 to build scop_info->loop_nest.
2666 (record_loop_in_sese): Remove.
2667 * sese.h (sese_info_t::loop_nest): Remove.
2668 * sese.c (new_sese_info): Do not allocate loop_nest.
2669 (free_sese_info): Do not free loop_nest.
2671 2017-09-27 Jakub Jelinek <jakub@redhat.com>
2674 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
2675 lhs from calls if the lhs has addressable type.
2677 2017-09-27 Richard Biener <rguenther@suse.de>
2679 * graphite.h (scop::max_alias_set): New member.
2680 * graphite-scop-detection.c: Remove references to non-existing
2681 --param in comments.
2682 (build_alias_sets): Record the maximum alias set used for drs.
2683 (build_scops): Support zero as unlimited for
2684 --param graphite-max-arrays-per-scop.
2685 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
2687 (build_poly_sr_1): ... here. Compute alias set based on the
2688 maximum alias set used for drs rather than
2689 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
2691 2017-09-27 Richard Biener <rguenther@suse.de>
2693 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
2694 --param loop-block-tile-size=0 to disable tiling.
2696 2017-09-27 Richard Biener <rguenther@suse.de>
2698 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
2699 (graphite-max-nb-scop-params): Document special value zero.
2700 * domwalk.h (dom_walker::STOP): New symbolical constant.
2701 (dom_walker::dom_walker): Add optional parameter for bb to
2703 (dom_walker::~dom_walker): Declare.
2704 (dom_walker::before_dom_children): Document STOP return value.
2705 (dom_walker::m_user_bb_to_rpo): New member.
2706 (dom_walker::m_bb_to_rpo): Likewise.
2707 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
2708 mapping here if not provided by the user.
2709 (dom_walker::~dom_walker): Free bb to RPO mapping if not
2710 provided by the user.
2711 (dom_walker::STOP): Define.
2712 (dom_walker::walk): Do not compute bb to RPO mapping here.
2713 Support STOP return value from before_dom_children to stop
2715 * graphite-optimize-isl.c (optimize_isl): If the schedule
2716 is the same still generate code if -fgraphite-identity
2717 or -floop-parallelize-all are given.
2718 * graphite-scop-detection.c: Include cfganal.h.
2719 (gather_bbs::gather_bbs): Get and pass through bb to RPO
2721 (gather_bbs::before_dom_children): Return STOP for BBs
2723 (build_scops): Compute bb to RPO mapping and pass it to
2724 the domwalk. Treat --param graphite-max-nb-scop-params=0
2725 as not limiting the number of params.
2726 * graphite.c (graphite_initialize): Remove limit on the
2727 number of basic-blocks in a function.
2728 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
2729 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
2730 default value of 10.
2732 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2734 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
2735 Adjust code to eliminate needing to do the shift right 32-bits
2736 operation after XSCVDPSPN.
2738 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2740 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
2741 ((X / Y) != 0 -> X >= Y): Likewise.
2743 2017-09-26 Carl Love <cel@us.ibm.com>
2745 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
2746 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
2747 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
2748 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
2749 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
2750 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
2751 definitions and overloading.
2752 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
2753 statement for P9V_BUILTIN_XST_LEN_R.
2754 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
2755 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
2756 define_expand and define_insn for the instructions and builtins.
2757 * doc/extend.texi: Update the built-in documentation file for the new
2759 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
2760 define_insn for the instructions
2762 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
2765 * gcc/config/netbsd-protos.h: New file.
2766 * gcc/config/netbsd.c: New file.
2767 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
2768 * gcc/config/t-netbsd: New file.
2769 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
2770 (tmake_file) Add t-netbsd.
2771 (extra_objs) Add netbsd.o.
2773 2017-09-26 Janus Weil <janus@gcc.gnu.org>
2777 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
2779 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2781 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
2782 sign extension from a vector register to a GPR by doing a 32-bit
2783 direct move and then an EXTSW.
2784 (extendsi<mode>2 splitter): Likewise.
2785 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
2786 right or vector extract after doing XSCVDPSPN. Use
2787 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
2789 (movdi_from_sf_zero_ext): Likewise.
2790 (reload_gpr_from_vsxsf): Likewise.
2791 (p8_mfvsrd_4_disf): Delete, no longer used.
2792 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
2793 then needing to move the SFmode to a GPR to use the XSCVDPSP
2794 instruction instead of FRSP and XSCVDPSPN.
2795 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
2796 it is adjacent to the other XSCVSPDP insns.
2797 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
2798 SFmode to be in traditional Altivec registers.
2799 (vsx_xscvdpspn): Eliminate useless alternative constraint.
2800 (vsx_xscvspdpn): Likewise.
2801 (vsx_xscvspdpn_scalar): Likewise.
2803 2017-09-26 Martin Jambor <mjambor@suse.cz>
2805 * tree-sra.c (compare_access_positions): Put integral types first,
2806 stabilize sorting of integral types, remove conditions putting
2807 non-full-precision integers last.
2808 (sort_and_splice_var_accesses): Disable scalarization if a
2809 non-integert would be represented by a non-full-precision integer.
2811 2017-09-26 Joseph Myers <joseph@codesourcery.com>
2813 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
2814 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
2815 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
2816 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
2817 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
2818 conditionals inside the function instead of around it. Call
2819 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
2820 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
2822 2017-09-26 Richard Biener <rguenther@suse.de>
2824 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
2826 (scop_detection::build_scop_breadth): ... this. Removed.
2827 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
2828 (scop_detection::harmful_stmt_in_bb): Likewise.
2829 (scop_detection::graphite_can_represent_stmt): Likewise.
2830 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
2831 (scop_detection::can_represent_loop): Remove recursion, fold in ...
2832 (scop_detection::can_represent_loop_1): ... this. Removed.
2833 (scop_detection::harmful_loop_in_region): Simplify after inlining
2834 the above and remove more quadraticness.
2835 (build_scops): Adjust.
2836 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
2839 2017-09-26 Jakub Jelinek <jakub@redhat.com>
2842 * config/i386/i386.c (ix86_print_operand_address_as): Only test
2843 REGNO (base) == SP_REG if base is a REG.
2846 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
2847 if it is different SSA_NAME.
2848 (optimize_range_tests_cmp_bitwise): New function.
2849 (optimize_range_tests): Call it.
2851 2017-09-26 Richard Biener <rguenther@suse.de>
2853 PR tree-optimization/82321
2854 * graphite.c (canonicalize_loop_closed_ssa): Properly check
2855 for the def being inside the loop.
2857 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2859 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
2861 * config/s390/s390-builtins.def: Fix constraint on op4.
2863 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2865 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
2866 independent expanders.
2867 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
2868 ("vec_ordered", "vec_unordered"): New expanders.
2870 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2872 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
2875 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2877 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
2878 vec_unpacks_lo_v16qi.
2879 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
2881 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2883 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
2884 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
2885 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
2887 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2889 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
2891 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
2893 ("*vec_slb<mode>"): New insn pattern.
2894 ("vec_shr_<mode>"): New expander.
2895 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
2896 and force the shift count operand to V16QImode.
2897 ("vec_srb<mode>"): Set shift count mode to V16QI.
2899 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2901 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
2902 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
2903 ("vec_widen_smult_hi_<mode>"): New expander definitions.
2905 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
2908 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
2910 2017-09-26 Richard Biener <rguenther@suse.de>
2912 PR tree-optimization/82320
2913 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
2916 2017-09-25 Jeff Law <law@redhat.com>
2918 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
2919 prototype for new argument.
2920 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
2921 mostly extracted from rs6000_emit_allocate_stack.
2922 (rs6000_emit_probe_stack_range_stack_clash): New function.
2923 (rs6000_emit_allocate_stack): Call
2924 rs6000_emit_probe_stack_range_stack_clash as needed.
2925 (rs6000_emit_probe_stack_range): Add additional argument
2926 to call to gen_probe_stack_range{si,di}.
2927 (output_probe_stack_range): New.
2928 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
2929 (output_probe_stack_range_stack_clash): New.
2930 (rs6000_emit_prologue): Emit notes into dump file as requested.
2931 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
2932 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
2933 Add additional operand and pass it to output_probe_stack_range.
2935 2017-09-25 Bin Cheng <bin.cheng@arm.com>
2937 PR tree-optimization/82163
2938 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
2939 (checking_verify_loop_closed_ssa): New parameter.
2940 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
2941 (check_loop_closed_ssa_stmt): Delete.
2942 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
2943 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
2944 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
2947 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
2949 * brig-builtins.def: Treat HSAIL barrier builtins as
2950 setjmp/longjump style functions.
2952 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2954 * target.def (constant_alignment): New hook.
2955 * defaults.h (CONSTANT_ALIGNMENT): Delete.
2956 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
2957 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
2958 * doc/tm.texi: Regenerate.
2959 * targhooks.h (default_constant_alignment): Declare.
2960 (constant_alignment_word_strings): Likewise.
2961 * targhooks.c (default_constant_alignment): New function.
2962 (constant_alignment_word_strings): Likewise.
2963 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
2964 instead of CONSTANT_ALIGNMENT.
2965 * varasm.c (align_variable, get_variable_align, build_constant_desc)
2966 (force_const_mem): Likewise.
2967 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
2968 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
2969 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
2970 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2971 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
2973 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
2974 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2975 constant_alignment_word_strings.
2976 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
2977 (CONSTANT_ALIGNMENT): Likewise.
2978 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2979 (arm_constant_alignment): New function.
2980 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
2981 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2982 constant_alignment_word_strings.
2983 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
2984 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2985 constant_alignment_word_strings.
2986 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
2987 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2988 (cris_constant_alignment): New function.
2989 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
2990 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2991 (epiphany_constant_alignment): New function.
2992 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
2993 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2994 constant_alignment_word_strings.
2995 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
2996 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2997 constant_alignment_word_strings.
2998 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
2999 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3000 constant_alignment_word_strings.
3001 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
3002 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
3003 * config/i386/i386.c (ix86_constant_alignment): Make static.
3004 Use the same interface as the target hook.
3005 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3006 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
3007 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3008 constant_alignment_word_strings.
3009 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
3010 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
3011 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3012 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
3013 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3014 constant_alignment_word_strings.
3015 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
3016 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3017 constant_alignment_word_strings.
3018 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
3019 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3020 constant_alignment_word_strings.
3021 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
3022 * config/microblaze/microblaze.c (microblaze_constant_alignment):
3024 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3025 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
3026 * config/mips/mips.c (mips_constant_alignment): New function.
3027 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3028 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
3029 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
3030 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3031 (mmix_constant_alignment): Make static. Use the same interface
3033 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
3034 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3035 constant_alignment_word_strings.
3036 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
3037 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3038 constant_alignment_word_strings.
3039 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
3040 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3041 constant_alignment_word_strings.
3042 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
3043 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3044 (rs6000_constant_alignment): New function.
3045 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
3046 * config/riscv/riscv.c (riscv_constant_alignment): New function.
3047 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3048 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
3049 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3050 (rs6000_constant_alignment): New function.
3051 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
3052 * config/s390/s390.c (s390_constant_alignment): New function.
3053 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3054 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
3055 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3056 constant_alignment_word_strings.
3057 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
3058 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3059 (sparc_constant_alignment): New function.
3060 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
3061 * config/spu/spu.c (spu_constant_alignment): New function.
3062 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3063 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
3064 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3065 constant_alignment_word_strings.
3066 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
3067 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3068 constant_alignment_word_strings.
3069 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
3070 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3071 constant_alignment_word_strings.
3072 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
3073 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3074 (visium_constant_alignment): New function.
3075 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
3076 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3077 (xtensa_constant_alignment): New function.
3078 * system.h (CONSTANT_ALIGNMENT): Poison.
3080 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
3082 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3083 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
3084 (rs6000_builtin_valid_without_lhs): New helper function.
3085 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3086 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
3088 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3090 * target.h (vec_perm_indices): Use unsigned short rather than
3092 (auto_vec_perm_indices): Likewise.
3093 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
3094 Use unsigned int rather than unsigned char.
3095 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3097 2017-09-25 Richard Biener <rguenther@suse.de>
3099 * cfgloop.h (sort_sibling_loops): Declare.
3100 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3101 (sort_sibling_loops): New function sorting the sibling loop list
3103 * graphite.c (graphite_transform_loops): Sort sibling loops.
3105 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3107 * target.def (vec_perm_const_ok): Change sel parameter to
3109 * optabs-query.c (can_vec_perm_p): Update accordingly.
3110 * doc/tm.texi: Regenerate.
3111 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3112 auto_vec_perm_indices and remove separate nelt field.
3113 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3114 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3115 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3116 (aarch64_expand_vec_perm_const): Update accordingly.
3117 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3118 to vec_perm_indices.
3119 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3120 auto_vec_perm_indices and remove separate nelt field.
3121 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3122 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3123 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3125 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3126 to vec_perm_indices.
3127 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3128 sel to vec_perm_indices.
3129 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3130 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3131 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3133 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3136 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3139 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3140 on the FUNCTION_DECL function context if it has a DIE that is a
3143 2017-09-25 Richard Biener <rguenther@suse.de>
3145 PR tree-optimization/82285
3146 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3149 2017-09-25 Tom de Vries <tom@codesourcery.com>
3153 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3156 2017-09-25 Richard Biener <rguenther@suse.de>
3158 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3159 ISL errors other than isl_error_quota happen. Dump if the
3160 schedule is the same.
3161 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3162 errors instead of aborting inside ISL.
3164 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3167 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3168 of libgcc_eh for m64.
3169 * config/i386/darwin64.h: Likewise.
3171 2017-09-25 Richard Biener <rguenther@suse.de>
3174 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
3175 attribute for incomplete types nor twice for complete ones.
3177 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
3180 * config/i386/i386.c (ix86_print_operand_address_as): Encode
3181 %esp as %rsp to avoid 0x67 prefix if there is no index or base
3184 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
3187 * config/i386/i386.opt (mprefer-avx256): Use
3188 ix86_target_flags variable.
3189 * config/i386/i386.c (ix86_target_string): Move
3190 -mprefer-avx256 to flag2_opts.
3192 2017-09-22 Jakub Jelinek <jakub@redhat.com>
3195 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
3196 and x != -1 | y != -1 into (x & y) != -1.
3198 2017-09-22 Steve Ellcey <sellcey@cavium.com>
3200 * config.gcc: Add new case statement to set
3201 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
3202 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
3203 s390x-*-linux* case statements. Added aarch64 to the list of
3204 supported architectures.
3206 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3208 PR tree-optimization/82289
3209 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
3210 STMT_VINFO_RELEVANT_P.
3212 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3213 Alan Hayward <alan.hayward@arm.com>
3214 David Sherwood <david.sherwood@arm.com>
3216 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
3217 for VR_RANGE only; don't allow VR_ANTI_RANGE.
3218 (extract_range_from_binary_expr_1): Don't call
3219 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
3221 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3222 Alan Hayward <alan.hayward@arm.com>
3223 David Sherwood <david.sherwood@arm.com>
3225 * target.def (preferred_vector_alignment): New hook.
3226 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
3228 * doc/tm.texi: Regenerate.
3229 * targhooks.h (default_preferred_vector_alignment): Declare.
3230 * targhooks.c (default_preferred_vector_alignment): New function.
3231 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
3233 (DR_TARGET_ALIGNMENT): New macro.
3234 (aligned_access_p): Update commentary.
3235 (vect_known_alignment_in_bytes): New function.
3236 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
3238 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
3239 Calculate the misalignment based on the target alignment rather than
3241 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
3242 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
3243 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
3244 the target alignment, rather than masking the element misalignment
3245 with the number of elements in a vector. Also use the target
3246 alignment when calculating the maximum number of peels.
3247 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
3248 instead of TYPE_ALIGN_UNIT.
3249 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
3250 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
3251 (vect_create_addr_base_for_vector_ref): Update call accordingly.
3252 (vect_create_data_ref_ptr): Likewise.
3253 (vect_setup_realignment): Realign by ANDing with
3254 -DR_TARGET_MISALIGNMENT.
3255 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
3256 the number of peels based on DR_TARGET_ALIGNMENT.
3257 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
3258 with the guaranteed alignment boundary when deciding whether
3260 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
3261 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
3262 (ensure_base_align): Remove stmt_info parameter. Get the
3263 target base alignment from DR_TARGET_ALIGNMENT.
3264 (vectorizable_store): Update call accordingly. Interpret
3265 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
3267 (vectorizable_load): Likewise.
3269 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3270 Alan Hayward <alan.hayward@arm.com>
3271 David Sherwood <david.sherwood@arm.com>
3273 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
3274 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
3275 (vect_enhance_data_refs_alignment): Likewise.
3277 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3279 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3280 error. Only quit immediately if parsing is complete.
3281 (BEGIN): Initialize fatal_err and parse_done.
3282 (begin fpu, end fpu): Check number of arguments.
3283 (begin arch, end arch): Likewise.
3284 (begin cpu, end cpu): Likewise.
3285 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3286 (optalias): Likewise.
3288 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3290 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3291 * config/arm/arm-isa.h: Delete. Move definitions to ...
3292 * arm-cpus.in: ... here. Use new feature and fgroup values.
3293 * config/arm/arm.c (arm_option_override): Use lower case for feature
3295 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3296 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3297 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3299 (print_isa_bits_for): New function.
3300 (gen_isa): New function.
3301 (gen_comm_data): Use print_isa_bits_for.
3302 (define feature): New keyword.
3303 (define fgroup): New keyword.
3304 * config/arm/t-arm (TM_H): Remove.
3305 (GTM_H): Add arm-isa.h.
3306 (arm-isa.h): Add rule to generate file.
3307 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3308 case for feature bit names.
3310 2017-09-22 Richard Biener <rguenther@suse.de>
3312 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
3314 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
3315 print of no dependency loops ...
3316 * graphite.c (graphite_transform_loops): ... here.
3317 (canonicalize_loop_closed_ssa_form): Work from inner to outer
3319 (same_close_phi_node, remove_duplicate_close_phi,
3320 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
3321 (canonicalize_loop_closed_ssa): ... here and simplify.
3322 * graphite-optimize-isl.c: Include tree-vectorizer.h.
3323 (optimize_isl): Use dump_printf_loc to tell when we stopped
3324 optimizing because of an ISL timeout.
3326 2017-09-22 Richard Biener <rguenther@suse.de>
3328 PR tree-optimization/82291
3329 * tree-if-conv.c (predicate_mem_writes): Make sure to
3330 remove writes in blocks predicated with false.
3332 2017-09-22 Richard Biener <rguenther@suse.de>
3334 * sese.c: Include cfganal.h.
3335 (if_region_set_false_region): Remove.
3336 (create_if_region_on_edge): Likewise.
3337 (move_sese_in_condition): Re-implement without destroying
3340 2017-09-22 Richard Biener <rguenther@suse.de>
3342 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
3343 Verify both BBs contain loop PHI nodes before dispatching to
3345 (graphite_regenerate_ast_isl): Do not recompute dominators,
3346 do not verify three times. Restructure for clarity.
3347 * graphite-scop-detection.c (same_close_phi_node,
3348 remove_duplicate_close_phi, make_close_phi_nodes_unique,
3349 defined_in_loop_p, canonicalize_loop_closed_ssa,
3350 canonicalize_loop_closed_ssa_form): Simplify, remove excess
3351 checking and SSA rewrite, move to ...
3352 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
3353 (graphite_initialize): Do not pass in ctx, do not reset the
3354 SCEV cache, compute only dominators.
3355 (graphite_transform_loops): Allocate ISL ctx after
3356 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
3357 Maintain post-dominators only around build_scops.
3358 * sese.c (if_region_set_false_region): Make static. Free
3359 and recompute dominators.
3360 (move_sese_in_condition): Assert we don't get called with
3361 post-dominators computed.
3362 * sese.h (if_region_set_false_region): Remove.
3364 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
3366 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
3367 mode attribute for TARGET_AVX512VL.
3369 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
3371 * config/i386/i386.opt (mprefer-avx256): New option.
3372 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
3374 (ix86_preferred_simd_mode): Return 256-bit AVX modes
3375 for TARGET_PREFER_AVX256.
3376 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
3378 2017-09-21 Jeff Law <law@redhat.com>
3380 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
3381 Fix dump output if the only stack space is for pushed registers.
3383 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3385 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
3388 2017-09-21 Martin Sebor <msebor@redhat.com>
3391 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
3392 code (in C++) or code that triggers warnings.
3394 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
3396 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
3398 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
3400 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
3401 * sched-rgn.c: Ditto.
3402 * sel-sched-ir.c: Ditto.
3404 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
3406 * toplev.h (set_random_seed): Adjust return type.
3407 * toplev.c (init_local_tick): Move eager initialization of random_seed
3408 to get_random_seed. Adjust comment.
3409 (init_random_seed): Inline to get_random_seed, delete.
3410 (get_random_seed): Initialize random_seed lazily.
3411 (set_random_seed): Do not return previous value.
3412 (print_switch_value): Do not call get_random_seed.
3414 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
3416 * cgraph.c (delete_function_version): New, broken out from...
3417 (cgraph_node::delete_function_version): ...here. Rename to
3418 cgraph_node::delete_function_version_by_decl. Update all uses.
3419 (cgraph_node::remove): Call delete_function_version.
3421 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3424 * tree-inline.c (expand_call_inline): Emit clobber stmts for
3425 VAR_DECLs to which addressable non-volatile parameters are mapped
3426 and for id->retvar after the return value assignment. Clear
3427 id->retval and id->retbnd after inlining.
3429 2017-09-21 Richard Biener <rguenther@suse.de>
3431 PR tree-optimization/82276
3432 PR tree-optimization/82244
3433 * tree-vrp.c (build_assert_expr_for): Set
3434 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
3436 (remove_range_assertions): Revert earlier change.
3438 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
3441 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
3443 2017-09-21 Richard Biener <rguenther@suse.de>
3445 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3446 Restore valid IL after code generation errors.
3447 * graphite.c (graphite_transform_loops): Diagnose code
3448 generation issues as MSG_MISSED_OPTIMIZATION and continue
3449 with processing SCOPs.
3451 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3452 Alan Hayward <alan.hayward@arm.com>
3453 David Sherwood <david.sherwood@arm.com>
3455 * calls.c (compute_argument_addresses): Use simplify_gen_binary
3456 rather than choosing between plus_constant and gen_rtx_<CODE>.
3457 * expr.c (emit_push_insn): Likewise.
3458 (expand_expr_real_2): Likewise.
3460 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3461 Alan Hayward <alan.hayward@arm.com>
3462 David Sherwood <david.sherwood@arm.com>
3464 * loop-unroll.c (split_iv): Call copy_rtx on the step.
3466 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3467 Alan Hayward <alan.hayward@arm.com>
3468 David Sherwood <david.sherwood@arm.com>
3470 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
3471 calling tree_to_uhwi.
3473 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3474 Alan Hayward <alan.hayward@arm.com>
3475 David Sherwood <david.sherwood@arm.com>
3477 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
3478 INTEGER_CST rather than a negative test for ADDR_EXPR.
3480 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3481 Alan Hayward <alan.hayward@arm.com>
3482 David Sherwood <david.sherwood@arm.com>
3484 * tree-vrp.c (extract_range_from_binary_expr_1): Check
3485 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
3487 2017-09-21 Richard Biener <rguenther@suse.de>
3489 PR tree-optimization/71351
3490 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
3491 graphite_create_new_loop_guard): Remove, fold remaining parts
3493 (translate_isl_ast_node_for): ... here and simplify.
3495 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3498 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
3499 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
3500 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
3501 alternative always use QI mode, for -Os imov (=R,R) alternative
3502 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
3505 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3506 Jeff Law <law@redhat.com>
3508 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
3509 (allocate_stack_space): New function, partially extracted from
3511 (s390_emit_prologue): Track offset to most recent stack probe.
3512 Code to allocate space moved into allocate_stack_space.
3513 Dump actions when no stack is allocated.
3514 (s390_prologue_plus_offset): New function.
3515 (s390_emit_stack_probe): Likewise.
3517 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
3519 * common.opt (Wa, Wl, Wp, g, gz=): Add
3521 (gno-column-info): Remove.
3522 (gcolumn-info): Drop RejectNegative.
3524 (gno-record-gcc-switches): Remove.
3525 (grecord-gcc-switches): Drop RejectNegative.
3526 (gno-split-dwarf): Remove.
3527 (gsplit-dwarf): Drop RejectNegative.
3528 (gno-strict-dwarf): Remove.
3529 (gstrict-dwarf): Drop RejectNegative.
3530 * config/darwin.opt (gfull, gused): Add RejectNegative.
3531 * dwarf2out.c (gen_producer_string): Drop
3532 gno-record-gcc-switches handler.
3533 * optc-gen.awk: Add g to prefixes with negative forms.
3534 * opts-common.c (remapping_prefix_p): New.
3535 (find_opt): Check it.
3536 (generate_canonical_option): Test g prefix.
3537 (option_map): Add -gno- mapping.
3538 (add_misspelling_candidates): Check remapping_prefix_p.
3540 2017-09-20 Jeff Law <law@redhat.com>
3542 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
3543 thinko in stack clash protection support.
3545 * explow.c (compute_stack_clash_protection_loop_data): Use
3546 CONST_INT_P instead of explicit test. Verify object is a
3547 CONST_INT_P before looking at INTVAL.
3548 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
3549 instead of explicit test.
3551 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3554 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
3555 address instead of to r1 and r11.
3557 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
3559 * config.gcc: Support "knm".
3560 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
3561 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3563 * config/i386/i386.c (m_KNM): Define.
3564 (processor_target_table): Add "knm".
3566 (ix86_option_override_internal): Add "knm".
3567 (ix86_issue_rate): Add PROCESSOR_KNM.
3568 (ix86_adjust_cost): Ditto.
3569 (ia32_multipass_dfa_lookahead): Ditto.
3570 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
3571 (fold_builtin_cpu): Add M_INTEL_KNM.
3572 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
3573 (processor_type): Add PROCESSOR_KNM.
3574 * config/i386/x86-tune.def: Add m_KNM.
3575 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
3577 2017-09-20 Richard Biener <rguenther@suse.de>
3579 PR tree-optimization/80213
3580 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
3581 are allowed in empty BBs as well.
3582 (canonicalize_loop_closed_ssa): Also look for other complex
3584 (scop_detection::get_sese): Include the loop-closed PHI block
3586 (scop_detection::merge_sese): Remove code adding extra blocks.
3587 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
3588 (build_scops): Assert the final returned scop is invalid.
3590 2017-09-20 Richard Biener <rguenther@suse.de>
3592 PR tree-optimization/82264
3593 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
3595 (vn_phi_lookup): Likewise.
3596 (vn_phi_insert): Likewise.
3598 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3600 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
3601 that fits into uhwi or shwi, add DW_AT_const_value regardless
3602 of early_dwarf without going through RTL, using add_AT_unsigned
3605 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
3606 (DEBUG_LTO_ABBREV_SECTION): Likewise.
3607 (DEBUG_LTO_MACINFO_SECTION): Likewise.
3608 (DEBUG_MACRO_SECTION): Likewise.
3609 (DEBUG_LTO_MACRO_SECTION): Likewise.
3610 (DEBUG_STR_DWO_SECTION): Likewise.
3611 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
3612 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
3613 (DEBUG_LTO_DWO_LINE_SECTION): Define.
3614 (DEBUG_LTO_LINE_STR_SECTION): Define.
3615 (init_sections_and_labels): Initialize debug_line_str_section
3616 variable. Initialize debug_loc_section for -gdwarf-5 to
3617 DEBUG_LOCLISTS_SECTION. Formatting fixes.
3619 2017-09-20 Richard Biener <rguenther@suse.de>
3621 * graphite-sese-to-poly.c (extract_affine): Properly handle
3622 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
3624 2017-09-20 Richard Biener <rguenther@suse.de>
3626 PR tree-optimization/81373
3627 * graphite-scop-detection.c (build_cross_bb_scalars_def):
3628 Force SESE live-out defs to be handled even if they are
3631 2017-09-19 Jeff Law <law@redhat.com>
3633 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
3634 nothing for stack adjustments with REG_STACK_CHECK.
3635 * sched-deps.c (parse_add_or_inc): Reject insns with
3636 REG_STACK_CHECK from dependency breaking.
3637 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
3638 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
3639 * reg-notes.def (STACK_CHECK): New note.
3641 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
3642 (ix86_expand_prologue): Dump stack clash info as needed.
3643 Call ix86_adjust_stack_and_probe_stack_clash as needed.
3645 * function.c (dump_stack_clash_frame_info): New function.
3646 * function.h (dump_stack_clash_frame_info): Prototype.
3647 (enum stack_clash_probes): New enum.
3649 * config/alpha/alpha.c (alpha_expand_prologue): Also check
3650 flag_stack_clash_protection.
3651 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
3652 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
3653 (arm_frame_pointer_required): Likewise.
3654 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
3655 (ia64_expand_prologue): Likewise.
3656 * config/mips/mips.c (mips_expand_prologue): Likewise.
3657 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
3658 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3659 (sparc_flat_expand_prologue): Likewise.
3660 * config/spu/spu.c (spu_expand_prologue): Likewise.
3662 * explow.c: Include "params.h".
3663 (anti_adjust_stack_and_probe_stack_clash): New function.
3664 (get_stack_check_protect): Likewise.
3665 (compute_stack_clash_protection_loop_data): Likewise.
3666 (emit_stack_clash_protection_loop_start): Likewise.
3667 (emit_stack_clash_protection_loop_end): Likewise.
3668 (allocate_dynamic_stack_space): Use get_stack_check_protect.
3669 Use anti_adjust_stack_and_probe_stack_clash.
3670 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
3671 (emit_stack_clash_protection_loop_start): Likewise.
3672 (emit_stack_clash_protection_loop_end): Likewise.
3673 * rtl.h (get_stack_check_protect): Prototype.
3674 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
3675 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
3676 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
3678 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
3680 * doc/tm.texi: Rebuilt.
3681 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3682 get_stack_check_protect.
3683 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
3684 * config/arm/arm.c (arm_expand_prologue): Likewise.
3685 (arm_frame_pointer_required): Likewise.
3686 * config/i386/i386.c (ix86_expand_prologue): Likewise.
3687 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
3688 * config/mips/mips.c (mips_expand_prologue): Likewise.
3689 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
3690 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
3691 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3692 (sparc_flat_expand_prologue): Likewise.
3694 * common.opt (-fstack-clash-protection): New option.
3695 * flag-types.h (enum stack_check_type): Note difference between
3696 -fstack-check= and -fstack-clash-protection.
3697 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
3698 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
3699 * toplev.c (process_options): Issue warnings/errors for cases
3700 not handled with -fstack-clash-protection.
3701 * doc/invoke.texi (-fstack-clash-protection): Document new option.
3702 (-fstack-check): Note additional problem with -fstack-check=generic.
3703 Note that -fstack-check is primarily for Ada and refer users
3704 to -fstack-clash-protection for stack-clash-protection.
3705 Document new params for stack clash protection.
3707 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
3709 * config/i386/i386.c (ix86_split_long_move): Do not handle
3710 address used for LEA in a special way.
3712 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3714 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
3716 2017-09-19 Martin Sebor <msebor@redhat.com>
3719 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
3720 of incompatible types.
3722 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
3724 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3725 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
3726 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3727 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
3729 2017-09-19 Richard Biener <rguenther@suse.de>
3731 PR tree-optimization/82244
3732 * tree-vrp.c (remove_range_assertions): Do not propagate
3733 a constant to abnormals but replace the assert with a copy.
3735 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
3737 PR rtl-optimization/57878
3738 PR rtl-optimization/68988
3739 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
3740 avoidance test involving non_reload_pseudos. Move frequency test
3741 below the general fragmentation avoidance test.
3743 2017-09-19 Richard Biener <rguenther@suse.de>
3745 PR tree-optimization/69728
3746 * graphite-sese-to-poly.c (schedule_error): New global.
3747 (add_loop_schedule): Handle empty domain by failing the
3749 (build_original_schedule): Handle schedule_error.
3751 2017-09-19 Richard Biener <rguenther@suse.de>
3753 * graphite-scop-detection.c (scop_detection::can_represent_loop):
3754 Do not iterate to sibling loops but only to siblings of inner
3757 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
3760 * config/m68k/m68k.md (moveq feeding equality comparison): Check
3761 that the registers are different.
3763 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
3765 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
3766 to processor_model and "amdfam17h" to arch_names_table.
3767 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
3769 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3772 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
3774 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3775 Alan Hayward <alan.hayward@arm.com>
3776 David Sherwood <david.sherwood@arm.com>
3778 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
3780 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
3782 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
3783 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
3784 vect_schedule_slp_instance.
3785 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
3786 Update call to vect_slp_analyze_node_operations. Simplify return
3788 (vect_slp_analyze_bb_1): Update call accordingly.
3789 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
3790 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
3791 (vect_schedule_slp): Update call accordingly.
3793 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3794 Alan Hayward <alan.hayward@arm.com>
3795 David Sherwood <david.sherwood@arm.com>
3797 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
3798 with types that aren't in fact scalar.
3800 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3802 * tree-vect-slp.c (vect_record_max_nunits): New function,
3804 (vect_build_slp_tree_1): ...here.
3805 (vect_build_slp_tree_2): Call it for phis too.
3807 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3809 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
3810 to vect_get_vec_def_for_operand when getting the mask operand.
3812 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3813 Alan Hayward <alan.hayward@arm.com>
3814 David Sherwood <david.sherwood@arm.com>
3816 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
3819 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3820 Alan Hayward <alan.hayward@arm.com>
3821 David Sherwood <david.sherwood@arm.com>
3823 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
3824 calculation for vector booleans.
3826 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3827 Alan Hayward <alan.hayward@arm.com>
3828 David Sherwood <david.sherwood@arm.com>
3830 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
3832 (vect_transform_stmt): ...here.
3833 (vect_analyze_stmt): Use it instead of calling
3834 vectorizable_live_operation directly.
3836 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
3838 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
3839 non-SIMT targets in acc vector loops.
3841 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
3843 * configure.ac: Add arc and check if assembler supports gdwarf2.
3844 * configure: Regenerate.
3846 2017-09-18 Richard Biener <rguenther@suse.de>
3848 PR tree-optimization/82220
3849 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
3850 epilogue niters from the min_profitable_iters compute.
3852 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3855 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
3856 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
3857 (ix86_init_pic_reg): Revert 2017-09-01 changes.
3859 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
3862 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
3863 switching to a new text section.
3865 2017-09-18 Richard Biener <rguenther@suse.de>
3867 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
3869 (build_alias_set): Reject aliases with no access function.
3871 2017-09-18 Richard Biener <rguenther@suse.de>
3873 PR tree-optimization/79622
3874 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
3876 (build_cross_bb_scalars_use): Likewise.
3878 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
3880 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
3882 2017-09-18 Alan Modra <amodra@gmail.com>
3885 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
3886 stack_pointer_rtx for count 0. Update comments. Break up
3887 large rtl expression.
3889 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
3891 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
3892 Increase to 20 bytes.
3893 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
3894 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
3895 or avx version of the stub.
3897 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
3900 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
3901 compute the minimum stack alignment. Also update preferred stack
3902 boundary for leaf functions.
3904 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
3906 PR tree-optimization/82228
3907 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
3910 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3912 * common/config/nds32/nds32-common.c
3913 (nds32_option_optimization_table): Refine formatting.
3914 (nds32_option_optimization_table): Use -fsched-pressure and
3915 -fomit-frame-pointer for specific optimization level.
3917 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3919 * config/nds32/nds32.c: Refine formatting and comments.
3920 * config/nds32/nds32.h: Likewise.
3921 * config/nds32/nds32.md: Likewise.
3922 * config/nds32/nds32-cost.c: Likewise.
3923 * config/nds32/nds32-isr.c: Likewise.
3924 * config/nds32/nds32-md-auxiliary.c: Likewise.
3925 * config/nds32/nds32-multiple.md: Likewise.
3926 * config/nds32/nds32-predicates.c: Likewise.
3928 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
3929 Jakub Jelinek <jakub@redhat.com>
3931 Add support for -std=c++2a.
3932 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
3934 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
3936 2017-09-15 Steve Ellcey <sellcey@cavium.com>
3939 * doc/extend.texi (Common Function Attributes): Add
3940 references to ARM, AArch64, and S/390 specific attributes.
3941 (Function Specific Option Pragmas): Add AArch64 and S/390
3942 to list of back ends that support the target pragma.
3944 2017-09-15 Nathan Sidwell <nathan@acm.org>
3946 * doc/standards.texi: Fix C++17 description. Update URLs for
3949 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3951 * common.opt (Wcast-align=strict): New warning option.
3952 * doc/invoke.texi: Document -Wcast-align=strict.
3954 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
3956 * cgraph.h (cgraph_thunk_info): Add comments.
3957 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
3958 assert for VIRTUAL_* arguments stricter.
3960 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
3962 PR tree-optimization/71026
3963 * match.pd: Move RDIV patterns from fold-const.c
3964 * fold-const.c (distribute_real_division): Removed.
3965 (fold_binary_loc): Remove calls to distribute_real_divison.
3967 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3969 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
3970 c++1z and gnu++1z as deprecated. Change other references to
3971 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
3972 Change -Wc++1z-compat to -Wc++17-compat.
3973 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
3974 * dwarf2out.c (highest_c_language): Handle C++17.
3975 (gen_compile_unit_die): Likewise.
3977 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3979 PR rtl-optimization/82192
3980 * combine.c (make_extraction): Don't look through non-paradoxical
3981 SUBREGs or TRUNCATE if pos + len is or might be bigger than
3984 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3985 Alan Hayward <alan.hayward@arm.com>
3986 David Sherwood <david.sherwood@arm.com>
3988 * target.def (function_arg_offset): New hook.
3989 * targhooks.h (default_function_arg_offset): Declare.
3990 * targhooks.c (default_function_arg_offset): New function.
3991 * function.c (locate_and_pad_parm): Use
3992 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
3993 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
3994 (TARGET_FUNCTION_ARG_OFFSET): ...this.
3995 * doc/tm.texi: Regenerate.
3996 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
3997 * config/spu/spu.c (spu_function_arg_offset): New function.
3998 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
3999 * system.h (FUNCTION_ARG_OFFSET): Poison.
4001 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4002 Alan Hayard <alan.hayward@arm.com>
4003 David Sherwood <david.sherwood@arm.com>
4005 * target.def (truly_noop_truncation): New hook.
4006 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
4007 than TRULY_NOOP_TRUNCATION.
4008 * hooks.h (hook_bool_uint_uint_true): Declare.
4009 * hooks.c (hook_bool_uint_uint_true): New function.
4010 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
4011 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
4012 * doc/tm.texi: Regenerate.
4013 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
4014 rather than TRULY_NOOP_TRUNCATION in comments.
4015 (simplify_comparison): Likewise.
4016 (record_truncated_value): Likewise.
4017 * expmed.c (extract_bit_field_1): Likewise.
4018 (extract_split_bit_field): Likewise.
4019 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
4020 instead of TRULY_NOOP_TRUNCATION.
4021 * function.c (assign_parm_setup_block): Likewise.
4022 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
4023 * rtlhooks.c: Include target.h.
4024 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
4025 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
4026 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
4027 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
4028 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
4029 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
4030 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
4031 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
4032 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
4033 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
4034 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
4035 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
4036 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
4037 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
4038 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
4039 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
4040 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
4041 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
4042 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
4043 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
4044 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
4045 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
4046 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
4047 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
4048 * config/mips/mips.c (mips_truly_noop_truncation): New function.
4049 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4050 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4051 rather than TRULY_NOOP_TRUNCATION in comments.
4052 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
4053 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
4054 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
4055 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
4056 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
4057 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
4058 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
4059 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
4060 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
4061 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
4062 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
4063 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4064 rather than TRULY_NOOP_TRUNCATION in comments.
4065 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
4066 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
4067 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
4068 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
4069 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
4070 TRULY_NOOP_TRUNCATION condition.
4071 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
4072 (TRULY_NOOP_TRUNCATION): Delete.
4073 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
4074 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
4075 * config/spu/spu.c (spu_truly_noop_truncation): New function.
4076 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4077 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
4078 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
4079 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
4080 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4081 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4082 rather than TRULY_NOOP_TRUNCATION in comments.
4083 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
4084 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
4085 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
4086 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
4087 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
4088 * system.h (TRULY_NOOP_TRUNCATION): Poison.
4090 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
4093 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
4094 (*cmp_ior): Likewise.
4095 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
4096 (*ior_scc_scc_cmp): Likewise.
4097 (*and_scc_scc): Likewise.
4098 (*and_scc_scc_cmp): Likewise.
4100 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4101 Alan Hayard <alan.hayward@arm.com>
4102 David Sherwood <david.sherwood@arm.com>
4104 * target.def (can_change_mode_class): New hook.
4105 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4106 (hard_regno_nregs): Likewise.
4107 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4108 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4109 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4110 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4111 (LOAD_EXTEND_OP): Update accordingly.
4112 * doc/tm.texi: Regenerate.
4113 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4114 CANNOT_CHANGE_MODE_CLASS.
4115 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4116 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4117 * combine.c (simplify_set): Update accordingly.
4118 * emit-rtl.c (validate_subreg): Likewise.
4119 * recog.c (general_operand): Likewise.
4120 * regcprop.c (mode_change_ok): Likewise.
4121 * reload1.c (choose_reload_regs): Likewise.
4122 (inherit_piecemeal_p): Likewise.
4123 * rtlanal.c (simplify_subreg_regno): Likewise.
4124 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4125 instead of CANNOT_CHANGE_MODE_CLASS.
4126 (reload_cse_simplify_operands): Likewise.
4127 * reload.c (push_reload): Use targetm.can_change_mode_class
4128 instead of CANNOT_CHANGE_MODE_CLASS.
4129 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4130 REG_CANNOT_CHANGE_MODE_P.
4131 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4132 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4133 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4134 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4135 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4136 (arm_can_change_mode_class): New function.
4137 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4138 than CANNOT_CHANGE_MODE_CLASS in comments.
4139 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4140 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4141 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4142 (ix86_can_change_mode_class): ...this new function, inverting the
4143 sense of the return value.
4144 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4145 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4146 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4147 (ia64_can_change_mode_class): New function.
4148 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4149 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4150 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4151 (m32c_can_change_mode_class): ...this new function, inverting the
4152 sense of the return value.
4153 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4154 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4155 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4156 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4157 (mips_can_change_mode_class): ...this new function, inverting the
4158 sense of the return value.
4159 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4160 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4161 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4162 (msp430_can_change_mode_class): New function.
4163 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4164 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4165 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4166 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4167 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4168 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4169 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4170 (pa_cannot_change_mode_class): Replace with...
4171 (pa_can_change_mode_class): ...this new function, inverting the
4172 sense of the return value.
4173 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4174 than CANNOT_CHANGE_MODE_CLASS in comments.
4175 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4176 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
4177 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4178 (pdp11_cannot_change_mode_class): Replace with...
4179 (pdp11_can_change_mode_class): ...this new function, inverting the
4180 sense of the return value.
4181 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4182 * config/powerpcspe/powerpcspe-protos.h
4183 (rs6000_cannot_change_mode_class_ptr): Delete.
4184 * config/powerpcspe/powerpcspe.c
4185 (rs6000_cannot_change_mode_class_ptr): Delete.
4186 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4187 (rs6000_option_override_internal): Assign to
4188 targetm.can_change_mode_class instead of
4189 rs6000_cannot_change_mode_class_ptr.
4190 (rs6000_cannot_change_mode_class): Replace with...
4191 (rs6000_can_change_mode_class): ...this new function, inverting the
4192 sense of the return value.
4193 (rs6000_debug_cannot_change_mode_class): Replace with...
4194 (rs6000_debug_can_change_mode_class): ...this new function.
4195 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4196 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
4197 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4198 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4199 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
4201 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
4202 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4203 (rs6000_option_override_internal): Assign to
4204 targetm.can_change_mode_class instead of
4205 rs6000_cannot_change_mode_class_ptr.
4206 (rs6000_cannot_change_mode_class): Replace with...
4207 (rs6000_can_change_mode_class): ...this new function, inverting the
4208 sense of the return value.
4209 (rs6000_debug_cannot_change_mode_class): Replace with...
4210 (rs6000_debug_can_change_mode_class): ...this new function.
4211 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4212 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
4213 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
4214 (s390_can_change_mode_class): ...this new function, inverting the
4215 sense of the return value.
4216 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4217 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4218 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
4219 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4220 (sh_cannot_change_mode_class): Replace with...
4221 (sh_can_change_mode_class): ...this new function, inverting the
4222 sense of the return value.
4223 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4224 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4225 (sparc_can_change_mode_class): New function.
4226 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4227 * config/spu/spu.c (spu_can_change_mode_class): New function.
4228 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4229 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4230 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4231 (visium_can_change_mode_class): New function.
4232 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
4234 2017-09-15 Richard Biener <rguenther@suse.de>
4236 PR tree-optimization/82217
4237 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
4238 but not undefined case.
4240 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4243 * postreload.c (reload_cse_simplify_operands): Skip
4244 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
4246 2017-09-15 Richard Biener <rguenther@suse.de>
4248 PR tree-optimization/68823
4249 * graphite-scop-detection.c (build_alias_set): If we have a
4250 possible dependence check whether we can handle them by just
4251 looking at the DRs DR_ACCESS_FNs.
4252 (build_scops): If build_alias_set fails, fail the SCOP.
4254 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4256 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4257 to support float128 built-in functions that require the ISA 3.0
4259 (BU_FLOAT128_3_HW): Likewise.
4260 (SQRTF128): Add support for the IEEE 128-bit square root and fma
4262 (FMAF128): Likewise.
4264 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4265 support for built-in functions that need the ISA 3.0 IEEE 128-bit
4266 floating point instructions.
4267 (rs6000_invalid_builtin): Likewise.
4268 (rs6000_builtin_mask_names): Likewise.
4269 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
4270 (RS6000_BTM_FLOAT128_HW): Likewise.
4271 (RS6000_BTM_COMMON): Likewise.
4272 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
4274 * doc/extend.texi (RS/6000 built-in functions): Document the
4275 IEEE 128-bit floating point square root and fused multiply-add
4278 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
4280 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
4281 reg (r2) isn't in the set of registers defined in the prologue.
4283 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4284 Alan Hayward <alan.hayward@arm.com>
4285 David Sherwood <david.sherwood@arm.com>
4287 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
4288 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
4289 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
4290 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
4291 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
4293 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
4294 max_vectorization_factor.
4295 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
4297 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4298 Alan Hayward <alan.hayward@arm.com>
4299 David Sherwood <david.sherwood@arm.com>
4301 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
4302 (vect_worthwhile_without_simd_p): Declare.
4303 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
4304 (vectorizable_reduction): Use it.
4305 * tree-vect-stmts.c (vectorizable_shift): Likewise.
4306 (vectorizable_operation): Likewise.
4308 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4309 Alan Hayward <alan.hayward@arm.com>
4310 David Sherwood <david.sherwood@arm.com>
4312 * tree-vectorizer.h (vect_get_num_copies): New function.
4313 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
4314 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4315 (vectorizable_induction): Likewise.
4316 (vectorizable_live_operation): Likewise.
4317 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4318 (vectorizable_bswap): Likewise.
4319 (vectorizable_call): Likewise.
4320 (vectorizable_conversion): Likewise.
4321 (vectorizable_assignment): Likewise.
4322 (vectorizable_shift): Likewise.
4323 (vectorizable_operation): Likewise.
4324 (vectorizable_store): Likewise.
4325 (vectorizable_load): Likewise.
4326 (vectorizable_condition): Likewise.
4327 (vectorizable_comparison): Likewise.
4328 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
4330 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4331 Alan Hayward <alan.hayward@arm.com>
4332 David Sherwood <david.sherwood@arm.com>
4334 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
4335 of vect_init_vector.
4337 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4338 Alan Hayward <alan.hayward@arm.com>
4339 David Sherwood <david.sherwood@arm.com>
4341 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
4342 an inline wrapper that provides a location.
4343 (gimple_build_vector): Likewise.
4344 * gimple-fold.c (gimple_build_vector_from_val): New function.
4345 (gimple_build_vector): Likewise.
4346 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
4347 functions to build the initial value. Always return a gimple value.
4348 (get_initial_defs_for_reduction): Likewise. Only compute
4350 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
4351 vect_init_vector on the results from get_initial_def(s)_for_reduction.
4352 (vectorizable_induction): Use gimple_build_vector rather than
4355 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4356 Alan Hayward <alan.hayward@arm.com>
4357 David Sherwood <david.sherwood@arm.com>
4359 * target.h (vec_perm_indices): New typedef.
4360 (auto_vec_perm_indices): Likewise.
4361 * optabs-query.h: Include target.h
4362 (can_vec_perm_p): Take a vec_perm_indices *.
4363 * optabs-query.c (can_vec_perm_p): Likewise.
4364 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
4365 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4366 * tree-vect-generic.c (lower_vec_perm): Likewise.
4367 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
4368 (vect_grouped_load_supported): Likewise.
4369 (vect_shift_permute_load_chain): Likewise.
4370 (vect_permute_store_chain): Use auto_vec_perm_indices.
4371 (vect_permute_load_chain): Likewise.
4372 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
4373 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
4374 Update uses of can_vec_perm_p.
4375 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
4376 mode with a number of elements. Take a vec_perm_indices *.
4377 (vect_create_epilog_for_reduction): Update accordingly.
4378 Use auto_vec_perm_indices.
4379 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
4380 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
4381 (vect_transform_slp_perm_load): Likewise.
4382 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
4383 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
4384 (vect_gen_perm_mask_checked): Likewise.
4385 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
4386 (vect_gen_perm_mask_checked): Likewise.
4387 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
4388 (vectorizable_store): Likewise.
4389 (vectorizable_load): Likewise.
4390 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
4391 (vectorizable_bswap): Likewise.
4393 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4394 Alan Hayward <alan.hayward@arm.com>
4395 David Sherwood <david.sherwood@arm.com>
4397 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
4398 * tree.c (build_vector): Likewise.
4399 (build_vector_from_ctor): Update accordingly.
4400 (build_vector_from_val): Likewise.
4401 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4402 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4403 * tree-vect-generic.c (add_rshift): Likewise.
4404 (expand_vector_divmod): Likewise.
4405 (optimize_vector_constructor): Likewise.
4406 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4407 (vect_transform_slp_perm_load): Likewise.
4408 (vect_schedule_slp_instance): Likewise.
4409 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
4410 (vectorizable_call): Likewise.
4411 (vect_gen_perm_mask_any): Likewise. Add elements in order.
4412 * expmed.c (make_tree): Likewise.
4413 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
4414 a vector passed to build_vector.
4415 (fold_convert_const): Likewise.
4416 (exact_inverse): Likewise.
4417 (fold_ternary_loc): Likewise.
4418 (fold_relational_const): Likewise.
4419 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
4420 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
4421 (const_unop): Likewise. Store the reduction accumulator in a
4422 variable rather than an array.
4423 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
4424 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
4425 the new vector, rather than constructing it after the input arrays.
4426 (native_interpret_vector): Use auto_vec<tree> when building
4427 a vector passed to build_vector. Add elements in order.
4428 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
4429 auto_vec<tree> when building a vector passed to build_vector.
4430 (vect_create_epilog_for_reduction): Likewise.
4431 (vectorizable_induction): Likewise.
4432 (get_initial_def_for_reduction): Likewise. Fix indentation of
4434 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
4436 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
4437 passed to build_vector.
4439 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4440 Alan Hayward <alan.hayward@arm.com>
4441 David Sherwood <david.sherwood@arm.com>
4443 * tree-core.h (tree_base::u): Add an "nelts" field.
4444 (tree_vector): Use VECTOR_CST_NELTS as the length.
4445 * tree.c (tree_size): Likewise.
4446 (make_vector): Initialize VECTOR_CST_NELTS.
4447 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
4448 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
4449 TYPE_VECTOR_SUBPARTS.
4450 * expr.c (const_vector_mask_from_tree): Consistently use "units"
4451 as the number of units, setting it from VECTOR_CST_NELTS.
4452 (const_vector_from_tree): Likewise.
4453 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
4454 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
4455 (fold_negate_expr_1): Likewise.
4456 (fold_convert_const): Likewise.
4457 (const_binop): Likewise. Differentiate the number of output and
4459 (const_unop): Likewise.
4460 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
4461 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
4462 in cases that did the opposite.
4464 2017-09-14 Richard Biener <rguenther@suse.de>
4466 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
4469 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
4471 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
4473 2017-09-14 Jakub Jelinek <jakub@redhat.com>
4476 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
4477 if and where to split a bb, except for splitting before debug insn
4478 sequences followed by non-label real insn. Delete debug insns
4479 in between basic blocks.
4481 * combine.c (make_compound_operation_int): Formatting fixes.
4483 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
4484 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4485 * config/netbsd.h (LINK_EH_SPEC): Likewise.
4486 * config/sol2.h (LINK_EH_SPEC): Likewise.
4487 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4488 * config/s390/linux.h (LINK_SPEC): Likewise.
4489 * config/freebsd.h (LINK_EH_SPEC): Likewise.
4490 * config/openbsd.h (LINK_EH_SPEC): Likewise.
4491 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4492 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
4493 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
4494 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4495 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
4496 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
4498 2017-09-13 Jakub Jelinek <jakub@redhat.com>
4500 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
4502 (ENDFILE_LINUX_SPEC): Likewise.
4503 (LINK_EH_SPEC): Likewise.
4504 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
4505 (LINK_OS_LINUX_SPEC32): Likewise.
4506 (LINK_OS_LINUX_SPEC64): Likewise.
4507 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
4508 (LINK_OS_LINUX_SPEC): Likewise.
4510 2017-09-13 Martin Liska <mliska@suse.cz>
4513 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
4514 CASE_HIGH is NULL_TREE.
4516 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4517 Alan Hayward <alan.hayward@arm.com>
4518 David Sherwood <david.sherwood@arm.com>
4520 * target.def (secondary_memory_needed): New hook.
4521 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
4522 instead of SECONDARY_MEMORY_NEEDED.
4523 (secondary_memory_needed_mode): Likewise.
4524 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
4525 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
4526 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
4527 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
4528 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
4529 * doc/tm.texi: Regenerate.
4530 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
4531 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
4532 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4533 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
4534 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
4535 * config/i386/i386.c (inline_secondary_memory_needed): Put the
4536 mode argument first and change the reg_class arguments to reg_class_t.
4537 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
4538 Make static. Update the call to inline_secondary_memory_needed.
4539 (ix86_register_move_cost): Update the call to
4540 inline_secondary_memory_needed.
4541 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4542 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
4544 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
4545 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4547 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
4548 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
4549 * config/mips/mips.c (mips_secondary_memory_needed): Make static
4550 and match hook interface. Add comment from mips.h.
4551 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4552 * config/mmix/mmix.md (truncdfsf2): Refer to
4553 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4555 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
4556 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
4557 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4558 (pa_secondary_memory_needed): New function.
4559 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
4560 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
4561 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4562 (pdp11_secondary_memory_needed): Make static and match hook interface.
4563 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
4564 * config/powerpcspe/powerpcspe-protos.h
4565 (rs6000_secondary_memory_needed_ptr): Delete.
4566 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
4568 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4569 (rs6000_option_override_internal): Assign to
4570 targetm.secondary_memory_needed rather than
4571 rs6000_secondary_memory_needed_ptr.
4572 (rs6000_secondary_memory_needed): Match hook interface.
4573 (rs6000_debug_secondary_memory_needed): Likewise.
4574 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
4575 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
4576 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
4577 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4578 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
4579 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
4581 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
4582 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4583 (rs6000_option_override_internal): Assign to
4584 targetm.secondary_memory_needed rather than
4585 rs6000_secondary_memory_needed_ptr.
4586 (rs6000_secondary_memory_needed): Match hook interface.
4587 (rs6000_debug_secondary_memory_needed): Likewise.
4588 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
4589 * config/s390/s390.c (s390_secondary_memory_needed): New function.
4590 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4591 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
4592 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4593 (sparc_secondary_memory_needed): New function.
4594 * lra-constraints.c (check_and_process_move): Refer to
4595 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4597 (curr_insn_transform): Likewise.
4598 (process_alt_operands): Use targetm.secondary_memory_needed
4599 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4600 (check_secondary_memory_needed_p): Likewise.
4601 (choose_split_class): Likewise.
4602 * reload.c: Unconditionally include code that was previously
4603 conditional on SECONDARY_MEMORY_NEEDED.
4604 (push_secondary_reload): Use targetm.secondary_memory_needed
4605 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4606 (push_reload): Likewise.
4607 * reload1.c: Unconditionally include code that was previously
4608 conditional on SECONDARY_MEMORY_NEEDED.
4609 (choose_reload_regs): Use targetm.secondary_memory_needed
4610 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4611 (gen_reload): Likewise.
4612 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
4614 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4615 Alan Hayward <alan.hayward@arm.com>
4616 David Sherwood <david.sherwood@arm.com>
4618 * target.def (secondary_memory_needed_mode): New hook:
4619 * targhooks.c (default_secondary_memory_needed_mode): Declare.
4620 * targhooks.h (default_secondary_memory_needed_mode): New function.
4621 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
4622 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
4623 * doc/tm.texi: Regenerate.
4624 * lra-constraints.c (check_and_process_move): Use
4625 targetm.secondary_memory_needed_mode instead of
4626 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
4627 (curr_insn_transform): Likewise.
4628 * reload.c (get_secondary_mem): Likewise.
4629 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4630 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
4632 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4633 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4634 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
4635 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4636 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
4638 * config/powerpcspe/powerpcspe-protos.h
4639 (rs6000_secondary_memory_needed_mode): Delete.
4640 * config/powerpcspe/powerpcspe.c
4641 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4642 (rs6000_secondary_memory_needed_mode): Make static.
4643 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4644 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
4646 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4648 (rs6000_secondary_memory_needed_mode): Make static.
4649 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4650 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
4651 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4652 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4653 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4655 (sparc_secondary_memory_needed_mode): New function.
4656 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
4658 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
4660 * config/aarch64/constraints.md (Umq): New constraint.
4661 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
4663 (mov<mode>): Update condition.
4665 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4667 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
4668 when bitposition is the same.
4670 2017-09-13 Richard Biener <rguenther@suse.de>
4672 * dwarf2out.c (output_die_symbol): Remove.
4673 (output_die): Do not output a DIEs symbol.
4675 2017-09-13 Richard Biener <rguenther@suse.de>
4678 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
4679 default-def to avoid breaking iterator update with the weird
4680 interaction with cgraph_update_edges_for_call_stmt_node.
4682 2017-09-13 Richard Biener <rguenther@suse.de>
4684 * tree-cfg.c (verify_gimple_assign_binary): Add verification
4685 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
4686 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
4687 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
4689 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
4691 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
4692 Disable pc relative literal load irrespective of
4693 TARGET_FIX_ERR_A53_84341 for default.
4695 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
4697 * config/sparc/sparc.c (output_return): Output the source location of
4698 the insn in the delay slot, if any.
4699 (output_sibcall): Likewise.
4701 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
4704 * common.opt (-static-pie): New alias.
4705 (shared): Negate static-pie.
4706 (-no-pie): Update help text.
4708 (static-pie): New option.
4709 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
4710 -static-pie support.
4711 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
4712 (LINK_EH_SPEC): Likewise.
4713 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4714 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4715 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4716 * gcc.c (LINK_COMMAND_SPEC): Likewise.
4717 (init_gcc_specs): Likewise.
4718 (init_spec): Likewise.
4719 (display_help): Update help message for -pie.
4720 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
4723 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
4725 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
4726 (movdi_aarch64): Likewise.
4727 (movti_aarch64): Likewise.
4729 2017-09-12 Simon Wright <simon@pushface.org>
4732 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
4733 calculation of the minor version, always output as 0.
4735 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4738 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4739 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
4740 on it early, rather than manual conversion late. For
4741 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
4742 instead of performing manual conversion.
4744 2017-09-12 Carl Love <cel@us.ibm.com>
4746 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
4747 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
4749 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4750 VMULOSW): Add definitions.
4751 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4752 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4753 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4754 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4755 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4757 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4759 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
4761 (movti_aarch64): Likewise.
4762 (movdf_aarch64): Likewise.
4763 (movtf_aarch64): Likewise.
4764 (load_pairdi): Likewise.
4765 (store_pairdi): Likewise.
4766 (load_pairdf): Likewise.
4767 (store_pairdf): Likewise.
4768 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
4769 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
4770 (ldr_got_small_<mode>): Likewise.
4771 (ldr_got_small_28k_<mode>): Likewise.
4772 (ldr_got_tiny): Likewise.
4773 * config/aarch64/iterators.md (ldst_sz): New.
4774 (ldpstp_sz): Likewise.
4775 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
4777 (thunderx_load): Split load_8 to load_16.
4778 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
4780 (thunderx2t99_storepair_basic): Split store_8 to store_16.
4781 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
4782 (xgene1_store_pair): Split store_8 to store_16.
4783 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
4784 (falkor_st_0_st_sd): Split store_8 to store_16.
4786 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4788 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
4789 and store1/2/3/4 to store_4/8/12/16.
4790 * config/aarch64/aarch64.md: Update for rename.
4791 * config/arm/arm.md: Likewise.: Likewise.
4792 * config/arm/arm.c: Likewise.
4793 * config/arm/thumb1.md: Likewise.
4794 * config/arm/thumb2.md: Likewise.
4795 * config/arm/vfp.md: Likewise.
4796 * config/arm/arm-generic.md: Likewise.
4797 * config/arm/arm1020e.md: Likewise.
4798 * config/arm/arm1026ejs.md: Likewise.
4799 * config/arm/arm1136jfs.md: Likewise.
4800 * config/arm/arm926ejs.md: Likewise.
4801 * config/arm/cortex-a15.md: Likewise.
4802 * config/arm/cortex-a17.md: Likewise.
4803 * config/arm/cortex-a5.md: Likewise.
4804 * config/arm/cortex-a53.md: Likewise.
4805 * config/arm/cortex-a57.md: Likewise.
4806 * config/arm/cortex-a7.md: Likewise.
4807 * config/arm/cortex-a8.md: Likewise.
4808 * config/arm/cortex-a9.md: Likewise.
4809 * config/arm/cortex-m4.md: Likewise.
4810 * config/arm/cortex-m7.md: Likewise.
4811 * config/arm/cortex-r4.md: Likewise.
4812 * config/arm/exynos-m1.md: Likewise.
4813 * config/arm/fa526.md: Likewise.
4814 * config/arm/fa606te.md: Likewise.
4815 * config/arm/fa626te.md: Likewise.
4816 * config/arm/fa726te.md: Likewise.
4817 * config/arm/fmp626.md: Likewise.
4818 * config/arm/iwmmxt.md: Likewise.
4819 * config/arm/ldmstm.md: Likewise.
4820 * config/arm/marvell-pj4.md: Likewise.
4821 * config/arm/xgene1.md: Likewise.
4822 * config/aarch64/thunderx.md: Likewise.
4823 * config/aarch64/thunderx2t99.md: Likewise.
4824 * config/aarch64/falkor.md: Likewise.
4826 2017-09-12 Martin Liska <mliska@suse.cz>
4828 * attribs.c (private_lookup_attribute): New function.
4829 * attribs.h (private_lookup_attribute): Declared here.
4830 (lookup_attribute): Called from this place.
4832 2017-09-12 Richard Biener <rguenther@suse.de>
4834 PR tree-optimization/82157
4835 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
4836 stmts with side-effects.
4838 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4839 Alan Hayward <alan.hayward@arm.com>
4840 David Sherwood <david.sherwood@arm.com>
4842 * target.def (hard_regno_nregs): New hook.
4843 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
4844 * targhooks.h (default_hard_regno_nregs): Declare.
4845 * targhooks.c (default_hard_regno_nregs): New function.
4846 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
4847 (TARGET_HARD_REGNO_NREGS): ...this hook.
4848 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
4849 (CLASS_MAX_NREGS): Likewise.
4850 * doc/tm.texi: Regenerate.
4851 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
4852 instead of HARD_REGNO_NREGS.
4853 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
4854 HARD_REGNO_NREGS in the comment.
4855 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
4856 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
4857 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
4858 Return an unsigned int.
4859 (TARGET_HARD_REGNO_NREGS): Redefine.
4860 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
4861 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
4862 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4863 (arc_hard_regno_nregs): New function.
4864 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
4865 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
4866 (arm_hard_regno_nregs): New function.
4867 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
4868 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
4869 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
4870 (TARGET_HARD_REGNO_NREGS): Redefine.
4871 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
4872 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
4873 (HARD_REGNO_NREGS): Delete.
4874 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
4875 (cr16_hard_regno_nregs): New function.
4876 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
4877 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
4878 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
4879 (cris_hard_regno_nregs): New function.
4880 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
4881 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
4882 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4883 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
4884 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
4885 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
4886 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
4887 (frv_hard_regno_nregs): Make static. Take and return an
4889 (frv_class_max_nregs): Remove outdated copy of documentation.
4890 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
4891 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
4892 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
4893 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
4894 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
4895 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
4896 (TARGET_HARD_REGNO_NREGS): Redefine.
4897 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
4898 (CLASS_MAX_NREGS): Update comment.
4899 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
4900 (ia64_hard_regno_nregs): New function.
4901 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
4902 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
4903 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
4904 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
4905 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
4907 (m32c_hard_regno_nregs): Likewise. Make static.
4908 (TARGET_HARD_REGNO_NREGS): Redefine.
4909 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
4910 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
4911 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
4912 (m68k_hard_regno_nregs): New function.
4913 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
4914 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
4915 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
4916 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
4917 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
4918 Take and return an unsigned int.
4919 (TARGET_HARD_REGNO_NREGS): Redefine.
4920 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
4921 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4922 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
4923 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
4924 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
4925 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
4926 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
4927 (msp430_hard_regno_nregs): Make static. Take and return an
4929 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
4930 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
4931 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
4932 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
4933 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
4934 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
4935 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
4936 (TARGET_HARD_REGNO_NREGS): Redefine.
4937 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
4938 (PA_HARD_REGNO_NREGS): ...this.
4939 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
4940 (PA_HARD_REGNO_NREGS): ...this.
4941 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4942 (pa_hard_regno_nregs): New function.
4943 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
4944 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
4945 (pdp11_hard_regno_nregs): New function.
4946 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
4947 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
4948 (rs6000_hard_regno_nregs_hook): New function.
4949 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
4950 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4951 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
4952 Take and return an unsigned int. Move earlier in file.
4953 (TARGET_HARD_REGNO_NREGS): Redefine.
4954 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
4955 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
4956 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
4957 (rl78_hard_regno_nregs): Make static. Take and return an
4959 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
4960 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
4961 (rs6000_hard_regno_nregs_hook): New function.
4962 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
4963 * config/rx/rx.c (rx_hard_regno_nregs): New function.
4964 (TARGET_HARD_REGNO_NREGS): Redefine.
4965 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
4966 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
4967 instead of HARD_REGNO_NREGS.
4968 (s390_hard_regno_nregs): New function.
4969 (s390_hard_regno_mode_ok): Add comment from s390.h.
4970 (TARGET_HARD_REGNO_NREGS): Redefine.
4971 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
4972 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
4973 (sh_hard_regno_nregs): New function.
4974 (sh_pass_in_reg_p): Use it.
4975 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
4976 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4977 (sparc_hard_regno_nregs): New function.
4978 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
4979 * config/spu/spu.c (spu_hard_regno_nregs): New function.
4980 (spu_function_arg_advance): Use it, supplying a valid register number.
4981 (TARGET_HARD_REGNO_NREGS): Redefine.
4982 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
4983 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
4984 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
4985 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
4986 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
4987 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
4988 (CLASS_MAX_NREGS): Remove copy of old documentation.
4989 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
4990 (visium_hard_regno_nregs): New function.
4991 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
4992 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
4993 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4994 xtensa_hard_regno_nregs): New function.
4995 * system.h (HARD_REGNO_NREGS): Poison.
4997 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4999 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
5000 hard_regno_nregs instead of HARD_REGNO_NREGS.
5001 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
5002 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
5003 (c6x_expand_epilogue): Likewise.
5004 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
5005 (frv_read_iacc_argument): Likewise.
5006 * config/sh/sh.c: Include regs.h.
5007 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5008 (regs_used): Likewise.
5009 (output_stack_adjust): Likewise.
5010 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
5011 * expmed.c: Include regs.h.
5012 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5013 * ree.c: Include regs.h.
5014 (combine_reaching_defs): Use hard_regno_nregs instead of
5016 (add_removable_extension): Likewise.
5018 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5020 * regs.h (hard_regno_nregs): Turn into a function.
5021 (end_hard_regno): Update accordingly.
5022 * caller-save.c (setup_save_areas): Likewise.
5023 (save_call_clobbered_regs): Likewise.
5024 (replace_reg_with_saved_mem): Likewise.
5025 (insert_restore): Likewise.
5026 (insert_save): Likewise.
5027 * combine.c (can_change_dest_mode): Likewise.
5028 (move_deaths): Likewise.
5029 (distribute_notes): Likewise.
5030 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
5031 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
5032 (rs6000_split_multireg_move): Likewise.
5033 (rs6000_register_move_cost): Likewise.
5034 (rs6000_memory_move_cost): Likewise.
5035 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
5036 (rs6000_split_multireg_move): Likewise.
5037 (rs6000_register_move_cost): Likewise.
5038 (rs6000_memory_move_cost): Likewise.
5039 * cselib.c (cselib_reset_table): Likewise.
5040 (cselib_lookup_1): Likewise.
5041 * emit-rtl.c (set_mode_and_regno): Likewise.
5042 * function.c (aggregate_value_p): Likewise.
5043 * ira-color.c (setup_profitable_hard_regs): Likewise.
5044 (check_hard_reg_p): Likewise.
5045 (calculate_saved_nregs): Likewise.
5046 (assign_hard_reg): Likewise.
5047 (improve_allocation): Likewise.
5048 (calculate_spill_cost): Likewise.
5049 * ira-emit.c (modify_move_list): Likewise.
5050 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
5051 (ira_hard_reg_in_set_p): Likewise.
5052 * ira.c (setup_reg_mode_hard_regset): Likewise.
5053 (clarify_prohibited_class_mode_regs): Likewise.
5054 (check_allocation): Likewise.
5055 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5056 (lra_setup_reg_renumber): Likewise.
5057 (setup_try_hard_regno_pseudos): Likewise.
5058 (spill_for): Likewise.
5059 (assign_hard_regno): Likewise.
5060 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
5061 * lra-constraints.c (in_class_p): Likewise.
5062 (lra_constraint_offset): Likewise.
5063 (simplify_operand_subreg): Likewise.
5064 (lra_constraints): Likewise.
5065 (split_reg): Likewise.
5066 (split_if_necessary): Likewise.
5067 (invariant_p): Likewise.
5068 (inherit_in_ebb): Likewise.
5069 * lra-lives.c (process_bb_lives): Likewise.
5070 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
5071 (get_hard_regs): Likewise.
5072 (do_remat): Likewise.
5073 * lra-spills.c (assign_spill_hard_regs): Likewise.
5074 * mode-switching.c (create_pre_exit): Likewise.
5075 * postreload.c (reload_combine_recognize_pattern): Likewise.
5076 * recog.c (peep2_find_free_register): Likewise.
5077 * regcprop.c (kill_value_regno): Likewise.
5078 (set_value_regno): Likewise.
5079 (copy_value): Likewise.
5080 (maybe_mode_change): Likewise.
5081 (find_oldest_value_reg): Likewise.
5082 (copyprop_hardreg_forward_1): Likewise.
5083 * regrename.c (check_new_reg_p): Likewise.
5084 (regrename_do_replace): Likewise.
5085 * reload.c (push_reload): Likewise.
5086 (combine_reloads): Likewise.
5087 (find_dummy_reload): Likewise.
5088 (operands_match_p): Likewise.
5089 (find_reloads): Likewise.
5090 (find_equiv_reg): Likewise.
5091 (reload_adjust_reg_for_mode): Likewise.
5092 * reload1.c (count_pseudo): Likewise.
5093 (count_spilled_pseudo): Likewise.
5094 (find_reg): Likewise.
5095 (clear_reload_reg_in_use): Likewise.
5096 (free_for_value_p): Likewise.
5097 (allocate_reload_reg): Likewise.
5098 (choose_reload_regs): Likewise.
5099 (reload_adjust_reg_for_temp): Likewise.
5100 (emit_reload_insns): Likewise.
5101 (delete_output_reload): Likewise.
5102 * rtlanal.c (subreg_get_info): Likewise.
5103 * sched-deps.c (sched_analyze_reg): Likewise.
5104 * sel-sched.c (init_regs_for_mode): Likewise.
5105 (mark_unavailable_hard_regs): Likewise.
5106 (choose_best_reg_1): Likewise.
5107 (verify_target_availability): Likewise.
5108 * valtrack.c (dead_debug_insert_temp): Likewise.
5109 * var-tracking.c (track_loc_p): Likewise.
5110 (emit_note_insn_var_location): Likewise.
5111 * varasm.c (make_decl_rtl): Likewise.
5112 * reginfo.c (choose_hard_reg_mode): Likewise.
5113 (init_reg_modes_target): Refer directly to
5114 this_target_regs->x_hard_regno_nregs.
5116 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5118 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5119 instead of hard_regno_nregs.
5121 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5123 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5124 end_hard_regno instead of hard_regno_nregs.
5125 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5126 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5127 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5128 * ira-color.c (improve_allocation): Likewise.
5129 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5130 * lra-lives.c (mark_regno_live): Likewise.
5131 (mark_regno_dead): Likewise.
5132 * lra-remat.c (operand_to_remat): Likewise.
5133 * lra.c (collect_non_operand_hard_regs): Likewise.
5134 * postreload.c (reload_combine_note_store): Likewise.
5135 (move2add_valid_value_p): Likewise.
5136 * reload.c (regno_clobbered_p): Likewise.
5138 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5140 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5142 * config/v850/v850.c (v850_reorg): Likewise.
5143 * reload.c (refers_to_regno_for_reload_p): Likewise.
5144 (find_equiv_reg): Likewise.
5145 * reload1.c (reload_reg_reaches_end_p): Likewise.
5147 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5149 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5151 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5152 * config/arm/arm.c (output_move_neon): Likewise.
5153 (arm_attr_length_move_neon): Likewise.
5154 (neon_split_vcombine): Likewise.
5155 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5156 (c6x_mark_reg_written): Likewise.
5157 (c6x_dwarf_register_span): Likewise.
5158 * config/i386/i386.c (ix86_save_reg): Likewise.
5159 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5160 (rws_access_reg): Likewise.
5161 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5162 * mode-switching.c (create_pre_exit): Likewise.
5163 * ree.c (combine_reaching_defs): Likewise.
5164 (add_removable_extension): Likewise.
5165 * regcprop.c (find_oldest_value_reg): Likewise.
5166 (copyprop_hardreg_forward_1): Likewise.
5167 * reload.c (reload_inner_reg_of_subreg): Likewise.
5168 (push_reload): Likewise.
5169 (combine_reloads): Likewise.
5170 (find_dummy_reload): Likewise.
5171 (reload_adjust_reg_for_mode): Likewise.
5172 * reload1.c (find_reload_regs): Likewise.
5173 (forget_old_reloads_1): Likewise.
5174 (reload_reg_free_for_value_p): Likewise.
5175 (reload_adjust_reg_for_temp): Likewise.
5176 (emit_reload_insns): Likewise.
5177 (delete_output_reload): Likewise.
5178 * sel-sched.c (choose_best_reg_1): Likewise.
5179 (choose_best_pseudo_reg): Likewise.
5181 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5182 Alan Hayward <alan.hayward@arm.com>
5183 David Sherwood <david.sherwood@arm.com>
5185 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
5186 * target.def (slow_unaligned_access): New hook.
5187 * targhooks.h (default_slow_unaligned_access): Declare.
5188 * targhooks.c (default_slow_unaligned_access): New function.
5189 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
5190 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
5191 * doc/tm.texi: Regenerate.
5192 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
5193 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
5194 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
5196 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
5197 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
5199 (rs6000_slow_unaligned_access): New function.
5200 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5201 (expand_block_compare): Likewise.
5202 (expand_strn_compare): Likewise.
5203 (rs6000_rtx_costs): Likewise.
5204 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
5205 (riscv_slow_unaligned_access): Likewise.
5206 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
5207 (riscv_slow_unaligned_access_p): ...this and make static.
5208 (riscv_option_override): Update accordingly.
5209 (riscv_slow_unaligned_access): New function.
5210 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5211 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
5212 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5213 (rs6000_slow_unaligned_access): New function.
5214 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5215 (rs6000_rtx_costs): Likewise.
5216 * config/rs6000/rs6000-string.c (expand_block_compare)
5217 (expand_strn_compare): Use targetm.slow_unaligned_access instead
5218 of SLOW_UNALIGNED_ACCESS.
5219 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
5220 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
5221 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
5222 of SLOW_UNALIGNED_ACCESS.
5223 * expmed.c (simple_mem_bitfield_p): Likewise.
5224 * expr.c (alignment_for_piecewise_move): Likewise.
5225 (emit_group_load_1): Likewise.
5226 (emit_group_store): Likewise.
5227 (copy_blkmode_from_reg): Likewise.
5228 (emit_push_insn): Likewise.
5229 (expand_assignment): Likewise.
5230 (store_field): Likewise.
5231 (expand_expr_real_1): Likewise.
5232 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5233 * lra-constraints.c (simplify_operand_subreg): Likewise.
5234 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5235 * gimple-ssa-store-merging.c: Likewise in block comment at start
5237 * tree-ssa-strlen.c: Include target.h.
5238 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
5239 of SLOW_UNALIGNED_ACCESS.
5240 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
5242 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5244 PR rtl-optimization/82185
5245 * expmed.c (emit_store_flag_int): Only test tem if it has been
5248 2017-09-12 Richard Biener <rguenther@suse.de>
5251 * match.pd ((FTYPE) N CMP CST): Fix typo.
5253 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
5255 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
5257 (mips_near_type_p): Add 'short_call' attribute as a synonym
5259 * doc/extend.texi (short_call): Document new function attribute.
5261 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5264 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
5265 assertion check that in the condition.
5266 (get_atomic_generic_size): Likewise. Before testing if parameter
5267 has pointer type, if it has array type, call for C++
5268 default_conversion to perform array-to-pointer conversion.
5270 2017-09-12 Richard Biener <rguenther@suse.de>
5272 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
5273 for operations we cannot scalarize.
5275 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
5277 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
5278 vectors heap vectors. Clean up comments.
5279 Make visited_bbs a reference.
5280 (profitable_jump_thread_path): Make GC
5281 vectors heap vectors. Clean up comments.
5283 (convert_and_register_jump_thread_path): Make GC vectors heap
5285 (check_subpath_and_update_thread_path): Same. Clean up comments.
5286 Make visited_bbs a reference.
5287 (handle_phi): Abstract common code to to
5288 register_jump_thread_path_if_profitable.
5289 Rename VAR_BB to DEF_BB.
5291 Make GC vectors heap vectors.
5292 Make visited_bbs a reference.
5293 (handle_assignment): Same.
5294 (register_jump_thread_path_if_profitable): New.
5295 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
5297 Make GC vectors heap vectors. Clean up comments.
5298 Make visited_bbs a reference.
5299 (find_jump_threads_backwards): Make visited_bbs live in the stack.
5300 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
5303 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
5306 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
5307 words of E_DImode object are reachable by xtensa_uimm8x4 access.
5309 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
5311 Revert r251800 and r251799.
5313 2017-09-11 Martin Jambor <mjambor@suse.cz>
5316 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
5317 arguments in advance.
5318 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
5319 use it to find predecessor edges.
5320 (naive_outof_ssa): Collect vector of predecessors.
5322 2017-09-08 Jason Merrill <jason@redhat.com>
5324 PR c++/70029 - ICE with ref-qualifier and -flto
5325 * langhooks.h (struct lang_hooks_for_types): Add
5326 copy_lang_qualifiers.
5327 * attribs.c (build_type_attribute_qual_variant): Use it.
5328 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
5330 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
5331 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
5333 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
5336 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
5337 (*mulsi3_sp64): New instruction.
5338 (mulsi3): New expander.
5340 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
5342 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
5344 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5346 * sancov.c: Include memmodel.h.
5348 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
5351 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
5354 2017-09-07 Carl Love <cel@us.ibm.com>
5356 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
5357 the sldi instruction.
5359 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
5361 * sancov.c: Include tm_p.h.
5363 2017-09-07 Jakub Jelinek <jakub@redhat.com>
5366 * output.h (switch_to_other_text_partition): New declaration.
5367 * varasm.c (switch_to_other_text_partition): New function.
5368 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
5369 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
5370 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
5371 to the other text partition before emitting LCL label and switch back
5372 after emitting the word after it.
5374 2017-09-07 Richard Biener <rguenther@suse.de>
5376 * passes.def (pass_split_crit_edges): Remove instance before PRE.
5377 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
5378 critical edges here, after loop init.
5379 (pass_data_pre): Remove PROP_no_crit_edges flags.
5380 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
5381 for valueization of call args to avoid leaking VN_TOP.
5382 (visit_use): Assert we do not visit default defs.
5383 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
5384 Use error_mark_node to more easily detect leaking VN_TOP.
5385 All default-defs are varying, not VN_TOP. Mark them visited.
5386 (run_scc_vn): Make code match comment.
5388 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5390 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
5391 OPTION_MASK_FLOAT128_KEYWORD.
5392 (POWERPC_MASKS): Likewise.
5393 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
5394 support for the -mfloat128-type option, and make -mfloat128
5395 default on PowerPC Linux systems. Define or undefine
5396 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
5397 Define __float128 to be __ieee128 if IEEE 128-bit support is
5398 enabled, or undefine it.
5399 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
5400 Delete defining __FLOAT128_TYPE__.
5401 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
5402 -mfloat128-type option and make -mfloat128 default on PowerPC
5404 (TARGET_FLOAT128_TYPE): Likewise.
5405 (-mfloat128-type): Likewise.
5406 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5407 Delete the -mfloat128-type option and make -mfloat128 default on
5408 PowerPC Linux systems. Always use __ieee128 to be the keyword for
5409 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
5410 128-bit floating point is enabled. Change tests from using
5411 -mfloat128-type to -mfloat128.
5412 (rs6000_mangle_type): Use the correct mangling for the __float128
5413 type even if normal long double is restricted to 64-bits.
5414 (floatn_mode): Enable the _Float128 type by default on VSX Linux
5416 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
5417 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
5418 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
5420 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
5421 documentation for -mfloat128.
5423 2017-09-06 Olivier Hainque <hainque@adacore.com>
5425 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
5427 2017-09-06 Wish Wu <wishwu007@gmail.com>
5428 Jakub Jelinek <jakub@redhat.com>
5430 * asan.c (initialize_sanitizer_builtins): Add
5431 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
5432 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
5433 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
5434 BT_FN_VOID_UINT64_PTR variables.
5435 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
5436 (BT_FN_VOID_UINT16_UINT16): Likewise.
5437 (BT_FN_VOID_UINT32_UINT32): Likewise.
5438 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
5439 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
5440 (BT_FN_VOID_UINT64_PTR): Likewise.
5441 * common.opt (flag_sanitize_coverage): New variable.
5442 (fsanitize-coverage=trace-pc): Remove.
5443 (fsanitize-coverage=): Add.
5444 * flag-types.h (enum sanitize_coverage_code): New enum.
5445 * fold-const.c (fold_range_test): Disable non-short-circuit
5446 optimization if flag_sanitize_coverage.
5447 (fold_truth_andor): Likewise.
5448 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
5449 * opts.c (COVERAGE_SANITIZER_OPT): Define.
5450 (coverage_sanitizer_opts): New array.
5451 (get_closest_sanitizer_option): Add OPTS argument, handle also
5452 OPT_fsanitize_coverage_.
5453 (parse_sanitizer_options): Adjusted to also handle
5454 OPT_fsanitize_coverage_.
5455 (common_handle_option): Add OPT_fsanitize_coverage_.
5456 * sancov.c (instrument_comparison, instrument_switch): New function.
5457 (sancov_pass): Add trace-cmp support.
5458 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
5459 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
5460 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
5461 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
5462 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
5463 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
5464 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
5465 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
5466 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
5467 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
5469 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5471 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5472 error. Only quit immediately if parsing is complete.
5473 (BEGIN): Initialize fatal_err and parse_done.
5474 (begin fpu, end fpu): Check number of arguments.
5475 (begin arch, end arch): Likewise.
5476 (begin cpu, end cpu): Likewise.
5477 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5478 (optalias): Likewise.
5480 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5482 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5483 * config/arm/arm-isa.h: Delete. Move definitions to ...
5484 * arm-cpus.in: ... here. Use new feature and fgroup values.
5485 * config/arm/arm.c (arm_option_override): Use lower case for feature
5487 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5488 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5489 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5491 (print_isa_bits_for): New function.
5492 (gen_isa): New function.
5493 (gen_comm_data): Use print_isa_bits_for.
5494 (define feature): New keyword.
5495 (define fgroup): New keyword.
5496 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
5497 (arm-isa.h): Add rule to generate file.
5498 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5499 case for feature bit names.
5501 2017-09-06 Richard Biener <rguenther@suse.de>
5503 * tree-ssa-pre.c (NECESSARY): Remove.
5504 (create_expression_by_pieces): Do not touch pass-local flags.
5505 (insert_into_preds_of_block): Likewise.
5506 (do_pre_regular_insertion): Likewise.
5507 (eliminate_insert): Likewise.
5508 (eliminate_dom_walker::before_dom_children): Likewise.
5509 (fini_eliminate): Do not look at inserted_exprs.
5510 (mark_operand_necessary): Remove.
5511 (remove_dead_inserted_code): Replace with simple work-list
5512 algorithm based on inserted_exprs and SSA uses.
5513 (pass_pre::execute): Re-order fini_eliminate and
5514 remove_dead_inserted_code.
5516 2017-09-06 Olivier Hainque <hainque@adacore.com>
5518 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
5519 for VxWorks 7. Adjust surrounding comments.
5521 2017-09-06 Richard Biener <rguenther@suse.de>
5523 * gimple-ssa-strength-reduction.c
5524 (find_candidates_dom_walker::before_dom_children): Also allow
5527 2017-09-06 Richard Biener <rguenther@suse.de>
5529 PR tree-optimization/82108
5530 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
5531 for gap in the non-permutation SLP case.
5533 2017-09-06 Martin Jambor <mjambor@suse.cz>
5535 PR tree-optimization/82078
5536 * tree-sra.c (sort_and_splice_var_accesses): Move call to
5537 add_access_to_work_queue...
5538 (build_accesses_from_assign): ...here.
5539 (propagate_all_subaccesses): Make sure racc is the group
5540 representative, if there is one.
5542 2017-09-06 Jakub Jelinek <jakub@redhat.com>
5545 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
5548 2017-09-06 Richard Biener <rguenther@suse.de>
5550 * gimple-ssa-strength-reduction.c
5551 (find_candidates_dom_walker::before_doom_children): Use a
5552 type and not a mode check.
5554 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5557 * config/arm/predicates.md (arm_general_adddi_operand): Create new
5559 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
5561 2017-09-05 Jeff Law <law@redhat.com>
5563 PR tree-optimization/64910
5564 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
5565 cases where we have 3 or more operands.
5567 2017-09-05 Jakub Jelinek <jakub@redhat.com>
5570 * omp-low.c (lower_omp_for): Recompute tree invariant if
5571 gimple_omp_for_initial/final is ADDR_EXPR.
5574 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
5575 into gimple val before gimplification fo the COND_EXPR.
5577 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
5579 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
5580 REGION_COPY argument.
5581 (thread_through_all_blocks): Remove unused argument to
5582 duplicate_thread_path.
5584 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5585 Alan Hayward <alan.hayward@arm.com>
5586 David Sherwood <david.sherwood@arm.com>
5588 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
5589 Take a scalar_mode rather than a machine_mode.
5590 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5591 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
5592 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5593 (aarch64_gen_adjusted_ldpstp): Likewise.
5594 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
5596 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5597 Alan Hayward <alan.hayward@arm.com>
5598 David Sherwood <david.sherwood@arm.com>
5600 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
5601 Take a scalar_int_mode instead of a machine_mode.
5602 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5603 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5604 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5605 (aarch64_simd_attr_length_rglist): Delete.
5606 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
5607 a scalar_int_mode instead of a machine_mode.
5608 (aarch64_add_offset): Likewise.
5609 (aarch64_internal_mov_immediate): Likewise
5610 (aarch64_add_constant_internal): Likewise.
5611 (aarch64_add_constant): Likewise.
5612 (aarch64_movw_imm): Likewise.
5613 (aarch64_rtx_arith_op_extract_p): Likewise.
5614 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5615 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5616 Remove assert that the mode isn't a vector.
5617 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5618 (aarch64_expand_mov_immediate): Update calls after above changes.
5619 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
5620 (aarch64_and_bitmask_imm): Check for scalar integer modes.
5621 (aarch64_move_imm): Likewise.
5622 (aarch64_can_const_movi_rtx_p): Likewise.
5623 (aarch64_strip_extend): Likewise.
5624 (aarch64_extr_rtx_p): Likewise.
5625 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
5626 a CONST_INT when the mode parameter is VOIDmode.
5627 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
5629 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5631 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
5632 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5633 (bitwise_type_for_mode): Update accordingly.
5635 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5637 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
5638 * stor-layout.c (mode_for_size_tree): Likewise.
5639 (mode_for_array): Update accordingly.
5640 (layout_decl): Likewise.
5641 (compute_record_mode): Likewise. Only set the mode once.
5643 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5645 * target.def (get_mask_mode): Change return type to opt_mode.
5647 * doc/tm.texi: Regenerate.
5648 * targhooks.h (default_get_mask_mode): Return an opt_mode.
5649 * targhooks.c (default_get_mask_mode): Likewise.
5650 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5651 * optabs-query.c (can_vec_mask_load_store_p): Update use of
5652 targetm.get_mask_mode.
5653 * tree.c (build_truth_vector_type): Likewise.
5655 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5657 * machmode.h (mode_for_vector): Return an opt_mode.
5658 * stor-layout.c (mode_for_vector): Likewise.
5659 (mode_for_int_vector): Update accordingly.
5660 (layout_type): Likewise.
5661 * config/i386/i386.c (emit_memmov): Likewise.
5662 (ix86_expand_set_or_movmem): Likewise.
5663 (ix86_expand_vector_init): Likewise.
5664 (ix86_get_mask_mode): Likewise.
5665 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
5667 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
5668 * expmed.c (extract_bit_field_1): Likewise.
5669 * expr.c (expand_expr_real_2): Likewise.
5670 * optabs-query.c (can_vec_perm_p): Likewise.
5671 (can_vec_mask_load_store_p): Likewise.
5672 * optabs.c (expand_vec_perm): Likewise.
5673 * targhooks.c (default_get_mask_mode): Likewise.
5674 * tree-vect-stmts.c (vectorizable_store): Likewise.
5675 (vectorizable_load): Likewise.
5676 (get_vectype_for_scalar_type_and_size): Likewise.
5678 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5680 * machmode.h (mode_for_int_vector): New function.
5681 * stor-layout.c (mode_for_int_vector): Likewise.
5682 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
5683 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
5684 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5685 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5686 (s390_expand_vcond): Likewise.
5688 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5690 * machmode.h (opt_machine_mode): New type.
5691 (opt_mode<T>): Allow construction from anything that can be
5693 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
5694 (mode_for_size): Return an opt_machine_mode.
5695 * stor-layout.c (mode_for_size): Likewise.
5696 (mode_for_size_tree): Update call accordingly.
5697 (bitwise_mode_for_mode): Likewise.
5698 (make_fract_type): Likewise.
5699 (make_accum_type): Likewise.
5700 * caller-save.c (replace_reg_with_saved_mem): Update call
5702 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5703 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5704 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5705 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5706 * expmed.c (extract_bit_field_1): Likewise.
5707 * reload.c (get_secondary_mem): Likewise.
5708 * varasm.c (assemble_integer): Likewise.
5709 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
5712 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5714 * machmode.h (decimal_float_mode_for_size): New function.
5715 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
5716 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
5717 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
5718 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
5719 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
5720 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
5722 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5724 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
5725 (get_builtin_sync_mode): Likewise.
5726 (expand_ifn_atomic_compare_exchange): Likewise.
5727 (expand_builtin_atomic_clear): Likewise.
5728 (expand_builtin_atomic_test_and_set): Likewise.
5729 (fold_builtin_atomic_always_lock_free): Likewise.
5730 * calls.c (compute_argument_addresses): Likewise.
5731 (emit_library_call_value_1): Likewise.
5732 (store_one_arg): Likewise.
5733 * combine.c (combine_instructions): Likewise.
5734 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
5735 * config/arm/arm.c (arm_function_value): Likewise.
5736 (aapcs_allocate_return_reg): Likewise.
5737 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
5738 * config/i386/i386.c (construct_container): Likewise.
5739 (ix86_gimplify_va_arg): Likewise.
5740 (ix86_expand_sse_cmp): Likewise.
5741 (emit_memmov): Likewise.
5742 (emit_memset): Likewise.
5743 (expand_small_movmem_or_setmem): Likewise.
5744 (ix86_expand_pextr): Likewise.
5745 (ix86_expand_pinsr): Likewise.
5746 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
5747 * config/microblaze/microblaze.c (microblaze_block_move_straight):
5749 * config/mips/mips.c (mips_function_value_1) Likewise.
5750 (mips_block_move_straight): Likewise.
5751 (mips_expand_ins_as_unaligned_store): Likewise.
5752 * config/powerpcspe/powerpcspe.c
5753 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5754 (rs6000_darwin64_record_arg_flush): Likewise.
5755 * config/rs6000/rs6000.c
5756 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5757 (rs6000_darwin64_record_arg_flush): Likewise.
5758 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
5759 (sparc_function_value_1): Likewise.
5760 * config/spu/spu.c (adjust_operand): Likewise.
5761 (spu_emit_branch_or_set): Likewise.
5762 (arith_immediate_p): Likewise.
5763 * emit-rtl.c (gen_lowpart_common): Likewise.
5764 * expr.c (expand_expr_real_1): Likewise.
5765 * function.c (assign_parm_setup_block): Likewise.
5766 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5767 * reload1.c (alter_reg): Likewise.
5768 * stor-layout.c (mode_for_vector): Likewise.
5769 (layout_type): Likewise.
5771 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5773 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
5774 (spu_convert_move): Likewise.
5775 * lower-subreg.c (resolve_simple_move): Likewise.
5777 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5780 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
5781 define_insn to a define_expand.
5782 (altivec_vsum2sws_direct): New define_insn.
5783 (altivec_vsumsws): Convert from a define_insn to a define_expand.
5785 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5787 * config/arm/arm.c (arm_option_params_internal): Improve setting of
5790 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
5795 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
5796 realign stack if stack alignment needed is less than incoming
5799 2017-09-05 Marek Polacek <polacek@redhat.com>
5802 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
5805 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5807 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
5809 2017-09-05 Richard Biener <rguenther@suse.de>
5811 PR tree-optimization/82084
5812 * fold-const.c (can_native_encode_string_p): Handle wide characters.
5814 2017-09-05 Richard Biener <rguenther@suse.de>
5816 PR tree-optimization/82102
5817 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
5819 2017-09-05 Martin Liska <mliska@suse.cz>
5821 PR tree-optimization/82032
5822 * tree-cfg.c (generate_range_test): New function.
5823 * tree-cfg.h (generate_range_test): Declared here.
5824 * tree-cfgcleanup.c (convert_single_case_switch): New function.
5825 (cleanup_control_expr_graph): Use it.
5826 * tree-switch-conversion.c (try_switch_expansion): Remove
5828 (emit_case_nodes): Use generate_range_test.
5830 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5833 * config/i386/i386.md (*<btsc><mode>_mask): Add
5834 TARGET_USE_BT to insn constraint.
5835 (*btr<mode>_mask): Ditto.
5837 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
5839 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
5840 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
5842 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5845 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
5846 TARGET_NEON and TARGET_IWMMXT.
5847 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
5848 TARGET_NEON and TARGET_IWMMXT.
5849 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
5851 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5853 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
5854 (ix86_rewrite_tls_address): Ditto.
5855 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
5856 (ix86_rewrite_tls_address_1): Ditto.
5857 (ix86_rewrite_tls_address): Ditto.
5858 * config/i386/predicates.md (tls_address_pattern): New predicate.
5859 * config/i386/i386.md (TLS address splitter): New splitter.
5861 2017-09-04 Richard Biener <rguenther@suse.de>
5863 PR tree-optimization/82084
5864 * fold-const.h (can_native_encode_string_p): Declare.
5865 * fold-const.c (can_native_encode_string_p): Factor out from ...
5866 (native_encode_string): ... here.
5867 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
5868 vectorizing stores from constants we later cannot handle.
5870 2017-09-04 Marek Polacek <polacek@redhat.com>
5873 * doc/invoke.texi: Update -Wtautological-compare documentation.
5875 2017-09-04 Jeff Law <law@redhat.com>
5877 PR tree-optimization/64910
5878 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
5879 swap the first and last operand if the last is a constant.
5881 2017-09-04 Marek Polacek <polacek@redhat.com>
5884 * convert.c (do_narrow): When sanitizing signed integer overflows,
5885 bail out for signed types.
5886 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
5888 2017-09-04 Richard Biener <rguenther@suse.de>
5890 PR tree-optimization/82060
5891 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5892 Move devirtualization after stmt folding and before EH/AB/noreturn
5893 cleanup to get the stmt refs canonicalized. Use a bool instead
5894 of gimple_modified_p since that doesn't work for NOPs. Schedule
5895 NOPs generated by folding for removal.
5897 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5898 Alan Hayward <alan.hayward@arm.com>
5899 David Sherwood <david.sherwood@arm.com>
5901 * coretypes.h (pad_direction): New enum.
5902 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
5903 (FUNCTION_ARG_PADDING): Likewise.
5904 * target.def (function_arg_padding): New hook.
5905 * targhooks.h (default_function_arg_padding): Declare.
5906 * targhooks.c (default_function_arg_padding): New function.
5907 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
5908 (TARGET_FUNCTION_ARG_PADDING): ...this.
5909 * doc/tm.texi: Regenerate.
5910 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
5911 instead of direction.
5912 (compute_argument_addresses): Likewise.
5913 (load_register_parameters): Likewise.
5914 (emit_library_call_value_1): Likewise.
5915 (store_one_arg): Use targetm.calls.function_arg_padding instead
5916 of FUNCTION_ARG_PADDING.
5917 (must_pass_in_stack_var_size_or_pad): Likewise.
5918 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
5919 (emit_group_store): Likewise.
5920 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
5921 instead of FUNCTION_ARG_PADDING.
5922 (emit_push_insn): Likewise, and propagate enum change throughout
5924 * function.h (direction): Delete.
5925 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
5927 * function.c (assign_parm_find_stack_rtl): Likewise.
5928 (assign_parm_setup_block_p): Likewise.
5929 (assign_parm_setup_block): Likewise.
5930 (gimplify_parameters): Likewise.
5931 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
5932 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
5934 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
5935 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5936 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
5937 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
5938 (aarch64_function_arg_padding): ...this new function.
5939 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
5940 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5941 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
5942 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5943 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
5944 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5945 (arm_pad_arg_upward): Replace with...
5946 (arm_function_arg_padding): ...this new function.
5947 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
5949 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
5950 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
5951 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5952 (ia64_hpux_function_arg_padding): Replace with...
5953 (ia64_function_arg_padding): ...this new function. Use pad_direction
5954 instead of direction. Check for TARGET_HPUX.
5955 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
5956 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5957 (iq2000_function_arg_padding): New function.
5958 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
5959 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
5960 (mips_function_arg_padding): ...this new function.
5961 (mips_pad_reg_upward): Update accordingly.
5962 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5963 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
5964 targetm.calls.function_arg_padding.
5965 (FUNCTION_ARG_PADDING): Delete.
5966 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5967 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
5968 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5969 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
5970 (nios2_block_reg_padding): Return pad_direction instead of direction.
5971 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
5972 instead of direction.
5973 (nios2_function_arg_padding): Likewise. Make static.
5974 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5975 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
5976 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
5977 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
5978 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5979 (pa_function_arg_padding): Make static. Return pad_direction instead
5981 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
5982 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5983 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
5984 instead of direction. Use targetm.calls.function_arg_padding.
5985 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
5986 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5987 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
5988 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
5989 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
5991 (function_arg_padding): Rename to...
5992 (rs6000_function_arg_padding): ...this. Make static. Return
5993 pad_direction instead of direction.
5994 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5995 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
5996 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5997 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
5998 instead of direction. Use targetm.calls.function_arg_padding.
5999 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
6000 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6001 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
6002 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
6003 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6004 (function_arg_padding): Rename to...
6005 (rs6000_function_arg_padding): ...this. Make static. Return
6006 pad_direction instead of direction.
6007 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6008 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
6009 * config/s390/s390.c (s390_function_arg_padding): New function.
6010 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6011 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
6012 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
6013 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6014 (function_arg_padding): Rename to...
6015 (sparc_function_arg_padding): ...this. Make static. Return
6016 pad_direction instead of direction.
6017 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
6018 * config/spu/spu.c (spu_function_arg_padding): New function.
6019 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6020 * system.h (FUNCTION_ARG_PADDING): Poison.
6022 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6023 Alan Hayward <alan.hayward@arm.com>
6024 David Sherwood <david.sherwood@arm.com>
6026 * target.def (modes_tieable_p): New hook.
6027 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
6028 (TARGET_MODES_TIEABLE_P): ...this.
6029 * doc/tm.texi.in: Regenerate.
6030 * hooks.h (hook_bool_mode_mode_true): Declare.
6031 * hooks.c (hook_bool_mode_mode_true): New function.
6032 * combine.c (subst): Use targetm.modes_tieable_p instead of
6034 * dse.c (find_shift_sequence): Likewise.
6035 * expmed.c (extract_low_bits): Likewise.
6036 * lower-subreg.c: Include target.h.
6037 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
6039 * rtlanal.c (rtx_cost): Likewise.
6040 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
6041 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
6042 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
6043 (TARGET_MODES_TIEABLE_P): Redefine.
6044 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
6045 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
6046 (TARGET_MODES_TIEABLE_P): Redefine.
6047 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
6048 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
6049 (arc_modes_tieable_p): New function.
6050 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
6051 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
6052 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
6053 (arm_modes_tieable_p): Make static.
6054 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
6055 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
6056 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
6057 (TARGET_MODES_TIEABLE_P): Redefine.
6058 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
6059 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
6060 (TARGET_MODES_TIEABLE_P): Redefine.
6061 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
6062 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
6063 (cr16_modes_tieable_p): New function.
6064 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
6065 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
6066 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
6067 (TRULY_NOOP_TRUNCATION): Update comment.
6068 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
6069 (TRULY_NOOP_TRUNCATION): Update comment.
6070 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
6071 (frv_modes_tieable_p): New function.
6072 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
6073 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
6074 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
6075 (TARGET_MODES_TIEABLE_P): Redefine.
6076 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
6077 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
6078 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
6079 (TARGET_MODES_TIEABLE_P): Redefine.
6080 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
6081 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
6082 (ia64_modes_tieable_p): New function.
6083 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
6084 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
6085 (iq2000_modes_tieable_p): New function.
6086 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
6087 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
6088 (lm32_modes_tieable_p): New function.
6089 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
6090 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
6091 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
6092 (TARGET_MODES_TIEABLE_P): Redefine.
6093 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
6094 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
6095 (m32r_modes_tieable_p): New function.
6096 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6097 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6098 (m68k_modes_tieable_p): New function.
6099 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6100 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6101 (mcore_modes_tieable_p): New function.
6102 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6103 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6105 (TARGET_MODES_TIEABLE_P): Redefine.
6106 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6107 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6108 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6109 (TARGET_MODES_TIEABLE_P): Redefine.
6110 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6111 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6112 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6113 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6114 (mn10300_modes_tieable_p): ...this and make static.
6115 (TARGET_MODES_TIEABLE_P): Redefine.
6116 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6117 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6118 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6119 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6120 (msp430_modes_tieable_p): Make static.
6121 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6122 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6123 (TARGET_MODES_TIEABLE_P): Redefine.
6124 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6125 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6126 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6127 (TARGET_MODES_TIEABLE_P): Redefine.
6128 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6129 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6130 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6131 (TARGET_MODES_TIEABLE_P): Redefine.
6132 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6133 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6134 (pdp11_modes_tieable_p): New function.
6135 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6136 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6137 (rs6000_modes_tieable_p): New function.
6138 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6139 * config/powerpcspe/powerpcspe.md: Update comment.
6140 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6141 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6142 (TARGET_MODES_TIEABLE_P): Redefine.
6143 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6144 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6145 (rl78_modes_tieable_p): New function.
6146 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6147 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6148 (rs6000_modes_tieable_p): New function.
6149 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6150 * config/rs6000/rs6000.md: Update comment.
6151 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6152 * config/rx/rx.c (rx_modes_tieable_p): New function.
6153 (TARGET_MODES_TIEABLE_P): Redefine.
6154 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6155 * config/s390/s390.c (s390_modes_tieable_p): New function.
6156 (TARGET_MODES_TIEABLE_P): Redefine.
6157 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6158 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6159 (sh_modes_tieable_p): New function.
6160 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6161 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6162 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6163 (sparc_modes_tieable_p): Make static.
6164 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6165 * config/spu/spu.c (spu_modes_tieable_p): New function.
6166 (TARGET_MODES_TIEABLE_P): Redefine.
6167 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6168 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6169 (TARGET_MODES_TIEABLE_P): Redefine.
6170 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6171 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6172 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6173 * config/v850/v850.c (v850_modes_tieable_p): New function.
6174 (TARGET_MODES_TIEABLE_P): Redefine.
6175 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
6176 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
6177 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
6178 (visium_modes_tieable_p): New function.
6179 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
6180 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
6181 (xtensa_modes_tieable_p): New function.
6182 * system.h (MODES_TIEABLE_P): Poison.
6184 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6185 Alan Hayward <alan.hayward@arm.com>
6186 David Sherwood <david.sherwood@arm.com>
6188 * target.def (hard_regno_mode_ok): New hook.
6189 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
6190 (TARGET_HARD_REGNO_MODE_OK): ...this.
6191 * doc/tm.texi.in: Regenerate.
6192 * hooks.h (hook_bool_uint_mode_true): Declare.
6193 * hooks.c (hook_bool_uint_mode_true): New function.
6194 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
6196 * genpreds.c (write_insn_preds_c): Add an include of target.h.
6197 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
6198 instead of HARD_REGNO_MODE_OK.
6199 * caller-save.c: Include target.h.
6200 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
6202 * combine.c (can_combine_p): Likewise.
6203 (combinable_i3pat): Likewise.
6204 (can_change_dest_mode): Likewise.
6205 * expr.c (init_expr_target): Likewise.
6206 (convert_move): Likewise.
6207 (convert_modes): Likewise.
6208 * ira.c (setup_prohibited_class_mode_regs): Likewise.
6209 (setup_prohibited_mode_move_regs): Likewise.
6210 * ira.h (target_ira): Likewise.
6211 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6212 * lra-constraints.c (process_alt_operands): Likewise.
6213 (split_reg): Likewise.
6214 * recog.c (peep2_find_free_register): Likewise.
6215 * ree.c (combine_reaching_defs): Likewise.
6216 * regcprop.c (maybe_mode_change): Likewise.
6217 * reginfo.c (init_reg_sets_1): Likewise.
6218 (choose_hard_reg_mode): Likewise.
6219 (simplifiable_subregs): Likewise.
6220 * regrename.c (check_new_reg_p): Likewise.
6221 * reload.c (find_valid_class): Likewise.
6222 (find_valid_class_1): Likewise.
6223 (reload_inner_reg_of_subreg): Likewise.
6224 (push_reload): Likewise.
6225 (combine_reloads): Likewise.
6226 (find_dummy_reload): Likewise.
6227 (find_reloads): Likewise.
6228 * reload1.c (find_reg): Likewise.
6229 (set_reload_reg): Likewise.
6230 (allocate_reload_reg): Likewise.
6231 (choose_reload_regs): Likewise.
6232 (reload_adjust_reg_for_temp): Likewise.
6233 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
6234 (simplify_subreg_regno): Likewise.
6235 * sel-sched.c (init_regs_for_mode): Likewise.
6236 * varasm.c (make_decl_rtl): Likewise.
6237 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
6238 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
6240 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
6241 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
6242 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6243 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
6244 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
6245 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6246 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
6247 (arc_mode_class): Delete.
6248 (HARD_REGNO_MODE_OK): Delete.
6249 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6250 (arc_hard_regno_mode_ok): Rename old array to...
6251 (arc_hard_regno_mode_ok_modes): ...this.
6252 (arc_conditional_register_usage): Update accordingly.
6253 (arc_mode_class): Make static.
6254 (arc_hard_regno_mode_ok): New function.
6255 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
6256 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
6257 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6258 (arm_hard_regno_mode_ok): Make static.
6259 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
6261 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
6262 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
6263 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
6265 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6266 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
6267 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
6268 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
6269 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
6270 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6271 * config/bfin/predicates.md (valid_reg_operand): Use
6272 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
6273 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
6274 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
6275 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6276 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
6277 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
6278 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6279 (cr16_hard_regno_mode_ok): Make static and return a bool.
6280 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
6281 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6282 (cris_hard_regno_mode_ok): New function.
6283 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
6284 (epiphany_mode_class): Delete.
6285 (HARD_REGNO_MODE_OK): Delete.
6286 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
6287 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6288 (hard_regno_mode_ok): Rename to...
6289 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
6290 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
6291 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
6293 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
6294 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
6295 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6296 (frv_hard_regno_mode_ok): Make static and return a bool.
6297 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
6299 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
6300 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
6301 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
6302 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
6304 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6305 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
6306 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
6307 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
6309 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6310 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
6311 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6312 (ia64_hard_regno_mode_ok): New function.
6313 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
6314 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6315 (iq2000_hard_regno_mode_ok): New function.
6316 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
6317 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6318 (lm32_hard_regno_mode_ok): New function.
6319 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
6320 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
6321 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
6322 instead of HARD_REGNO_MODE_OK.
6323 (m32c_hard_regno_ok): Rename to...
6324 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
6325 (m32c_cannot_change_mode_class): Update accordingly.
6326 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6327 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
6328 (m32r_mode_class): Delete.
6329 (HARD_REGNO_MODE_OK): Delete.
6330 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6331 (m32r_hard_regno_mode_ok): Rename to...
6332 (m32r_hard_regno_modes): ...this.
6333 (m32r_mode_class): Make static.
6334 (m32r_hard_regno_mode_ok): New function.
6335 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
6336 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
6337 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6338 (m68k_hard_regno_mode_ok): Make static.
6339 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
6340 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6341 (mcore_hard_regno_mode_ok): New function.
6342 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
6343 (HARD_REGNO_MODE_OK): Delete.
6344 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
6346 (microblaze_hard_regno_mode_ok_p): ...this and make static.
6347 (microblaze_hard_regno_mode_ok): New function.
6348 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6349 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
6350 (mips_hard_regno_mode_ok): Delete.
6351 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
6352 (mips_hard_regno_mode_ok_p): ...this and make static.
6353 (mips_hard_regno_mode_ok_p): Rename to...
6354 (mips_hard_regno_mode_ok_uncached): ...this.
6355 (mips_hard_regno_mode_ok): New function.
6356 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
6357 of HARD_REGNO_MODE_OK.
6358 (mips_option_override): Update after above name changes.
6359 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6360 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
6361 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
6362 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
6363 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
6364 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6365 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
6366 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
6367 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
6368 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6369 (msp430_hard_regno_mode_ok): Make static and return a bool.
6370 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
6371 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
6372 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
6374 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6375 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
6376 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
6377 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
6378 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
6379 (PA_HARD_REGNO_MODE_OK): ...this
6380 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
6381 (PA_HARD_REGNO_MODE_OK): ...this.
6382 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6383 (pa_hard_regno_mode_ok): New function.
6384 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
6385 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6386 (pdp11_hard_regno_mode_ok): New function.
6387 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
6388 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
6390 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
6392 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6393 (rs6000_hard_regno_mode_ok): Rename to...
6394 (rs6000_hard_regno_mode_ok_uncached): ...this.
6395 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6396 (rs6000_hard_regno_mode_ok): New function.
6397 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
6398 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
6399 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
6400 (riscv_hard_regno_mode_ok): ...this and make static.
6401 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6402 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
6403 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
6404 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6405 (rl78_hard_regno_mode_ok): Make static and return bool.
6406 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
6407 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
6409 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
6410 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6411 (rs6000_hard_regno_mode_ok): Rename to...
6412 (rs6000_hard_regno_mode_ok_uncached): ...this.
6413 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6414 (rs6000_hard_regno_mode_ok): New function.
6415 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
6416 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
6417 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6418 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
6419 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
6420 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
6421 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6422 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
6423 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
6424 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6425 (sh_hard_regno_mode_ok): Make static.
6426 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
6427 instead of HARD_REGNO_MODE_OK.
6428 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
6429 (sparc_mode_class): Delete.
6430 (HARD_REGNO_MODE_OK): Delete.
6431 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6432 (hard_regno_mode_classes): Make static.
6433 (sparc_mode_class): Likewise.
6434 (sparc_hard_regno_mode_ok): New function.
6435 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
6436 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
6437 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
6439 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6440 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
6441 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
6442 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
6443 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
6444 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6445 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
6446 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
6447 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6448 (visium_hard_regno_mode_ok): New function.
6449 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
6450 instead of HARD_REGNO_MODE_OK.
6451 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
6452 (HARD_REGNO_MODE_OK): Delete.
6453 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
6454 (xtensa_hard_regno_mode_ok_p): ...this and make static.
6455 (xtensa_option_override): Update accordingly.
6456 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6457 (xtensa_hard_regno_mode_ok): New function.
6458 * system.h (HARD_REGNO_MODE_OK): Poison.
6460 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6461 Alan Hayward <alan.hayward@arm.com>
6462 David Sherwood <david.sherwood@arm.com>
6464 * target.def (hard_regno_call_part_clobbered): New hook.
6465 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
6466 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
6467 * doc/tm.texi: Regenerate.
6468 * hooks.h (hook_bool_uint_mode_false): Declare.
6469 * hooks.c (hook_bool_uint_mode_false): New function.
6470 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6471 * cselib.c (cselib_process_insn): Use
6472 targetm.hard_regno_call_part_clobbered instead of
6473 HARD_REGNO_CALL_PART_CLOBBERED.
6474 * ira-conflicts.c (ira_build_conflicts): Likewise.
6475 * ira-costs.c (ira_tune_allocno_costs): Likewise.
6476 * lra-constraints.c (need_for_call_save_p): Likewise.
6477 * lra-lives.c: Include target.h.
6478 (check_pseudos_live_through_calls): Use
6479 targetm.hard_regno_call_part_clobbered instead of
6480 HARD_REGNO_CALL_PART_CLOBBERED.
6481 * regcprop.c: Include target.h.
6482 (copyprop_hardreg_forward_1): Use
6483 targetm.hard_regno_call_part_clobbered instead of
6484 HARD_REGNO_CALL_PART_CLOBBERED.
6485 * reginfo.c (choose_hard_reg_mode): Likewise.
6486 * regrename.c (check_new_reg_p): Likewise.
6487 * reload.c (find_equiv_reg): Likewise.
6488 * reload1.c (emit_reload_insns): Likewise.
6489 * sched-deps.c (deps_analyze_insn): Likewise.
6490 * sel-sched.c (init_regs_for_mode): Likewise.
6491 (mark_unavailable_hard_regs): Likewise.
6492 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
6493 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6494 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
6496 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6497 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6498 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
6500 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
6502 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6503 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6504 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
6506 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6507 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6508 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
6510 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6511 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
6513 * config/powerpcspe/powerpcspe.c
6514 (rs6000_hard_regno_call_part_clobbered): New function.
6515 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6516 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6517 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
6519 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6520 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6521 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
6523 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6524 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6525 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
6527 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6528 Alan Hayward <alan.hayward@arm.com>
6529 David Sherwood <david.sherwood@arm.com>
6531 * rtl.h (subreg_memory_offset): Declare.
6532 * emit-rtl.c (subreg_memory_offset): New function.
6533 * expmed.c (store_bit_field_1): Use it.
6534 * expr.c (undefined_operand_subword_p): Likewise.
6535 * simplify-rtx.c (simplify_subreg): Likewise.
6537 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
6539 PR rtl-optimization/57448
6542 * optabs.c (expand_atomic_load): Place compiler memory barriers if
6543 using atomic_load pattern.
6544 (expand_atomic_store): Likewise.
6546 2017-09-04 Jakub Jelinek <jakub@redhat.com>
6549 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
6550 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
6551 handling. Use replace_call_with_value with NULL instead of
6552 gsi_replace, unlink_stmt_vdef and release_defs.
6554 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
6557 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
6559 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6562 * rtl.h (emit_library_call_value_1): Declare.
6563 (emit_library_call): Replace declaration with a series of overloads.
6564 Remove the parameter count argument.
6565 (emit_library_call_value): Likewise.
6566 * calls.c (emit_library_call_value_1): Make global. Replace varargs
6567 with an "rtx_mode_t *".
6568 (emit_library_call_value): Delete.
6569 (emit_library_call): Likewise.
6570 * asan.c (asan_emit_stack_protection): Update calls accordingly.
6571 (asan_emit_allocas_unpoison): Likewise.
6572 * builtins.c (expand_builtin_powi): Likewise.
6573 (expand_asan_emit_allocas_unpoison): Likewise.
6574 * cfgexpand.c (expand_main_function): Likewise.
6575 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
6576 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
6577 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
6578 * config/arm/arm.c (arm_trampoline_init): Likewise.
6579 (arm_call_tls_get_addr): Likewise.
6580 (arm_expand_divmod_libfunc): Likewise.
6581 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
6582 (smulsi3_highpart): Likewise.
6583 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
6584 (c6x_expand_compare): Likewise.
6585 (c6x_expand_movmem): Likewise.
6586 * config/frv/frv.c (frv_trampoline_init): Likewise.
6587 * config/i386/i386.c (ix86_trampoline_init): Likewise.
6588 (ix86_expand_divmod_libfunc): Likewise.
6589 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
6590 (ia64_expand_compare): Likewise.
6591 (ia64_profile_hook): Likewise.
6592 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
6593 (nonlocal_goto): Likewise.
6594 (restore_stack_nonlocal): Likewise.
6595 * config/m32r/m32r.c (block_move_call): Likewise.
6596 (m32r_trampoline_init): Likewise.
6597 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
6598 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
6599 (m68k_call_m68k_read_tp): Likewise.
6600 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
6601 (microblaze_expand_divide): Likewise.
6602 * config/mips/mips.h (mips_args): Likewise.
6603 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
6604 (MIPS_ICACHE_SYNC): Likewise.
6605 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
6606 (nios2_trampoline_init): Likewise.
6607 * config/pa/pa.c (hppa_tls_call): Likewise.
6608 (pa_trampoline_init): Likewise.
6609 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
6610 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
6611 (expand_strn_compare): Likewise.
6612 (rs6000_generate_compare): Likewise.
6613 (rs6000_expand_float128_convert): Likewise.
6614 (output_profile_hook): Likewise.
6615 (rs6000_trampoline_init): Likewise.
6616 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
6617 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
6618 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
6619 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
6620 (rs6000_generate_compare): Likewise.
6621 (rs6000_expand_float128_convert): Likewise.
6622 (output_profile_hook): Likewise.
6623 (rs6000_trampoline_init): Likewise.
6624 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
6625 * config/sh/sh.c (sh_trampoline_init): Likewise.
6626 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
6627 (sparc_emit_float_lib_cmp): Likewise.
6628 (sparc32_initialize_trampoline): Likewise.
6629 (sparc64_initialize_trampoline): Likewise.
6630 (sparc_profile_hook): Likewise.
6631 * config/spu/spu.c (ea_load_store): Likewise.
6632 * config/spu/spu.md (floatunssidf2): Likewise.
6633 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
6634 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
6635 * config/visium/visium.c (expand_block_move_4): Likewise.
6636 (expand_block_move_2): Likewise.
6637 (expand_block_move_1): Likewise.
6638 (expand_block_set_4): Likewise.
6639 (expand_block_set_2): Likewise.
6640 (expand_block_set_1): Likewise.
6641 (visium_trampoline_init): Likewise.
6642 (visium_profile_hook): Likewise.
6643 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
6644 (xtensa_setup_frame_addresses): Likewise.
6645 (xtensa_trampoline_init): Likewise.
6646 * except.c (sjlj_emit_function_enter): Likewise.
6647 (sjlj_emit_function_exit): Likewise.
6648 * explow.c (allocate_dynamic_stack_space): Likewise.
6649 (probe_stack_range): Likewise.
6650 * expr.c (convert_mode_scalar): Likewise.
6651 * optabs.c (expand_binop): Likewise.
6652 (expand_twoval_binop_libfunc): Likewise.
6653 (expand_unop): Likewise.
6654 (prepare_cmp_insn): Likewise.
6655 (prepare_float_lib_cmp): Likewise.
6656 (expand_float): Likewise.
6657 (expand_fix): Likewise.
6658 (expand_fixed_convert): Likewise.
6659 (maybe_emit_sync_lock_test_and_set): Likewise.
6660 (expand_atomic_compare_and_swap): Likewise.
6661 (expand_mem_thread_fence): Likewise.
6662 (expand_atomic_fetch_op): Likewise.
6664 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
6666 * doc/generic.texi (OpenACC): Adjust URL.
6667 * doc/invoke.texi (C Dialect Options): Ditto.
6669 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
6671 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
6672 predicate for operand 1. Add (m,<S>) constraint.
6673 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
6674 Prevent memory operand 1 with register operand 2.
6676 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
6678 PR rtl-optimization/82024
6679 * combine.c (try_combine): If the combination result is a PARALLEL,
6680 and we only need to retain the SET in there that would be placed
6681 at I2, check that we can place that at I3 instead, before doing so.
6683 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6686 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
6688 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
6689 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
6692 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
6693 Jeff Law <law@redhat.com>
6695 * varasm.c (bss_initializer_p): Do not put constants into .bss
6696 (categorize_decl_for_section): Handle bss_initializer_p returning
6697 false when DECL_INITIAL is NULL.
6699 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6702 * config/s390/s390.c (s390_can_inline_p): New function.
6704 2017-09-01 Jeff Law <law@redhat.com>
6706 PR tree-optimization/82052
6707 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
6708 Always initialize the returned slot after a hash table miss
6709 when INSERT is true.
6711 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
6713 * config/s390/s390.md (mem_signal_fence): Remove.
6714 * doc/md.texi (mem_signal_fence): Remove.
6715 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
6717 * target-insns.def (mem_signal_fence): Remove.
6719 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6722 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
6725 * asan.c (create_odr_indicator): Strip name encoding from assembler
6726 name before appending it after __odr_asan_.
6728 2017-09-01 Martin Liska <mliska@suse.cz>
6730 PR tree-optimization/82059
6731 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
6732 frequency only when an edge is redirected.
6734 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6736 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
6737 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
6738 (arc_conditional_register_usage): Remove ARC600 lp_count
6740 (arc_file_start): Emit Tag_ARC_CPU_variation.
6741 (arc_can_use_doloop_p): New conditions to use ZOLs.
6742 (hwloop_fail): New function.
6743 (hwloop_optimize): Likewise.
6744 (hwloop_pattern_reg): Likewise.
6745 (arc_doloop_hooks): New struct, to be used with reorg_loops.
6746 (arc_reorg_loops): New function, calls reorg_loops.
6747 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
6748 (arc600_corereg_hazard): Remove ZOL checking, case handled by
6750 (arc_loop_hazard): Remove function, functionality moved into
6752 (arc_hazard): Remove arc_loop_hazard call.
6753 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
6754 into hwloop_optimize.
6755 (arc_label_align): Remove ZOL handling.
6756 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
6757 * config/arc/arc.md (doloop_begin): Remove pattern.
6758 (doloop_begin_i): Likewise.
6759 (doloop_end_i): Likewise.
6760 (doloop_fallback): Likewise.
6761 (doloop_fallback_m): Likewise.
6762 (doloop_end): Reimplement expand.
6763 (arc_lp): New pattern for LP instruction.
6764 (loop_end): New pattern.
6765 (loop_fail): Likewise.
6766 (decrement_and_branch_until_zero): Likewise.
6767 * config/arc/arc.opt (mlpc-width): New option.
6768 * doc/invoke.texi (mlpc-width): Document option.
6770 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6772 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
6773 (arc_ccfsm_advance): Fix checking for delay slots.
6774 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
6776 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6778 * config/arc/arc.md (movqi_insn): Add stores to save constant long
6780 (movhi_insn): Update store instruction constraint which are saving
6781 6-bit short immediates.
6782 (movsi_insn): Consider also short scaled load operations.
6783 (zero_extendhisi2_i): Use Usd constraint instead of T.
6784 (extendhisi2_i): Add q constraint.
6785 (arc_clzsi2): Add type and length attributes.
6786 (arc_ctzsi2): Likewise.
6787 * config/arc/constraints.md (Usc): Update constraint, the
6788 assembler can parse two relocations for a single instruction.
6790 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6792 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
6793 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
6795 2017-08-31 Olivier Hainque <hainque@adacore.com>
6797 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
6798 match as powerpc-wrs-vxworks*.
6800 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
6802 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
6803 register constraint for by-element operand.
6804 (aarch64_mls_elt_merge<mode>): Likewise.
6806 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6808 * config/arc/arc.c (arc_can_follow_jump): Check for short
6811 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6813 * config.gcc: Use g.opt for arc.
6814 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
6815 functionality moved to ...
6816 (legitimate_scaled_address_p): New function, ...here.
6817 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
6818 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
6819 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
6821 (arc_override_options): Handle G option.
6822 (arc_output_pic_addr_const): Correct function definition.
6823 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
6824 (arc_decl_anon_ns_mem_p): Delete.
6825 (arc_in_small_data_p): Overhaul this function to take into
6826 consideration the value given via G option.
6827 (arc_rewrite_small_data_1): Renamed and corrected old
6828 arc_rewrite_small_data function.
6829 (arc_rewrite_small_data): New function.
6830 (small_data_pattern): Don't use pic_offset_table_rtx.
6831 * config/arc/arc.h (CC1_SPEC): Recognize G option.
6832 * config/arc/simdext.md (movmisalignv2hi): Use
6833 prepare_move_operands function.
6835 (movmisalign*): Likewise.
6836 * doc/invoke.texi (ARC options): Document -G option.
6838 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6840 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
6842 * config/arc/arc.c (arc_print_operand): Output scalled address for
6843 sdata whenever is possible.
6844 (arc_in_small_data_p): Allow sdata for 64bit datum when double
6845 load/stores are available.
6846 (compact_sda_memory_operand): Check for the alignment required by
6847 code density instructions.
6848 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
6850 * config/arc/constraints.md (Usd): Update constraint.
6851 (Us0): New constraint.
6852 (Usc): Update constraint.
6854 2017-08-31 Richard Biener <rguenther@suse.de>
6857 * dwarf2out.c (dwarf2out_early_global_decl): Process each
6860 2017-08-31 Tamar Christina <tamar.christina@arm.com>
6862 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
6863 Resize type_signature.
6865 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6866 Alan Hayward <alan.hayward@arm.com>
6867 David Sherwood <david.sherwood@arm.com>
6869 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
6870 subregs whose inner modes can be stored in GPRs.
6871 (aarch64_classify_index): Likewise.
6873 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6874 Alan Hayward <alan.hayward@arm.com>
6875 David Sherwood <david.sherwood@arm.com>
6877 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6878 (V_INT_EQUIV): ...this.
6879 (v_cmp_result): Rename to...
6880 (v_int_equiv): ...this.
6881 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
6882 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
6883 (copysign<mode>3): Likewise.
6884 (aarch64_simd_bsl<mode>_internal): Likewise.
6885 (aarch64_simd_bsl<mode>): Likewise.
6886 (vec_cmp<mode><mode>): Likewise.
6887 (vcond<mode><mode>): Likewise.
6888 (vcond<v_cmp_mixed><mode>): Likewise.
6889 (vcondu<mode><v_cmp_mixed>): Likewise.
6890 (aarch64_cm<optab><mode>): Likewise.
6891 (aarch64_cmtst<mode>): Likewise.
6892 (aarch64_fac<optab><mode>): Likewise.
6893 (vec_perm_const<mode>): Likewise.
6894 (vcond_mask_<mode><v_cmp_result>): Rename to...
6895 (vcond_mask_<mode><v_int_equiv>): ...this.
6896 (vec_cmp<mode><v_cmp_result>): Rename to...
6897 (vec_cmp<mode><v_int_equiv>): ...this.
6899 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6900 Alan Hayward <alan.hayward@arm.com>
6901 David Sherwood <david.sherwood@arm.com>
6903 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
6905 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
6906 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
6907 (UNSPEC_LD4_DREG): New unspecs.
6908 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
6909 (aarch64_ld2<mode>_dreg_be): Replace with...
6910 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
6912 (aarch64_ld3<mode>_dreg_le)
6913 (aarch64_ld3<mode>_dreg_be): Replace with...
6914 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
6916 (aarch64_ld4<mode>_dreg_le)
6917 (aarch64_ld4<mode>_dreg_be): Replace with...
6918 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
6921 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6923 PR tree-optimization/81987
6924 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
6925 insert an initializer in a location not dominated by the stride
6928 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6930 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
6931 on the entire header of the finally block in the fallthru case.
6933 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6935 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
6937 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
6939 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
6940 rs6000_emit_move_from_cr and call renamed function.
6941 (rs6000_emit_prologue): Call renamed functions.
6942 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
6943 movesi_from_cr, remove volatile CRs.
6945 2017-08-30 Jon Beniston <jon@beniston.com>
6946 Richard Biener <rguenther@suse.de>
6948 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
6949 of VECTOR_MODE_P check.
6950 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
6951 element vector types.
6953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6955 * df.h (df_read_modify_subreg_p): Remove in favor of...
6956 * rtl.h (read_modify_subreg_p): ...this new function. Take a
6957 const_rtx instead of an rtx.
6958 * cprop.c (local_cprop_find_used_regs): Update accordingly.
6959 * df-problems.c (df_word_lr_mark_ref): Likewise.
6960 * ira-lives.c (mark_pseudo_reg_live): Likewise.
6961 (mark_pseudo_reg_dead): Likewise.
6962 (mark_ref_dead): Likewise.
6963 * reginfo.c (init_subregs_of_mode): Likewise.
6964 * sched-deps.c (sched_analyze_1): Likewise.
6965 * df-scan.c (df_def_record_1): Likewise.
6966 (df_uses_record): Likewise.
6967 (df_read_modify_subreg_p): Remove in favor of...
6968 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
6969 const_rtx instead of an rtx.
6971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6972 Alan Hayward <alan.hayward@arm.com>
6973 David Sherwood <david.sherwood@arm.com>
6975 * rtl.h (partial_subreg_p): New function.
6976 * caller-save.c (save_call_clobbered_regs): Use it.
6977 * calls.c (expand_call): Likewise.
6978 * combine.c (combinable_i3pat): Likewise.
6979 (simplify_set): Likewise.
6980 (make_extraction): Likewise.
6981 (make_compound_operation_int): Likewise.
6982 (gen_lowpart_or_truncate): Likewise.
6983 (force_to_mode): Likewise.
6984 (make_field_assignment): Likewise.
6985 (reg_truncated_to_mode): Likewise.
6986 (record_truncated_value): Likewise.
6987 (move_deaths): Likewise.
6988 * cse.c (record_jump_cond): Likewise.
6989 (cse_insn): Likewise.
6990 * cselib.c (cselib_lookup_1): Likewise.
6991 * expmed.c (extract_bit_field_using_extv): Likewise.
6992 * function.c (assign_parm_setup_reg): Likewise.
6993 * ifcvt.c (noce_convert_multiple_sets): Likewise.
6994 * ira-build.c (create_insn_allocnos): Likewise.
6995 * lra-coalesce.c (merge_pseudos): Likewise.
6996 * lra-constraints.c (match_reload): Likewise.
6997 (simplify_operand_subreg): Likewise.
6998 (curr_insn_transform): Likewise.
6999 * lra-lives.c (process_bb_lives): Likewise.
7000 * lra.c (new_insn_reg): Likewise.
7001 (lra_substitute_pseudo): Likewise.
7002 * regcprop.c (mode_change_ok): Likewise.
7003 (maybe_mode_change): Likewise.
7004 (copyprop_hardreg_forward_1): Likewise.
7005 * reload.c (push_reload): Likewise.
7006 (find_reloads): Likewise.
7007 (find_reloads_subreg_address): Likewise.
7008 * reload1.c (alter_reg): Likewise.
7009 (eliminate_regs_1): Likewise.
7010 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7012 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
7014 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
7015 back to if statements, including unpack.
7017 2017-08-30 Martin Liska <mliska@suse.cz>
7020 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
7022 (func_checker::compare_asm_inputs_outputs): ... this function.
7023 (func_checker::compare_gimple_asm): Use the function to compare
7024 also ASM constrains.
7025 * ipa-icf-gimple.h: Rename the function.
7027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7028 Alan Hayward <alan.hayward@arm.com>
7029 David Sherwood <david.sherwood@arm.com>
7031 * coretypes.h (complex_mode): New type.
7032 * gdbhooks.py (build_pretty_printer): Handle it.
7033 * machmode.h (complex_mode): New class.
7034 (complex_mode::includes_p): New function.
7035 (is_complex_int_mode): Likewise.
7036 (is_complex_float_mode): Likewise.
7037 * genmodes.c (get_mode_class): Handle complex mode classes.
7038 * function.c (expand_function_end): Use is_complex_int_mode.
7040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7041 Alan Hayward <alan.hayward@arm.com>
7042 David Sherwood <david.sherwood@arm.com>
7044 * coretypes.h (scalar_mode_pod): New typedef.
7045 * gdbhooks.py (build_pretty_printer): Handle it.
7046 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
7047 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
7048 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
7049 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
7052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7053 Alan Hayward <alan.hayward@arm.com>
7054 David Sherwood <david.sherwood@arm.com>
7056 * machmode.h (mode_for_vector): Take a scalar_mode instead
7058 * stor-layout.c (mode_for_vector): Likewise.
7059 * explow.c (promote_mode): Use as_a <scalar_mode>.
7060 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
7062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7063 Alan Hayward <alan.hayward@arm.com>
7064 David Sherwood <david.sherwood@arm.com>
7066 * target.def (preferred_simd_mode): Take a scalar_mode
7067 instead of a machine_mode.
7068 * targhooks.h (default_preferred_simd_mode): Likewise.
7069 * targhooks.c (default_preferred_simd_mode): Likewise.
7070 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
7071 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
7072 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
7073 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
7074 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
7075 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
7076 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
7077 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
7079 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
7080 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
7081 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
7082 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
7083 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
7084 * doc/tm.texi: Regenerate.
7085 * optabs-query.c (can_vec_mask_load_store_p): Return false for
7088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7089 Alan Hayward <alan.hayward@arm.com>
7090 David Sherwood <david.sherwood@arm.com>
7092 * target.def (scalar_mode_supported_p): Take a scalar_mode
7093 instead of a machine_mode.
7094 * targhooks.h (default_scalar_mode_supported_p): Likewise.
7095 * targhooks.c (default_scalar_mode_supported_p): Likewise.
7096 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7097 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7098 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7099 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7100 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7101 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7102 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7103 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7104 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7105 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7106 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7107 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7109 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7110 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7111 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7112 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7113 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7115 * doc/tm.texi: Regenerate.
7117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7118 Alan Hayward <alan.hayward@arm.com>
7119 David Sherwood <david.sherwood@arm.com>
7121 * coretypes.h (opt_scalar_mode): New typedef.
7122 * gdbhooks.py (build_pretty_printers): Handle it.
7123 * machmode.h (mode_iterator::get_2xwider): Add overload for
7125 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7127 * expr.c (convert_mode_scalar): Likewise.
7128 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7129 * optabs.c (expand_float): Likewise.
7130 (expand_fix): Likewise.
7131 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7134 Alan Hayward <alan.hayward@arm.com>
7135 David Sherwood <david.sherwood@arm.com>
7137 * optabs.c (expand_float): Explicitly check for scalars before
7138 using a branching expansion.
7139 (expand_fix): Likewise.
7141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7142 Alan Hayward <alan.hayward@arm.com>
7143 David Sherwood <david.sherwood@arm.com>
7145 * expr.c (convert_mode): Split scalar handling out into...
7146 (convert_mode_scalar): ...this new function. Treat the modes
7149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7150 Alan Hayward <alan.hayward@arm.com>
7151 David Sherwood <david.sherwood@arm.com>
7153 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7155 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7158 Alan Hayward <alan.hayward@arm.com>
7159 David Sherwood <david.sherwood@arm.com>
7161 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7162 rather than a machine_mode.
7163 (fixed_from_string): Likewise.
7164 (fixed_convert): Likewise.
7165 (fixed_convert_from_int): Likewise.
7166 (fixed_convert_from_real): Likewise.
7167 (real_convert_from_fixed): Likewise.
7168 * fixed-value.c (fixed_from_double_int): Likewise.
7169 (fixed_from_string): Likewise.
7170 (fixed_convert): Likewise.
7171 (fixed_convert_from_int): Likewise.
7172 (fixed_convert_from_real): Likewise.
7173 (real_convert_from_fixed): Likewise.
7174 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
7176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7177 Alan Hayward <alan.hayward@arm.com>
7178 David Sherwood <david.sherwood@arm.com>
7180 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
7181 of separate mode class checks. Do not allow vector modes here.
7182 (immed_wide_int_const): Use as_a <scalar_mode>.
7183 * explow.c (trunc_int_for_mode): Likewise.
7184 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
7185 (wi::shwi): Likewise.
7186 (wi::min_value): Likewise.
7187 (wi::max_value): Likewise.
7188 * dwarf2out.c (loc_descriptor): Likewise.
7189 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
7192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7193 Alan Hayward <alan.hayward@arm.com>
7194 David Sherwood <david.sherwood@arm.com>
7196 * tree.h (SCALAR_TYPE_MODE): New macro.
7197 * expr.c (expand_expr_addr_expr_1): Use it.
7198 (expand_expr_real_2): Likewise.
7199 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
7200 (fold_convert_const_fixed_from_int): Likewise.
7201 (fold_convert_const_fixed_from_real): Likewise.
7202 (native_encode_fixed): Likewise
7203 (native_encode_complex): Likewise
7204 (native_encode_vector): Likewise.
7205 (native_interpret_fixed): Likewise.
7206 (native_interpret_real): Likewise.
7207 (native_interpret_complex): Likewise.
7208 (native_interpret_vector): Likewise.
7209 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
7210 (simd_clone_adjust_argument_types): Likewise.
7211 (simd_clone_init_simd_arrays): Likewise.
7212 (simd_clone_adjust): Likewise.
7213 * stor-layout.c (layout_type): Likewise.
7214 * tree.c (build_minus_one_cst): Likewise.
7215 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7216 * tree-inline.c (estimate_move_cost): Likewise.
7217 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
7218 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
7219 (vectorizable_reduction): Likewise.
7220 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
7221 (vect_recog_mixed_size_cond_pattern): Likewise.
7222 (check_bool_pattern): Likewise.
7223 (adjust_bool_pattern): Likewise.
7224 (search_type_for_mask_1): Likewise.
7225 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7226 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7227 (vectorizable_load): Likewise.
7228 (vectorizable_store): Likewise.
7229 * ubsan.c (ubsan_encode_value): Likewise.
7230 * varasm.c (output_constant): Likewise.
7232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7233 Alan Hayward <alan.hayward@arm.com>
7234 David Sherwood <david.sherwood@arm.com>
7236 * coretypes.h (scalar_mode): New class.
7237 * machmode.h (scalar_mode): Likewise.
7238 (scalar_mode::includes_p): New function.
7239 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
7240 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
7241 * genmodes.c (get_mode_class): Handle remaining scalar modes.
7242 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
7243 * expmed.c (store_bit_field_1): Likewise.
7244 (extract_bit_field_1): Likewise.
7245 * expr.c (write_complex_part): Likewise.
7246 (read_complex_part): Likewise.
7247 (emit_move_complex_push): Likewise.
7248 (expand_expr_real_2): Likewise.
7249 * function.c (assign_parm_setup_reg): Likewise.
7250 (assign_parms_unsplit_complex): Likewise.
7251 * optabs.c (expand_binop): Likewise.
7252 * rtlanal.c (subreg_get_info): Likewise.
7253 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7254 * varasm.c (output_constant_pool_2): Likewise.
7256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7257 Alan Hayward <alan.hayward@arm.com>
7258 David Sherwood <david.sherwood@arm.com>
7260 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7262 (expmed_mult_highpart_optab): Likewise.
7263 (expmed_mult_highpart): Likewise.
7265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7266 Alan Hayward <alan.hayward@arm.com>
7267 David Sherwood <david.sherwood@arm.com>
7269 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
7270 instead of a machine_mode.
7271 (builtin_memset_read_str): Likewise.
7272 * builtins.c (c_readstr): Likewise.
7273 (builtin_memcpy_read_str): Likewise.
7274 (builtin_strncpy_read_str): Likewise.
7275 (builtin_memset_read_str): Likewise.
7276 (builtin_memset_gen_str): Likewise.
7277 (expand_builtin_signbit): Use scalar_int_mode for local variables.
7278 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
7279 instead of a machine_mode.
7280 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
7282 (make_extraction): Likewise.
7283 (try_widen_shift_mode): Take and return scalar_int_modes instead
7285 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
7286 a scalar_int_mode instead of a machine_mode.
7287 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
7288 (avr_addr_space_pointer_mode): Likewise.
7289 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
7290 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
7291 (msp430_unwind_word_mode): Likewise.
7292 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
7293 (spu_addr_space_pointer_mode): Likewise.
7294 (spu_addr_space_address_mode): Likewise.
7295 (spu_libgcc_cmp_return_mode): Likewise.
7296 (spu_libgcc_shift_count_mode): Likewise.
7297 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
7298 (rl78_addr_space_pointer_mode): Likewise.
7299 (fl78_unwind_word_mode): Likewise.
7300 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
7302 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
7303 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
7304 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
7305 (mips_valid_pointer_mode): Likewise.
7306 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
7307 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
7308 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
7310 (ft32_addr_space_address_mode): Likewise.
7311 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
7312 scalar_int_mode instead of a machine_mode.
7313 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
7315 (m32c_addr_space_address_mode): Likewise.
7316 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
7317 (rs6000_eh_return_filter_mode): Likewise.
7318 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
7319 (rs6000_eh_return_filter_mode): Likewise.
7320 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
7321 (s390_libgcc_shift_count_mode): Likewise.
7322 (s390_unwind_word_mode): Likewise.
7323 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
7325 * target.def (mode_rep_extended): Likewise.
7326 (valid_pointer_mode): Likewise.
7327 (addr_space.valid_pointer_mode): Likewise.
7328 (eh_return_filter_mode): Return a scalar_int_mode rather than
7330 (libgcc_cmp_return_mode): Likewise.
7331 (libgcc_shift_count_mode): Likewise.
7332 (unwind_word_mode): Likewise.
7333 (addr_space.pointer_mode): Likewise.
7334 (addr_space.address_mode): Likewise.
7335 * doc/tm.texi: Regenerate.
7336 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
7338 (do_jump): Use scalar_int_mode for local variables.
7339 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
7340 rather than a machine_mode.
7341 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
7342 (scompare_loc_descriptor_wide): Likewise.
7343 (scompare_loc_descriptor_narrow): Likewise.
7344 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
7346 * except.c (sjlj_emit_dispatch_table): Likewise.
7347 (expand_builtin_eh_copy_values): Likewise.
7348 * explow.c (convert_memory_address_addr_space_1): Likewise.
7349 Take a scalar_int_mode rather than a machine_mode.
7350 (convert_memory_address_addr_space): Take a scalar_int_mode rather
7351 than a machine_mode.
7352 (memory_address_addr_space): Use scalar_int_mode for local variables.
7353 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
7354 rather than a machine_mode.
7355 * expmed.c (mask_rtx): Likewise.
7356 (init_expmed_one_conv): Likewise.
7357 (expand_mult_highpart_adjust): Likewise.
7358 (extract_high_half): Likewise.
7359 (expmed_mult_highpart_optab): Likewise.
7360 (expmed_mult_highpart): Likewise.
7361 (expand_smod_pow2): Likewise.
7362 (expand_sdiv_pow2): Likewise.
7363 (emit_store_flag_int): Likewise.
7364 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
7366 (extract_low_bits): Likewise.
7367 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
7369 * expr.c (pieces_addr::adjust): Likewise.
7370 (can_store_by_pieces): Likewise.
7371 (store_by_pieces): Likewise.
7372 (clear_by_pieces_1): Likewise.
7373 (expand_expr_addr_expr_1): Likewise.
7374 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
7375 (expand_expr_real_1): Likewise.
7376 (try_casesi): Likewise.
7377 * final.c (shorten_branches): Likewise.
7378 * fold-const.c (fold_convert_const_int_from_fixed): Change the
7379 type of "mode" to machine_mode.
7380 * internal-fn.c (expand_arith_overflow_result_store): Take a
7381 scalar_int_mode rather than a machine_mode.
7382 (expand_mul_overflow): Use scalar_int_mode for local variables.
7383 * loop-doloop.c (doloop_modify): Likewise.
7384 (doloop_optimize): Likewise.
7385 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
7386 than a machine_mode.
7387 (expand_doubleword_shift_condmove): Likewise.
7388 (expand_doubleword_shift): Likewise.
7389 (expand_doubleword_clz): Likewise.
7390 (expand_doubleword_popcount): Likewise.
7391 (expand_doubleword_parity): Likewise.
7392 (expand_absneg_bit): Use scalar_int_mode for local variables.
7393 (prepare_float_lib_cmp): Likewise.
7394 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
7395 rather than a machine_mode.
7396 (convert_memory_address_addr_space): Likewise.
7397 (get_mode_bounds): Likewise.
7398 (get_address_mode): Return a scalar_int_mode rather than a
7400 * rtlanal.c (get_address_mode): Likewise.
7401 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
7402 than a machine_mode.
7403 * targhooks.c (default_mode_rep_extended): Likewise.
7404 (default_valid_pointer_mode): Likewise.
7405 (default_addr_space_valid_pointer_mode): Likewise.
7406 (default_eh_return_filter_mode): Return a scalar_int_mode rather
7407 than a machine_mode.
7408 (default_libgcc_cmp_return_mode): Likewise.
7409 (default_libgcc_shift_count_mode): Likewise.
7410 (default_unwind_word_mode): Likewise.
7411 (default_addr_space_pointer_mode): Likewise.
7412 (default_addr_space_address_mode): Likewise.
7413 * targhooks.h (default_eh_return_filter_mode): Likewise.
7414 (default_libgcc_cmp_return_mode): Likewise.
7415 (default_libgcc_shift_count_mode): Likewise.
7416 (default_unwind_word_mode): Likewise.
7417 (default_addr_space_pointer_mode): Likewise.
7418 (default_addr_space_address_mode): Likewise.
7419 (default_mode_rep_extended): Take a scalar_int_mode rather than
7421 (default_valid_pointer_mode): Likewise.
7422 (default_addr_space_valid_pointer_mode): Likewise.
7423 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
7425 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
7426 rather than a machine_mode.
7427 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
7428 for local variables.
7429 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7430 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
7431 than a machine_mode.
7433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7434 Alan Hayward <alan.hayward@arm.com>
7435 David Sherwood <david.sherwood@arm.com>
7437 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
7438 the mode argument to scalar_int_mode.
7439 (do_jump_by_parts_zero_rtx): Likewise.
7440 (do_jump_by_parts_equality_rtx): Likewise.
7441 (do_jump_by_parts_greater): Take a mode argument.
7442 (do_jump_by_parts_equality): Likewise.
7443 (do_jump_1): Update calls accordingly.
7445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7446 Alan Hayward <alan.hayward@arm.com>
7447 David Sherwood <david.sherwood@arm.com>
7449 * is-a.h (safe_dyn_cast): New function.
7450 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
7451 (jump_table_for_label): Likewise.
7452 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
7453 instead of an rtx_insn *.
7454 (shorten_branches): Use dyn_cast instead of LABEL_P and
7455 JUMP_TABLE_DATA_P. Use jump_table_for_label and
7456 rtx_jump_table_data::get_data_mode.
7457 (final_scan_insn): Likewise.
7459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7460 Alan Hayward <alan.hayward@arm.com>
7461 David Sherwood <david.sherwood@arm.com>
7463 * combine.c (try_combine): Use is_a <scalar_int_mode> when
7464 trying to combine a full-register integer set with a subreg
7467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7468 Alan Hayward <alan.hayward@arm.com>
7469 David Sherwood <david.sherwood@arm.com>
7471 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
7472 that is always either address_mode or pointer_mode.
7474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7475 Alan Hayward <alan.hayward@arm.com>
7476 David Sherwood <david.sherwood@arm.com>
7478 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
7479 when the two are known to be equal.
7481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7482 Alan Hayward <alan.hayward@arm.com>
7483 David Sherwood <david.sherwood@arm.com>
7485 * simplify-rtx.c (simplify_const_unary_operation): Use
7486 is_a <scalar_int_mode> instead of checking for a nonzero
7487 precision. Forcibly convert op_mode to a scalar_int_mode
7488 in that case. More clearly differentiate the operand and
7489 result modes and use the former when deciding what the value
7490 of a count-bits operation should be. Use is_int_mode instead
7491 of checking for a MODE_INT. Remove redundant check for whether
7492 this mode has a zero precision.
7494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7495 Alan Hayward <alan.hayward@arm.com>
7496 David Sherwood <david.sherwood@arm.com>
7498 * optabs.c (widen_leading): Change the type of the mode argument
7499 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
7500 (widen_bswap): Likewise.
7501 (expand_parity): Likewise.
7502 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
7503 (expand_ffs): Likewise.
7504 (epand_unop): Check for scalar integer modes before calling the
7507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7508 Alan Hayward <alan.hayward@arm.com>
7509 David Sherwood <david.sherwood@arm.com>
7511 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
7513 (expand_expr_real_1): Update call accordingly.
7515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7516 Alan Hayward <alan.hayward@arm.com>
7517 David Sherwood <david.sherwood@arm.com>
7519 * expmed.c (store_bit_field_using_insv): Add op0_mode and
7520 value_mode arguments. Use scalar_int_mode internally.
7521 (store_bit_field_1): Rename the new integer mode from imode
7522 to op0_mode and use it instead of GET_MODE (op0). Update calls
7523 to store_split_bit_field, store_bit_field_using_insv and
7524 store_fixed_bit_field.
7525 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
7526 Use scalar_int_mode internally. Use a bit count rather than a mode
7527 when calculating the largest bit size for get_best_mode.
7528 Update calls to store_split_bit_field and store_fixed_bit_field_1.
7529 (store_fixed_bit_field_1): Add mode and value_mode arguments.
7530 Remove assertion that OP0 has a scalar integer mode.
7531 (store_split_bit_field): Add op0_mode and value_mode arguments.
7532 Update calls to extract_fixed_bit_field.
7533 (extract_bit_field_using_extv): Add an op0_mode argument.
7534 Use scalar_int_mode internally.
7535 (extract_bit_field_1): Rename the new integer mode from imode to
7536 op0_mode and use it instead of GET_MODE (op0). Update calls to
7537 extract_split_bit_field, extract_bit_field_using_extv and
7538 extract_fixed_bit_field.
7539 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
7540 to extract_split_bit_field and extract_fixed_bit_field_1.
7541 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
7542 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
7544 (extract_split_bit_field): Add an op0_mode argument. Update call
7545 to extract_fixed_bit_field.
7547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7548 Alan Hayward <alan.hayward@arm.com>
7549 David Sherwood <david.sherwood@arm.com>
7551 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
7552 * explow.c (hard_function_value): Likewise.
7553 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
7554 convert_to_mode call outside the loop.
7555 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
7556 for the mode iterator. Require the mode specified by max_pieces
7558 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
7560 (copy_blkmode_to_reg): Likewise.
7561 (set_storage_via_setmem): Likewise.
7562 * optabs.c (prepare_cmp_insn): Likewise.
7563 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7564 * stor-layout.c (finish_bitfield_representative): Likewise.
7566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7567 Alan Hayward <alan.hayward@arm.com>
7568 David Sherwood <david.sherwood@arm.com>
7570 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
7571 * expr.c (convert_move): Use them.
7572 (convert_modes): Likewise.
7573 (store_expr_with_bounds): Likewise.
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 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
7580 parameter for the mode of "x". Remove the "known_x", "known_mode"
7581 and "known_ret" arguments. Change the type of the mode argument
7583 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
7584 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
7585 (reg_num_sign_bit_copies_for_combine): Likewise.
7586 * rtlanal.c (nonzero_bits1): Likewise.
7587 (num_sign_bit_copies1): Likewise.
7588 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
7589 (reg_num_sign_bit_copies_general): Likewise.
7590 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
7591 (reg_nonzero_bits_general): Likewise.
7593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7594 Alan Hayward <alan.hayward@arm.com>
7595 David Sherwood <david.sherwood@arm.com>
7597 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
7598 than in subroutines. Return 1 for non-integer modes.
7599 (cached_num_sign_bit_copies): Change the type of the mode parameter
7601 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
7602 classes. Handle CONST_INT_P first and then check whether X also
7603 has a scalar integer mode. Check the same thing for inner registers
7604 of a SUBREG and for values that are being extended or truncated.
7606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7607 Alan Hayward <alan.hayward@arm.com>
7608 David Sherwood <david.sherwood@arm.com>
7610 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
7611 in subroutines. Return the mode mask for non-integer modes.
7612 (cached_nonzero_bits): Change the type of the mode parameter
7614 (nonzero_bits1): Likewise. Remove early exit for other mode
7615 classes. Handle CONST_INT_P first and then check whether X
7616 also has a scalar integer mode.
7618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7619 Alan Hayward <alan.hayward@arm.com>
7620 David Sherwood <david.sherwood@arm.com>
7622 * expr.c (widest_int_mode_for_size): Make the comment match the code.
7623 Return a scalar_int_mode and assert that the size is greater than
7625 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
7626 (op_by_pieces_d::op_by_pieces_d): Likewise.
7627 (op_by_pieces_d::run): Likewise.
7628 (can_store_by_pieces): Likewise.
7630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7631 Alan Hayward <alan.hayward@arm.com>
7632 David Sherwood <david.sherwood@arm.com>
7634 * combine.c (extract_left_shift): Add a mode argument and update
7636 (make_compound_operation_int): Change the type of the mode parameter
7637 to scalar_int_mode and update the call to extract_left_shift.
7639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7640 Alan Hayward <alan.hayward@arm.com>
7641 David Sherwood <david.sherwood@arm.com>
7643 * combine.c (simplify_and_const_int): Change the type of the mode
7644 parameter to scalar_int_mode.
7645 (simplify_and_const_int_1): Likewise. Update recursive call.
7647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7648 Alan Hayward <alan.hayward@arm.com>
7649 David Sherwood <david.sherwood@arm.com>
7651 * combine.c (simplify_compare_const): Check that the mode is a
7652 scalar_int_mode (rather than VOIDmode) before testing its
7654 (simplify_comparison): Move COMPARISON_P handling out of the
7655 loop and restrict the latter part of the loop to scalar_int_modes.
7656 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
7657 and when considering SUBREG_REGs. Use is_int_mode instead of
7658 checking GET_MODE_CLASS against MODE_INT.
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 * combine.c (try_widen_shift_mode): Move check for equal modes to...
7665 (simplify_shift_const_1): ...here. Use scalar_int_mode for
7666 shift_unit_mode and for modes involved in scalar shifts.
7668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7669 Alan Hayward <alan.hayward@arm.com>
7670 David Sherwood <david.sherwood@arm.com>
7672 * combine.c (force_int_to_mode): New function, split out from...
7673 (force_to_mode): ...here. Keep xmode up-to-date and use it
7674 instead of GET_MODE (x).
7676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7677 Alan Hayward <alan.hayward@arm.com>
7678 David Sherwood <david.sherwood@arm.com>
7680 * optabs-query.h (extraction_insn::struct_mode): Change type to
7681 opt_scalar_int_mode and update comment.
7682 (extraction_insn::field_mode): Change type to scalar_int_mode.
7683 (extraction_insn::pos_mode): Likewise.
7684 * combine.c (make_extraction): Update accordingly.
7685 * optabs-query.c (get_traditional_extraction_insn): Likewise.
7686 (get_optab_extraction_insn): Likewise.
7687 * recog.c (simplify_while_replacing): Likewise.
7688 * expmed.c (narrow_bit_field_mem): Change the type of the mode
7689 parameter to opt_scalar_int_mode.
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 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
7696 to a scalar_int_mode instead of a machine_mode.
7697 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
7698 (get_best_mode): Return a boolean and use a pointer argument to store
7699 the selected mode. Replace the limit mode parameter with a bit limit.
7700 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
7701 for the values returned by bit_field_mode_iterator::next_mode.
7702 (store_bit_field): Update call to get_best_mode.
7703 (store_fixed_bit_field): Likewise.
7704 (extract_fixed_bit_field): Likewise.
7705 * expr.c (optimize_bitfield_assignment_op): Likewise.
7706 * fold-const.c (optimize_bit_field_compare): Likewise.
7707 (fold_truth_andor_1): Likewise.
7708 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
7709 Update for new type of m_mode.
7710 (get_best_mode): As above.
7712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7713 Alan Hayward <alan.hayward@arm.com>
7714 David Sherwood <david.sherwood@arm.com>
7716 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
7717 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
7718 (store_bit_field): Check is_a <scalar_int_mode> before calling
7719 strict_volatile_bitfield_p.
7720 (extract_bit_field): Likewise.
7722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7723 Alan Hayward <alan.hayward@arm.com>
7724 David Sherwood <david.sherwood@arm.com>
7726 * target.def (cstore_mode): Return a scalar_int_mode.
7727 * doc/tm.texi: Regenerate.
7728 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
7729 * targhooks.h (default_cstore_mode): Likewise.
7730 * targhooks.c (default_cstore_mode): Likewise, using a forced
7732 * expmed.c (emit_cstore): Expect the target of the cstore to be
7735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7736 Alan Hayward <alan.hayward@arm.com>
7737 David Sherwood <david.sherwood@arm.com>
7739 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
7741 (niter_desc): Likewise mode.
7742 (iv_analyze): Add a mode parameter.
7744 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
7745 and change its type to scalar_int_mode.
7746 * loop-iv.c: Update commentary at head of file.
7747 (iv_constant): Pass the mode paraeter before the rtx it describes
7748 and change its type to scalar_int_mode. Remove VOIDmode handling.
7749 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
7750 (iv_extend): Likewise.
7751 (shorten_into_mode): Likewise.
7752 (iv_add): Use scalar_int_mode.
7753 (iv_mult): Likewise.
7754 (iv_shift): Likewise.
7755 (canonicalize_iv_subregs): Likewise.
7756 (get_biv_step_1): Pass the outer_mode parameter before the rtx
7757 it describes and change its mode to scalar_int_mode. Also change
7758 the type of the returned inner_mode to scalar_int_mode.
7759 (get_biv_step): Likewise, turning outer_mode from a pointer
7760 into a direct parameter. Update call to get_biv_step_1.
7761 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
7762 iv_constant and get_biv_step.
7763 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
7764 and change its type to scalar_int_mode. Don't initialise iv->mode
7765 to VOIDmode and remove later checks for its still being VOIDmode.
7766 Update calls to iv_analyze_op and iv_analyze_expr. Check
7767 is_a <scalar_int_mode> when changing the mode under consideration.
7768 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
7769 Update call to iv_analyze_expr.
7770 (iv_analyze_op): Add a mode parameter. Reject subregs whose
7771 inner register is not also a scalar_int_mode. Update call to
7773 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
7774 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
7775 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
7776 separate mode class checks. Update calls to iv_analyze. Remove
7777 fix-up of VOIDmodes after iv_analyze_biv.
7778 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
7779 don't have a scalar_int_mode. Update call to biv_p.
7781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7782 Alan Hayward <alan.hayward@arm.com>
7783 David Sherwood <david.sherwood@arm.com>
7785 * cfgexpand.c (convert_debug_memory_address): Use
7786 as_a <scalar_int_mode>.
7787 * combine.c (expand_compound_operation): Likewise.
7788 (make_extraction): Likewise.
7789 (change_zero_ext): Likewise.
7790 (simplify_comparison): Likewise.
7791 * cse.c (cse_insn): Likewise.
7792 * dwarf2out.c (minmax_loc_descriptor): Likewise.
7793 (mem_loc_descriptor): Likewise.
7794 (loc_descriptor): Likewise.
7795 * expmed.c (init_expmed_one_mode): Likewise.
7796 (synth_mult): Likewise.
7797 (emit_store_flag_1): Likewise.
7798 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
7799 of a comparison with size.
7800 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
7801 (reduce_to_bit_field_precision): Likewise.
7802 * function.c (expand_function_end): Likewise.
7803 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
7804 * loop-doloop.c (doloop_modify): Likewise.
7805 * optabs.c (expand_binop): Likewise.
7806 (expand_unop): Likewise.
7807 (expand_copysign_absneg): Likewise.
7808 (prepare_cmp_insn): Likewise.
7809 (maybe_legitimize_operand): Likewise.
7810 * recog.c (const_scalar_int_operand): Likewise.
7811 * rtlanal.c (get_address_mode): Likewise.
7812 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7813 (simplify_cond_clz_ctz): Likewise.
7814 * tree-nested.c (get_nl_goto_field): Likewise.
7815 * tree.c (build_vector_type_for_mode): Likewise.
7816 * var-tracking.c (use_narrower_mode): Likewise.
7818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7819 Alan Hayward <alan.hayward@arm.com>
7820 David Sherwood <david.sherwood@arm.com>
7822 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
7823 * builtins.c (expand_builtin_signbit): Use it.
7824 * cfgexpand.c (expand_debug_expr): Likewise.
7825 * dojump.c (do_jump): Likewise.
7826 (do_compare_and_jump): Likewise.
7827 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
7828 * expmed.c (make_tree): Likewise.
7829 * expr.c (expand_expr_real_2): Likewise.
7830 (expand_expr_real_1): Likewise.
7831 (try_casesi): Likewise.
7832 * fold-const-call.c (fold_const_call_ss): Likewise.
7833 * fold-const.c (unextend): Likewise.
7834 (extract_muldiv_1): Likewise.
7835 (fold_single_bit_test): Likewise.
7836 (native_encode_int): Likewise.
7837 (native_encode_string): Likewise.
7838 (native_interpret_int): Likewise.
7839 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
7840 * internal-fn.c (expand_addsub_overflow): Likewise.
7841 (expand_neg_overflow): Likewise.
7842 (expand_mul_overflow): Likewise.
7843 (expand_arith_overflow): Likewise.
7844 * match.pd: Likewise.
7845 * stor-layout.c (layout_type): Likewise.
7846 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7847 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
7848 * tree-ssanames.c (get_range_info): Likewise.
7849 * tree-switch-conversion.c (array_value_type) Likewise.
7850 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
7851 (vect_recog_divmod_pattern): Likewise.
7852 (vect_recog_mixed_size_cond_pattern): Likewise.
7853 * tree-vrp.c (extract_range_basic): Likewise.
7854 (simplify_float_conversion_using_ranges): Likewise.
7855 * tree.c (int_fits_type_p): Likewise.
7856 * ubsan.c (instrument_bool_enum_load): Likewise.
7857 * varasm.c (mergeable_string_section): Likewise.
7858 (narrowing_initializer_constant_valid_p): Likewise.
7859 (output_constant): Likewise.
7861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7862 Alan Hayward <alan.hayward@arm.com>
7863 David Sherwood <david.sherwood@arm.com>
7865 * machmode.h (NARROWEST_INT_MODE): New macro.
7866 * expr.c (alignment_for_piecewise_move): Use it instead of
7867 GET_CLASS_NARROWEST_MODE (MODE_INT).
7868 (push_block): Likewise.
7869 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
7871 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7873 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7874 Alan Hayward <alan.hayward@arm.com>
7875 David Sherwood <david.sherwood@arm.com>
7877 * postreload.c (move2add_valid_value_p): Change the type of the
7878 mode parameter to scalar_int_mode.
7879 (move2add_use_add2_insn): Add a mode parameter and use it instead
7881 (move2add_use_add3_insn): Likewise.
7882 (reload_cse_move2add): Update accordingly.
7884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7885 Alan Hayward <alan.hayward@arm.com>
7886 David Sherwood <david.sherwood@arm.com>
7888 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
7890 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
7891 * optabs.c (expand_unop): Likewise.
7893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7894 Alan Hayward <alan.hayward@arm.com>
7895 David Sherwood <david.sherwood@arm.com>
7897 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
7898 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
7899 (popcount_loc_descriptor): Likewise.
7900 (bswap_loc_descriptor): Likewise.
7901 (rotate_loc_descriptor): Likewise.
7902 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
7903 calling the functions above.
7905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7906 Alan Hayward <alan.hayward@arm.com>
7907 David Sherwood <david.sherwood@arm.com>
7909 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
7911 (try_combine): Likewise.
7912 (simplify_if_then_else): Likewise.
7913 * cse.c (cse_insn): Likewise.
7914 * dwarf2out.c (mem_loc_descriptor): Likewise.
7915 * emit-rtl.c (gen_lowpart_common): Likewise.
7916 * simplify-rtx.c (simplify_truncation): Likewise.
7917 (simplify_binary_operation_1): Likewise.
7918 (simplify_const_relational_operation): Likewise.
7919 (simplify_ternary_operation): Likewise.
7920 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
7922 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7923 Alan Hayward <alan.hayward@arm.com>
7924 David Sherwood <david.sherwood@arm.com>
7926 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
7927 * reload.c (push_reload): Likewise.
7928 (find_reloads): 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 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
7935 (make_compound_operation_int): Likewise.
7936 (change_zero_ext): Likewise.
7937 * expr.c (convert_move): Likewise.
7938 (convert_modes): Likewise.
7939 * fwprop.c (forward_propagate_subreg): Likewise.
7940 * loop-iv.c (get_biv_step_1): Likewise.
7941 * optabs.c (widen_operand): Likewise.
7942 * postreload.c (move2add_valid_value_p): Likewise.
7943 * recog.c (simplify_while_replacing): Likewise.
7944 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7945 (simplify_binary_operation_1): Likewise. Remove redundant
7946 mode equality check.
7948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7949 Alan Hayward <alan.hayward@arm.com>
7950 David Sherwood <david.sherwood@arm.com>
7952 * combine.c (combine_simplify_rtx): Add checks for
7953 is_a <scalar_int_mode>.
7954 (simplify_if_then_else): Likewise.
7955 (make_field_assignment): Likewise.
7956 (simplify_comparison): Likewise.
7957 * ifcvt.c (noce_try_bitop): Likewise.
7958 * loop-invariant.c (canonicalize_address_mult): Likewise.
7959 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7962 Alan Hayward <alan.hayward@arm.com>
7963 David Sherwood <david.sherwood@arm.com>
7965 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
7966 is_a <scalar_int_mode> instead of != BLKmode.
7968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7969 Alan Hayward <alan.hayward@arm.com>
7970 David Sherwood <david.sherwood@arm.com>
7972 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
7973 instead of != VOIDmode.
7974 * combine.c (if_then_else_cond): Likewise.
7975 (change_zero_ext): Likewise.
7976 * dwarf2out.c (mem_loc_descriptor): Likewise.
7977 (loc_descriptor): Likewise.
7978 * rtlanal.c (canonicalize_condition): Likewise.
7979 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
7981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7982 Alan Hayward <alan.hayward@arm.com>
7983 David Sherwood <david.sherwood@arm.com>
7985 * simplify-rtx.c (simplify_binary_operation_1): Use
7986 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
7988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7989 Alan Hayward <alan.hayward@arm.com>
7990 David Sherwood <david.sherwood@arm.com>
7992 * wide-int.h (int_traits<unsigned char>) New class.
7993 (int_traits<unsigned short>) Likewise.
7994 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
7995 Use GET_MODE_UNIT_PRECISION and remove redundant test for
7997 * combine.c (set_nonzero_bits_and_sign_copies): Use
7998 is_a <scalar_int_mode>.
7999 (find_split_point): Likewise.
8000 (combine_simplify_rtx): Likewise.
8001 (simplify_logical): Likewise.
8002 (expand_compound_operation): Likewise.
8003 (expand_field_assignment): Likewise.
8004 (make_compound_operation): Likewise.
8005 (extended_count): Likewise.
8006 (change_zero_ext): Likewise.
8007 (simplify_comparison): Likewise.
8008 * dwarf2out.c (scompare_loc_descriptor): Likewise.
8009 (ucompare_loc_descriptor): Likewise.
8010 (minmax_loc_descriptor): Likewise.
8011 (mem_loc_descriptor): Likewise.
8012 (loc_descriptor): Likewise.
8013 * expmed.c (init_expmed_one_mode): Likewise.
8014 * lra-constraints.c (lra_constraint_offset): Likewise.
8015 * optabs.c (prepare_libcall_arg): Likewise.
8016 * postreload.c (move2add_note_store): Likewise.
8017 * reload.c (operands_match_p): Likewise.
8018 * rtl.h (load_extend_op): Likewise.
8019 * rtlhooks.c (gen_lowpart_general): Likewise.
8020 * simplify-rtx.c (simplify_truncation): Likewise.
8021 (simplify_unary_operation_1): Likewise.
8022 (simplify_binary_operation_1): Likewise.
8023 (simplify_const_binary_operation): Likewise.
8024 (simplify_const_relational_operation): Likewise.
8025 (simplify_subreg): Likewise.
8026 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8027 * var-tracking.c (adjust_mems): Likewise.
8028 (prepare_call_arguments): Likewise.
8030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8031 Alan Hayward <alan.hayward@arm.com>
8032 David Sherwood <david.sherwood@arm.com>
8034 * machmode.h (is_int_mode): New fuction.
8035 * combine.c (find_split_point): Use it.
8036 (combine_simplify_rtx): Likewise.
8037 (simplify_if_then_else): Likewise.
8038 (simplify_set): Likewise.
8039 (simplify_shift_const_1): Likewise.
8040 (simplify_comparison): Likewise.
8041 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
8042 * cse.c (notreg_cost): Likewise.
8043 (cse_insn): Likewise.
8044 * cselib.c (cselib_lookup_1): Likewise.
8045 * dojump.c (do_jump_1): Likewise.
8046 (do_compare_rtx_and_jump): Likewise.
8047 * dse.c (get_call_args): Likewise.
8048 * dwarf2out.c (rtl_for_decl_init): Likewise.
8049 (native_encode_initializer): Likewise.
8050 * expmed.c (emit_store_flag_1): Likewise.
8051 (emit_store_flag): Likewise.
8052 * expr.c (convert_modes): Likewise.
8053 (store_field): Likewise.
8054 (expand_expr_real_1): Likewise.
8055 * fold-const.c (fold_read_from_constant_string): Likewise.
8056 * gimple-ssa-sprintf.c (get_format_string): Likewise.
8057 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
8058 * optabs.c (expand_binop): Likewise.
8059 (expand_unop): Likewise.
8060 (expand_abs_nojump): Likewise.
8061 (expand_one_cmpl_abs_nojump): Likewise.
8062 * simplify-rtx.c (mode_signbit_p): Likewise.
8063 (val_signbit_p): Likewise.
8064 (val_signbit_known_set_p): Likewise.
8065 (val_signbit_known_clear_p): Likewise.
8066 (simplify_relational_operation_1): Likewise.
8067 * tree.c (vector_type_mode): Likewise.
8069 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8070 Alan Hayward <alan.hayward@arm.com>
8071 David Sherwood <david.sherwood@arm.com>
8073 * machmode.h (smallest_mode_for_size): Fix formatting.
8074 (smallest_int_mode_for_size): New function.
8075 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
8076 instead of smallest_mode_for_size.
8077 * combine.c (make_extraction): Likewise.
8078 * config/arc/arc.c (arc_expand_movmem): Likewise.
8079 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
8080 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8081 * config/s390/s390.c (s390_expand_insv): Likewise.
8082 * config/sparc/sparc.c (assign_int_registers): Likewise.
8083 * config/spu/spu.c (spu_function_value): Likewise.
8084 (spu_function_arg): Likewise.
8085 * coverage.c (get_gcov_type): Likewise.
8086 (get_gcov_unsigned_t): Likewise.
8087 * dse.c (find_shift_sequence): Likewise.
8088 * expmed.c (store_bit_field_1): Likewise.
8089 * expr.c (convert_move): Likewise.
8090 (store_field): Likewise.
8091 * internal-fn.c (expand_arith_overflow): Likewise.
8092 * optabs-query.c (get_best_extraction_insn): Likewise.
8093 * optabs.c (expand_twoval_binop_libfunc): Likewise.
8094 * stor-layout.c (layout_type): Likewise.
8095 (initialize_sizetypes): Likewise.
8096 * targhooks.c (default_get_mask_mode): Likewise.
8097 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8100 Alan Hayward <alan.hayward@arm.com>
8101 David Sherwood <david.sherwood@arm.com>
8103 * machmode.h (opt_mode::else_blk): New function.
8104 (int_mode_for_mode): Declare.
8105 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8106 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8108 * cfgexpand.c (expand_debug_expr): Likewise.
8109 * combine.c (gen_lowpart_or_truncate): Likewise.
8110 (gen_lowpart_for_combine): Likewise.
8111 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8112 * config/avr/avr.c (avr_to_int_mode): Likewise.
8113 (avr_out_plus_1): Likewise.
8114 (avr_out_plus): Likewise.
8115 (avr_out_round): Likewise.
8116 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8117 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8118 (s390_expand_vcond): Likewise.
8119 * config/spu/spu.c (spu_split_immediate): Likewise.
8120 (spu_expand_mov): Likewise.
8121 * dse.c (get_stored_val): Likewise.
8122 * expmed.c (store_bit_field_1): Likewise.
8123 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8125 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8126 (extract_low_bits): Likewise.
8127 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8128 handling rather than repeating the check.
8129 (emit_group_store): Likewise.
8130 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8131 * optabs.c (expand_absneg_bit): Likewise.
8132 (expand_copysign_absneg): Likewise.
8133 (expand_copysign_bit): Likewise.
8134 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8135 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8136 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8137 * var-tracking.c (prepare_call_arguments): Likewise.
8138 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8139 int_mode_for_mode instead of mode_for_size.
8140 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8143 Alan Hayward <alan.hayward@arm.com>
8144 David Sherwood <david.sherwood@arm.com>
8146 * machmode.h (int_mode_for_size): New function.
8147 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8148 instead of mode_for_size.
8149 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8151 * combine.c (expand_field_assignment): Use int_mode_for_size
8152 instead of mode_for_size.
8153 (make_extraction): Likewise.
8154 (simplify_shift_const_1): Likewise.
8155 (simplify_comparison): Likewise.
8156 * dojump.c (do_jump): Likewise.
8157 * dwarf2out.c (mem_loc_descriptor): Likewise.
8158 * emit-rtl.c (init_derived_machine_modes): Likewise.
8159 * expmed.c (flip_storage_order): Likewise.
8160 (convert_extracted_bit_field): Likewise.
8161 * expr.c (copy_blkmode_from_reg): Likewise.
8162 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8163 * internal-fn.c (expand_mul_overflow): Likewise.
8164 * lower-subreg.c (simple_move): Likewise.
8165 * optabs-libfuncs.c (init_optabs): Likewise.
8166 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8167 * tree.c (vector_type_mode): Likewise.
8168 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8169 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8170 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8171 * tree-vect-stmts.c (vectorizable_load): Likewise.
8172 (vectorizable_store): Likewise.
8174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8175 Alan Hayward <alan.hayward@arm.com>
8176 David Sherwood <david.sherwood@arm.com>
8178 * coretypes.h (pod_mode): New type.
8179 (scalar_int_mode_pod): New typedef.
8180 * machmode.h (pod_mode): New class.
8181 (int_n_data_t::m): Change type to scalar_int_mode_pod.
8182 * genmodes.c (emit_mode_int_n): Update accordingly.
8183 * lower-subreg.h (target_lower_subreg): Change type to
8184 scalar_int_mode_pod.
8185 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
8186 scalar_int_mode_pod.
8188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8189 Alan Hayward <alan.hayward@arm.com>
8190 David Sherwood <david.sherwood@arm.com>
8192 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
8193 machine_mode to scalar_int_mode.
8194 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
8195 (rs6000_option_override_internal): Remove cast to int.
8196 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
8197 machine_mode to scalar_int_mode.
8198 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
8199 (rs6000_option_override_internal): Remove cast to int.
8200 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
8201 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
8203 * config/s390/s390.c (s390_expand_builtin): Likewise.
8204 * coretypes.h (scalar_int_mode): New type.
8205 (opt_scalar_int_mode): New typedef.
8206 * machmode.h (scalar_int_mode): New class.
8207 (scalar_int_mode::includes_p): New function.
8208 (byte_mode): Change type to scalar_int_mode.
8209 (word_mode): Likewise.
8210 (ptr_mode): Likewise.
8211 * emit-rtl.c (byte_mode): Likewise.
8212 (word_mode): Likewise.
8213 (ptr_mode): Likewise.
8214 (init_derived_machine_modes): Update accordingly.
8215 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
8216 and MODE_PARTIAL_INT.
8217 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
8218 opt_scalar_int_mode.
8220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8221 Alan Hayward <alan.hayward@arm.com>
8222 David Sherwood <david.sherwood@arm.com>
8224 * target.def (libgcc_floating_mode_supported_p): Take a
8226 * doc/tm.texi: Regenerate.
8227 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
8229 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8230 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
8233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8234 Alan Hayward <alan.hayward@arm.com>
8235 David Sherwood <david.sherwood@arm.com>
8237 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
8238 * doc/tm.texi: Regenerate.
8239 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
8240 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
8241 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
8242 * targhooks.h (default_floatn_mode): Likewise.
8243 * targhooks.c (default_floatn_mode): Likewise.
8244 * tree.c (build_common_tree_nodes): Update accordingly.
8246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8247 Alan Hayward <alan.hayward@arm.com>
8248 David Sherwood <david.sherwood@arm.com>
8250 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
8251 (mode_iterator::iterate_p): Likewise.
8252 (mode_iterator::get_wider): Likewise.
8253 * expr.c (init_expr_target): Use opt_scalar_float_mode.
8255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8256 Alan Hayward <alan.hayward@arm.com>
8257 David Sherwood <david.sherwood@arm.com>
8259 * coretypes.h (opt_scalar_float_mode): New typedef.
8260 * machmode.h (float_mode_for_size): New function.
8261 * emit-rtl.c (double_mode): Delete.
8262 (init_emit_once): Use float_mode_for_size.
8263 * stor-layout.c (layout_type): Likewise.
8264 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
8266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8267 Alan Hayward <alan.hayward@arm.com>
8268 David Sherwood <david.sherwood@arm.com>
8270 * output.h (assemble_real): Take a scalar_float_mode.
8271 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
8272 * config/arm/arm.md (consttable_4): Likewise.
8273 (consttable_8): Likewise.
8274 (consttable_16): Likewise.
8275 * config/mips/mips.md (consttable_float): Likewise.
8276 * config/s390/s390.c (s390_output_pool_entry): Likewise.
8277 * varasm.c (assemble_real): Take a scalar_float_mode.
8278 (output_constant_pool_2): Update accordingly.
8279 (output_constant): Likewise.
8281 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8282 Alan Hayward <alan.hayward@arm.com>
8283 David Sherwood <david.sherwood@arm.com>
8285 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
8286 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
8287 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
8288 (native_encode_real): Likewise.
8289 (native_interpret_real): Likewise.
8290 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
8291 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8294 Alan Hayward <alan.hayward@arm.com>
8295 David Sherwood <david.sherwood@arm.com>
8297 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
8298 <scalar_float_mode>. Simplify.
8299 (gen_extend_conv_libfunc): Likewise.
8301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8302 Alan Hayward <alan.hayward@arm.com>
8303 David Sherwood <david.sherwood@arm.com>
8305 * coretypes.h (scalar_float_mode): New type.
8306 * machmode.h (mode_traits::from_int): Use machine_mode if
8307 USE_ENUM_MODES is defined.
8308 (is_a): New function.
8310 (dyn_cast): Likewise.
8311 (scalar_float_mode): New class.
8312 (scalar_float_mode::includes_p): New function.
8313 (is_float_mode): Likewise.
8314 * gdbhooks.py (MachineModePrinter): New class.
8315 (build_pretty_printer): Use it for scalar_float_mode.
8316 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
8317 (format_helper::format_helper): Turn into a template.
8318 * genmodes.c (get_mode_class): New function.
8319 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
8320 or machine_mode if none.
8321 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
8322 as_a <scalar_float_mode>.
8323 * dwarf2out.c (mem_loc_descriptor): Likewise.
8324 (insert_float): Likewise.
8325 (add_const_value_attribute): Likewise.
8326 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8327 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
8328 (expand_unop): Update accordingly.
8329 (expand_abs_nojump): Likewise.
8330 (expand_copysign_absneg): Take a scalar_float_mode.
8331 (expand_copysign_bit): Likewise.
8332 (expand_copysign): Update accordingly.
8334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8335 Alan Hayward <alan.hayward@arm.com>
8336 David Sherwood <david.sherwood@arm.com>
8338 * coretypes.h (opt_mode): New class.
8339 * machmode.h (opt_mode): Likewise.
8340 (opt_mode::else_void): New function.
8341 (opt_mode::require): Likewise.
8342 (opt_mode::exists): Likewise.
8343 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
8344 (GET_MODE_2XWIDER_MODE): Likewise.
8345 (mode_iterator::get_wider): Update accordingly.
8346 (mode_iterator::get_2xwider): Likewise.
8347 (mode_iterator::get_known_wider): Likewise, turning into a template.
8348 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
8349 forcing a wider mode to exist.
8350 * config/cr16/cr16.h (LONG_REG_P): Likewise.
8351 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8352 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
8353 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
8354 * lower-subreg.c (init_lower_subreg): Likewise.
8355 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
8356 on the final iteration.
8357 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
8358 a wider mode exists before asking for a move pattern.
8359 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
8360 forcing a wider mode to exist.
8361 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
8362 returning false if no such mode exists.
8363 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
8364 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
8365 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
8366 Avoid checking for a MODE_INT if we already know the mode is not a
8368 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
8369 forcing a wider mode to exist.
8370 (expmed_mult_highpart_optab): Likewise.
8371 (expmed_mult_highpart): Likewise.
8372 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
8374 * lto-streamer-in.c (lto_input_mode_table): Likewise.
8375 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
8376 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8377 * internal-fn.c (expand_mul_overflow): Update use of
8378 GET_MODE_2XWIDER_MODE.
8379 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8380 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
8381 GET_MODE_WIDER_MODE.
8382 (convert_plusminus_to_widen): Likewise.
8383 * tree-switch-conversion.c (array_value_type): Likewise.
8384 * var-tracking.c (emit_note_insn_var_location): Likewise.
8385 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8386 Return false inside rather than outside the loop if no wider mode
8388 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
8389 and GET_MODE_2XWIDER_MODE
8390 (can_compare_p): Use else_void.
8391 * gdbhooks.py (OptMachineModePrinter): New class.
8392 (build_pretty_printer): Use it for opt_mode.
8394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8395 Alan Hayward <alan.hayward@arm.com>
8396 David Sherwood <david.sherwood@arm.com>
8398 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
8399 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
8401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8402 Alan Hayward <alan.hayward@arm.com>
8403 David Sherwood <david.sherwood@arm.com>
8405 * machmode.h (mode_traits): New structure.
8406 (get_narrowest_mode): New function.
8407 (mode_iterator::start): Likewise.
8408 (mode_iterator::iterate_p): Likewise.
8409 (mode_iterator::get_wider): Likewise.
8410 (mode_iterator::get_known_wider): Likewise.
8411 (mode_iterator::get_2xwider): Likewise.
8412 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
8413 (FOR_EACH_MODE): Likewise.
8414 (FOR_EACH_MODE_FROM): Likewise.
8415 (FOR_EACH_MODE_UNTIL): Likewise.
8416 (FOR_EACH_WIDER_MODE): Likewise.
8417 (FOR_EACH_2XWIDER_MODE): Likewise.
8418 * builtins.c (expand_builtin_strlen): Use new mode iterators.
8419 * combine.c (simplify_comparison): Likewise
8420 * config/i386/i386.c (type_natural_mode): Likewise.
8421 * cse.c (cse_insn): Likewise.
8422 * dse.c (find_shift_sequence): Likewise.
8423 * emit-rtl.c (init_derived_machine_modes): Likewise.
8424 (init_emit_once): Likewise.
8425 * explow.c (hard_function_value): Likewise.
8426 * expmed.c (extract_fixed_bit_field_1): Likewise.
8427 (extract_bit_field_1): Likewise.
8428 (expand_divmod): Likewise.
8429 (emit_store_flag_1): Likewise.
8430 * expr.c (init_expr_target): Likewise.
8431 (convert_move): Likewise.
8432 (alignment_for_piecewise_move): Likewise.
8433 (widest_int_mode_for_size): Likewise.
8434 (emit_block_move_via_movmem): Likewise.
8435 (copy_blkmode_to_reg): Likewise.
8436 (set_storage_via_setmem): Likewise.
8437 (compress_float_constant): Likewise.
8438 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8439 * optabs-query.c (get_best_extraction_insn): Likewise.
8440 * optabs.c (expand_binop): Likewise.
8441 (expand_twoval_unop): Likewise.
8442 (expand_twoval_binop): Likewise.
8443 (widen_leading): Likewise.
8444 (widen_bswap): Likewise.
8445 (expand_parity): Likewise.
8446 (expand_unop): Likewise.
8447 (prepare_cmp_insn): Likewise.
8448 (prepare_float_lib_cmp): Likewise.
8449 (expand_float): Likewise.
8450 (expand_fix): Likewise.
8451 (expand_sfix_optab): Likewise.
8452 * postreload.c (move2add_use_add2_insn): Likewise.
8453 * reg-stack.c (reg_to_stack): Likewise.
8454 * reginfo.c (choose_hard_reg_mode): Likewise.
8455 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8456 * stor-layout.c (mode_for_size): Likewise.
8457 (smallest_mode_for_size): Likewise.
8458 (mode_for_vector): Likewise.
8459 (finish_bitfield_representative): Likewise.
8460 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
8461 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
8462 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8463 * var-tracking.c (prepare_call_arguments): Likewise.
8465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8466 Alan Hayward <alan.hayward@arm.com>
8467 David Sherwood <david.sherwood@arm.com>
8469 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
8470 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
8471 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
8472 * machmode.h (mode_size): Move earlier in file.
8473 (mode_precision): Likewise.
8474 (mode_inner): Likewise.
8475 (mode_nunits): Likewise.
8476 (mode_unit_size): Likewise.
8477 (unit_unit_precision): Likewise.
8478 (mode_wider): Likewise.
8479 (mode_2xwider): Likewise.
8480 (machine_mode): New class.
8481 (mode_to_bytes): New function.
8482 (mode_to_bits): Likewise.
8483 (mode_to_precision): Likewise.
8484 (mode_to_inner): Likewise.
8485 (mode_to_unit_size): Likewise.
8486 (mode_to_unit_precision): Likewise.
8487 (mode_to_nunits): Likewise.
8488 (GET_MODE_SIZE): Use mode_to_bytes.
8489 (GET_MODE_BITSIZE): Use mode_to_bits.
8490 (GET_MODE_PRECISION): Use mode_to_precision.
8491 (GET_MODE_INNER): Use mode_to_inner.
8492 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
8493 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
8494 (GET_MODE_NUNITS): Use mode_to_nunits.
8495 * system.h (ALWAYS_INLINE): New macro.
8496 * config/powerpcspe/powerpcspe-c.c
8497 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
8498 int for arg1_mode and arg2_mode.
8500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8501 Alan Hayward <alan.hayward@arm.com>
8502 David Sherwood <david.sherwood@arm.com>
8504 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8505 Prefix mode names with E_ in case statements.
8506 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8507 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
8508 (aarch64_split_simd_move): Likewise.
8509 (aarch64_gen_storewb_pair): Likewise.
8510 (aarch64_gen_loadwb_pair): Likewise.
8511 (aarch64_gen_store_pair): Likewise.
8512 (aarch64_gen_load_pair): Likewise.
8513 (aarch64_get_condition_code_1): Likewise.
8514 (aarch64_constant_pool_reload_icode): Likewise.
8515 (get_rsqrte_type): Likewise.
8516 (get_rsqrts_type): Likewise.
8517 (get_recpe_type): Likewise.
8518 (get_recps_type): Likewise.
8519 (aarch64_gimplify_va_arg_expr): Likewise.
8520 (aarch64_simd_container_mode): Likewise.
8521 (aarch64_emit_load_exclusive): Likewise.
8522 (aarch64_emit_store_exclusive): Likewise.
8523 (aarch64_expand_compare_and_swap): Likewise.
8524 (aarch64_gen_atomic_cas): Likewise.
8525 (aarch64_emit_bic): Likewise.
8526 (aarch64_emit_atomic_swap): Likewise.
8527 (aarch64_emit_atomic_load_op): Likewise.
8528 (aarch64_evpc_trn): Likewise.
8529 (aarch64_evpc_uzp): Likewise.
8530 (aarch64_evpc_zip): Likewise.
8531 (aarch64_evpc_ext): Likewise.
8532 (aarch64_evpc_rev): Likewise.
8533 (aarch64_evpc_dup): Likewise.
8534 (aarch64_gen_ccmp_first): Likewise.
8535 (aarch64_gen_ccmp_next): Likewise.
8536 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8537 (alpha_emit_xfloating_libcall): Likewise.
8538 (emit_insxl): Likewise.
8539 (alpha_arg_type): Likewise.
8540 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
8541 (arc_preferred_simd_mode): Likewise.
8542 (arc_secondary_reload): Likewise.
8543 (get_arc_condition_code): Likewise.
8544 (arc_print_operand): Likewise.
8545 (arc_legitimate_constant_p): Likewise.
8546 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8547 * config/arc/arc.md (casesi_load): Likewise.
8548 (casesi_compact_jump): Likewise.
8549 * config/arc/predicates.md (proper_comparison_operator): Likewise.
8550 (cc_use_register): Likewise.
8551 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8552 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8553 (arm_init_iwmmxt_builtins): Likewise.
8554 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
8555 (neon_expand_vector_init): Likewise.
8556 (arm_attr_length_move_neon): Likewise.
8557 (maybe_get_arm_condition_code): Likewise.
8558 (arm_emit_vector_const): Likewise.
8559 (arm_preferred_simd_mode): Likewise.
8560 (arm_output_iwmmxt_tinsr): Likewise.
8561 (thumb1_output_casesi): Likewise.
8562 (thumb2_output_casesi): Likewise.
8563 (arm_emit_load_exclusive): Likewise.
8564 (arm_emit_store_exclusive): Likewise.
8565 (arm_expand_compare_and_swap): Likewise.
8566 (arm_evpc_neon_vuzp): Likewise.
8567 (arm_evpc_neon_vzip): Likewise.
8568 (arm_evpc_neon_vrev): Likewise.
8569 (arm_evpc_neon_vtrn): Likewise.
8570 (arm_evpc_neon_vext): Likewise.
8571 (arm_validize_comparison): Likewise.
8572 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
8573 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
8574 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
8575 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
8576 (c6x_preferred_simd_mode): Likewise.
8577 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
8578 (epiphany_rtx_costs): Likewise.
8579 * config/epiphany/predicates.md (proper_comparison_operator):
8581 * config/frv/frv.c (condexec_memory_operand): Likewise.
8582 (frv_emit_move): Likewise.
8583 (output_move_single): Likewise.
8584 (output_condmove_single): Likewise.
8585 (frv_hard_regno_mode_ok): Likewise.
8586 (frv_matching_accg_mode): Likewise.
8587 * config/h8300/h8300.c (split_adds_subs): Likewise.
8588 (h8300_rtx_costs): Likewise.
8589 (h8300_print_operand): Likewise.
8590 (compute_mov_length): Likewise.
8591 (output_logical_op): Likewise.
8592 (compute_logical_op_length): Likewise.
8593 (compute_logical_op_cc): Likewise.
8594 (h8300_shift_needs_scratch_p): Likewise.
8595 (output_a_shift): Likewise.
8596 (compute_a_shift_length): Likewise.
8597 (compute_a_shift_cc): Likewise.
8598 (expand_a_rotate): Likewise.
8599 (output_a_rotate): Likewise.
8600 * config/i386/i386.c (classify_argument): Likewise.
8601 (function_arg_advance_32): Likewise.
8602 (function_arg_32): Likewise.
8603 (function_arg_64): Likewise.
8604 (function_value_64): Likewise.
8605 (ix86_gimplify_va_arg): Likewise.
8606 (ix86_legitimate_constant_p): Likewise.
8607 (put_condition_code): Likewise.
8608 (split_double_mode): Likewise.
8609 (ix86_avx256_split_vector_move_misalign): Likewise.
8610 (ix86_expand_vector_logical_operator): Likewise.
8611 (ix86_split_idivmod): Likewise.
8612 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
8613 (ix86_build_const_vector): Likewise.
8614 (ix86_build_signbit_mask): Likewise.
8615 (ix86_match_ccmode): Likewise.
8616 (ix86_cc_modes_compatible): Likewise.
8617 (ix86_expand_branch): Likewise.
8618 (ix86_expand_sse_cmp): Likewise.
8619 (ix86_expand_sse_movcc): Likewise.
8620 (ix86_expand_int_sse_cmp): Likewise.
8621 (ix86_expand_vec_perm_vpermi2): Likewise.
8622 (ix86_expand_vec_perm): Likewise.
8623 (ix86_expand_sse_unpack): Likewise.
8624 (ix86_expand_int_addcc): Likewise.
8625 (ix86_split_to_parts): Likewise.
8626 (ix86_vectorize_builtin_gather): Likewise.
8627 (ix86_vectorize_builtin_scatter): Likewise.
8628 (avx_vpermilp_parallel): Likewise.
8629 (inline_memory_move_cost): Likewise.
8630 (ix86_tieable_integer_mode_p): Likewise.
8631 (x86_maybe_negate_const_int): Likewise.
8632 (ix86_expand_vector_init_duplicate): Likewise.
8633 (ix86_expand_vector_init_one_nonzero): Likewise.
8634 (ix86_expand_vector_init_one_var): Likewise.
8635 (ix86_expand_vector_init_concat): Likewise.
8636 (ix86_expand_vector_init_interleave): Likewise.
8637 (ix86_expand_vector_init_general): Likewise.
8638 (ix86_expand_vector_set): Likewise.
8639 (ix86_expand_vector_extract): Likewise.
8640 (emit_reduc_half): Likewise.
8641 (ix86_emit_i387_round): Likewise.
8642 (ix86_mangle_type): Likewise.
8643 (ix86_expand_round_sse4): Likewise.
8644 (expand_vec_perm_blend): Likewise.
8645 (canonicalize_vector_int_perm): Likewise.
8646 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
8647 (expand_vec_perm_1): Likewise.
8648 (expand_vec_perm_interleave3): Likewise.
8649 (expand_vec_perm_even_odd_pack): Likewise.
8650 (expand_vec_perm_even_odd_1): Likewise.
8651 (expand_vec_perm_broadcast_1): Likewise.
8652 (ix86_vectorize_vec_perm_const_ok): Likewise.
8653 (ix86_expand_vecop_qihi): Likewise.
8654 (ix86_expand_mul_widen_hilo): Likewise.
8655 (ix86_expand_sse2_abs): Likewise.
8656 (ix86_expand_pextr): Likewise.
8657 (ix86_expand_pinsr): Likewise.
8658 (ix86_preferred_simd_mode): Likewise.
8659 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
8660 * config/i386/sse.md (*andnot<mode>3): Likewise.
8661 (<mask_codefor><code><mode>3<mask_name>): Likewise.
8662 (*<code><mode>3): Likewise.
8663 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
8664 (ia64_expand_atomic_op): Likewise.
8665 (ia64_arg_type): Likewise.
8666 (ia64_mode_to_int): Likewise.
8667 (ia64_scalar_mode_supported_p): Likewise.
8668 (ia64_vector_mode_supported_p): Likewise.
8669 (expand_vec_perm_broadcast): Likewise.
8670 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
8671 (iq2000_function_arg_advance): Likewise.
8672 (iq2000_function_arg): Likewise.
8673 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
8674 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
8675 (m68k_libcall_value): Likewise.
8676 (m68k_function_value): Likewise.
8677 (sched_attr_op_type): Likewise.
8678 * config/mcore/mcore.c (mcore_output_move): Likewise.
8679 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
8681 (microblaze_function_arg): Likewise.
8682 * config/mips/mips.c (mips16_build_call_stub): Likewise.
8683 (mips_print_operand): Likewise.
8684 (mips_mode_ok_for_mov_fmt_p): Likewise.
8685 (mips_vector_mode_supported_p): Likewise.
8686 (mips_preferred_simd_mode): Likewise.
8687 (mips_expand_vpc_loongson_even_odd): Likewise.
8688 (mips_expand_vec_unpack): Likewise.
8689 (mips_expand_vi_broadcast): Likewise.
8690 (mips_expand_vector_init): Likewise.
8691 (mips_expand_vec_reduc): Likewise.
8692 (mips_expand_msa_cmp): Likewise.
8693 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
8694 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
8695 (cc_flags_for_mode): Likewise.
8696 * config/msp430/msp430.c (msp430_print_operand): Likewise.
8697 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
8698 (nds32_output_casesi_pc_relative): Likewise.
8699 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8700 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
8701 (nvptx_gen_unpack): Likewise.
8702 (nvptx_gen_pack): Likewise.
8703 (nvptx_gen_shuffle): Likewise.
8704 (nvptx_gen_wcast): Likewise.
8705 (nvptx_preferred_simd_mode): Likewise.
8706 * config/pa/pa.c (pa_secondary_reload): Likewise.
8707 * config/pa/predicates.md (base14_operand): Likewise.
8708 * config/powerpcspe/powerpcspe-c.c
8709 (altivec_resolve_overloaded_builtin): Likewise.
8710 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
8712 (rs6000_preferred_simd_mode): Likewise.
8713 (output_vec_const_move): Likewise.
8714 (rs6000_expand_vector_extract): Likewise.
8715 (rs6000_split_vec_extract_var): Likewise.
8716 (reg_offset_addressing_ok_p): Likewise.
8717 (rs6000_legitimate_offset_address_p): Likewise.
8718 (rs6000_legitimize_address): Likewise.
8719 (rs6000_emit_set_const): Likewise.
8720 (rs6000_const_vec): Likewise.
8721 (rs6000_emit_move): Likewise.
8722 (spe_build_register_parallel): Likewise.
8723 (rs6000_darwin64_record_arg_recurse): Likewise.
8724 (swap_selector_for_mode): Likewise.
8725 (spe_init_builtins): Likewise.
8726 (paired_init_builtins): Likewise.
8727 (altivec_init_builtins): Likewise.
8728 (do_load_for_compare): Likewise.
8729 (rs6000_generate_compare): Likewise.
8730 (rs6000_expand_float128_convert): Likewise.
8731 (emit_load_locked): Likewise.
8732 (emit_store_conditional): Likewise.
8733 (rs6000_output_function_epilogue): Likewise.
8734 (rs6000_handle_altivec_attribute): Likewise.
8735 (rs6000_function_value): Likewise.
8736 (emit_fusion_gpr_load): Likewise.
8737 (emit_fusion_p9_load): Likewise.
8738 (emit_fusion_p9_store): Likewise.
8739 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
8740 (fusion_gpr_mem_load): Likewise.
8741 (fusion_addis_mem_combo_load): Likewise.
8742 (fusion_addis_mem_combo_store): Likewise.
8743 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
8744 (fusion_gpr_mem_load): Likewise.
8745 (fusion_addis_mem_combo_load): Likewise.
8746 (fusion_addis_mem_combo_store): Likewise.
8747 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8749 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
8750 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8751 (rs6000_preferred_simd_mode): Likewise.
8752 (output_vec_const_move): Likewise.
8753 (rs6000_expand_vector_extract): Likewise.
8754 (rs6000_split_vec_extract_var): Likewise.
8755 (reg_offset_addressing_ok_p): Likewise.
8756 (rs6000_legitimate_offset_address_p): Likewise.
8757 (rs6000_legitimize_address): Likewise.
8758 (rs6000_emit_set_const): Likewise.
8759 (rs6000_const_vec): Likewise.
8760 (rs6000_emit_move): Likewise.
8761 (rs6000_darwin64_record_arg_recurse): Likewise.
8762 (swap_selector_for_mode): Likewise.
8763 (paired_init_builtins): Likewise.
8764 (altivec_init_builtins): Likewise.
8765 (rs6000_expand_float128_convert): Likewise.
8766 (emit_load_locked): Likewise.
8767 (emit_store_conditional): Likewise.
8768 (rs6000_output_function_epilogue): Likewise.
8769 (rs6000_handle_altivec_attribute): Likewise.
8770 (rs6000_function_value): Likewise.
8771 (emit_fusion_gpr_load): Likewise.
8772 (emit_fusion_p9_load): Likewise.
8773 (emit_fusion_p9_store): Likewise.
8774 * config/rx/rx.c (rx_gen_move_template): Likewise.
8775 (flags_from_mode): Likewise.
8776 * config/s390/predicates.md (s390_alc_comparison): Likewise.
8777 (s390_slb_comparison): Likewise.
8778 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
8779 (s390_vector_mode_supported_p): Likewise.
8780 (s390_cc_modes_compatible): Likewise.
8781 (s390_match_ccmode_set): Likewise.
8782 (s390_canonicalize_comparison): Likewise.
8783 (s390_emit_compare_and_swap): Likewise.
8784 (s390_branch_condition_mask): Likewise.
8785 (s390_rtx_costs): Likewise.
8786 (s390_secondary_reload): Likewise.
8787 (__SECONDARY_RELOAD_CASE): Likewise.
8788 (s390_expand_cs): Likewise.
8789 (s390_preferred_simd_mode): Likewise.
8790 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
8791 * config/sh/sh.c (sh_print_operand): Likewise.
8792 (dump_table): Likewise.
8793 (sh_secondary_reload): Likewise.
8794 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8795 * config/sh/sh.md (casesi_worker_1): Likewise.
8796 (casesi_worker_2): Likewise.
8797 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
8798 (fcc_comparison_operator): Likewise.
8799 * config/sparc/sparc.c (sparc_expand_move): Likewise.
8800 (emit_soft_tfmode_cvt): Likewise.
8801 (sparc_preferred_simd_mode): Likewise.
8802 (output_cbranch): Likewise.
8803 (sparc_print_operand): Likewise.
8804 (sparc_expand_vec_perm_bmask): Likewise.
8805 (vector_init_bshuffle): Likewise.
8806 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8807 (spu_vector_mode_supported_p): Likewise.
8808 (spu_expand_insv): Likewise.
8809 (spu_emit_branch_or_set): Likewise.
8810 (spu_handle_vector_attribute): Likewise.
8811 (spu_builtin_splats): Likewise.
8812 (spu_builtin_extract): Likewise.
8813 (spu_builtin_promote): Likewise.
8814 (spu_expand_sign_extend): Likewise.
8815 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8816 (tilegx_simd_int): Likewise.
8817 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
8818 (tilepro_simd_int): Likewise.
8819 * config/v850/v850.c (const_double_split): Likewise.
8820 (v850_print_operand): Likewise.
8821 (ep_memory_offset): Likewise.
8822 * config/vax/vax.c (vax_rtx_costs): Likewise.
8823 (vax_output_int_move): Likewise.
8824 (vax_output_int_add): Likewise.
8825 (vax_output_int_subtract): Likewise.
8826 * config/visium/predicates.md (visium_branch_operator): Likewise.
8827 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
8828 (visium_print_operand_address): Likewise.
8829 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8830 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
8831 (xtensa_expand_conditional_branch): Likewise.
8832 (xtensa_copy_incoming_a7): Likewise.
8833 (xtensa_output_literal): Likewise.
8834 * dfp.c (decimal_real_maxval): Likewise.
8835 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8838 Alan Hayward <alan.hayward@arm.com>
8839 David Sherwood <david.sherwood@arm.com>
8841 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
8842 (mode_nunits_inline): Likewise.
8843 (mode_inner_inline): Likewise.
8844 (mode_unit_size_inline): Likewise.
8845 (mode_unit_precision_inline): Likewise.
8846 (emit_insn_modes_h): Likewise. Also emit a #define of the
8848 (emit_mode_wider): Add an E_ prefix to mode names.
8849 (emit_mode_complex): Likewise.
8850 (emit_mode_inner): Likewise.
8851 (emit_mode_adjustments): Likewise.
8852 (emit_mode_int_n): Likewise.
8853 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
8854 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
8855 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
8856 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
8857 (CRC32_BUILTIN, ENTRY): Likewise.
8858 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
8859 (aarch64_pop_regs): Likewise.
8860 (aarch64_process_components): Likewise.
8861 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
8862 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
8863 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
8864 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
8865 * config/arm/arm.c (arm_init_libfuncs): Likewise.
8866 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
8868 * config/i386/i386-builtin.def (pcmpestr): Likewise.
8869 (pcmpistr): Likewise.
8870 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
8871 * config/mmix/mmix.c (mmix_output_condition): Likewise.
8872 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8874 * config/rl78/rl78.c (mduc_regs): Likewise.
8875 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8876 (htm_expand_builtin): Likewise.
8877 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
8878 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
8879 * config/xtensa/xtensa.c (print_operand): Likewise.
8880 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
8881 (NUM_MODE_VECTOR_INT): Likewise.
8882 * genoutput.c (null_operand): Likewise.
8883 (output_operand_data): Likewise.
8884 * genrecog.c (print_parameter_value): Likewise.
8885 * lra.c (debug_operand_data): Likewise.
8887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8889 * dwarf2out.c (scompare_loc_descriptor_wide)
8890 (scompare_loc_descriptor_narrow): New functions, split out from...
8891 (scompare_loc_descriptor): ...here.
8892 * expmed.c (emit_store_flag_int): New function, split out from...
8893 (emit_store_flag): ...here.
8895 2017-08-30 Richard Biener <rguenther@suse.de>
8897 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
8898 (dwarf2out_early_finish): Move setting of AT_pubnames from
8899 early debug output to early finish.
8901 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8903 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
8904 and -mdata-region to the assembler.
8906 2017-08-30 Richard Biener <rguenther@suse.de>
8908 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
8910 (gen_subprogram_die): Add DW_AT_object_pointer only early.
8911 (dwarf2out_early_global_decl): Only generate a DIE for the
8912 abstract origin if it doesn't already exist or is a declaration DIE.
8913 (resolve_addr): Do not add the linkage name twice when
8914 generating a stub DIE for the DW_TAG_GNU_call_site target.
8916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8918 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8919 Use machine_mode rather than int for arg1_mode.
8921 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8924 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
8925 that the second argument of the built-in functions to unpack
8926 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
8927 switch statement instead a lot of if statements.
8928 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
8929 Allow 64-bit values to be in Altivec registers as well as
8930 traditional floating point registers.
8931 (pack<mode>, FMOVE128_VSX iterator): Likewise.
8933 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
8935 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
8936 MAX_REGS_PER_ADDRESS == 1.
8938 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
8940 * config/i386/i386.opt (flag_fentry): Do not init to -1.
8941 * config/i386/i386.c (ix86_option_override_internal): Simplify
8942 setting of opts->x_flag_entry.
8944 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8945 Jakub Jelinek <jakub@redhat.com>
8946 Richard Biener <rguenther@suse.de>
8948 PR tree-optimization/81503
8949 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
8950 folded constant fits in the target type; reorder tests for clarity.
8952 2017-08-29 Martin Liska <mliska@suse.cz>
8954 * passes.def: Include pass_lower_switch.
8955 * stmt.c (dump_case_nodes): Remove and move to
8956 tree-switch-conversion.
8957 (case_values_threshold): Likewise.
8958 (expand_switch_as_decision_tree_p): Likewise.
8959 (emit_case_decision_tree): Likewise.
8960 (expand_case): Likewise.
8961 (balance_case_nodes): Likewise.
8962 (node_has_low_bound): Likewise.
8963 (node_has_high_bound): Likewise.
8964 (node_is_bounded): Likewise.
8965 (emit_case_nodes): Likewise.
8966 (struct simple_case_node): New struct.
8967 (add_case_node): Remove.
8968 (emit_case_dispatch_table): Use vector instead of case_list.
8969 (reset_out_edges_aux): Remove.
8970 (compute_cases_per_edge): Likewise.
8971 (expand_case): Build list of simple_case_node.
8972 (expand_sjlj_dispatch_table): Use it.
8973 * tree-switch-conversion.c (struct case_node): Moved from
8974 stmt.c and adjusted.
8975 (emit_case_nodes): Likewise.
8976 (node_has_low_bound): Likewise.
8977 (node_has_high_bound): Likewise.
8978 (node_is_bounded): Likewise.
8979 (case_values_threshold): Likewise.
8980 (reset_out_edges_aux): Likewise.
8981 (compute_cases_per_edge): Likewise.
8982 (add_case_node): Likewise.
8983 (dump_case_nodes): Likewise.
8984 (balance_case_nodes): Likewise.
8985 (expand_switch_as_decision_tree_p): Likewise.
8986 (emit_jump): Likewise.
8987 (emit_case_decision_tree): Likewise.
8988 (try_switch_expansion): Likewise.
8989 (do_jump_if_equal): Likewise.
8990 (emit_cmp_and_jump_insns): Likewise.
8991 (fix_phi_operands_for_edge): New function.
8992 (record_phi_operand_mapping): Likewise.
8993 (class pass_lower_switch): New pass.
8994 (pass_lower_switch::execute): New function.
8995 (make_pass_lower_switch): Likewise.
8996 (conditional_probability):
8997 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
8998 * tree-pass.h: Add make_pass_lower_switch.
9000 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9003 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
9006 2017-08-29 Richard Biener <rguenther@suse.de>
9008 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
9009 we do not add a DW_AT_inline attribute twice.
9010 (gen_subprogram_die): Remove code setting DW_AT_inline on
9011 DECL_ABSTRACT_P nodes.
9013 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
9015 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
9016 calls to internal functions.
9017 (gimplify_modify_expr): Likewise.
9018 * tree-call-cdce.c (use_internal_fn): Likewise.
9019 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9020 (convert_to_divmod): Set the nothrow flag.
9021 * tree-if-conv.c (predicate_mem_writes): Likewise.
9022 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9023 (vectorizable_call): Likewise.
9024 (vectorizable_store): Likewise.
9025 (vectorizable_load): Likewise.
9026 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
9027 (vect_recog_mask_conversion_pattern): Likewise.
9029 2017-08-29 Martin Liska <mliska@suse.cz>
9032 * gcc.c (driver_handle_option): Add new argument.
9033 * opts-common.c (handle_option): Pass
9034 target_option_override_hook.
9035 * opts-global.c (lang_handle_option): Add new option.
9036 (set_default_handlers): Add new argument.
9037 (decode_options): Likewise.
9038 * opts.c (target_handle_option): Likewise.
9039 (common_handle_option): Call target_option_override_hook.
9040 * opts.h (struct cl_option_handler_func): Add hook for
9041 target option override.
9042 (struct cl_option_handlers): Likewise.
9043 (set_default_handlers): Add new argument.
9044 (decode_options): Likewise.
9045 (common_handle_option): Likewise.
9046 (target_handle_option): Likewise.
9047 * toplev.c (toplev::main): Pass targetm.target_option.override
9050 2017-08-29 Richard Biener <rguenther@suse.de>
9051 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9053 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
9054 life to the active subtree.
9056 2017-08-28 Jeff Law <law@redhat.com>
9058 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
9059 derive_equivalences.
9060 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
9062 (edge_info::derive_equivalences): New private member function
9064 * tree-ssa-dom.c (class edge_info): Changed from a struct
9065 to a class. Add ctor/dtor, methods and data members.
9066 (edge_info::edge_info): Renamed from allocate_edge_info.
9067 Initialize additional members.
9068 (edge_info::~edge_info): New.
9069 (free_dom_edge_info): Delete the edge info.
9070 (record_edge_info): Use new class & associated member functions.
9071 Tighten forms for testing for edge equivalences.
9072 (record_temporary_equivalences): Iterate over the simple
9073 equivalences rather than assuming there's only one per edge.
9074 (cprop_into_successor_phis): Iterate over the simple
9075 equivalences rather than assuming there's only one per edge.
9076 (optimize_stmt): Use operand_equal_p rather than pointer
9077 equality for mini-DSE code.
9079 2017-08-28 Nathan Sidwell <nathan@acm.org>
9081 * gcc.c (execute): Fold SIGPIPE handling into switch
9082 statement. Adjust internal error message.
9084 2017-08-28 Richard Biener <rguenther@suse.de>
9087 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
9088 Do nothing for removed DIEs.
9090 2017-08-28 Richard Biener <rguenther@suse.de>
9092 PR tree-optimization/81977
9093 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
9096 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9099 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9100 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9101 using targetm.gen_mem_thread_fence.
9103 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9106 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9107 predicate to register_operand. Reorder operands.
9108 (*btr<mode>): Ditto.
9109 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9110 (*btr<mode>_mask): Ditto.
9112 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9114 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9115 * config/rs6000/xmmintrin.h: New file.
9116 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9118 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9121 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9122 parameter and_insn and return it.
9123 (recombine_lvx_pattern): Insert a copy to ensure availability of
9124 the base register of the copied masking operation at the point of
9125 the instruction replacement.
9126 (recombine_stvx_pattern): Likewise.
9128 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9130 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9131 undocumented switches.
9132 (-mpower9-dform-vector): Likewise.
9133 (-mpower9-dform): Likewise.
9134 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9135 comments to delete references to -mpower9-dform* switches.
9136 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9137 Delete reference to -mpower9-dform* switches, test for
9138 -mpower9-vector instead.
9139 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9140 (OTHER_P9_VECTOR_MASKS): Likewise.
9141 (POWERPC_MASKS): Likewise.
9142 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9143 tests against -mpower9-dform* to -mpower9-vector. Delete code
9144 that checked for -mpower9-dform* consistancy with other options.
9145 Add test for -mpower9-misc to enable other power9 switches.
9146 (rs6000_init_hard_regno_mode_ok): Likewise.
9147 (rs6000_option_override_internal): Likewise.
9148 (rs6000_emit_prologue): Likewise.
9149 (rs6000_emit_epilogue): Likewise.
9150 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9151 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9152 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9154 (emit_fusion_p9_store): Likewise.
9155 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9156 resetting these macros if the assembler does not support ISA 3.0
9158 (TARGET_P9_DFORM_VECTOR): Likewise.
9159 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9160 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9162 2017-08-25 Alan Modra <amodra@gmail.com>
9164 PR rtl-optimization/81747
9165 * cse.c (cse_extended_basic_block): Don't attempt to record
9166 equivalences for degenerate conditional branches that branch
9167 to their fall-through.
9169 2017-08-24 Martin Sebor <msebor@redhat.com>
9172 * gimple-fold.c (size_must_be_zero_p): New function.
9173 (gimple_fold_builtin_memory_op): Call it.
9175 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
9177 * config/rs6000/mm_malloc.h: New file.
9179 2017-08-24 Bin Cheng <bin.cheng@arm.com>
9181 PR tree-optimization/81913
9182 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
9183 analysis when either IVs in condition can wrap.
9185 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
9187 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
9188 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
9190 2017-08-24 Richard Biener <rguenther@suse.de>
9193 * targhooks.c (default_target_can_inline_p): Properly
9194 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
9195 is present and always compare.
9196 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
9197 infer -mfpmath=sse from TARGET_SSE_P.
9198 (ix86_can_inline_p): Properly use target_option_default_node when
9199 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
9201 2017-08-24 Richard Biener <rguenther@suse.de>
9204 * dwarf2out.c (output_die): Handle flag_generate_offload like
9206 (output_comp_unit): Likewise.
9207 (gen_array_type_die): Likewise.
9208 (dwarf2out_early_finish): Likewise.
9209 (note_variable_value_in_expr): Likewise.
9210 (dwarf2out_finish): Likewise. Adjust assert.
9211 * cgraphunit.c (symbol_table::compile): Move setting of
9212 flag_generate_offload earlier ...
9213 (symbol_table::finalize_compilation_unit): ... here, before
9214 early debug finalization.
9216 2017-08-24 Richard Biener <rguenther@suse.de>
9218 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
9219 and ipa-fnsummary.h.
9220 (ix86_can_inline_p): When ix86_fpmath flags do not match
9221 check whether the callee uses FP math at all.
9223 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
9226 * tree-ssanames.c (get_nonzero_bits): Use element_precision
9227 instead of TYPE_PRECISION.
9229 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
9230 Alan Hayward <alan.hayward@arm.com>
9231 David Sherwood <david.sherwood@arm.com>
9233 * combine.c (make_extraction): Use subreg_offset_from_lsb.
9235 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
9237 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
9239 (ix86_frame::stack_realign_allocate): New field.
9240 (struct machine_frame_state): Modify comments.
9241 (machine_frame_state::sp_realigned_fp_end): New field.
9242 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
9244 (sp_valid_at): Add assertion to assure no attempt to access invalid
9245 offset of a realigned stack.
9246 (fp_valid_at): Likewise.
9247 (choose_baseaddr): Modify comments.
9248 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
9249 ix86_expand_prologue.
9250 (ix86_expand_prologue): Modify stack realignment and allocation.
9251 (ix86_expand_epilogue): Modify comments.
9252 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
9253 avx2_runtime, avx512f, and avx512f_runtime.
9255 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
9257 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
9258 (mstackrealign): Do not init to -1.
9259 * config/i386/i386.c (ix86_option_override_internal):
9260 Check opts_set, not opts when setting default value of
9261 opts->x_ix86_force_align_arg_pointer.
9263 2017-08-23 Richard Biener <rguenther@suse.de>
9265 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
9266 lang_hooks.decl_printable_name.
9267 * print-rtl-function.c (print_rtx_function): Likewise.
9268 * tree-pretty-print.c (dump_function_header): Likewise.
9270 2017-08-23 Richard Biener <rguenther@suse.de>
9273 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
9274 -g0 at compile-time.
9276 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9279 * doc/sourcebuild.texi (Other hardware attributes):
9282 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9285 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
9286 Fix single-use check.
9288 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
9290 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
9292 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9294 * doc/install.texi: Modify to add more details on running selected
9297 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9299 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
9300 is combined with -mabi=ms.
9301 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
9304 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9306 PR tree-optimization/81488
9307 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
9308 and cached_basis fields.
9309 (MAX_SPREAD): New constant.
9310 (alloc_cand_and_find_basis): Initialize new fields.
9311 (clear_visited): New function.
9312 (create_phi_basis_1): Rename from create_phi_basis, set visited
9313 and cached_basis fields.
9314 (create_phi_basis): New wrapper function.
9315 (phi_add_costs_1): Rename from phi_add_costs, add spread
9316 parameter, set visited field, short-circuit when limits reached.
9317 (phi_add_costs): New wrapper function.
9318 (record_phi_increments_1): Rename from record_phi_increments, set
9320 (record_phi_increments): New wrapper function.
9321 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
9322 (phi_incr_cost): New wrapper function.
9323 (all_phi_incrs_profitable_1): Rename from
9324 all_phi_incrs_profitable, set visited field.
9325 (all_phi_incrs_profitable): New wrapper function.
9327 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9328 Alan Hayward <alan.hayward@arm.com>
9329 David Sherwood <david.sherwood@arm.com>
9331 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
9332 that takes the outer and inner modes.
9333 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
9334 comparison as the canonical test for a paradoxical subreg.
9335 * combine.c (simplify_set): Use paradoxical_subreg_p.
9336 (make_extraction): Likewise.
9337 (force_to_mode): Likewise.
9338 (rtx_equal_for_field_assignment_p): Likewise.
9339 (gen_lowpart_for_combine): Likewise.
9340 (simplify_comparison): Likewise.
9341 * cse.c (equiv_constant): Likewise.
9342 * expmed.c (store_bit_field_1): Likewise.
9343 * final.c (alter_subreg): Likewise.
9344 * fwprop.c (propagate_rtx): Likewise.
9345 (forward_propagate_subreg): Likewise.
9346 * ira-conflicts.c (ira_build_conflicts): Likewise.
9347 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
9348 * lra-constraints.c (curr_insn_transform): Likewise.
9349 (split_reg): Likewise.
9350 * lra-eliminations.c (move_plus_up): Likewise.
9351 (lra_eliminate_regs_1): Likewise.
9352 * recog.c (general_operand): Likewise.
9353 * ree.c (combine_reaching_defs): Likewise.
9354 * reload.c (push_reload): Likewise.
9355 (find_reloads): Likewise.
9356 * reload1.c (elimination_effects): Likewise.
9357 (compute_reload_subreg_offset): Likewise.
9358 (choose_reload_regs): Likewise.
9359 * rtlanal.c (subreg_lsb_1): Likewise.
9360 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9361 (simplify_subreg): Likewise.
9362 * var-tracking.c (track_loc_p): Likewise.
9363 * emit-rtl.c (byte_lowpart_offset): Likewise.
9364 (paradoxical_subreg_p): Delete out-of-line definition.
9366 2017-08-22 Jeff Law <law@redhat.com>
9368 PR tree-optimization/81741
9369 PR tree-optimization/71947
9370 * tree-ssa-dom.c: Include tree-inline.h.
9371 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
9372 equivalences if one is more expensive to compute than the other.
9373 * tree-ssa-scopedtables.h (class const_or_copies): Make
9374 record_const_or_copy_raw method private.
9375 (class avail_exprs_stack): New method simplify_binary_operation.
9376 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
9377 avail_exprs_stack::simplify_binary_operation as needed.
9378 (avail_exprs_stack::simplify_binary_operation): New function.
9380 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
9382 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
9383 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
9384 (DOT_SYMBOLS): Likewise.
9385 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
9386 (RELOCATABLE_NEEDS_FIXUP): Likewise.
9387 (RS6000_ABI_NAME): Likewise.
9388 (TARGET_CMODEL): Likewise.
9389 (TOC_SECTION_ASM_OP): Likewise.
9390 (SET_CMODEL): New macro.
9391 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
9393 2017-08-22 Richard Biener <rguenther@suse.de>
9395 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
9396 to allow for free-lang-data replacements similar to verify_type_variant.
9398 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9399 Alan Hayward <alan.hayward@arm.com>
9400 David Sherwood <david.sherwood@arm.com>
9402 * config/aarch64/aarch64.md (casesi): Use DImode rather than
9403 VOIDmode for the LABEL_REF.
9405 2017-08-22 Richard Biener <rguenther@suse.de>
9407 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
9409 2017-08-22 Richard Biener <rguenther@suse.de>
9411 * common.opt (feliminate-dwarf2-dups): Ignore.
9412 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
9413 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
9414 same_die_p_wrap, compute_section_prefix,
9415 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
9416 (comdat_symbol_id, comdat_symbol_number): Likewise.
9417 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
9419 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
9420 (output_die): Mark unreachable path unreachable.
9421 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
9422 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
9423 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
9424 (dwarf2out_early_finish): Likewise.
9426 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
9428 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
9430 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
9433 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
9434 not VAR_P. Filter attribute warnings with OPT_Wattributes.
9435 (avr_attribute_table) <io, io_low, address>: Initialize
9436 .decl_required with true.
9438 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
9440 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
9441 undocumented debugging options.
9442 (-mvsx-scalar-double): Likewise.
9443 (-mallow-df-permute): Likewise.
9444 (-mvectorize-builtins): Likewise.
9445 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9446 (rs6000_builtin_vectorized_function): Likewise.
9447 (rs6000_builtin_md_vectorized_function): Likewise.
9448 (rs6000_opt_vars): Likewise.
9450 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
9453 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
9454 (*btrq_imm): Rename from *btrq.
9455 (*btcq_imm): Rename from *btcq.
9456 (btsc): New code attribute.
9457 (*<btsc><mode>): New insn pattern.
9458 (*btr<mode>): Ditto.
9459 (*<btsc><mode>_mask): New insn_and_split pattern.
9460 (*btr<mode>_mask): Ditto.
9462 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9463 Alan Hayward <alan.hayward@arm.com>
9464 David Sherwood <david.sherwood@arm.com>
9466 * function.c (pad_below): Simplify padding calculation.
9468 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9469 Alan Hayward <alan.hayward@arm.com>
9470 David Sherwood <david.sherwood@arm.com>
9472 * target.def (function_prologue): Remove frame size argument.
9473 (function_epilogue): Likewise.
9474 * doc/tm.texi: Regenerate.
9475 * final.c (final_start_function): Update call to function_prologue.
9476 (final_end_function): Update call to function_epilogue.
9477 (default_function_pro_epilogue): Remove frame size argument.
9478 * output.h (default_function_pro_epilogue): Likewise.
9479 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
9480 (arm_output_function_prologue): Likewise.
9481 * config/frv/frv.c (frv_function_prologue): Likewise.
9482 (frv_function_epilogue): Likewise.
9483 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
9484 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
9485 (ia64_output_function_epilogue): Likewise.
9486 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
9487 (m32r_output_function_epilogue): Likewise.
9488 * config/microblaze/microblaze.c (microblaze_function_prologue)
9489 (microblaze_function_epilogue): Likewise.
9490 * config/mips/mips.c (mips_output_function_prologue): Likewise.
9491 (mips_output_function_epilogue): Likewise.
9492 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
9493 (mmix_target_asm_function_epilogue): Likewise.
9494 * config/msp430/msp430.c (msp430_start_function): Likewise.
9495 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
9496 (nds32_asm_function_epilogue): Likewise.
9497 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
9498 * config/pa/pa.c (pa_output_function_prologue): Likewise.
9499 (pa_output_function_epilogue): Likewise.
9500 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
9501 (rs6000_output_function_epilogue): Likewise.
9502 * config/rl78/rl78.c (rl78_start_function): Likewise.
9503 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
9504 (rs6000_output_function_epilogue): Likewise.
9505 * config/rx/rx.c (rx_output_function_prologue): Likewise.
9506 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
9507 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
9508 (sparc_asm_function_epilogue): Likewise.
9510 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9512 * tree.h (type_has_mode_precision_p): New function.
9513 * convert.c (convert_to_integer_1): Use it.
9514 * expr.c (expand_expr_real_2): Likewise.
9515 (expand_expr_real_1): Likewise.
9516 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
9517 * match.pd: Likewise.
9518 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
9519 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
9520 * tree-tailcall.c (process_assignment): Likewise.
9521 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9522 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
9523 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
9524 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9525 (vectorizable_assignment): Likewise.
9526 (vectorizable_shift): Likewise.
9527 (vectorizable_operation): Likewise.
9528 * tree-vrp.c (register_edge_assert_for_2): Likewise.
9530 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
9532 * match.pd: Add pow (C, x) simplification.
9534 2017-08-21 Richard Biener <rguenther@suse.de>
9536 PR tree-optimization/81900
9537 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
9538 for blocks with abnormal predecessors.
9539 (compute_antic): Do not set visited flag prematurely.
9541 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
9544 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
9546 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9548 * stor-layout.h (vector_type_mode): Move to...
9549 * tree.h (vector_type_mode): ...here.
9550 * stor-layout.c (vector_type_mode): Move to...
9551 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
9553 2017-08-21 Richard Biener <rguenther@suse.de>
9555 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
9556 register_external_die hooks.
9557 (debug_false_tree_charstarstar_uhwistar): Declare.
9558 (debug_nothing_tree_charstar_uhwi): Likewise.
9559 * debug.c (do_nothing_debug_hooks): Adjust.
9560 (debug_false_tree_charstarstar_uhwistar): New do nothing.
9561 (debug_nothing_tree_charstar_uhwi): Likewise.
9562 * dbxout.c (dbx_debug_hooks): Adjust.
9563 (xcoff_debug_hooks): Likewise.
9564 * sdbout.c (sdb_debug_hooks): Likewise.
9565 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9566 * dwarf2out.c (macinfo_label_base): New global.
9567 (dwarf2out_register_external_die): New function for the
9568 register_external_die hook.
9569 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
9570 (dwarf2_debug_hooks): Use them.
9571 (dwarf2_lineno_debug_hooks): Adjust.
9572 (struct die_struct): Add with_offset flag.
9573 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
9574 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
9575 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
9576 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
9577 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
9578 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
9579 defining section names for the early LTO debug variants.
9580 (reset_indirect_string): New helper.
9581 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
9582 (print_dw_val): Add support for offsetted symbol references.
9583 (get_ultimate_context): Split out from is_cxx.
9584 (is_cxx): Use get_ultimate_context.
9585 (is_fortran): Add decl overload.
9586 (compute_comp_unit_symbol): Split out worker from
9587 compute_section_prefix.
9588 (compute_section_prefix): Call compute_comp_unit_symbol and
9589 set comdat_type_p here.
9590 (output_die): Skip DIE symbol output for the LTO added one.
9591 Handle DIE symbol references with offset.
9592 (output_comp_unit): Guard section name mangling properly.
9593 For LTO debug sections emit a symbol at the section beginning
9594 which we use to refer to its DIEs.
9595 (add_abstract_origin_attribute): For DIEs registered via
9596 dwarf2out_register_external_die directly refer to the early
9597 DIE rather than indirectly through the shadow one we created.
9598 Remove obsolete call to dwarf2out_abstract_function for
9599 non-function/block origins.
9600 (gen_array_type_die): When generating early LTO debug do
9601 not emit DW_AT_string_length.
9602 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
9603 late when in LTO. As suggested place a gcc_unreachable for
9604 the DECL_ABSTRACT_P case.
9605 (gen_subprogram_die): Avoid another specification DIE
9606 for early built declarations/definitions for the late LTO case.
9607 (gen_variable_die): Add type references for late duplicated VLA dies
9609 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
9610 we have the abstract instance already.
9611 (process_scope_var): Adjust decl DIE contexts in LTO which
9612 first puts them in limbo.
9613 (gen_decl_die): Do not generate type DIEs late apart from
9614 types for VLAs or for decls we do not yet have a DIE. Do not
9615 call dwarf2out_abstract_function late.
9616 (dwarf2out_early_global_decl): Make sure to create DIEs
9617 for abstract instances of a decl first.
9618 (dwarf2out_late_global_decl): Adjust comment.
9619 (output_macinfo_op): With multiple macro sections use
9620 macinfo_label_base to distinguish labels.
9621 (output_macinfo): Likewise. Update macinfo_label_base.
9622 Pass in the line info label.
9623 (note_variable_value_in_expr): When generating LTO resolve
9624 all variable values here by generating DIEs as needed.
9625 (init_sections_and_labels): Add early LTO debug flag parameter
9626 and generate different sections and names if set. Add generation
9627 counter for the labels so we can have multiple of them.
9628 (reset_dies): Helper to allow DIEs to be output multiple times.
9629 (dwarf2out_finish): When outputting DIEs to the fat part of an
9630 LTO object first reset DIEs.
9631 (dwarf2out_early_finish): Output early DIEs when generating LTO.
9632 (modified_type_die): Check for decl_ultimate_origin being self
9634 (gen_type_die_with_usage): Likewise.
9635 (gen_typedef_die): Allow decl_ultimate_origin being self.
9636 (set_decl_abstract_flags): Remove.
9637 (set_block_abstract_flags): Likewise.
9638 (dwarf2out_abstract_function): Treat the early generated DIEs
9639 as the abstract copy and only add DW_AT_inline and
9640 DW_AT_artificial here and call set_decl_origin_self.
9641 If the DIE has an abstract origin don't do anything.
9642 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
9643 if we have none yet (Go fails to build one, PR78628).
9644 (variably_modified_type_p): Prevent endless recursion for Ada
9645 cyclic pointer types.
9646 * lto-streamer-in.c: Include debug.h.
9647 (dref_queue): New global.
9648 (lto_read_tree_1): Stream in DIE references.
9649 (lto_input_tree): Register DIE references.
9650 (input_function): Stream DECL_DEBUG_ARGS.
9651 * lto-streamer-out.c: Include debug.h.
9652 (lto_write_tree_1): Output DIE references.
9653 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
9654 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
9655 (output_function): Stream DECL_DEBUG_ARGS.
9656 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
9657 Stream DECL_ABSTRACT_ORIGIN.
9658 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
9659 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
9660 DECL_CONTEXT for file-scope decls.
9661 * lto-streamer.h (struct dref_entry): Declare.
9662 (dref_queue): Likewise.
9663 * cfgexpand.c (pass_expand::execute): Do not call the
9664 outlining_inline_function hook here.
9665 * lto-wrapper.c (debug_obj): New global.
9666 (tool_cleanup): Unlink it if required.
9667 (debug_objcopy): New function.
9668 (run_gcc): Handle early debug sections in the IL files by
9669 extracting them to separate files, partially linkin them and
9670 feeding the result back as result to the linker.
9671 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
9672 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
9673 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
9674 sections into a separate segment.
9675 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
9677 (darwin_asm_dwarf_section): Likewise.
9678 (darwin_asm_output_dwarf_offset): Likewise.
9679 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
9681 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9682 Alan Hayward <alan.hayward@arm.com>
9683 David Sherwood <david.sherwood@arm.com>
9685 * read-md.h (md_reader::record_potential_iterator_use): Replace
9686 pointer argument with an rtx and an index.
9687 * read-rtl.c (iterator_group::apply_iterator): Likewise.
9688 (apply_mode_iterator): Likewise.
9689 (apply_code_iterator): Likewise.
9690 (apply_int_iterator): Likewise.
9691 (apply_subst_iterator): Likewise.
9692 (record_iterator_use): Likewise.
9693 (record_attribute_use): Likewise.
9694 (md_reader::record_potential_iterator_use): Likewise. Update calls
9695 to record_iterator_use and apply_iterator.
9696 (iterator_use): Replace ptr with x and index.
9697 (attribute_use): Likewise.
9698 (apply_attribute_uses): Update calls to apply_iterator.
9699 (apply_iterators): Likewise. Update initialization of iterator_use.
9700 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
9701 and record_potential_iterator_use.
9702 (rtx_reader::read_rtx_operand): Likewise.
9704 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9705 Alan Hayward <alan.hayward@arm.com>
9706 David Sherwood <david.sherwood@arm.com>
9708 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
9711 2017-08-21 Richard Biener <rguenther@suse.de>
9714 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
9715 at struct end conservatively when comparing common bases.
9717 2017-08-21 Richard Biener <rguenther@suse.de>
9719 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
9720 (mem_ref_in_stmt): Remove.
9721 (determine_max_movement): Use ref index to get at the reference.
9722 (invariantness_dom_walker::before_dom_children): Deal with
9723 lim data already initialized.
9724 (gather_mem_refs_stmt): Initialize lim data and record ref index.
9726 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9728 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
9729 (TARGET_ISA_ROUND): Ditto.
9730 (TARGET_ROUND): Ditto.
9731 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
9732 * config/i386/i386.md: Ditto.
9733 * config/i386/sse.md: Ditto.
9734 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
9735 with OPTION_MASK_ISA_SSE4_1.
9737 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9740 * doc/extend.texi (x86 Built-in Functions): Correct the name of
9741 __builtin_ia32_lzcnt_u16.
9743 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9746 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
9747 (rs6000_set_current_function): Rewrite function to use it.
9749 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
9752 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
9753 and TYPE_WARN_IF_NOT_ALIGN.
9754 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
9755 (handle_warn_if_not_align): New.
9756 (place_union_field): Call handle_warn_if_not_align.
9757 (place_field): Call handle_warn_if_not_align.
9758 Copy TYPE_WARN_IF_NOT_ALIGN.
9759 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
9760 (layout_type): Likewise.
9761 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
9763 (tree_decl_common): Add warn_if_not_align.
9764 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
9765 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
9766 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
9767 (DECL_WARN_IF_NOT_ALIGN): Likewise.
9768 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
9769 * doc/extend.texi: Document warn_if_not_aligned attribute.
9770 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
9772 2017-08-17 Martin Liska <mliska@suse.cz>
9775 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
9776 (get_data_dependence): Use it as pointer type.
9777 (distribute_loop): Likewise.
9779 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9781 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
9782 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
9783 (p8_vmrgow_<mode>_direct): New define_insn.
9784 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
9785 handle endianness for vmrgew and vmrgow permute patterns.
9787 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9789 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
9790 * config/rs6000/rs6000-cpus.def: Remove comment.
9791 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
9792 (POWERPC_MASKS): Likewise.
9793 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
9794 use of TARGET_VSX_TIMODE.
9795 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
9796 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
9797 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
9798 (rs6000_option_override_internal): Remove dead code.
9799 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
9800 (rs6000_legitimize_reload_address): Likewise.
9801 (rs6000_legitimate_address_p): Likewise.
9802 (rs6000_opt_masks): Delete "vsx-timode".
9803 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
9804 from function comment.
9805 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
9806 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
9807 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
9809 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
9810 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
9811 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
9814 (define_peephole2): Likewise.
9816 2017-08-17 Martin Sebor <msebor@redhat.com>
9819 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
9820 past the end of an array.
9821 (test_pp_format): Add test cases.
9823 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
9825 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
9826 missing ECF_NOTHROW flags.
9828 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9831 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
9832 operands residing in integer registers.
9833 (*vsx_le_perm_load_<mode>): Likewise.
9834 (*vsx_le_perm_store_<mode>): Likewise.
9835 (define_peephole2): Add peepholes to optimize the above.
9837 2017-08-17 Marek Polacek <polacek@redhat.com>
9840 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
9841 to mimic what shorten_compare did. Change the return type to bool.
9842 (fold_cond_expr_with_comparison): Update call to
9843 operand_equal_for_comparison_p.
9844 (fold_ternary_loc): Likewise.
9846 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
9848 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
9850 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
9852 2017-08-17 Richard Biener <rguenther@suse.de>
9854 * tree-ssa-structalias.c (solve_graph): When propagating
9855 to successors update the graphs succ edges and avoid duplicate work.
9857 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
9860 * config/i386/i386.c (ix86_option_override_internal): Save target
9861 specific options after ix86_stack_protector_guard_reg was changed.
9863 2017-08-17 Richard Biener <rguenther@suse.de>
9865 PR tree-optimization/81827
9866 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
9867 (new_var_info): Initialize it conservatively.
9868 (get_call_vi): Mark register vars.
9869 (new_scalar_tmp_constraint_exp): Likewise.
9870 (handle_rhs_call): Likewise.
9871 (handle_const_call): Likewise.
9872 (create_function_info_for): Likewise.
9873 (solve_constraints): Sort varinfos to separate register from
9874 non-register vars to pack points-to solution bitmaps during
9877 2017-08-17 Marek Polacek <polacek@redhat.com>
9879 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
9881 2017-08-17 Richard Biener <rguenther@suse.de>
9883 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
9884 to true when overflow is undefined and we saturated the result.
9886 2017-08-17 Alan Modra <amodra@gmail.com>
9889 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
9890 Don't use store multiple if only one reg needs saving.
9891 (interesting_frame_related_regno): New function.
9892 (rs6000_frame_related): Don't emit frame info for regs that
9894 (rs6000_emit_epilogue): Likewise.
9896 2017-08-16 Nathan Sidwell <nathan@acm.org>
9898 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
9899 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
9900 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
9901 * tree.c (free_lang_data_in_type): Use else-if chain. Always
9902 clear TYPE_LANG_1. Remove obsolete member-function stripping.
9903 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
9904 (verify_type): Adjust for TYPE_BINFO move.
9905 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
9906 process TYPE_BINFO directly.
9907 (hash_tree): Likewise.
9908 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9910 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9913 2017-08-16 David Malcolm <dmalcolm@redhat.com>
9915 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
9917 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
9920 * config/i386/i386.md (*anddi_1_btr): Change predicates of
9921 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
9922 Add ix86_binary_operator_ok to insn constraint.
9923 (*iordi_1_bts): Ditto.
9924 (*xordi_1_btc): Ditto.
9925 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
9926 Update corresponding peephole2 pattern.
9930 2017-08-16 Bin Cheng <bin.cheng@arm.com>
9932 PR tree-optimization/81832
9933 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
9934 copy loop header which has IFN_LOOP_DIST_ALIAS call.
9936 2017-08-16 Marek Polacek <polacek@redhat.com>
9939 * fold-const.c (fold_indirect_ref_1): Restore original behavior
9940 regarding size_zero_node.
9942 2017-08-16 Martin Liska <mliska@suse.cz>
9945 * config.gcc: Respect previously set extra_objs in case
9948 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
9950 PR tree-optimization/81835
9951 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
9952 the phi SSA_NAME. Check that the condition in a COND_EXPR does
9953 not depend on the phi.
9955 2017-08-16 Alan Modra <amodra@gmail.com>
9957 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
9960 2017-08-16 Alan Modra <amodra@gmail.com>
9962 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
9963 (save_reg_p): ..into this. Update all callers.
9964 (first_reg_to_save): Simplify.
9966 2017-08-16 Alan Modra <amodra@gmail.com>
9968 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
9971 2017-08-15 Joseph Myers <joseph@codesourcery.com>
9975 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
9976 constant count if that count is less than 32.
9978 2017-08-15 Nathan Sidwell <nathan@acm.org>
9980 * gcc.c (execute): Emit friendlier message if inferior is killed
9981 by an external cause.
9983 2017-08-15 Richard Biener <rguenther@suse.de>
9985 PR tree-optimization/81790
9986 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
9987 CONSTRUCTORs from simplifying and VN.
9989 2017-08-14 Martin Sebor <msebor@redhat.com>
9991 * builtin-attrs.def: Add comments.
9993 2017-08-14 Martin Sebor <msebor@redhat.com>
9996 * doc/extend.texi (attribute nonstring): Document new attribute.
9998 2017-08-14 Martin Sebor <msebor@redhat.com>
10001 * tree-diagnostic.c (default_tree_printer): Handle %G.
10002 * gimple-pretty-print.h (percent_G_format): Declare new function.
10003 * gimple-pretty-print.c (percent_G_format): Define.
10004 * tree-pretty-print.c (percent_K_format): Add argument.
10006 2017-08-14 Martin Sebor <msebor@redhat.com>
10008 PR translation/79998
10009 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
10010 Remove a stray space.
10012 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
10015 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
10016 (*iordi_1_bts): Ditto.
10017 (*xordi_1_btc): Ditto.
10019 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10022 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
10023 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10025 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
10026 quoted strings, and make more translator-friendly.
10027 (darwin_rs6000_override_options): Likewise.
10028 (rs6000_option_override_internal): Likewise.
10029 (rs6000_return_in_memory): Fix overlong line.
10030 (init_cmulative_args): Use quoted strings, and make more
10031 translator-friendly.
10032 (rs6000_pass_by_reference): Fix overlong line.
10033 (def_builtin): Use quoted strings.
10034 (altivec_expand_predicate_builtin): Use quoted strings, and make
10035 more translator-friendly.
10036 (htm_expand_builtin): Use quoted strings.
10037 (cpu_expand_builtin): Use quoted strings, and make more
10038 translator-friendly.
10039 (altivec_expand_builtin): Likewise.
10040 (paired_expand_predicate_builtin): Likewise.
10041 (rs6000_invalid_builtin): Likewise.
10042 (builtin_function_type): Use quoted strings.
10043 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
10044 more translator-friendly.
10045 (rs6000_trampoline_init): Likewise.
10046 (rs6000_handle_altivec_attribute): Likewise.
10047 (rs6000_inner_target_options): Use quoted strings.
10048 (rs6000_disable_incompatible_switches): Likewise.
10049 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
10050 strings, and make more translator-friendly.
10051 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
10053 2017-08-14 Bin Cheng <bin.cheng@arm.com>
10055 PR tree-optimization/81799
10056 * tree-loop-distribution.c (version_loop_by_alias_check): Force
10057 cond_expr to simple gimple operand.
10059 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
10061 PR middle-end/46932
10062 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
10064 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
10068 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
10069 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
10070 TARGET_GASISR_PROLOGUES.
10071 * config/avr/avr.c (avr_option_override): Same.
10072 (avr_pass_pre_proep::execute): Same.
10074 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
10077 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
10078 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
10080 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
10082 * config/i386/i386.md (*load_tp_<mode>): Redefine as
10083 define_insn_and_split. Split to a memory load from 0 in
10084 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
10085 using PTR mode iterator.
10086 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
10087 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
10088 (*add_tp_<mode>): Redefine as define_insn_and_split.
10089 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
10090 address space. Merge with *add_tp_x32 using PTR mode iterator.
10091 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
10092 Split to an add with a memory load from 0 in
10093 DEFAULT_TLS_SEG_REG address space.
10095 2017-08-12 Andrew Pinski <apinski@cavium.com>
10097 * config/aarch64/aarch64-option-extensions.def (rdma):
10098 Fix feature string to what Linux prints out in /proc/cpuinfo.
10100 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10103 * dwarf2out.c (modified_type_die): For C typedef types that have
10104 an ultimate origin, process the ultimate origin instead of the
10106 (gen_typedef_die): Assert that input DECLs have no ultimate
10108 (gen_type_die_with_usage): For typedef variants that have an
10109 ultimate origin, just call gen_decl_die on the original DECL.
10110 (process_scope_var): Avoid creating DIEs for local typedefs and
10111 concrete static variables.
10113 2017-08-12 Alan Modra <amodra@gmail.com>
10117 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10118 match gnu-user.h startfile.
10119 (ENDFILE_LINUX_SPEC): Similarly.
10121 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10124 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10126 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10128 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10129 * config/aarch64/aarch64.md (mov<mode>): Change.
10130 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10131 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10132 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10134 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10136 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10137 for storage order barriers.
10139 2017-08-11 Martin Liska <mliska@suse.cz>
10142 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10143 variables of void type.
10145 2017-08-11 Martin Liska <mliska@suse.cz>
10147 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10148 TARGET_SUPPORTS_ALIASES.
10149 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10150 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10151 (optimize_weakref): Likewise.
10152 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10153 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10154 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10156 2017-08-11 Martin Liska <mliska@suse.cz>
10159 * config/i386/i386.c (make_resolver_func): Do complete
10160 refactoring of the function.
10162 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10165 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10166 * config/i386/i386.c (ix86_stack_protect_guard): Use
10167 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10168 * doc/invoke.texi (x86 Options): Document
10169 -mstack-protector-guard-symbol= option.
10171 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10173 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10174 * config/i386/i386.c (ix86_split_stack_guard): New function.
10175 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
10176 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
10177 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
10178 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
10179 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
10180 (split_stack_space_check): Call ix86_split_stack_guard.
10182 2017-08-10 Martin Sebor <msebor@redhat.com>
10184 * print-tree.c (print_node): Print location using the established
10186 Replace spaces with colons.
10187 (debug_raw, debug): Ditto.
10189 2017-08-10 Martin Sebor <msebor@redhat.com>
10192 * pretty-print.c (pp_format): Correct the handling of %s precision.
10194 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
10197 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10199 (ix86_finalize_stack_frame_flags): This. Also clear
10200 frame_pointer_needed if -fno-omit-frame-pointer is used without
10202 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10203 with ix86_finalize_stack_frame_flags.
10204 (ix86_expand_epilogue): Likewise.
10205 (ix86_expand_split_stack_prologue): Likewise.
10206 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
10208 2017-08-10 Martin Liska <mliska@suse.cz>
10211 * c-attribs.c (handle_target_attribute):
10212 Report warning for an empty string argument of target attribute.
10214 2017-08-09 Jakub Jelinek <jakub@redhat.com>
10217 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
10219 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
10220 or DECL_NONLOCAL labels.
10221 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
10222 or DECL_NONLOCAL labels here.
10224 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
10226 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
10227 to indicate when early gimple folding has been disabled.
10228 (rs6000_gimple_fold_builtin): Add debug content.
10229 (rs6000_invalid_builtin): Fix whitespace.
10230 (rs6000_expand_builtin): Fix whitespace.
10231 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
10233 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
10236 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
10237 SAVE_MULTIPLE if not all the registers that saves, should be saved.
10239 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
10241 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
10242 (qdf24xx): Likewise.
10243 * config/aarch64/aarch64.md: Include falkor.md.
10244 * config/aarch64/falkor.md: New.
10246 2017-08-09 Marek Polacek <polacek@redhat.com>
10249 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
10250 * diagnostic.c (emit_diagnostic): Add a comment.
10251 (emit_diagnostic_valist): New function.
10253 2017-08-09 Marek Polacek <polacek@redhat.com>
10256 * input.c (make_location): New overload.
10257 * input.h (make_location): Declare.
10259 2017-08-08 Alan Modra <amodra@gmail.com>
10260 H.J. Lu <hongjiu.lu@intel.com>
10263 * gcc.c (NO_PIE_SPEC): Delete.
10264 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
10266 (LINK_PIE_SPEC): ..to here.
10267 (LINK_COMMAND_SPEC): Support -no-pie.
10268 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
10269 chain of crtbegin*.o selection, update for PIE_SPEC changes and
10271 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
10272 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
10273 (ENDFILE_CRTEND_SPEC): Similarly.
10275 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
10278 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
10279 (mstack-protector-guard-offset=): Ditto.
10280 * config/i386/i386.c (ix86_option_override): Handle
10281 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
10283 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
10284 ix86_stack_protect_guard_offset variables.
10285 (TARGET_STACK_PROTECT_GUARD): Always define.
10286 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
10287 and -mstack-protector-guard-offset= options.
10289 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10291 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
10292 boundary case for the last candidate.
10294 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10296 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
10297 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
10299 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10301 PR middle-end/19706
10302 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
10303 * config/aarch64/aarch64-builtins.c
10304 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
10305 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
10306 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
10308 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10309 Andrew Pinski <pinskia@gmail.com>
10311 PR middle-end/19706
10312 * internal-fn.def (XORSIGN): New.
10313 * optabs.def (xorsign_optab): New.
10314 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
10315 (convert_expand_mult_copysign): New.
10316 (pass_optimize_widening_mul::execute): Call
10317 convert_expand_mult_copysign.
10319 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10321 PR tree-optimization/81354
10322 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
10323 Insert on edges rather than explicitly creating landing pads.
10324 (analyze_candidates_and_replace): Commit edge inserts.
10326 2017-08-08 Richard Biener <rguenther@suse.de>
10328 PR middle-end/81719
10329 * tree-ssa-loop-niter.c: Include tree-dfa.h.
10330 (expand_simple_operations): Also look through ADDR_EXPRs with
10331 MEM_REF bases treating them as POINTER_PLUS_EXPR.
10333 2017-08-08 Richard Biener <rguenther@suse.de>
10335 PR tree-optimization/81723
10336 * tree-vect-slp.c (struct bst_traits): New hash traits.
10337 (bst_fail): New global.
10338 (vect_build_slp_tree_2): New worker, split out from ...
10339 (vect_build_slp_tree): ... this now wrapping it with using
10340 bst_fail set to cache SLP tree build fails. Properly handle
10342 (vect_analyze_slp_instance): Allocate and free bst_fail.
10344 2017-08-08 Martin Liska <mliska@suse.cz>
10347 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
10348 LABEL_DECLs that can be from a different function.
10350 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10352 PR tree-optimization/81744
10353 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
10354 loop's number of iterations.
10356 2017-08-08 Martin Liska <mliska@suse.cz>
10358 * asan.c: Include header files.
10359 * attribs.c (build_decl_attribute_variant): New function moved
10361 (build_type_attribute_qual_variant): Likewise.
10362 (cmp_attrib_identifiers): Likewise.
10363 (simple_cst_list_equal): Likewise.
10364 (omp_declare_simd_clauses_equal): Likewise.
10365 (attribute_value_equal): Likewise.
10366 (comp_type_attributes): Likewise.
10367 (build_type_attribute_variant): Likewise.
10368 (lookup_ident_attribute): Likewise.
10369 (remove_attribute): Likewise.
10370 (merge_attributes): Likewise.
10371 (merge_type_attributes): Likewise.
10372 (merge_decl_attributes): Likewise.
10373 (merge_dllimport_decl_attributes): Likewise.
10374 (handle_dll_attribute): Likewise.
10375 (attribute_list_equal): Likewise.
10376 (attribute_list_contained): Likewise.
10377 * attribs.h (lookup_attribute): New function moved from tree.[ch].
10378 (lookup_attribute_by_prefix): Likewise.
10379 * bb-reorder.c: Include header files.
10380 * builtins.c: Likewise.
10381 * calls.c: Likewise.
10382 * cfgexpand.c: Likewise.
10383 * cgraph.c: Likewise.
10384 * cgraphunit.c: Likewise.
10385 * convert.c: Likewise.
10386 * dwarf2out.c: Likewise.
10387 * final.c: Likewise.
10388 * fold-const.c: Likewise.
10389 * function.c: Likewise.
10390 * gimple-expr.c: Likewise.
10391 * gimple-fold.c: Likewise.
10392 * gimple-pretty-print.c: Likewise.
10393 * gimple.c: Likewise.
10394 * gimplify.c: Likewise.
10395 * hsa-common.c: Likewise.
10396 * hsa-gen.c: Likewise.
10397 * internal-fn.c: Likewise.
10398 * ipa-chkp.c: Likewise.
10399 * ipa-cp.c: Likewise.
10400 * ipa-devirt.c: Likewise.
10401 * ipa-fnsummary.c: Likewise.
10402 * ipa-inline.c: Likewise.
10403 * ipa-visibility.c: Likewise.
10405 * lto-cgraph.c: Likewise.
10406 * omp-expand.c: Likewise.
10407 * omp-general.c: Likewise.
10408 * omp-low.c: Likewise.
10409 * omp-offload.c: Likewise.
10410 * omp-simd-clone.c: Likewise.
10411 * opts-global.c: Likewise.
10412 * passes.c: Likewise.
10413 * predict.c: Likewise.
10414 * sancov.c: Likewise.
10415 * sanopt.c: Likewise.
10416 * symtab.c: Likewise.
10417 * toplev.c: Likewise.
10418 * trans-mem.c: Likewise.
10419 * tree-chkp.c: Likewise.
10420 * tree-eh.c: Likewise.
10421 * tree-into-ssa.c: Likewise.
10422 * tree-object-size.c: Likewise.
10423 * tree-parloops.c: Likewise.
10424 * tree-profile.c: Likewise.
10425 * tree-ssa-ccp.c: Likewise.
10426 * tree-ssa-live.c: Likewise.
10427 * tree-ssa-loop.c: Likewise.
10428 * tree-ssa-sccvn.c: Likewise.
10429 * tree-ssa-structalias.c: Likewise.
10430 * tree-ssa.c: Likewise.
10431 * tree-streamer-in.c: Likewise.
10432 * tree-vectorizer.c: Likewise.
10433 * tree-vrp.c: Likewise.
10434 * tsan.c: Likewise.
10435 * ubsan.c: Likewise.
10436 * varasm.c: Likewise.
10437 * varpool.c: Likewise.
10438 * tree.c: Remove functions moved to attribs.[ch].
10439 * tree.h: Likewise.
10440 * config/aarch64/aarch64.c: Add attrs.h header file.
10441 * config/alpha/alpha.c: Likewise.
10442 * config/arc/arc.c: Likewise.
10443 * config/arm/arm.c: Likewise.
10444 * config/avr/avr.c: Likewise.
10445 * config/bfin/bfin.c: Likewise.
10446 * config/c6x/c6x.c: Likewise.
10447 * config/cr16/cr16.c: Likewise.
10448 * config/cris/cris.c: Likewise.
10449 * config/darwin.c: Likewise.
10450 * config/epiphany/epiphany.c: Likewise.
10451 * config/fr30/fr30.c: Likewise.
10452 * config/frv/frv.c: Likewise.
10453 * config/ft32/ft32.c: Likewise.
10454 * config/h8300/h8300.c: Likewise.
10455 * config/i386/winnt.c: Likewise.
10456 * config/ia64/ia64.c: Likewise.
10457 * config/iq2000/iq2000.c: Likewise.
10458 * config/lm32/lm32.c: Likewise.
10459 * config/m32c/m32c.c: Likewise.
10460 * config/m32r/m32r.c: Likewise.
10461 * config/m68k/m68k.c: Likewise.
10462 * config/mcore/mcore.c: Likewise.
10463 * config/microblaze/microblaze.c: Likewise.
10464 * config/mips/mips.c: Likewise.
10465 * config/mmix/mmix.c: Likewise.
10466 * config/mn10300/mn10300.c: Likewise.
10467 * config/moxie/moxie.c: Likewise.
10468 * config/msp430/msp430.c: Likewise.
10469 * config/nds32/nds32-isr.c: Likewise.
10470 * config/nds32/nds32.c: Likewise.
10471 * config/nios2/nios2.c: Likewise.
10472 * config/nvptx/nvptx.c: Likewise.
10473 * config/pa/pa.c: Likewise.
10474 * config/pdp11/pdp11.c: Likewise.
10475 * config/powerpcspe/powerpcspe.c: Likewise.
10476 * config/riscv/riscv.c: Likewise.
10477 * config/rl78/rl78.c: Likewise.
10478 * config/rx/rx.c: Likewise.
10479 * config/s390/s390.c: Likewise.
10480 * config/sh/sh.c: Likewise.
10481 * config/sol2.c: Likewise.
10482 * config/sparc/sparc.c: Likewise.
10483 * config/spu/spu.c: Likewise.
10484 * config/stormy16/stormy16.c: Likewise.
10485 * config/tilegx/tilegx.c: Likewise.
10486 * config/tilepro/tilepro.c: Likewise.
10487 * config/v850/v850.c: Likewise.
10488 * config/vax/vax.c: Likewise.
10489 * config/visium/visium.c: Likewise.
10490 * config/xtensa/xtensa.c: Likewise.
10492 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10495 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
10496 constraints since the -mupper-regs-* switches have been
10498 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
10499 into a vector from a double word element that was extracted from
10500 another vector, and eliminate extra XXPERMDI instructions.
10501 (vsx_concat_<mode>_2): Likewise.
10502 (vsx_concat_<mode>_3): Likewise.
10503 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
10504 concat to allow optimizing inserts from previous extracts.
10506 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
10508 * config/i386/i386.c (ix86_stack_protect_guard): Generate
10509 memory reference to a SSP offset in TLS address space.
10510 (ix86_print_operand) <case '@'>: Remove.
10511 (ix86_print_operand_punct_valid_p): Remove '@' code.
10512 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
10513 UNSPEC_SP_TLS_TEST.
10514 (stack_tls_protect_set_<mode>): Remove.
10515 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
10516 (stack_tls_protect_test_<mode>): Remove.
10517 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
10519 2017-08-07 Olivier Hainque <hainque@adacore.com>
10522 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
10524 2017-08-07 Douglas Rupp <rupp@adacore.com>
10526 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
10527 variable was referenced as multidir in command.
10529 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10532 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
10535 2017-08-07 Martin Liska <mliska@suse.cz>
10537 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
10538 * config/rl78/rl78.c: Add include of attribs.h.
10539 * config/sh/sh.c: Likewise.
10540 * config/v850/v850.c: Likewise.
10542 2017-08-07 Tom de Vries <tom@codesourcery.com>
10544 PR middle-end/78266
10545 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
10547 2017-08-07 Martin Liska <mliska@suse.cz>
10549 * config/mips/mips.c: Include attribs.h.
10551 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
10554 * doc/invoke.texi: Document change in behvaior for -Ofast for
10557 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
10559 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
10561 (aarch64_pop_regs): Likewise.
10562 (aarch64_gen_load_pair): Likewise.
10563 (aarch64_save_callee_saves): Likewise.
10564 (aarch64_restore_callee_saves): Likewise.
10566 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10568 * config/i386/i386.c: Revert the last change.
10570 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10573 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10575 (ix86_finalize_stack_frame_flags): This. Also clear
10576 frame_pointer_needed if -fno-omit-frame-pointer is used without
10578 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10579 with ix86_finalize_stack_frame_flags.
10580 (ix86_expand_epilogue): Likewise.
10581 (ix86_expand_split_stack_prologue): Likewise.
10583 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10586 * config/i386/i386.c (get_builtin_code_for_version): Set priority
10587 to P_AES for Westmere.
10589 2017-08-07 Jonathan Yong <10walls@gmail.com>
10591 * config/i386/mingw.opt (fset-stack-executable): Removed.
10592 * config/i386/cygming.opt (fset-stack-executable): Moved
10594 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
10596 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
10598 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
10600 2017-08-07 Marek Polacek <polacek@redhat.com>
10602 PR middle-end/81737
10603 * fold-const.c (fold_indirect_ref_1): Check type_domain.
10605 2017-08-07 Martin Liska <mliska@suse.cz>
10607 * attribs.h (canonicalize_attr_name): New function.
10608 (cmp_attribs): Move from c-format.c and adjusted.
10609 (is_attribute_p): Moved from tree.h.
10610 * tree-inline.c: Add new includes.
10611 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
10612 (private_is_attribute_p): Remove.
10613 (private_lookup_attribute): Likewise.
10614 (private_lookup_attribute_by_prefix): Simplify.
10615 (remove_attribute): Use is_attribute_p.
10616 * tree.h: Remove removed declarations.
10618 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10620 PR middle-end/81698
10621 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
10622 instead of computing it in the function. Formatting fix.
10623 (expand_case): Don't rely on default_edge being the first edge,
10624 clear it if removing it, pass default_edge to
10625 emit_case_dispatch_table.
10626 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
10629 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
10631 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
10632 insn in the function, emit NOP after the insn.
10634 2017-08-06 Tom de Vries <tom@codesourcery.com>
10636 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
10639 2017-08-06 Tom de Vries <tom@codesourcery.com>
10641 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
10644 2017-08-04 Yury Gribov <tetra2005@gmail.com>
10646 PR tree-optimization/57371
10647 * match.pd: New pattern.
10649 2017-08-04 Marek Polacek <polacek@redhat.com>
10651 PR middle-end/81695
10652 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
10653 perform the computation in offset_int.
10655 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10657 PR tree-optimization/81136
10658 * tree-vectorizer.h: Include tree-hash-traits.h.
10659 (vec_base_alignments): New typedef.
10660 (vec_info): Add a base_alignments field.
10661 (vect_record_base_alignments): Declare.
10662 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
10664 (DR_IS_CONDITIONAL_IN_STMT): New macro.
10665 (create_data_ref): Add an is_conditional_in_stmt argument.
10666 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
10667 the is_conditional_in_stmt field.
10668 (data_ref_loc): Add an is_conditional_in_stmt field.
10669 (get_references_in_stmt): Set the is_conditional_in_stmt field.
10670 (find_data_references_in_stmt): Update call to create_data_ref.
10671 (graphite_find_data_references_in_stmt): Likewise.
10672 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
10673 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
10674 (vect_record_base_alignment): New function.
10675 (vect_record_base_alignments): Likewise.
10676 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
10677 for nested statements even if we fail to compute a misalignment.
10678 Use pooled base alignments for unconditional references.
10679 (vect_find_same_alignment_drs): Compare base addresses instead
10681 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
10682 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
10684 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10686 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
10687 Add an explicit name for the enum. Use auto_vec for slp_instances
10688 and grouped_stores.
10689 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
10691 (_bb_vec_info): Add a constructor and destructor.
10692 (vinfo_for_stmt): Return NULL for uids of -1 as well.
10693 (destroy_loop_vec_info): Delete.
10694 (vect_destroy_datarefs): Likewise.
10695 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
10696 (vec_info::vec_info): New function.
10697 (vec_info::~vec_info): Likewise.
10698 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
10699 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
10700 destroy_loop_vec_info.
10701 * tree-vect-loop.c (new_loop_vec_info): Replace with...
10702 (_loop_vec_info::_loop_vec_info): ...this.
10703 (destroy_loop_vec_info): Replace with...
10704 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
10705 the stmt_vec_infos. Leave handling of vec_info information to its
10706 destructor. Remove explicit vector releases.
10707 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
10708 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
10709 * tree-vect-slp.c (new_bb_vec_info): Replace with...
10710 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
10711 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
10712 (destroy_bb_vec_info): Replace with...
10713 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
10714 information to its destructor.
10715 (vect_slp_analyze_bb_1): Use new and delete instead of
10716 new_bb_vec_info and destroy_bb_vec_info.
10717 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
10720 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10722 * tree-data-ref.h (subscript): Add access_fn field.
10723 (data_dependence_relation): Add could_be_independent_p.
10724 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
10725 (same_access_functions): Move to tree-data-ref.c.
10726 * tree-data-ref.c (ref_contains_union_access_p): New function.
10727 (access_fn_component_p): Likewise.
10728 (access_fn_components_comparable_p): Likewise.
10729 (dr_analyze_indices): Add a reference to access_fn_component_p.
10730 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
10732 (constant_access_functions): Likewise.
10733 (add_other_self_distances): Likewise.
10734 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
10735 (initialize_data_dependence_relation): Use XCNEW and remove
10736 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
10737 of access functions that have the same type. Allow the
10738 subsequence to end with different bases in some circumstances.
10739 Record the chosen access functions in SUB_ACCESS_FN.
10740 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
10741 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
10742 (subscript_dependence_tester_1): Likewise dra and drb.
10743 (build_classic_dist_vector): Update calls accordingly.
10744 (subscript_dependence_tester): Likewise.
10745 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
10746 DDR_COULD_BE_INDEPENDENT_P.
10747 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
10748 comp_alias_ddrs instead of may_alias_ddrs.
10749 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
10751 (vect_analyze_data_ref_dependence): Use it if
10752 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
10753 distance vectors if that fails.
10754 (dependence_distance_ge_vf): New function.
10755 (vect_prune_runtime_alias_test_list): Use it. Don't clear
10756 LOOP_VINFO_MAY_ALIAS_DDRS.
10758 2017-08-04 Richard Biener <rguenther@suse.de>
10760 PR middle-end/81705
10761 * fold-const.c (fold_binary_loc): Properly restrict
10762 minus_var0 && minus_var1 case when associating undefined overflow
10765 2017-08-04 Tom de Vries <tom@codesourcery.com>
10767 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
10769 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10771 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10772 Don't start diagnostic messages with a capital letter.
10773 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10775 (rs6000_invalid_builtin): Likewise.
10776 (rs6000_trampoline_init): Likewise.
10778 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10781 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
10782 after setting changeable df flags.
10784 2017-08-03 Richard Biener <rguenther@suse.de>
10786 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
10787 up if the use is in USE - X.
10789 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
10791 * toplev.c (dumpfile.h): New include.
10792 (internal_error_reentered): New static function. Use it...
10793 (internal_error_function): ...here to handle reentered internal_error.
10795 2017-08-03 Richard Biener <rguenther@suse.de>
10797 PR middle-end/81148
10798 * fold-const.c (split_tree): Add minus_var and minus_con
10799 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
10800 here but always use minus_*.
10801 (associate_trees): Assert we never associate with MINUS_EXPR
10802 and NULL first operand. Do not recurse for PLUS_EXPR operands
10803 when associating as MINUS_EXPR either.
10804 (fold_binary_loc): Track minus_var and minus_con.
10806 2017-08-03 Tom de Vries <tom@codesourcery.com>
10809 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
10810 ACCEL_COMPILER, apply finish_options on
10811 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
10813 2017-08-03 Tom de Vries <tom@codesourcery.com>
10816 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
10817 function_entry_patch_area_size > 0.
10819 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10822 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
10823 instead of 10??LU, perform unit multiplication in wide_int,
10824 don't change alloc_object_size_limit if the limit is larger
10827 PR tree-optimization/81655
10828 PR tree-optimization/81588
10829 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
10830 the case when ranges[i].low and high are 1 for unsigned type with
10833 PR middle-end/81052
10834 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
10835 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
10837 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10839 * tree-vrp.h: Add include guard.
10841 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
10844 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
10845 (ud2): New insn pattern.
10846 * config/i386/i386.c (ix86_expand_epilogue):
10847 For naked functions, generate ud2 instead of trap insn.
10849 2017-08-02 Marek Polacek <polacek@redhat.com>
10852 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
10854 2017-08-02 Tom de Vries <tom@codesourcery.com>
10855 Cesar Philippidis <cesar@codesourcery.com>
10857 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
10858 Add missing edge probabilities.
10860 2017-08-02 Tamar Christina <tamar.christina@arm.com>
10862 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
10863 Correct endianness.
10865 2017-08-02 Jakub Jelinek <jakub@redhat.com>
10867 PR middle-end/79499
10868 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
10869 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
10870 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
10871 prologue_seq sequences - if any.
10873 2017-08-02 Richard Biener <rguenther@suse.de>
10875 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
10876 via vectors if supported, integer extracts via punning if supported
10877 or otherwise vector extracts.
10879 2017-08-02 Richard Biener <rguenther@suse.de>
10881 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
10883 (bitmap_insert_into_set): ... this.
10885 2017-08-02 Richard Biener <rguenther@suse.de>
10887 PR tree-optimization/81633
10889 2015-08-17 Alan Hayward <alan.hayward@arm.com>
10891 PR tree-optimization/71752
10892 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
10894 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
10896 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
10897 (machine_function::call_ms2sysv_pad_out): Remove field.
10898 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
10899 (ix86_compute_frame_layout): Likewise.
10901 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
10904 * config/i386/i386.c (ix86_set_func_type): Disallow naked
10905 attribute with interrupt attribute.
10907 2017-08-01 Andrew Pinski <apinski@cavium.com>
10909 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
10910 BIT_INSERT_EXPR's operand 1
10911 to see if the types precision matches.
10913 2017-08-01 Martin Liska <mliska@suse.cz>
10915 PR middle-end/70140
10916 * builtins.c (expand_builtin_memcpy_args): Remove.
10917 (expand_builtin_memcpy): Call newly added function
10918 expand_builtin_memory_copy_args.
10919 (expand_builtin_memcpy_with_bounds): Likewise.
10920 (expand_builtin_mempcpy): Remove last argument.
10921 (expand_builtin_mempcpy_with_bounds): Likewise.
10922 (expand_builtin_memory_copy_args): New function created from
10923 expand_builtin_mempcpy_args with small modifications.
10924 (expand_builtin_mempcpy_args): Remove.
10925 (expand_builtin_stpcpy): Remove unused argument.
10926 (expand_builtin): Likewise.
10927 (expand_builtin_with_bounds): Likewise.
10929 2017-08-01 Martin Liska <mliska@suse.cz>
10932 Make mempcpy more optimal (PR middle-end/70140).
10934 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10937 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10938 __builtin_vec_cmpne verify both arguments are compatible vectors
10939 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
10940 verify arg1_type is a pointer or array type. For __builtin_vec_st,
10941 move computation of aligned to after checking the argument types.
10945 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10946 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
10949 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
10950 Doug Rupp <rupp@adacore.com>
10951 Olivier Hainque <hainque@adacore.com>
10953 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
10954 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
10956 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
10957 for TARGET_OS_CPP_BUILTIN.
10958 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
10959 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
10961 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
10962 passing required abi options to the assembler for EABI configurations.
10963 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
10964 of .text.hot and .text.unlikely sections for kernel modules when
10965 using ARM style exceptions.
10966 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
10967 options. Add EXTRA_CC1_SPEC.
10968 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
10970 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
10972 (ARM_TARGET2_DWARF_FORMAT): Define.
10973 * config/arm/t-vxworks: Adjust multilib control to removal of the
10974 Diab command line options.
10976 2017-08-01 Martin Liska <mliska@suse.cz>
10978 PR gcov-profile/81561
10979 * gcov.c (unblock): Make unblocking safe as we need to preserve
10980 index correspondence of blocks and block_lists.
10982 2017-08-01 Richard Biener <rguenther@suse.de>
10984 PR tree-optimization/81181
10985 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10986 (compute_antic): ... end of iteration here.
10988 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
10990 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
10991 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
10992 (ftree-slp-vectorize): Likewise.
10993 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
10994 can no longer be set independent of flag_tree_loop_vectorize.
10995 * omp-general.c (emp_max_vf): Likewise.
10996 * opts.c (enable_fdo_optimizations): Remove references to
10997 flag_tree_vectorize, these are now implicit.
10998 (common_handle_option): Remove handling for OPT_ftree_vectorize,
10999 and leave it for the options machinery.
11001 2017-08-01 Martin Liska <mliska@suse.cz>
11003 PR middle-end/70140
11004 * builtins.c (expand_builtin_memcpy_args): Remove.
11005 (expand_builtin_memcpy): Call newly added function
11006 expand_builtin_memory_copy_args.
11007 (expand_builtin_memcpy_with_bounds): Likewise.
11008 (expand_builtin_mempcpy): Remove last argument.
11009 (expand_builtin_mempcpy_with_bounds): Likewise.
11010 (expand_builtin_memory_copy_args): New function created from
11011 expand_builtin_mempcpy_args with small modifications.
11012 (expand_builtin_mempcpy_args): Remove.
11013 (expand_builtin_stpcpy): Remove unused argument.
11014 (expand_builtin): Likewise.
11015 (expand_builtin_with_bounds): Likewise.
11017 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11020 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
11021 print "ds:" only for immediates in generic address space.
11023 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11026 * config/i386/i386.c (ix86_funciton_naked): New prototype.
11027 (ix86_function_ok_for_sibcall): Return false for naked functions.
11029 2017-08-01 Richard Biener <rguenther@suse.de>
11031 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
11032 (compute_antic): Seed worklist with exit block predecessors.
11033 * cfganal.c (dfs_find_deadend): For a cycle return the source
11034 of the edge closing it.
11036 2017-08-01 Tamar Christina <tamar.christina@arm.com>
11038 * config/aarch64/aarch64.c
11039 (aarch64_can_const_movi_rtx_p): Move 0 check.
11041 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11043 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
11044 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
11046 * match.pd: Ditto in address comparison pattern.
11048 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11050 PR tree-optimization/81627
11051 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
11052 closed ssa form for store-store chain.
11054 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11056 PR tree-optimization/81620
11057 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
11058 for store-store chain.
11060 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11062 PR tree-optimization/81588
11063 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
11064 ranges[i].in_p, invert comparison code ccode. For >/>=,
11065 swap rhs1 and rhs2 and comparison code unconditionally,
11066 for </<= don't do that. Don't swap rhs1/rhs2 again if
11067 ranges[i].in_p, instead invert comparison code ccode if
11068 opcode or oe->rank is BIT_IOR_EXPR.
11071 * optabs.def (vec_extract_optab, vec_init_optab): Change from
11072 a direct optab to conversion optab.
11073 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
11074 with GET_MODE_INNER as last argument instead of optab_handler.
11075 * expmed.c (extract_bit_field_1): Likewise. Use vector from
11076 vector extraction if possible and optab is available.
11077 * expr.c (store_constructor): Use convert_optab_handler instead
11078 of optab_handler. Use vector initialization from smaller
11079 vectors if possible and optab is available.
11080 * tree-vect-stmts.c (vectorizable_load): Likewise.
11081 * doc/md.texi (vec_extract, vec_init): Document that the optabs
11082 now have two modes.
11083 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
11084 of vec_init from half-sized vectors with the same element mode.
11085 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
11086 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
11087 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
11088 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
11089 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
11090 after mode in gen_vec_extract* calls.
11091 (vec_extract<mode>): Renamed to ...
11092 (vec_extract<mode><ssescalarmodelower>): ... this.
11093 (vec_extract<mode><ssehalfvecmodelower>): New expander.
11094 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
11095 element mode after mode in gen_vec_init* calls.
11096 (VEC_INIT_HALF_MODE): New mode iterator.
11097 (vec_init<mode>): Renamed to ...
11098 (vec_init<mode><ssescalarmodelower>): ... this.
11099 (vec_init<mode><ssehalfvecmodelower>): New expander.
11100 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11101 (vec_extractv2sfsf): ... this.
11102 (vec_initv2sf): Renamed to ...
11103 (vec_initv2sfsf): ... this.
11104 (vec_extractv2si): Renamed to ...
11105 (vec_extractv2sisi): ... this.
11106 (vec_initv2si): Renamed to ...
11107 (vec_initv2sisi): ... this.
11108 (vec_extractv4hi): Renamed to ...
11109 (vec_extractv4hihi): ... this.
11110 (vec_initv4hi): Renamed to ...
11111 (vec_initv4hihi): ... this.
11112 (vec_extractv8qi): Renamed to ...
11113 (vec_extractv8qiqi): ... this.
11114 (vec_initv8qi): Renamed to ...
11115 (vec_initv8qiqi): ... this.
11116 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
11117 (vec_init<mode>): Renamed to ...
11118 (vec_init<mode><VEC_base_l>): ... this.
11119 (vec_extract<mode>): Renamed to ...
11120 (vec_extract<mode><VEC_base_l>): ... this.
11121 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
11122 (vec_initv2sfsf): ... this.
11123 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
11124 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11125 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
11126 element mode after mode in gen_vec_init* calls.
11127 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11128 (vec_init<mode><Vel>): ... this.
11129 (vec_extract<mode>): Renamed to ...
11130 (vec_extract<mode><Vel>): ... this.
11131 * config/aarch64/iterators.md (Vel): New mode attribute.
11132 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11133 Add element mode after mode in gen_vec_extract* calls.
11134 * config/s390/vector.md (non_vec_l): New mode attribute.
11135 (vec_extract<mode>): Renamed to ...
11136 (vec_extract<mode><non_vec_l>): ... this.
11137 (vec_init<mode>): Renamed to ...
11138 (vec_init<mode><non_vec_l>): ... this.
11139 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11140 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11142 * config/arm/iterators.md (V_elem_l): New mode attribute.
11143 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11144 (vec_extract<mode><V_elem_l>): ... this.
11145 (vec_extractv2di): Renamed to ...
11146 (vec_extractv2didi): ... this.
11147 (vec_init<mode>): Renamed to ...
11148 (vec_init<mode><V_elem_l>): ... this.
11149 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11150 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11151 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11152 Add element mode after gen_vec_extract* calls.
11153 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11154 (vec_init<mode><unitmode>): ... this.
11155 (vec_extract<mode>): Renamed to ...
11156 (vec_extract<mode><unitmode>): ... this.
11157 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11158 (vec_init<mode><unitmode>): ... this.
11159 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11160 (vec_initv2sfsf): ... this.
11161 (vec_extractv2sf): Renamed to ...
11162 (vec_extractv2sfsf): ... this.
11163 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11164 Add element mode after gen_vec_extract* calls.
11165 * config/mips/mips.md (unitmode): New mode iterator.
11166 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11167 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11168 * config/spu/spu.md (inner_l): New mode attribute.
11169 (vec_init<mode>): Renamed to ...
11170 (vec_init<mode><inner_l>): ... this.
11171 (vec_extract<mode>): Renamed to ...
11172 (vec_extract<mode><inner_l>): ... this.
11173 * config/sparc/sparc.md (veltmode): New mode iterator.
11174 (vec_init<VMALL:mode>): Renamed to ...
11175 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
11176 * config/ia64/vect.md (vec_initv2si): Renamed to ...
11177 (vec_initv2sisi): ... this.
11178 (vec_initv2sf): Renamed to ...
11179 (vec_initv2sfsf): ... this.
11180 (vec_extractv2sf): Renamed to ...
11181 (vec_extractv2sfsf): ... this.
11182 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
11183 (vec_init<mode>): Renamed to ...
11184 (vec_init<mode><VEC_base_l>): ... this.
11185 (vec_extract<mode>): Renamed to ...
11186 (vec_extract<mode><VEC_base_l>): ... this.
11187 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
11188 (vec_initv2sfsf): ... this.
11189 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
11190 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11191 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
11192 gen_vec_init* calls.
11194 2017-08-01 Richard Biener <rguenther@suse.de>
11196 PR tree-optimization/81297
11197 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
11198 TREE_OVERFLOW from INTEGER_CSTs.
11200 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
11202 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
11204 2017-07-31 Carl Love <cel@us.ibm.com>
11206 * config/rs6000/rs6000-c: Add support for built-in functions
11207 vector signed char vec_xl_be (signed long long, signed char *);
11208 vector unsigned char vec_xl_be (signed long long, unsigned char *);
11209 vector signed int vec_xl_be (signed long long, signed int *);
11210 vector unsigned int vec_xl_be (signed long long, unsigned int *);
11211 vector signed long long vec_xl_be (signed long long, signed long long *);
11212 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
11213 vector signed short vec_xl_be (signed long long, signed short *);
11214 vector unsigned short vec_xl_be (signed long long, unsigned short *);
11215 vector double vec_xl_be (signed long long, double *);
11216 vector float vec_xl_be (signed long long, float *);
11217 * config/rs6000/altivec.h (vec_xl_be): Add #define.
11218 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
11219 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
11221 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
11222 (altivec_expand_builtin): Add switch statement to call
11223 altivec_expand_xl_be for each builtin.
11224 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
11225 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
11226 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
11227 __builtin_vsx_le_be_v16qi.
11228 * doc/extend.texi: Update the built-in documentation file for the
11229 new built-in functions.
11231 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11234 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
11236 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11238 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11240 * config.gcc: Add z14.
11241 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
11242 CPU model numbers for z13s and z14.
11243 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
11245 * config/s390/s390-opts.h (enum processor_type): Rename
11246 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11247 * config/s390/s390.c (processor_table): Add field for CPU name to
11248 be passed to Binutils.
11249 (s390_asm_output_machine_for_arch): Use the new field in
11250 processor_table for Binutils.
11251 (s390_expand_builtin): Replace arch12 with z14.
11252 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11253 (s390_get_sched_attrmask): Likewise.
11254 (s390_get_unit_mask): Likewise.
11255 * config/s390/s390.opt: Add z14 to processor_type enum.
11257 2017-07-31 Martin Jambor <mjambor@suse.cz>
11260 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
11261 regardless of optimization level.
11263 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
11264 Martin Liska <mliska@suse.cz>
11266 * predict.def: Remove old comment and adjust probability.
11267 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
11268 PREDICT statements.
11270 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11273 * config/i386/i386.c (ix86_function_naked): New function.
11274 (ix86_can_use_return_insn_p): Return false for naked functions.
11275 (ix86_expand_prologue): Skip prologue for naked functions.
11276 (ix86_expand_epilogue): Skip epilogue for naked functions
11277 and emit trap instruction.
11278 (ix86_warn_func_return): New function.
11279 (ix86_attribute_table): Add "naked" attribute specification.
11280 (TARGET_WARN_FUNC_RETURN): Define.
11281 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
11283 2017-07-31 Martin Liska <mliska@suse.cz>
11285 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
11286 (dump_gimple_bb_header): Always dump BB info.
11287 (pp_cfg_jump): Do not append info about BB when dumping a jump.
11289 2017-07-31 Martin Liska <mliska@suse.cz>
11292 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
11293 also with current_function_decl non-null equality.
11295 2017-07-31 Jakub Jelinek <jakub@redhat.com>
11298 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
11299 change type to the element type, instead add eltype variable and
11300 use it where we are interested in the element type.
11302 PR tree-optimization/81603
11303 * ipa-polymorphic-call.c
11304 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
11305 offset arithmetic in offset_int, bail out if the resulting bit offset
11306 doesn't fit into shwi.
11308 2017-07-31 Martin Liska <mliska@suse.cz>
11310 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
11311 (gimplify_save_expr): Fix comment.
11313 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
11316 * config/i386/i386.c (ix86_function_arg): Update arguments for
11318 (ix86_compute_frame_layout): Set the initial stack offset to
11319 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
11320 INCOMING_FRAME_SP_OFFSET.
11321 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
11322 stack before exception handler returns.
11323 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
11324 the 'ERROR_CODE' for exception handler.
11326 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
11328 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
11329 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
11330 (ASM_OUTPUT_REG_POP): Ditto.
11331 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
11332 instead of asm_fprintf to output pure string.
11334 2017-07-29 Jakub Jelinek <jakub@redhat.com>
11336 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
11337 to imported_module_or_decl hook.
11338 (debug_nothing_tree_tree_tree_bool): Remove.
11339 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
11340 * debug.c (do_nothing_debug_hooks): Use
11341 debug_nothing_tree_tree_tree_bool_bool instead of
11342 debug_nothing_tree_tree_tree_bool.
11343 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11344 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
11345 * sdbout.c (sdb_debug_hooks): Likewise.
11346 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
11347 (gen_namespace_die): Add DW_AT_export_symbols attribute if
11349 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
11350 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
11351 attribute, don't add anything.
11353 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11355 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
11356 (fold_build2_stat_loc): Likewise.
11357 (fold_build3_stat_loc): Likewise.
11358 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
11359 (fold_build1_loc): Remove macro.
11360 (fold_build2_loc): Likewise.
11361 (fold_build3_loc): Likewise.
11363 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11365 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
11366 (gimple_build_debug_bind_source_stat): Likewise.
11367 * gimple.h (gimple_build_debug_bind): Remove macro.
11368 (gimple_build_debug_bind_source): Likewise.
11370 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11372 * bitmap.c (bitmap_alloc): Adjust.
11373 (bitmap_gc_alloc): Likewise.
11374 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
11376 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11378 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
11379 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
11380 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
11381 (bitmap_gc_alloc_stat): Likewise.
11382 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
11384 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11386 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
11387 * rtl.h (shallow_copy_rtx): Remove macro.
11389 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11391 * emit-rtl.c (gen_raw_REG): Adjust.
11392 * gengenrtl.c (gendef): Likewise.
11393 * rtl.c (rtx_alloc_stat): Remove _stat from name.
11394 * rtl.h (rtx_alloc): Remove macro.
11396 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11398 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
11399 (build_tree_list_stat): Likewise.
11400 * tree.h (build_tree_list): Remove macro.
11401 (build_tree_list_vec): Likewise.
11403 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11405 * tree.c (make_vector_stat): Remove _stat from name.
11406 (build_vector_stat): Likewise.
11407 * tree.h (make_vector_stat): Remove macro.
11408 (build_vector_stat): Likewise.
11410 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11412 * tree.h (build_var_debug_value): Remove prototype.
11414 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11416 * tree.c (tree_cons_stat): Remove _stat from name.
11417 * tree.h (tree_cons): Remove macro.
11419 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11421 * tree.c (build_vl_exp_stat): Remove _stat from name.
11422 * tree.h (build_vl_exp): Remove macro.
11424 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11426 * tree.c (build_decl_stat): Remove _stat from name.
11427 * tree.h (build_decl): Remove macro.
11429 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11431 * gimple.c (gimple_build_with_ops_stat): Adjust.
11432 (gimple_alloc_stat): Remove _stat from name.
11433 * gimple.h (gimple_alloc): Remove macro.
11435 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11437 * tree.c (make_tree_vec_stat): Remove _stat from name.
11438 (grow_tree_vec_stat): Likewise.
11439 * tree.h (make_tree_vec_stat): Adjust prototype.
11440 (grow_tree_vec_stat): Likewise.
11441 (make_tree_vec): Remove macro.
11442 (grow_tree_vec): Likewise.
11444 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11446 * fold-const.c (fold_build1_stat_loc): Adjust.
11447 (fold_build2_stat_loc): Likewise.
11448 (fold_build3_stat_loc): Likewise.
11449 * tree.c (build0_stat): Remove _stat from name.
11450 (build1_stat): Likewise.
11451 (build2_stat): Likewise.
11452 (build3_stat): Likewise.
11453 (build4_stat): Likewise.
11454 (build5_stat): Likewise.
11455 * tree.h (build1_loc): Remove macro, and rename _stat function
11457 (build2_loc): Likewise.
11458 (build3_loc): Likewise.
11459 (build4_loc): Likewise.
11460 (build5_loc): Likewise.
11462 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11464 * tree.c (make_int_cst_stat): Remove _stat from name.
11465 * tree.h (make_int_cst_stat): Adjust prototype.
11466 (make_int_cst): Remove macro.
11468 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11470 * tree.c (make_tre_binfo_stat): Remove _stat from name.
11471 * tree.h (make_tree_binfo_stat): Adjust prototype.
11472 (make_tree_binfo): Remove.
11474 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11476 * tree.c (copy_node_stat): Rename to copy_node.
11477 (build_distinct_type_copy): Adjust.
11478 * tree.h (copy_node_stat): Adjust prototype.
11479 (copy_node): Remove macro.
11481 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11483 * tree.c (make_node_stat): rename to make_node.
11484 (build_tree_list_stat): Adjust.
11485 (build0_stat): Likewise.
11486 (build2_stat): Likewise.
11487 (build3_stat): Likewise.
11488 (build4_stat): Likewise.
11489 (build5_stat): Likewise.
11490 (build_decl_stat): Likewise.
11491 * tree.h (make_node_stat): Adjust prototype.
11492 (make_node): remove macro.
11494 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
11496 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
11497 (PPC_FEATURE2_SCV): Likewise.
11498 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
11500 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11502 * config/aarch64/aarch64.c
11503 (aarch64_internal_mov_immediate): Add new special pattern.
11504 * config/aarch64/aarch64.md (*movdi_aarch64):
11505 Add reg/32bit const mov case.
11507 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11508 Richard Sandiford <richard.sandiford@linaro.org>
11510 * config/aarch64/aarch64.md (mov<mode>): Generalize.
11511 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11512 Add integer and movi cases.
11513 (movi-split-hf-df-sf split, fp16): New.
11514 (enabled): Added TARGET_FP_F16INST.
11515 * config/aarch64/iterators.md (GPF_HF): New.
11516 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
11518 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11520 * config/aarch64/aarch64.c
11521 (aarch64_simd_container_mode): Add prototype.
11522 (aarch64_expand_mov_immediate): Add HI support.
11523 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
11524 (aarch64_can_const_movi_rtx_p): New.
11525 (aarch64_preferred_reload_class):
11526 Remove restrictions of using FP registers for certain SIMD operations.
11527 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
11528 (aarch64_valid_floating_const): Add integer move validation.
11529 (aarch64_simd_imm_scalar_p): Remove.
11530 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
11531 (aarch64_legitimate_constant_p): Expand list of supported cases.
11532 * config/aarch64/aarch64-protos.h
11533 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
11534 (aarch64_reinterpret_float_as_int): New.
11535 (aarch64_simd_imm_scalar_p): Remove.
11536 * config/aarch64/constraints.md (Uvi): New.
11537 (Dd): Split into Ds and new Dd.
11538 * config/aarch64/aarch64.md (*movsi_aarch64):
11540 (*movdi_aarch64): Add SIMD mov case.
11542 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11544 * tree-predcom.c: (struct chain): Handle store-store chain in which
11545 stores for elimination only store loop invariant values.
11546 (execute_pred_commoning_chain): Ditto.
11547 (prepare_initializers_chain_store_elim): Ditto.
11548 (prepare_finalizers): Ditto.
11549 (is_inv_store_elimination_chain): New function.
11550 (initialize_root_vars_store_elim_1): New function.
11552 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11554 * tree-predcom.c: Revise general description of the pass.
11555 (enum chain_type): New enum type for store elimination.
11556 (struct chain): New field supporting store elimination.
11557 (struct component): Ditto.
11558 (dump_chain): Dump store-stores chain.
11559 (release_chain): Release resources.
11560 (split_data_refs_to_components): Compute and create component
11561 contains only stores for elimination.
11562 (get_chain_last_ref_at): New function.
11563 (make_invariant_chain): Initialization.
11564 (make_rooted_chain): Specify chain type in parameter and record it.
11565 (add_looparound_copies): Skip for store-stores chain.
11566 (determine_roots_comp): Compute type of chain and pass it to
11568 (initialize_root_vars_store_elim_2): New function.
11569 (finalize_eliminated_stores): New function.
11570 (remove_stmt): Handle store for elimination.
11571 (execute_pred_commoning_chain): Execute predictive commoning on
11572 store-store chains.
11573 (determine_unroll_factor): Skip unroll for store-stores chain.
11574 (prepare_initializers_chain_store_elim): New function.
11575 (prepare_initializers_chain): Hanlde store-store chain.
11576 (prepare_finalizers_chain, prepare_finalizers): New function.
11577 (tree_predictive_commoning_loop): Return integer value indicating
11578 if loop is unrolled or lcssa form is corrupted.
11579 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
11581 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11583 * tree-predcom.c (initialize_root): Delete.
11584 (execute_pred_commoning_chain): Initialize root vars and replace
11585 reference of non-combined chain directly, rather than call above
11588 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11590 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
11591 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
11593 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11595 * tree-predcom.c (struct chain): New field init_seq.
11596 (release_chain): Release init_seq.
11597 (prepare_initializers_chain): Record intialization stmts in above
11599 (insert_init_seqs): New function.
11600 (tree_predictive_commoning_loop): Call insert_init_seqs.
11602 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11604 * tree-predcom.c (determine_roots_comp): Skip trivial components.
11606 2017-07-28 Richard Biener <rguenther@suse.de>
11608 * match.pd: Remove superfluous :c.
11609 * genmatch.c (simplify::id): Add member.
11610 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
11612 (current_id): New global.
11613 (dt_node::parent): Move from ...
11614 (dt_operand::parent): ... here. Add for_id member.
11615 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
11616 (decision_tree::find_node): Relax order requirement when
11617 merging DT_TRUE nodes to ones inbetween the current simplify
11618 and the one we try to merge with. Add diagnostic whenever
11619 we need to enforce pattern order by not merging.
11620 (decision_tree::insert): Set current_id.
11621 (decision_tree::print_node): Dump parent node and for_id.
11622 (parser::last_id): Add member.
11623 (parser::push_simplify): Assign unique id.
11624 (parser::parser): Initialize last_id.
11626 2017-07-28 Martin Liska <mliska@suse.cz>
11629 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
11630 gimple_build_debug_bind.
11632 2017-07-28 Richard Biener <rguenther@suse.de>
11634 PR tree-optimization/81502
11635 * match.pd: Add pattern combining BIT_INSERT_EXPR with
11637 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
11639 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
11640 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
11641 for BIT_FIELD_REF args.
11642 * fold-const.c (make_bit_field_ref): Likewise.
11643 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
11645 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11648 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
11649 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
11650 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
11651 Or it into SANITIZER_UNDEFINED.
11652 * ubsan.c: Include gimple-fold.h and varasm.h.
11653 (ubsan_expand_ptr_ifn): New function.
11654 (instrument_pointer_overflow): New function.
11655 (maybe_instrument_pointer_overflow): New function.
11656 (instrument_object_size): Formatting fix.
11657 (pass_ubsan::execute): Call instrument_pointer_overflow
11658 and maybe_instrument_pointer_overflow.
11659 * internal-fn.c (expand_UBSAN_PTR): New function.
11660 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
11661 * sanitizer.def (__ubsan_handle_pointer_overflow,
11662 __ubsan_handle_pointer_overflow_abort): New builtins.
11663 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
11664 * internal-fn.def (UBSAN_PTR): New internal function.
11665 * opts.c (sanitizer_opts): Add pointer-overflow.
11666 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
11667 * fold-const.c (build_range_check): Compute pointer range check in
11668 integral type if pointer arithmetics would be needed. Formatting
11671 2017-07-28 Martin Liska <mliska@suse.cz>
11674 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
11675 parameters that are of a variable-length.
11677 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
11679 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
11681 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
11682 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
11683 (CRT_CALL_STATIC_FUNCTION): Likewise.
11684 (ASM_DEFAULT_SPEC): New define.
11685 (ASM_SPEC32): Likewise.
11686 (ASM_SPEC64): Likewise.
11687 (ASM_SPEC_COMMON): Likewise.
11688 (ASM_SPEC): Likewise.
11689 (INVALID_64BIT): Likewise.
11690 (LINK_OS_DEFAULT_SPEC): Likewise.
11691 (LINK_OS_SPEC32): Likewise.
11692 (LINK_OS_SPEC64): Likewise.
11693 (POWERPC_LINUX): Likewise.
11694 (PTRDIFF_TYPE): Likewise.
11695 (RESTORE_FP_PREFIX): Likewise.
11696 (RESTORE_FP_SUFFIX): Likewise.
11697 (SAVE_FP_PREFIX): Likewise.
11698 (SAVE_FP_SUFFIX): Likewise.
11699 (SIZE_TYPE): Likewise.
11700 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
11701 (TARGET_64BIT): Likewise.
11702 (TARGET_64BIT): Likewise.
11703 (TARGET_AIX): Likewise.
11704 (WCHAR_TYPE_SIZE): Likewise.
11705 (WCHAR_TYPE): Undefine.
11706 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
11707 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
11708 (CPP_OS_RTEMS_SPEC): Delete.
11709 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
11710 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
11712 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
11714 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11716 PR tree-optimization/81578
11717 * tree-parloops.c (build_new_reduction): Bail out if
11718 reduction_code isn't one of the standard OpenMP reductions.
11719 Move the details printing after that decision.
11721 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11723 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
11724 related to reload_in_progress.
11725 (splat_input_operand): Likewise.
11726 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
11728 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
11730 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
11731 (TARGET_INSTANTIATE_DECLS): Likewise.
11732 (legitimate_indexed_address_p): Delete reload_in_progress code.
11733 (rs6000_debug_legitimate_address_p): Likewise.
11734 (rs6000_eliminate_indexed_memrefs): Likewise.
11735 (rs6000_emit_le_vsx_store): Likewise.
11736 (rs6000_emit_move_si_sf_subreg): Likewise.
11737 (rs6000_emit_move): Likewise.
11738 (register_to_reg_type): Likewise.
11739 (rs6000_pre_atomic_barrier): Likewise.
11740 (rs6000_machopic_legitimize_pic_address): Likewise.
11741 (rs6000_allocate_stack_temp): Likewise.
11742 (rs6000_address_for_fpconvert): Likewise.
11743 (rs6000_address_for_altivec): Likewise.
11744 (rs6000_secondary_memory_needed_rtx): Delete function.
11745 (rs6000_check_sdmode): Likewise.
11746 (rs6000_alloc_sdmode_stack_slot): Likewise.
11747 (rs6000_instantiate_decls): Likewise.
11748 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
11749 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
11750 Delete reload_in_progress.
11751 (*vec_reload_and_plus_<mptrsize>): Likewise.
11752 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
11753 (vsx_div_v2di): Likewise.
11754 (vsx_udiv_v2di): Likewise.
11756 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11758 * config/rs6000/rs6000.opt (mlra): Replace with stub.
11759 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
11760 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
11761 (rs6000_debug_reg_global): Delete print of LRA status.
11762 (rs6000_option_override_internal): Delete dead LRA related code.
11763 (rs6000_lra_p): Delete function.
11764 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
11766 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
11768 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
11769 * config/riscv/rtems.h: New file.
11771 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11772 Sudakshina Das <sudi.das@arm.com>
11774 * config/aarch64/aarch64.md
11775 (define_split for and<mode>3nr_compare): Move
11776 non aarch64_logical_operand to a register.
11777 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
11778 register immediate operand to a register.
11779 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
11781 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11783 PR middle-end/81564
11784 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
11786 2017-07-27 Richard Biener <rguenther@suse.de>
11788 PR tree-optimization/81573
11789 PR tree-optimization/81494
11790 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
11791 multi defuse cycle case.
11793 2017-07-27 Richard Biener <rguenther@suse.de>
11795 PR tree-optimization/81571
11796 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
11799 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
11801 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
11802 earlier and only if MASK_FPU is set. Adjust formatting.
11804 2017-07-27 Martin Liska <mliska@suse.cz>
11806 * opt-functions.awk: Add validation of value of Init.
11807 * optc-gen.awk: Pass new argument.
11809 2017-07-27 Martin Liska <mliska@suse.cz>
11811 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11812 Fix wrong condition.
11814 2017-07-27 Martin Liska <mliska@suse.cz>
11816 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
11817 BBs and edges seen by autoFDO.
11819 2017-07-27 Richard Biener <rguenther@suse.de>
11821 PR tree-optimization/81502
11822 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
11823 with incompatible but same sized type.
11824 (execute_update_addresses_taken): Likewise.
11826 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
11828 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
11829 flag_tree_loop_vectorize rather than flag_tree_vectorize.
11831 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11834 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
11835 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
11836 Change s_operand to memory_operand.
11838 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
11840 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11841 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
11842 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
11843 Emit instructions rather than returning an expression. Handle TFmode
11844 and KFmode by casting to TImode.
11845 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
11846 (rs6000_emit_le_vsx_store): Likewise.
11847 * config/rs6000/vsx.md (VSX_TI): New iterator.
11848 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
11849 (*vsx_le_undo_permute_<mode>): Likewise.
11850 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
11851 emit the split sequence.
11852 (*vsx_le_perm_store_<mode>): Likewise.
11854 2017-07-27 Jakub Jelinek <jakub@redhat.com>
11856 PR tree-optimization/81555
11857 PR tree-optimization/81556
11858 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
11859 if true, force CHANGED for the recursive invocation.
11860 (reassociate_bb): Remember original length of ops array, pass
11861 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
11863 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
11864 attributes for noipa attribute. For naked attribute use
11865 lookup_attribute first before lookup_attribute_spec.
11866 * final.c (rest_of_handle_final): Disable IPA RA for functions with
11868 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
11869 for functions with noipa attribute.
11870 (cgraph_externally_visible_p): Return true for functions with noipa
11872 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
11873 for functions with noipa attribute.
11874 * doc/extend.texi: Document noipa function attribute.
11875 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
11876 also for functions with noipa attribute.
11877 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
11879 2017-07-26 Andrew Pinski <apinski@cavium.com>
11881 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
11882 vec_unalign_load_cost and vec_unalign_store_cost.
11884 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11886 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
11887 -mvsx-small-integer option.
11888 (ISA_3_0_MASKS_IEEE): Likewise.
11889 (OTHER_VSX_VECTOR_MASKS): Likewise.
11890 (POWERPC_MASKS): Likewise.
11891 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
11892 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
11893 code, only testing for DImode being allowed in non-VSX floating
11895 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
11896 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
11898 (rs6000_option_override_internal): Delete -mvsx-small-integer.
11899 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
11900 TARGET_P8_VECTOR test.
11901 (rs6000_secondary_reload_simple_move): Likewise.
11902 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
11903 since TARGET_P9_VECTOR was already tested.
11904 (rs6000_opt_masks): Remove -mvsx-small-integer.
11905 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
11906 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11908 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
11909 test for TARGET_VEXTRACTUB was used, and that uses
11911 (p9 extract splitter): Likewise.
11912 (vsx_extract_<mode>_di_p9): Likewise.
11913 (vsx_extract_<mode>_store_p9): Likewise.
11914 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
11915 for TARGET_P9_VECTOR was used. Delete code that is now dead with
11916 the elimination of TARGET_VSX_SMALL_INTEGER.
11917 (vsx_extract_<mode>_p8): Likewise.
11918 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
11919 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
11920 (vsx_set_<mode>_p9): Likewise.
11921 (vsx_set_v4sf_p9): Likewise.
11922 (vsx_set_v4sf_p9_zero): Likewise.
11923 (vsx_insert_extract_v4sf_p9): Likewise.
11924 (vsx_insert_extract_v4sf_p9_2): Likewise.
11925 * config/rs6000/rs6000.md (sign extend splitter): Change
11926 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
11927 (floatsi<mode>2_lfiwax_mem): Likewise.
11928 (floatunssi<mode>2_lfiwzx_mem): Likewise.
11929 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
11930 since a test for TARGET_P9_VECTOR was used.
11931 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11932 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
11933 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11934 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11935 TARGET_P8_VECTOR test.
11936 (fix_trunc<mode>si2_stfiwx): Likewise.
11937 (fix_trunc<mode>si2_internal): Likewise.
11938 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
11939 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11941 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11942 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11943 TARGET_P8_VECTOR test.
11944 (fixuns_trunc<mode>si2_stfiwx): Likewise.
11945 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
11946 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11948 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11949 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
11950 since a test for TARGET_P9_VECTOR was used.
11951 (splitter for loading small constants): Likewise.
11953 2017-07-26 Andrew Pinski <apinski@cavium.com>
11955 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
11958 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
11961 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
11962 (fp_valid_at): Likewise.
11964 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11966 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
11967 (qdf24xx_addrcost_table): Likewise.
11968 (cortexa57_tunings): Update to use generic_branch_cost.
11969 (cortexa72_tunings): Likewise.
11970 (cortexa73_tunings): Likewise.
11971 (qdf24xx_tunings): Likewise.
11973 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11975 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
11976 (thunderx2t99_branch_cost): Likewise.
11977 (cortexa35_tunings): Update to use generic_branch_cost.
11978 (cortexa53_tunings): Likewise.
11979 (cortexa57_tunings): Likewise.
11980 (cortexa72_tunings): Likewise.
11981 (cortexa73_tunings): Likewise.
11982 (thunderx2t99_tunings): Likewise.
11984 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11986 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
11987 (sparc_option_override): Honour MASK_FSMULD.
11988 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
11989 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
11990 * config/sparc/sparc.opt (mfsmuld): New option.
11991 * doc/invoke.texi (mfsmuld): Document option.
11993 2017-07-26 Marek Polacek <polacek@redhat.com>
11995 PR middle-end/70992
11996 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
11998 2017-07-26 Richard Biener <rguenther@suse.de>
12000 * gimple-match-head.c (do_valueize): Return OP if valueize
12002 (get_def): New helper to get at the def stmt of a SSA name
12003 if valueize allows.
12004 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
12005 do_valueize to get at the def stmt.
12006 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
12008 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
12010 PR middle-end/46932
12011 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
12013 2017-07-26 Martin Liska <mliska@suse.cz>
12016 * function.c (expand_function_start): Make expansion of
12017 nonlocal_goto_save_area after parm_birth_insn.
12019 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12021 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
12022 from all CPU target flags enable members.
12024 2017-07-26 Richard Biener <rguenther@suse.de>
12026 * genmatch.c (dt_simplify::gen): Make iterator vars const.
12027 (decision_tree::gen): Make 'type' const.
12028 (write_predicate): Likewise.
12030 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12032 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
12033 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
12034 (rs6000_option_override_internal): Likewise.
12035 (rs6000_expand_vector_set): Likewise.
12036 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
12037 (TARGET_UPPER_REGS_SF): Likewise.
12038 (TARGET_UPPER_REGS_DI): Likewise.
12039 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
12040 (TARGET_DIRECT_MOVE_64BIT): Likewise.
12041 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
12042 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12043 (Splitters for DI constants in Altivec registers): Likewise.
12044 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
12045 (vsx_set_v4sf_p9): Likewise.
12046 (vsx_set_v4sf_p9_zero): Likewise.
12047 (vsx_insert_extract_v4sf_p9): Likewise.
12048 (vsx_insert_extract_v4sf_p9_2): Likewise.
12050 2017-07-25 Carl Love <cel@us.ibm.com>
12052 * doc/extend.texi: Update the built-in documentation file for the
12053 existing built-in functions
12054 vector signed char vec_cnttz (vector signed char);
12055 vector unsigned char vec_cnttz (vector unsigned char);
12056 vector signed short vec_cnttz (vector signed short);
12057 vector unsigned short vec_cnttz (vector unsigned short);
12058 vector signed int vec_cnttz (vector signed int);
12059 vector unsigned int vec_cnttz (vector unsigned int);
12060 vector signed long long vec_cnttz (vector signed long long);
12061 vector unsigned long long vec_cnttz (vector unsigned long long);
12063 2017-07-25 Andrew Pinski <apinski@cavium.com>
12065 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
12066 accesses where the use is for the first operand of a BIT_INSERT.
12068 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12071 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12072 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12074 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12076 * config/i386/gstabs.h: Delete.
12077 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12079 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
12081 * config/i386/i386.c (ix86_decompose_address): Do not check for
12082 register RTX when looking at index_reg or base_reg.
12083 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
12085 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
12087 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
12088 to update EH info here.
12090 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12092 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
12094 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12096 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12098 2017-07-25 Torsten Duwe <duwe@suse.de>
12100 * common.opt: Introduce -fpatchable-function-entry
12101 command line option, and its variables function_entry_patch_area_size
12102 and function_entry_patch_area_start.
12103 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12104 including a two-value parser.
12105 * target.def (print_patchable_function_entry): New target hook.
12106 * targhooks.h (default_print_patchable_function_entry): New function.
12107 * targhooks.c (default_print_patchable_function_entry): Likewise.
12108 * toplev.c (process_options): Switch off IPA-RA if
12109 patchable function entries are being generated.
12110 * varasm.c (assemble_start_function): Look at the
12111 patchable-function-entry command line switch and current
12112 function attributes and maybe generate NOP instructions by
12113 calling the print_patchable_function_entry hook.
12114 * doc/extend.texi: Document patchable_function_entry attribute.
12115 * doc/invoke.texi: Document -fpatchable_function_entry
12116 command line option.
12117 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12119 * doc/tm.texi: Re-generate.
12121 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12124 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12125 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12127 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12129 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12131 2017-07-25 Richard Biener <rguenther@suse.de>
12133 PR tree-optimization/81455
12134 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12135 not walk in cycles when looking for guards.
12137 2017-07-25 Richard Biener <rguenther@suse.de>
12139 PR tree-optimization/81529
12140 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12141 when optimizing backedge uses.
12143 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12145 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12147 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12148 to dl_section_ref. On AIX, append an expression to subtract
12149 the size of the section length to dl_section_ref.
12151 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12153 * configure.ac: If any of the config.* scripts fail, exit 1.
12154 * configure: Regenerate.
12156 2017-07-25 Richard Biener <rguenther@suse.de>
12158 PR middle-end/81546
12159 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12160 of immediate uses, be more verbose on errors.
12162 2017-07-25 Richard Biener <rguenther@suse.de>
12164 PR tree-optimization/81510
12165 * tree-vect-loop.c (vect_is_simple_reduction): When the
12166 reduction stmt is not inside the loop bail out.
12168 2017-07-25 Richard Biener <rguenther@suse.de>
12170 PR tree-optimization/81303
12171 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12172 profitability check against LOOP_VINFO_NITERSM1.
12174 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12176 * domwalk.c (cmp_bb_postorder): Simplify.
12177 (sort_bbs_postorder): New function. Use it...
12178 (dom_walker::walk): ...here to optimize common cases.
12180 2017-07-25 Martin Liska <mliska@suse.cz>
12183 * ipa-visibility.c (function_and_variable_visibility): Make the
12184 redirection just on target that supports aliasing.
12185 Fix GNU coding style.
12187 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
12190 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
12192 * config/arm/rtems.h: Likewise.
12193 * config/bfin/rtems.h: Likewise.
12194 * config/i386/rtemself.h: Likewise.
12195 * config/lm32/rtems.h: Likewise.
12196 * config/m32c/rtems.h: Likewise.
12197 * config/m68k/rtemself.h: Likewise.
12198 * config/microblaze/rtems.h: Likewise.
12199 * config/mips/rtems.h: Likewise.
12200 * config/moxie/rtems.h: Likewise.
12201 * config/nios2/rtems.h: Likewise.
12202 * config/powerpcspe/rtems.h: Likewise.
12203 * config/rs6000/rtems.h: Likewise.
12204 * config/rtems.h: Likewise.
12205 * config/sh/rtems.h: Likewise.
12206 * config/sh/rtemself.h: Likewise.
12207 * config/sparc/rtemself.h: Likewise.
12209 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12212 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
12213 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
12214 * tree-ssa-structalias.c (alias_get_name): Same.
12216 2017-07-25 Bin Cheng <bin.cheng@arm.com>
12219 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
12220 instructions if no du chain is found.
12222 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12224 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
12226 2017-07-25 Richard Biener <rguenther@suse.de>
12228 PR middle-end/81505
12229 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
12232 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12234 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
12235 upper-regs options.
12236 (ISA_2_7_MASKS_SERVER): Likewise.
12237 (ISA_3_0_MASKS_IEEE): Likewise.
12238 (OTHER_P8_VECTOR_MASKS): Likewise.
12239 (OTHER_VSX_VECTOR_MASKS): Likewise.
12240 (POWERPC_MASKS): Likewise.
12241 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
12242 duplicate list of options.
12243 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
12244 explicit -mupper-regs options.
12245 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
12246 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
12247 alias for -mupper-regs-df.
12248 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12249 (rs6000_init_hard_regno_mode_ok): Likewise.
12250 (rs6000_option_override_internal): Likewise.
12251 (rs6000_opt_masks): Likewise.
12252 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
12253 options in terms of whether -mvsx or -mpower8-vector was used.
12254 (TARGET_UPPER_REGS_DI): Likewise.
12255 (TARGET_UPPER_REGS_SF): Likewise.
12256 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
12257 -mupper-regs-* options.
12259 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12261 * passes.c (emergency_dump_function): Print some empty lines and a
12262 header before the RTL dump.
12264 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12266 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
12268 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
12271 * config/aarch64/aarch64.c (aarch64_classify_symbol):
12272 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
12274 2017-07-24 Carl Love <cel@us.ibm.com>
12276 * config/rs6000/rs6000-c.c: Add support for built-in functions
12277 vector float vec_extract_fp32_from_shorth (vector unsigned short);
12278 vector float vec_extract_fp32_from_shortl (vector unsigned short);
12279 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
12280 vec_extract_fp_from_shortl): Add defines for the two builtins.
12281 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
12282 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
12284 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
12285 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
12286 * doc/extend.texi: Update the built-in documentation file for the
12287 new built-in function.
12289 2017-07-24 Jakub Jelinek <jakub@redhat.com>
12292 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
12294 * doc/generic.texi: Likewise.
12295 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12296 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12298 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
12300 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
12301 (aarch64_mls_elt_merge<mode>): Likewise.
12303 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
12305 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12306 having __cxa_atexit.
12308 2017-07-23 Michael Collison <michael.collison@arm.com>
12310 * config/arm/arm.c (arm_option_override): Deprecate
12311 use of -mstructure-size-boundary.
12312 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
12313 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
12315 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12318 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12319 Reduce cost estimate for direct moves.
12321 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
12324 * config/i386/i386.c (ix86_option_override_internal): Disable
12325 BMI, BMI2 and TBM instructions for -m16.
12327 2017-07-21 Carl Love <cel@us.ibm.com>
12329 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12330 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12331 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12332 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12333 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12334 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12335 VMULOSW): New enum "unspec" values.
12336 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12337 altivec_vmulosw): New patterns.
12338 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12339 VMULOSW): Add definitions.
12341 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
12343 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
12344 (qdf24xx): Likewise.
12345 * config/aarch64/aarch64-options-extensions.def (rdma); New.
12346 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
12347 (AARCH64_FL_V8_1): Renumber.
12348 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
12349 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
12350 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
12351 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
12352 rdma to feature modifiers list.
12354 2017-07-21 Yury Gribov <tetra2005@gmail.com>
12356 PR middle-end/56727
12357 * ipa-visibility (function_and_variable_visibility): Convert
12358 recursive PLT call to direct call if appropriate.
12360 2017-07-21 Andrew Pinski <apinski@cavium.com>
12362 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
12363 operand 1 to see if the types precision matches.
12364 * fold-const.c (operand_equal_p): Likewise.
12366 2017-07-21 Richard Biener <rguenther@suse.de>
12368 PR tree-optimization/81303
12369 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
12370 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
12371 (vect_peeling_hash_get_lowest_cost): Adjust.
12372 (vect_enhance_data_refs_alignment): Likewise. Use
12373 vect_get_peeling_costs_all_drs to compute the penalty for no
12374 peeling to match up costs.
12376 2017-07-21 Richard Biener <rguenther@suse.de>
12378 PR tree-optimization/81500
12379 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
12380 we didn't identify a reduction path.
12382 2017-07-21 Tom de Vries <tom@codesourcery.com>
12383 Cesar Philippidis <cesar@codesourcery.com>
12385 PR gcov-profile/81442
12386 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
12389 2017-07-21 Tom de Vries <tom@codesourcery.com>
12392 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
12394 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
12395 nvptx_override_options_after_change.
12397 2017-07-21 Ulrich Drepper <drepper@redhat.com>
12399 * dwarf2out.c (output_file_names): Avoid double testing for
12400 dwarf_version >= 5.
12402 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
12404 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
12406 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12408 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
12410 (try_crossjump_to_edge): Do not punt on partitioned functions.
12412 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12414 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
12415 Put all BBs reachable only via paths crossing cold region to cold
12417 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
12419 2016-07-21 Richard Biener <rguenther@suse.de>
12421 PR tree-optimization/81303
12422 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
12423 into account prologue and epilogue iterations when raising
12424 min_profitable_iters to sth at least covering one vector iteration.
12426 2017-07-21 Tamar Christina <tamar.christina@arm.com>
12428 * config/arm/arm.c (arm_test_cpu_arch_dat):
12431 2017-07-20 Nathan Sidwell <nathan@acm.org>
12433 Remove TYPE_METHODS.
12434 * tree.h (TYPE_METHODS): Delete.
12435 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
12436 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
12437 (dbxout_type_methods): Scan TYPE_FIELDS.
12438 (dbxout_type): Don't check TYPE_METHODS here.
12439 * function.c (use_register_for_decl): Always ignore register for
12440 class types when not optimizing.
12441 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
12442 * tree.c (free_lang_data_in_type): Stitch out member functions and
12443 templates from TYPE_FIELDS.
12444 (build_distinct_type_copy, verify_type_variant,
12445 verify_type): Member fns are on TYPE_FIELDS.
12446 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
12447 * tree-pretty-print.c (dump_generic_node): Likewise.
12449 2017-07-20 Jakub Jelinek <jakub@redhat.com>
12452 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
12453 V2TImode and V4TImode.
12454 (ix86_expand_vector_extract): Likewise.
12455 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
12456 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
12457 (ssescalarmode): Handle V4TImode and V2TImode.
12458 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
12459 (*vec_extractv2ti, *vec_extractv4ti): New insns.
12460 (VEXTRACTI128_MODE): New mode iterator.
12461 (splitter for *vec_extractv?ti first element): New.
12462 (VEC_INIT_MODE): New mode iterator.
12463 (vec_init<mode>): Consolidate 3 expanders into one using
12464 VEC_INIT_MODE mode iterator.
12466 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12468 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
12469 non_spilled_static_chain_regno_p.
12471 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12473 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
12475 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
12477 * bb-reorder.c (connect_traces): Allow copying of blocks within
12480 2017-07-20 Richard Biener <rguenther@suse.de>
12482 * gimple.h (gimple_phi_result): Add gphi * overload.
12483 (gimple_phi_result_ptr): Likewise.
12484 (gimple_phi_arg): Likewise. Adjust index assert to only
12485 allow actual argument accesses rather than all slots available
12487 (gimple_phi_arg_def): Add gphi * overload.
12488 * tree-phinodes.c (make_phi_node): Initialize only actual
12490 (resize_phi_node): Clear memory not covered by old node,
12491 do not initialize excess argument slots.
12492 (reserve_phi_args_for_new_edge): Initialize new argument slot
12495 2017-07-20 Bin Cheng <bin.cheng@arm.com>
12497 PR tree-optimization/81388
12499 2016-07-21 Bin Cheng <bin.cheng@arm.com>
12501 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
12502 by removing computation of may_be_zero.
12504 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12505 Tom de Vries <tom@codesourcery.com>
12507 PR middle-end/81030
12508 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
12509 when gimple level profile disagrees with what RTL expander did.
12511 2017-07-20 Richard Biener <rguenther@suse.de>
12513 PR tree-optimization/61171
12514 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
12515 (vect_analyze_stmt): Add slp instance parameter.
12516 (vectorizable_reduction): Likewise.
12517 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
12518 (vect_is_simple_reduction): Deal with chains not detected
12519 as SLP reduction chain, specifically not properly associated
12520 chains containing a mix of plus/minus.
12521 (get_reduction_op): Remove.
12522 (get_initial_defs_for_reduction): Simplify, pass in whether
12523 this is a reduction chain, pass in the SLP node for the PHIs.
12524 (vect_create_epilog_for_reduction): Get the SLP instance as
12526 (vectorizable_reduction): Get the SLP instance as arg.
12527 During analysis remember the SLP node with the PHIs in the
12528 instance. Simplify getting at the vectorized reduction PHIs.
12529 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
12530 through SLP instance.
12531 (vect_slp_analyze_operations): Likewise.
12532 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
12533 (vect_transform_stmt): Likewise.
12535 2017-07-20 Tom de Vries <tom@codesourcery.com>
12537 PR tree-optimization/81489
12538 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
12539 read of phi arg location to before loop that modifies phi.
12541 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12543 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
12546 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12548 PR middle-end/81331
12549 * except.c (execute): Fix ordering issue.
12551 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12553 PR rtl-optimization/81423
12554 * combine.c (make_compound_operation_int): Don't try to optimize
12555 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
12557 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12559 PR rtl-optimization/81423
12560 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
12561 with a constant that is -1 in the truncated to mode.
12563 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12565 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
12566 (determine_unlikely_bbs): ... here.
12567 * predict.h (propagate_unlikely_bbs_forward): Declare.
12568 * cfgexpand.c (pass_expand::execute): Use it.
12569 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
12571 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
12572 propagate_unlikely_bbs_forward.
12574 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12576 PR middle-end/81331
12577 * except.c (maybe_add_nop_after_section_switch): New function.
12580 2017-07-19 Tom de Vries <tom@codesourcery.com>
12582 * gimple.h (gimple_phi_set_arg): Make assert more strict.
12584 2017-07-19 Tom de Vries <tom@codesourcery.com>
12586 * gimple.h (gimple_phi_arg): Make assert more strict.
12588 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
12590 * config.gcc (powerpc*-*-*): Add mmintrin.h.
12591 * config/rs6000/mmintrin.h: New file.
12592 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
12594 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12596 PR tree-optimization/81346
12597 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
12599 2017-07-19 Tom de Vries <tom@codesourcery.com>
12601 * config/nvptx/nvptx.md (VECIM): Add V2DI.
12603 2017-07-19 Tom de Vries <tom@codesourcery.com>
12605 * config/nvptx/nvptx-modes.def: Add V2DImode.
12606 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
12607 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
12608 (nvptx_output_mov_insn): Handle lack of mov.b128.
12609 (nvptx_print_operand): Handle 'H' and 'L' codes.
12610 (nvptx_vector_mode_supported): Allow V2DImode.
12611 (nvptx_preferred_simd_mode): New function.
12612 (nvptx_data_alignment): New function.
12613 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
12614 nvptx_preferred_simd_mode.
12615 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
12617 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
12619 2017-07-19 Tom de Vries <tom@codesourcery.com>
12621 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
12622 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
12623 (nvptx_vector_mode_supported): New function. Allow V2SImode.
12624 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
12625 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
12626 (mov<VECIM>_insn): New define_insn.
12627 (define_expand "mov<VECIM>): New define_expand.
12629 2017-07-19 Tom de Vries <tom@codesourcery.com>
12631 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
12633 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12635 PR tree-optimization/81346
12636 * fold-const.h (fold_div_compare, range_check_type): Declare.
12637 * fold-const.c (range_check_type): New function.
12638 (build_range_check): Use range_check_type.
12639 (fold_div_compare): No longer static, rewritten into
12640 a match.pd helper function.
12641 (fold_comparison): Don't call fold_div_compare here.
12642 * match.pd (X / C1 op C2): New optimization using fold_div_compare
12643 as helper function.
12645 2017-07-19 Nathan Sidwell <nathan@acm.org>
12647 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
12648 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
12649 * tree.c (find_decls_types_r, verify_type): Use
12650 TYPE_{MIN,MAX}_VALUE_RAW.
12651 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
12652 (hash_tree): Likewise.
12653 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12655 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12658 2017-07-18 Tom de Vries <tom@codesourcery.com>
12660 PR middle-end/81464
12661 * omp-expand.c (expand_omp_for_static_chunk): Handle
12662 equal-argument loop exit phi.
12664 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
12667 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
12668 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
12669 operand 2 predicate.
12670 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
12671 operand 2 predicate.
12672 (ror,rol -> rorx splitters): Use const_int_operand as
12673 operand 2 predicate.
12675 2017-06-18 Richard Biener <rguenther@suse.de>
12677 PR tree-optimization/81410
12678 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
12679 the gap in the ! slp_perm SLP case after each group.
12681 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12683 PR middle-end/81463
12684 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
12687 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12689 PR middle-end/81462
12690 * predict.c (set_even_probabilities): Cleanup; do not affect
12691 probabilities that are already known.
12692 (combine_predictions_for_bb): Call even when count is set.
12694 2017-07-18 Nathan Sidwell <nathan@acm.org>
12696 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
12699 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12702 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
12703 optimization for loop niter analysis.
12705 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
12708 * config/avr/avr.c (avr_optimize_casesi): Don't use
12709 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
12711 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12713 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
12714 body_cost_vec from _vect_peel_extended_info.
12715 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
12716 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
12719 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12721 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
12723 2017-07-18 Richard Biener <rguenther@suse.de>
12725 PR tree-optimization/80620
12726 PR tree-optimization/81403
12727 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
12728 info when re-using a VN table entry.
12730 2017-07-18 Richard Biener <rguenther@suse.de>
12732 PR tree-optimization/81418
12733 * tree-vect-loop.c (vectorizable_reduction): Properly compute
12734 vectype_in. Verify that with lane-reducing reduction operations
12735 we have a single def-use cycle.
12737 2017-07-17 Carl Love <cel@us.ibm.com>
12739 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
12741 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12742 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12743 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12744 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12745 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12746 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12747 VMULOSW): New enum "unspec" values.
12748 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12749 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12750 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12751 altivec_vmulosw): New patterns.
12752 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12753 VMULOSW): Add definitions.
12754 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
12756 * config/alpha/alpha.c: Include predict.h.
12758 2017-07-17 Yury Gribov <tetra2005@gmail.com>
12760 * tree-vrp.c (compare_assert_loc): Fix comparison function
12761 to return predictable results.
12763 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12765 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
12767 (subdi3): Likewise.
12768 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
12769 * doc/invoke.texi (mexpand-adddi): Update text.
12771 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12773 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
12774 that also clobbers the CC register. The old expand code is moved
12776 (*arc_clzsi2): ... here.
12777 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
12778 the CC register. The old expand code is moved to ...
12779 (arc_ctzsi2): ... here.
12781 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12783 * config/arc/arc.opt (mindexed-loads): Use initial value
12784 TARGET_INDEXED_LOADS_DEFAULT.
12785 (mauto-modify-reg): Use initial value
12786 TARGET_AUTO_MODIFY_REG_DEFAULT.
12787 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12788 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12789 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12790 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12792 2017-07-17 Martin Liska <mliska@suse.cz>
12795 * opts.c (finish_options): Do not allow -fgnu-tm
12796 w/ -fsanitize={kernel-,}address. Say sorry.
12798 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12801 * tree-loop-distribution.c (classify_partition): Only assert on
12802 numer of iterations.
12803 (merge_dep_scc_partitions): Delete prameter. Update function call.
12804 (distribute_loop): Remove code handling loop with unknown niters.
12805 (pass_loop_distribution::execute): Skip loop with unknown niters.
12807 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12810 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
12811 function sort_partitions_by_post_order.
12813 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12815 PR tree-optimization/81374
12816 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
12817 the max index of basic blocks, rather than number of basic blocks.
12819 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12821 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
12823 (arc_legitimate_pic_operand_p): Likewise.
12824 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
12826 (arc_needs_pcl_p): Likewise.
12827 (arc_legitimate_pc_offset_p): Likewise.
12828 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
12829 function is also used in constrains.md.
12830 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
12831 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
12832 PLUS. Only return true/false in known cases, otherwise assert.
12833 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
12834 is already called in arc_legitimate_constant_p.
12835 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
12837 (LEGITIMATE_PIC_OPERAND_P): Use
12838 arc_raw_symbolic_reference_mentioned_p function.
12839 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
12844 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12845 Andrew Burgess <andrew.burgess@embecosm.com>
12847 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
12848 (arc_return_address_register): New function.
12849 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
12850 (arc_handle_fndecl_attribute): Add naked attribute.
12851 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12852 (TARGET_WARN_FUNC_RETURN): Likewise.
12853 (arc_allocate_stack_slots_for_args): New function.
12854 (arc_warn_func_return): Likewise.
12855 (machine_function): Change type fn_type.
12856 (arc_compute_function_type): Consider new naked function type,
12857 change function return type.
12858 (arc_must_save_register): Adapt to handle new
12859 arc_compute_function_type's return type.
12860 (arc_expand_prologue): Likewise.
12861 (arc_expand_epilogue): Likewise.
12862 (arc_return_address_regs): Delete.
12863 (arc_return_address_register): New function.
12864 (arc_epilogue_uses): Use above function.
12865 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
12866 (arc_function_type): Change encoding, add naked type.
12867 (ARC_INTERRUPT_P): Change to handle the new encoding.
12868 (ARC_FAST_INTERRUPT_P): Likewise.
12869 (ARC_NORMAL_P): Define.
12870 (ARC_NAKED_P): Likewise.
12871 (arc_compute_function_type): Delete prototype.
12872 * config/arc/arc.md (in_ret_delay_slot): Use
12873 arc_return_address_register function.
12874 (simple_return): Likewise.
12875 (p_return_i): Likewise.
12877 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12879 PR tree-optimization/81428
12880 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
12881 can't be built for those types.
12883 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12885 Remove stuff dead since r239246.
12887 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
12888 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
12889 (avr_inform_devices): Remove dead stuff.
12891 2017-07-17 Tamar Christina <tamar.christina@arm.com>
12893 * config/arm/arm_neon.h: Fix softp typo.
12895 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12897 PR tree-optimization/81365
12898 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
12899 aggregate moves onto bb predecessor edges, make sure there are no
12900 loads that could alias the lhs in between the start of bb and the
12903 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12906 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
12907 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
12908 [LSHIFTRT, outer_code = TRUNCATE]: Same.
12910 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12912 PR tree-optimization/81396
12913 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
12914 (init_symbolic_number): Initialize it to 1.
12915 (perform_symbolic_merge): Add n_ops from both operands into the new
12917 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
12918 without base_addr as useless if they need more than one operation.
12919 (bswap_replace): Handle !bswap case for NULL base_addr.
12921 2017-07-17 Tom de Vries <tom@codesourcery.com>
12924 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
12927 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
12929 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
12930 conditional builtin define __FIX_LEON3FT_B2BST.
12932 2017-07-17 Daniel Cederman <cederman@gaisler.com>
12934 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
12935 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
12938 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12940 PR rtl-optimization/81424
12941 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
12942 to remove potential trapping from operands if -fnon-call-exceptions.
12944 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12946 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
12947 profile_proability for scalling.
12948 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
12950 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12952 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
12954 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12956 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
12957 fixpoint arithmetics.
12959 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12961 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
12962 fixpoint arithmetics.
12964 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12966 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
12967 fixpoint arithmetics.
12969 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12971 * profile-count.h (profile_probability::from_reg_br_prob_note,
12972 profile_probability::to_reg_br_prob_note): New functions.
12973 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
12974 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
12975 * predict.c (probability_reliable_p): Update.
12976 (edge_probability_reliable_p): Update.
12977 (br_prob_note_reliable_p): Update.
12978 (invert_br_probabilities): Update.
12979 (add_reg_br_prob_note): New function.
12980 (combine_predictions_for_insn): Update.
12981 * asan.c (asan_clear_shadow): Update.
12982 * cfgbuild.c (compute_outgoing_frequencies): Update.
12983 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12984 (update_br_prob_note): Update.
12985 (rtl_verify_edges): Update.
12986 (purge_dead_edges): Update.
12987 (fixup_reorder_chain): Update.
12988 * emit-rtl.c (try_split): Update.
12989 * ifcvt.c (cond_exec_process_insns): Update.
12990 (cond_exec_process_if_block): Update.
12991 (dead_or_predicable): Update.
12992 * internal-fn.c (expand_addsub_overflow): Update.
12993 (expand_neg_overflow): Update.
12994 (expand_mul_overflow): Update.
12995 * loop-doloop.c (doloop_modify): Update.
12996 * loop-unroll.c (compare_and_jump_seq): Update.
12997 * optabs.c (emit_cmp_and_jump_insn_1): Update.
12998 * predict.h: Update.
12999 * reorg.c (mostly_true_jump): Update.
13001 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
13002 * config/alpha/alpha.c (emit_unlikely_jump): Update.
13003 * config/arc/arc.c: (emit_unlikely_jump): Update.
13004 * config/arm/arm.c: (emit_unlikely_jump): Update.
13005 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
13006 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
13007 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
13008 (ix86_print_operand): Update.
13009 (ix86_split_fp_branch): Update.
13010 (predict_jump): Update.
13011 * config/ia64/ia64.c (ia64_print_operand): Update.
13012 * config/mmix/mmix.c (mmix_print_operand): Update.
13013 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
13014 (rs6000_expand_split_stack_prologue): Update.
13015 * config/rs6000/rs6000.c: Update.
13016 * config/s390/s390.c (s390_expand_vec_strlen): Update.
13017 (s390_expand_vec_movstr): Update.
13018 (s390_expand_cs_tdsi): Update.
13019 (s390_expand_split_stack_prologue): Update.
13020 * config/sh/sh.c (sh_print_operand): Update.
13021 (expand_cbranchsi4): Update.
13022 (expand_cbranchdi4): Update.
13023 * config/sparc/sparc.c (output_v9branch): Update.
13024 * config/spu/spu.c (get_branch_target): Update.
13025 (ea_load_store_inline): Update.
13026 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
13027 * config/tilepro/tilepro.c: Update.
13029 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13031 * gimplify.c (mostly_copy_tree_r): Revert latest change.
13032 (gimplify_save_expr): Likewise.
13034 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13036 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
13038 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13040 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
13042 * timevar.def (TV_IPA_FNSUMMARY): Define.
13044 2017-07-16 Daniel Cederman <cederman@gaisler.com>
13046 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
13047 to back store errata sensitive sequence from being generated.
13048 (sqrtdf2_fix): Likewise.
13050 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13052 * tree-ssa-threadupdate.c (compute_path_counts,
13053 update_joiner_offpath_counts): Use profile_probability.
13055 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
13058 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13060 * config/arm/arm-c.c (arm_cpu_builtins): Define
13061 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13063 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13065 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13066 array entries to represent __ieee128 versions of the
13067 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
13068 scalar_extract_sig, and scalar_insert_exp built-in functions.
13069 (altivec_resolve_overloaded_builtin): Add special case handling
13070 for the __builtin_scalar_insert_exp function, as represented by
13071 the P9V_BUILTIN_VEC_VSIEDP constant.
13072 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
13073 exponent support for __ieee128 argument.
13074 (VSESQP): Add scalar extract signature support for __ieee128
13076 (VSTDCNQP): Add scalar test negative support for __ieee128
13078 (VSIEQP): Add scalar insert exponent support for __int128 argument
13079 with __ieee128 result.
13080 (VSIEQPF): Add scalar insert exponent support for __ieee128
13081 argument with __ieee128 result.
13082 (VSTDCQP): Add scalar test data class support for __ieee128
13084 (VSTDCNQP): Add overload support for scalar test negative with
13085 __ieee128 argument.
13086 (VSTDCQP): Add overload support for scalar test data class
13087 __ieee128 argument.
13088 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
13089 UNSPEC_VSX_SXSIGDP.
13090 (UNSPEC_VSX_SIEXPQP): New constant.
13091 (xsxexpqp): New insn for VSX scalar extract exponent quad
13093 (xsxsigqp): New insn for VSX scalar extract significand quad
13095 (xsiexpqpf): New insn for VSX scalar insert exponent quad
13096 precision with floating point argument.
13097 (xststdcqp): New expand for VSX scalar test data class quad
13099 (xststdcnegqp): New expand for VSX scalar test negative quad
13101 (xststdcqp): New insn to match expansions for VSX scalar test data
13102 class quad precision and VSX scalar test negative quad precision.
13103 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13104 special case operand checking to enforce that second operand of
13105 VSX scalar test data class with quad precision argument is a 7-bit
13107 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13108 prototypes and descriptions of __ieee128 versions of
13109 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13110 scalar_test_data_class, and scalar_test_neg built-in functions.
13112 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13114 PR tree-optimization/81162
13115 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13116 replace a negate with an add.
13118 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13120 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13122 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13124 * config/arm/arm-c.c (arm_cpu_builtins): Define
13125 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13127 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13129 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13130 (armv8-r): Set ARM Cortex-R52 as default CPU.
13131 * config/arm/arm-tables.opt: Regenerate.
13132 * config/arm/arm-tune.md: Regenerate.
13133 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13135 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13136 extension for -mcpu=cortex-r52.
13138 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13140 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13141 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13142 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13143 (fp-armv8): Define it as FP_ARMv8 only.
13144 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13145 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13147 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13148 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13149 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13150 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13151 than TARGET_FPU_ARMV8.
13152 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13153 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13154 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13156 * config/arm/neon.md (neon_vrint): Likewise.
13157 (neon_vcvt): Likewise.
13158 (neon_<fmaxmin_op><mode>): Likewise.
13159 (<fmaxmin><mode>3): Likewise.
13160 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13161 * config/arm/predicates.md (arm_cond_move_operator): Check against
13162 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13164 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13166 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13167 to top of function.
13169 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13171 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13172 loop in comment with memset.
13174 2017-07-14 Martin Liska <mliska@suse.cz>
13176 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
13177 * dwarf2out.c (is_java): Remove the function.
13178 (output_pubname): Remove usage of the function.
13179 (lower_bound_default): Remove usage of DW_LANG_Java.
13180 (gen_compile_unit_die): Likewise.
13181 * gcc.c: Remove compiler defaults for .java and .zip files.
13182 * gimple-expr.c (remove_suffix): Change as there's no longer
13183 extension than 4-letter one.
13184 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
13185 (gimplify_save_expr): Likewise.
13186 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
13187 as it's possible even for other languages than Java.
13188 * langhooks.h (struct lang_hooks): Remove Java from a comment.
13189 * lto-opts.c (lto_write_options): Remove reference to Java.
13190 * opts.c (strip_off_ending): Update file extension handling.
13191 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
13192 * tree-eh.c (lower_resx): Likewise.
13193 * tree.c (free_lang_data_in_type): Remove dead code.
13194 (find_decls_types_r): Likewise.
13195 (build_common_builtin_nodes): Remove Java from a comment.
13196 (verify_type): Remove dead code.
13197 * varasm.c (assemble_external): Remove Java from a comment.
13199 2017-07-14 Martin Liska <mliska@suse.cz>
13201 * opts.c (finish_options): Add quotes.
13202 (common_handle_option): Likewise.
13204 2017-07-14 Martin Liska <mliska@suse.cz>
13206 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
13207 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
13208 Remove N_SO_PASCAL.
13209 * dwarf2out.c (lower_bound_default): Do not handle
13211 (gen_compile_unit_die): Likewise.
13212 * gcc.c: Remove default extension binding for GNU Pascal.
13213 * stmt.c: Remove Pascal language from a comment.
13214 * xcoffout.c: Likewise.
13216 2017-07-13 David Malcolm <dmalcolm@redhat.com>
13219 * diagnostic-show-locus.c (fixit_cmp): New function.
13220 (layout::layout): Sort m_fixit_hints.
13221 (column_range::column_range): Assert that the values are valid.
13222 (struct char_span): New struct.
13223 (correction::overwrite): New method.
13224 (struct source_line): New struct.
13225 (line_corrections::add_hint): Add assertions. Reimplement memcpy
13226 calls in terms of classes source_line and char_span, and
13227 correction::overwrite.
13228 (selftest::test_overlapped_fixit_printing_2): New function.
13229 (selftest::diagnostic_show_locus_c_tests): Call it.
13231 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
13233 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
13234 early if there is no lhs.
13236 2017-07-13 Martin Liska <mliska@suse.cz>
13238 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
13239 (gen_reference_type_die): Likewise.
13240 * stor-layout.c: Remove Pascal-related comment.
13242 2017-07-13 Martin Liska <mliska@suse.cz>
13244 * opts.c (finish_options): Add quotes to error messages.
13245 (parse_sanitizer_options): Likewise.
13247 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
13249 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
13251 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
13253 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
13255 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
13257 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
13259 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
13261 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
13264 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
13265 provides the hardware capability bits, define the macro
13266 __BUILTIN_CPU_SUPPORTS__.
13267 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
13268 if GLIBC does not provide the hardware capability bits. Add a
13269 gcc_unreachable call if the built-in cpu function is neither
13270 __builtin_cpu_is nor __builtin_cpu_supports.
13271 (rs6000_get_function_versions_dispatcher): Change the warning
13272 that an old GLIBC is used which does not export the capability
13273 bits to be an error.
13274 * doc/extend.texi (target_clones attribute): Document the
13275 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
13276 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
13277 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
13278 the macros defined by GCC if the newer GLIBC is available.
13280 2017-07-12 Jeff Law <law@redhat.com>
13282 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
13283 remaining includes slightly.
13284 * config/riscv/riscv-builtins.c: Include profile-count.h.
13286 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13289 * config/avr/avr.c (avr_set_current_function): In diagnostic
13290 messages: Quote keywords and (parts of) identifiers.
13291 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
13294 2017-07-12 Carl Love <cel@us.ibm.com>
13296 * config/rs6000/rs6000-c.c: Add support for built-in functions
13297 vector bool char vec_revb (vector bool char);
13298 vector bool short vec_revb (vector short char);
13299 vector bool int vec_revb (vector bool int);
13300 vector bool long long vec_revb (vector bool long long);
13301 * doc/extend.texi: Update the built-in documentation file for the
13302 new built-in functions.
13304 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13306 * config/s390/s390.md: Remove movcc splitter.
13308 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13310 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
13311 load/store on condition.
13313 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13316 * config/avr/avr.c (avr_encode_section_info)
13317 [progmem && !TREE_READONLY]: Error if progmem object needs
13320 2017-07-11 Michael Collison <michael.collison@arm.com>
13322 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
13325 2017-07-11 Carl Love <cel@us.ibm.com>
13327 * config/rs6000/rs6000-c.c: Add support for builtins
13328 vector unsigned int vec_parity_lsbb (vector signed int);
13329 vector unsigned int vec_parity_lsbb (vector unsigned int);
13330 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
13331 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
13332 vector unsigned long long vec_parity_lsbb (vector signed long long);
13333 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
13334 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
13335 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
13336 * doc/extend.texi: Update the built-in documentation file for the
13337 new built-in functions.
13339 2017-07-11 David Malcolm <dmalcolm@redhat.com>
13341 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
13342 (layout::m_primary_loc): New field.
13343 (layout::layout): Initialize new field. Move location filtering
13344 logic from here to...
13345 (layout::maybe_add_location_range): ...this new method. Add
13346 support for filtering to just the lines already specified by other
13348 (layout::will_show_line_p): New method.
13349 (gcc_rich_location::add_location_if_nearby): New method.
13350 (selftest::test_add_location_if_nearby): New test function.
13351 (selftest::diagnostic_show_locus_c_tests): Call it.
13352 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
13355 2017-07-11 Tom de Vries <tom@codesourcery.com>
13357 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
13358 (bb_first_real_insn): New function.
13359 (nvptx_single): Add extra initialization of broadcasted condition
13362 2017-07-11 Nathan Sidwell <nathan@acm.org>
13364 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
13366 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
13368 * doc/extend.texi (AVR Function Attributes): Remove weblink to
13369 Binutils doc as TEXI will mess them up.
13370 * doc/invoke.texi (AVR Options): Same here.
13372 2017-07-11 Daniel Cederman <cederman@gaisler.com>
13374 * config/sparc/sparc.opt (mfix-ut700): New option.
13375 (mfix-gr712rc): Likewise.
13376 (sparc_fix_b2bst): New variable.
13377 * doc/invoke.texi (SPARC options): Document them.
13378 (ARM options): Fix warnings.
13379 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
13380 instructions to prevent sequences that can trigger the store-store
13381 errata for certain LEON3FT processors.
13382 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
13383 (sparc_option_override): Set sparc_fix_b2bst appropriately.
13384 * config/sparc/sparc.md (fix_b2bst): New attribute.
13385 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
13387 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
13390 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
13392 (*rsqrtsf2_sse): Ditto.
13394 (div<mode>3): Macroize insn from divdf3 and divsf3
13395 using MODEF mode iterator.
13397 2017-07-10 Martin Sebor <msebor@redhat.com>
13399 PR tree-optimization/80397
13400 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
13401 instead of testing for equality to INTEGER_TYPE.
13403 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
13405 * config.gcc: Remove uclibc from arc target spec.
13407 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
13409 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
13411 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13414 * lto-wrapper.c (remove_option): New function.
13415 (merge_and_complain): Merge PIC/PIE options more realistically.
13417 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13419 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
13423 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
13424 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
13425 * config.in: Regenerate.
13426 * configure: Regenerate.
13427 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
13428 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
13429 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
13430 (TARGET_GASISR_PROLOGUES): ...target mask.
13431 * common/config/avr/avr-common.c
13432 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
13433 Set -mgas-isr-prologues.
13434 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
13435 INSERT_PASS_BEFORE for it.
13436 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
13437 * config/avr/avr.c (avr_option_override)
13438 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
13439 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
13440 (avr_attribute_table) <no_gccisr>: Add new function attribute.
13441 (avr_set_current_function) <is_no_gccisr>: Init machine field.
13442 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
13444 (make_avr_pass_pre_proep): New function.
13445 (emit_push_sfr) <treg>: Add argument to function and use it
13446 instead of TMP_REG.
13447 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
13448 and set machine->gasisr.yes.
13449 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
13450 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
13451 __gcc_isr.n_pushed to .L__stack_usage.
13452 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
13453 (avr_asm_final_postscan_insn): ...this new static function.
13454 * config/avr/avr.h (machine_function)
13455 <is_no_gccisr, use_L__stack_usage>: New fields.
13456 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
13457 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
13458 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
13459 (gasisr, *gasisr): New expander and insn.
13460 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13461 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
13462 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
13464 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
13466 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
13469 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13471 Move jump-tables out of .text again.
13474 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
13475 (ASM_OUTPUT_ADDR_VEC): New function.
13476 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
13477 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
13478 INSN_ADDRESSes as asm comment.
13479 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
13480 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
13481 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
13482 * config/avr/avr.md (*tablejump): Adjust comment.
13483 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
13484 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
13486 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
13487 (avr_output_addr_vec): New proto.
13488 (avr_log_t) <insn_addresses>: New field.
13490 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
13493 * config/i386/i386.c (ix86_function_arg_advance): Set
13494 outgoing_args_on_stack to true if there are outgoing arguments
13496 (ix86_function_arg): Likewise.
13497 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
13498 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
13499 * config/i386/i386.h (machine_function): Add
13500 outgoing_args_on_stack.
13502 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
13504 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13505 supporting pthreds.
13506 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
13508 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
13510 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
13511 (REAL_H): Remove $(MACHMODE_H).
13512 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
13514 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
13515 $(MACHMODE_H) and double-int.h.
13516 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
13518 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
13521 2017-07-07 Andrew Pinski <apinski@cavium.com>
13523 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
13524 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
13526 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13528 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
13529 Add warning if GCC was not configured to link against a GLIBC that
13530 exports the hardware capability bits.
13531 (make_resolver_func): Make resolver function private and not a
13532 COMDAT function. Create the name with clone_function_name instead
13533 of make_unique_name.
13536 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
13537 correct operand in doing the split.
13539 2017-07-07 Carl Love <cel@us.ibm.com>
13541 * config/rs6000/rs6000-c: Add support for built-in function
13542 vector unsigned short vec_pack_to_short_fp32 (vector float,
13544 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
13545 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
13546 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
13547 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
13548 (convert_4f32_8i16): Add define_expand.
13549 * doc/extend.texi: Update the built-in documentation file for the
13550 new built-in function.
13552 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13554 * config/sparc/m8.md: New file.
13555 * config/sparc/sparc.md: Include m8.md.
13557 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13559 * config/sparc/sparc.opt: New option -mvis4b.
13560 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
13561 (sparc_option_override): Handle VIS4B.
13562 (enum sparc_builtins): Define
13563 SPARC_BUILTIN_DICTUNPACK{8,16,32},
13564 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
13565 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
13566 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
13567 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
13568 (check_constant_argument): New function.
13569 (sparc_vis_init_builtins): Define builtins
13570 __builtin_vis_dictunpack{8,16,32},
13571 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
13572 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
13573 __builtin_vis_fpcmpde{8,16,32}shl and
13574 __builtin_vis_fpcmpur{8,16,32}shl.
13575 (sparc_expand_builtin): Check that the constant operands to
13576 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
13577 constant and in range.
13578 * config/sparc/sparc-c.c (sparc_target_macros): Handle
13580 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
13581 (SPARC_IMM5_P): Likewise.
13582 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
13583 (enabled): Handle vis4b.
13584 (UNSPEC_DICTUNPACK): New unspec.
13585 (UNSPEC_FPCMPSHL): Likewise.
13586 (UNSPEC_FPUCMPSHL): Likewise.
13587 (UNSPEC_FPCMPDESHL): Likewise.
13588 (UNSPEC_FPCMPURSHL): Likewise.
13589 (cpu_feature): New CPU feature `vis4b'.
13590 (dictunpack{8,16,32}): New insns.
13591 (FPCSMODE): New mode iterator.
13592 (fpcscond): New code iterator.
13593 (fpcsucond): Likewise.
13594 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
13595 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
13596 (fpcmpde{8,16,32}{si,di}shl): Likewise.
13597 (fpcmpur{8,16,32}{si,di}shl): Likewise.
13598 * config/sparc/constraints.md: Define constraints `q' for unsigned
13599 2-bit integer constants and `t' for unsigned 5-bit integer
13601 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
13603 (imm5_operand_dictunpack16): Likewise.
13604 (imm5_operand_dictunpack32): Likewise.
13605 (imm2_operand): Likewise.
13606 * doc/invoke.texi (SPARC Options): Document -mvis4b.
13607 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
13608 ditunpack* and fpcmp*shl builtins.
13610 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13612 * config.gcc: Handle m8 in --with-{cpu,tune} options.
13613 * config.in: Add HAVE_AS_SPARC6 define.
13614 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
13616 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
13618 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
13619 (CPP_CPU_SPEC): Handle m8.
13620 (ASM_CPU_SPEC): Likewise.
13621 * config/sparc/sparc-opts.h (enum processor_type): Add
13623 * config/sparc/sparc.c (m8_costs): New struct.
13624 (sparc_option_override): Handle TARGET_CPU_m8.
13625 (sparc32_initialize_trampoline): Likewise.
13626 (sparc64_initialize_trampoline): Likewise.
13627 (sparc_issue_rate): Likewise.
13628 (sparc_register_move_cost): Likewise.
13629 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
13630 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
13631 (ASM_CPU64_DEFAULT_SPEC): Likewise.
13632 (CPP_CPU_SPEC): Handle M8.
13633 (ASM_CPU_SPEC): Likewise.
13634 (AS_M8_FLAG): Define.
13635 * config/sparc/sparc.md: Add m8 to the cpu attribute.
13636 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
13637 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
13639 * configure: Regenerate.
13640 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
13643 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13645 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
13647 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
13648 ("*movdi_insn_sp32"): Do not set v3pipe.
13649 ("*movsi_insn"): Likewise.
13650 ("*movdi_insn_sp64"): Likewise.
13651 ("*movsf_insn"): Likewise.
13652 ("*movdf_insn_sp32"): Likewise.
13653 ("*movdf_insn_sp64"): Likewise.
13654 ("*zero_extendsidi2_insn_sp64"): Likewise.
13655 ("*sign_extendsidi2_insn"): Likewise.
13656 ("*mov<VM32:mode>_insn"): Likewise.
13657 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13658 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13659 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13660 ("<vlop:code><VL:mode>3"): Likewise.
13661 ("*not_<vlop:code><VL:mode>3"): Likewise.
13662 ("*nand<VL:mode>_vis"): Likewise.
13663 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
13664 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
13665 ("one_cmpl<VL:mode>2"): Likewise.
13666 ("faligndata<VM64:mode>_vis"): Likewise.
13667 ("alignaddrsi_vis"): Likewise.
13668 ("alignaddrdi_vis"): Likweise.
13669 ("alignaddrlsi_vis"): Likewise.
13670 ("alignaddrldi_vis"): Likewise.
13671 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13672 ("bmaskdi_vis"): Likewise.
13673 ("bmasksi_vis"): Likewise.
13674 ("bshuffle<VM64:mode>_vis"): Likewise.
13675 ("cmask8<P:mode>_vis"): Likewise.
13676 ("cmask16<P:mode>_vis"): Likewise.
13677 ("cmask32<P:mode>_vis"): Likewise.
13678 ("pdistn<P:mode>_vis"): Likewise.
13679 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13681 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13683 * config/sparc/sparc.md ("subtype"): New insn attribute.
13684 ("*wrgsr_sp64"): Set insn subtype.
13685 ("*rdgsr_sp64"): Likewise.
13686 ("alignaddrsi_vis"): Likewise.
13687 ("alignaddrdi_vis"): Likewise.
13688 ("alignaddrlsi_vis"): Likewise.
13689 ("alignaddrldi_vis"): Likewise.
13690 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13691 ("fexpand_vis"): Likewise.
13692 ("fpmerge_vis"): Likewise.
13693 ("faligndata<VM64:mode>_vis"): Likewise.
13694 ("bshuffle<VM64:mode>_vis"): Likewise.
13695 ("cmask8<P:mode>_vis"): Likewise.
13696 ("cmask16<P:mode>_vis"): Likewise.
13697 ("cmask32<P:mode>_vis"): Likewise.
13698 ("fchksm16_vis"): Likewise.
13699 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
13700 ("fmean16_vis"): Likewise.
13701 ("fp<plusminus_insn>64_vis"): Likewise.
13702 ("<plusminus_insn>v8qi3"): Likewise.
13703 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13704 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
13705 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
13706 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
13707 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
13708 ("*movqi_insn"): Likewise.
13709 ("*movhi_insn"): Likewise.
13710 ("*movsi_insn"): Likewise.
13711 ("movsi_pic_gotdata_op"): Likewise.
13712 ("*movdi_insn_sp32"): Likewise.
13713 ("*movdi_insn_sp64"): Likewise.
13714 ("movdi_pic_gotdata_op"): Likewise.
13715 ("*movsf_insn"): Likewise.
13716 ("*movdf_insn_sp32"): Likewise.
13717 ("*movdf_insn_sp64"): Likewise.
13718 ("*zero_extendhisi2_insn"): Likewise.
13719 ("*zero_extendqihi2_insn"): Likewise.
13720 ("*zero_extendqisi2_insn"): Likewise.
13721 ("*zero_extendqidi2_insn"): Likewise.
13722 ("*zero_extendhidi2_insn"): Likewise.
13723 ("*zero_extendsidi2_insn_sp64"): Likewise.
13724 ("ldfsr"): Likewise.
13725 ("prefetch_64"): Likewise.
13726 ("prefetch_32"): Likewise.
13727 ("tie_ld32"): Likewise.
13728 ("tie_ld64"): Likewise.
13729 ("*tldo_ldub_sp32"): Likewise.
13730 ("*tldo_ldub1_sp32"): Likewise.
13731 ("*tldo_ldub2_sp32"): Likewise.
13732 ("*tldo_ldub_sp64"): Likewise.
13733 ("*tldo_ldub1_sp64"): Likewise.
13734 ("*tldo_ldub2_sp64"): Likewise.
13735 ("*tldo_ldub3_sp64"): Likewise.
13736 ("*tldo_lduh_sp32"): Likewise.
13737 ("*tldo_lduh1_sp32"): Likewise.
13738 ("*tldo_lduh_sp64"): Likewise.
13739 ("*tldo_lduh1_sp64"): Likewise.
13740 ("*tldo_lduh2_sp64"): Likewise.
13741 ("*tldo_lduw_sp32"): Likewise.
13742 ("*tldo_lduw_sp64"): Likewise.
13743 ("*tldo_lduw1_sp64"): Likewise.
13744 ("*tldo_ldx_sp64"): Likewise.
13745 ("*mov<VM32:mode>_insn"): Likewise.
13746 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13747 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13749 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13751 * config/sparc/sparc.md ("type"): New insn type viscmp.
13752 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
13754 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
13755 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
13756 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13757 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
13759 ("n7_vis_logical_11cycle"): Likewise.
13760 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
13761 * config/sparc/niagara2.md ("niag3_vis": Likewise.
13762 * config/sparc/niagara.md ("niag_vis"): Likewise.
13763 * config/sparc/ultra3.md ("us3_fga"): Likewise.
13764 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
13766 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13768 * config/sparc/sparc.md: New instruction type `bmask'.
13769 (bmaskdi_vis): Use the `bmask' type.
13770 (bmasksi_vis): Likewise.
13771 * config/sparc/ultra3.md (us3_array): Likewise.
13772 * config/sparc/niagara7.md (n7_array): Likewise.
13773 * config/sparc/niagara4.md (n4_array): Likewise.
13774 * config/sparc/niagara2.md (niag2_vis): Likewise.
13775 (niag3_vis): Likewise.
13776 * config/sparc/niagara.md (niag_vis): Likewise.
13778 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13780 * ipa-comdats.c: Remove optimize check from gate.
13781 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
13782 for functions not optimized.
13783 (ipa_fn_summary_read): Skip optimize check.
13784 (ipa_fn_summary_write): Likewise.
13785 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
13787 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
13789 (can_inline_edge_p): Check flag_pcc_struct_return for match.
13790 (check_callers): Give up on caller which is not optimized.
13791 (inline_small_functions): Likewise.
13792 (ipa_inline): Do not give up when not optimizing.
13793 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
13794 away unoptimizes cdtors.
13795 (whole_program_function_and_variable_visibility): Do
13796 ipa_discover_readonly_nonaddressable_vars in LTO mode.
13797 * ipa.c (process_references): Do not check optimize.
13798 (symbol_table::remove_unreachable_nodes): Update optimize check.
13799 (set_writeonly_bit): Update optimize check.
13800 (pass_ipa_cdtor_merge::gate): Do not check optimize.
13801 (pass_ipa_single_use::gate): Remove.
13803 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13805 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
13806 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
13807 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
13808 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
13809 permute_load, permute_store, adjust_extract, adjust_splat,
13810 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
13811 replace_swap_with_copy, dump_swap_insn_table,
13812 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
13813 recombine_lvx_pattern, recombine_stvx_pattern,
13814 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
13815 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
13816 to file rs6000-p8swap.c.
13817 * config/rs6000/rs6000-p8swap.c: New file.
13818 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
13819 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
13820 and rs6000*-*-* targets.
13822 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13824 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
13826 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13828 * lto-wrapper.c (merge_and_complain): Do not merge
13829 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
13830 fsigned_zeros, ftrapping_math, fwrapv.
13831 (append_compiler_options): Do not track these options.
13832 (append_linker_options): Likewie
13834 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13836 * cgraphunit.c (cgraph_node::finalize_function): When
13837 !flag_toplevel_reorde set no_reorder flag.
13838 (varpool_node::finalize_decl): Likewise.
13839 (symbol_table::compile): Drop no toplevel reorder path.
13841 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13843 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
13844 edges; zero probability is not better than uninitialized.
13846 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13848 * asan.h (asan_sanitize_allocas_p): Declare.
13849 * asan.c (asan_sanitize_allocas_p): New function.
13850 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
13851 (handle_builtin_alloca): Likewise.
13852 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
13853 if !asan_sanitize_allocas_p.
13854 * params.def (asan-instrument-allocas): Add new option.
13855 * params.h (ASAN_PROTECT_ALLOCAS): Define.
13856 * opts.c (common_handle_option): Disable allocas sanitization for
13859 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13861 * asan.c: Include gimple-fold.h.
13862 (get_last_alloca_addr): New function.
13863 (handle_builtin_stackrestore): Likewise.
13864 (handle_builtin_alloca): Likewise.
13865 (asan_emit_allocas_unpoison): Likewise.
13866 (get_mem_refs_of_builtin_call): Add new parameter, remove const
13867 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
13868 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
13869 (instrument_builtin_call): Pass gimple iterator to
13870 get_mem_refs_of_builtin_call.
13871 (last_alloca_addr): New global.
13872 * asan.h (asan_emit_allocas_unpoison): Declare.
13873 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
13874 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
13875 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
13876 if function calls alloca.
13877 * gimple-fold.c (replace_call_with_value): Remove static keyword.
13878 * gimple-fold.h (replace_call_with_value): Declare.
13879 * internal-fn.c: Include asan.h.
13880 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
13881 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
13883 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13885 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
13886 (C_SELFTEST_FLAGS): New.
13887 (CPP_SELFTEST_FLAGS): New.
13888 (SELFTEST_DEPS): New, from deps of s-selftest.
13889 (C_SELFTEST_DEPS): New, from deps of s-selftest.
13890 (CPP_SELFTEST_DEPS): New.
13891 (selftest): Add dependency on s-selftest-c++.
13892 (s-selftest): Rename to...
13893 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
13894 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
13895 than SELFTEST_FLAGS.
13896 (selftest-gdb): Rename to...
13897 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
13899 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
13900 (selftest-valgrind): Rename to...
13901 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
13903 (selftest-valgrind): Reintroduce as an alias for
13904 selftest-c-valgrind.
13905 (s-selftest-c++): New.
13906 (selftest-c++-gdb): New.
13907 (selftest-c++-valgrind): New.
13909 2017-07-06 Olivier Hainque <hainque@adacore.com>
13911 * gcc.c (process_command): When deciding if undefined variables
13912 should be ignored when processing specs, accept "gcc -v" as well.
13914 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13916 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
13917 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
13919 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13921 * config/arm/arm-cpus.in (armv8-r): Add new entry.
13922 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
13923 * config/arm/arm-tables.opt: Regenerate.
13924 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
13926 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
13928 2017-07-06 Carl Love <cel@us.ibm.com>
13930 * ChangeLog: Clean up from mid air collision
13932 2017-07-06 Carl Love <cel@us.ibm.com>
13934 * config/rs6000/rs6000-c.c: Add support for built-in functions
13935 vector signed int vec_subc (vector signed int, vector signed int);
13936 vector signed __int128 vec_subc (vector signed __int128,
13937 vector signed __int128);
13938 vector unsigned __int128 vec_subc (vector unsigned __int128,
13939 vector unsigned __int128);
13940 vector signed int vec_sube (vector signed int, vector signed int,
13941 vector signed int);
13942 vector unsigned int vec_sube (vector unsigned int,
13943 vector unsigned int,
13944 vector unsigned int);
13945 vector signed __int128 vec_sube (vector signed __int128,
13946 vector signed __int128,
13947 vector signed__int128);
13948 vector unsigned __int128 vec_sube (vector unsigned __int128,
13949 vector unsigned __int128,
13950 vector unsigned __int128);
13951 vector signed int vec_subec (vector signed int, vector signed int,
13952 vector signed int);
13953 vector unsigned int vec_subec (vector unsigned int,
13954 vector unsigned int,
13955 vector unsigned int);
13956 vector signed __int128 vec_subec (vector signed __int128,
13957 vector signed __int128,
13958 vector signed__int128);
13959 vector unsigned __int128 vec_subec (vector unsigned __int128,
13960 vector unsigned __int128,
13961 vector unsigned __int128);
13962 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
13963 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
13964 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
13965 BU_ALTIVEC_OVERLOAD_X definitions.
13966 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
13967 * doc/extend.texi: Update the built-in documentation file for the new
13968 built-in functions.
13970 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13973 * diagnostic-show-locus.c (layout::layout): Use start and finish
13974 spelling location for the start and finish of each range.
13975 * genmatch.c (linemap_client_expand_location_to_spelling_point):
13976 Add unused aspect param.
13977 * input.c (expand_location_1): Add "aspect" param, and use it
13978 to access the correct part of the location.
13979 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
13981 (expand_location_to_spelling_point): Likewise.
13982 (linemap_client_expand_location_to_spelling_point): Add "aspect"
13983 param, and pass it to expand_location_1.
13985 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
13987 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
13988 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
13989 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
13990 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
13991 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
13992 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
13993 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
13994 _mm_maskz_getmant_ss): New intrinsics.
13995 (__builtin_ia32_getexpss128_mask): Changed to ...
13996 __builtin_ia32_getexpss128_round ... this.
13997 (__builtin_ia32_getexpsd128_mask): Changed to ...
13998 __builtin_ia32_getexpsd128_round ... this.
13999 * config/i386/i386-builtin-types.def
14000 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
14001 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
14002 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
14003 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
14004 __builtin_ia32_getmantss_mask_round): New builtins.
14005 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
14006 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
14007 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
14008 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
14009 * config/i386/sse.md
14010 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
14011 avx512f_sgetexp<mode><mask_scalar_name>
14012 <round_saeonly_scalar_name> ... this.
14013 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
14014 %0, %1, %2<round_saeonly_op3>}): Changed to ...
14015 vgetexp<ssescalarmodesuffix>
14016 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14017 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
14018 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
14019 avx512f_vgetmant<mode><mask_scalar_name>
14020 <round_saeonly_scalar_name> ... this.
14021 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14022 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
14023 vgetmant<ssescalarmodesuffix>
14024 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
14025 %0<mask_scalar_operand4>, %1, %2
14026 <round_saeonly_scalar_mask_op4>, %3} ... this.
14027 * config/i386/subst.md (mask_scalar_operand4,
14028 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
14029 round_saeonly_scalar_nimm_predicate): New subst attributes.
14031 2017-07-06 Julia Koval <julia.koval@intel.com>
14033 * config/i386/i386.c (ix86_erase_embedded_rounding):
14034 Remove code for old rounding pattern.
14036 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
14038 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
14040 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
14042 * doc/sourcebuild.texi (Test Directives, Variants of
14043 dg-require-support): Add documentation for dg-require-stack-check.
14045 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
14047 * config/i386/subst.md (mask_scalar, round_scalar,
14048 round_saeonly_scalar): New meta-templates.
14049 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
14050 round_scalar_mask_operand3, round_scalar_mask_op3,
14051 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
14052 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
14053 round_saeonly_scalar_constraint,
14054 round_saeonly_scalar_prefix): New subst attribute.
14055 * config/i386/sse.md
14056 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
14057 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
14058 <round_scalar_name> ... this.
14059 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
14060 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
14061 <round_scalar_name> ... this.
14062 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
14063 <sse>_vm<code><mode>3<mask_scalar_name>
14064 <round_saeonly_scalar_name> ... this.
14065 (v<plusminus_mnemonic><ssescalarmodesuffix>
14066 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14067 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14068 v<plusminus_mnemonic><ssescalarmodesuffix>
14069 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14070 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14071 (v<multdiv_mnemonic><ssescalarmodesuffix>
14072 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14073 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14074 v<multdiv_mnemonic><ssescalarmodesuffix>
14075 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14076 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14077 (v<maxmin_float><ssescalarmodesuffix>
14078 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
14079 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
14080 v<maxmin_float><ssescalarmodesuffix>
14081 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14082 %0<mask_scalar_operand3>, %1, %<iptr>2
14083 <round_saeonly_scalar_mask_op3>} ... this.
14085 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
14087 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
14088 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
14090 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
14091 Alan Hayward <alan.hayward@arm.com>
14092 David Sherwood <david.sherwood@arm.com>
14094 * combine.c (simplify_if_then_else): Remove "enum" before
14096 * compare-elim.c (can_eliminate_compare): Likewise.
14097 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14099 (aarch64_lookup_simd_builtin_type): Likewise.
14100 (aarch64_simd_builtin_type): Likewise.
14101 (aarch64_init_simd_builtin_types): Likewise.
14102 (aarch64_simd_expand_args): Likewise.
14103 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14105 (aarch64_reverse_mask): Likewise.
14106 (aarch64_simd_emit_reg_reg_move): Likewise.
14107 (aarch64_gen_adjusted_ldpstp): Likewise.
14108 (aarch64_ccmp_mode_to_code): Likewise.
14109 (aarch64_operands_ok_for_ldpstp): Likewise.
14110 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14111 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14113 (aarch64_min_divisions_for_recip_mul): Likewise.
14114 (aarch64_reassociation_width): Likewise.
14115 (aarch64_get_condition_code_1): Likewise.
14116 (aarch64_simd_emit_reg_reg_move): Likewise.
14117 (aarch64_simd_attr_length_rglist): Likewise.
14118 (aarch64_reverse_mask): Likewise.
14119 (aarch64_operands_ok_for_ldpstp): Likewise.
14120 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14121 (aarch64_gen_adjusted_ldpstp): Likewise.
14122 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14124 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14125 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14126 (arm_lookup_simd_builtin_type): Likewise.
14127 (arm_simd_builtin_type): Likewise.
14128 (arm_init_simd_builtin_types): Likewise.
14129 (arm_expand_builtin_args): Likewise.
14130 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14131 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14132 (ft32_setup_incoming_varargs): Likewise.
14133 (ft32_function_arg): Likewise.
14134 (ft32_function_arg_advance): Likewise.
14135 (ft32_pass_by_reference): Likewise.
14136 (ft32_arg_partial_bytes): Likewise.
14137 (ft32_valid_pointer_mode): Likewise.
14138 (ft32_addr_space_pointer_mode): Likewise.
14139 (ft32_addr_space_legitimate_address_p): Likewise.
14140 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14142 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14143 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14144 (iamcu_alignment): Likewise.
14145 (canonicalize_vector_int_perm): Likewise.
14146 (ix86_noce_conversion_profitable_p): Likewise.
14147 (ix86_mpx_bound_mode): Likewise.
14148 (ix86_operands_ok_for_move_multiple): Likewise.
14149 * config/microblaze/microblaze-protos.h
14150 (microblaze_expand_conditional_branch_reg): Likewise.
14151 * config/microblaze/microblaze.c
14152 (microblaze_expand_conditional_branch_reg): Likewise.
14153 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14155 (rs6000_reassociation_width): Likewise.
14156 (rs6000_invalid_binary_op): Likewise.
14157 (fusion_p9_p): Likewise.
14158 (emit_fusion_p9_load): Likewise.
14159 (emit_fusion_p9_store): Likewise.
14160 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14162 (riscv_hard_regno_mode_ok_p): Likewise.
14163 (riscv_address_insns): Likewise.
14164 (riscv_split_symbol): Likewise.
14165 (riscv_legitimize_move): Likewise.
14166 (riscv_function_value): Likewise.
14167 (riscv_hard_regno_nregs): Likewise.
14168 (riscv_expand_builtin): Likewise.
14169 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14170 (riscv_build_integer): Likewise.
14171 (riscv_split_integer): Likewise.
14172 (riscv_legitimate_constant_p): Likewise.
14173 (riscv_cannot_force_const_mem): Likewise.
14174 (riscv_regno_mode_ok_for_base_p): Likewise.
14175 (riscv_valid_base_register_p): Likewise.
14176 (riscv_valid_offset_p): Likewise.
14177 (riscv_valid_lo_sum_p): Likewise.
14178 (riscv_classify_address): Likewise.
14179 (riscv_legitimate_address_p): Likewise.
14180 (riscv_address_insns): Likewise.
14181 (riscv_load_store_insns): Likewise.
14182 (riscv_force_binary): Likewise.
14183 (riscv_split_symbol): Likewise.
14184 (riscv_force_address): Likewise.
14185 (riscv_legitimize_address): Likewise.
14186 (riscv_move_integer): Likewise.
14187 (riscv_legitimize_const_move): Likewise.
14188 (riscv_legitimize_move): Likewise.
14189 (riscv_address_cost): Likewise.
14190 (riscv_subword): Likewise.
14191 (riscv_output_move): Likewise.
14192 (riscv_canonicalize_int_order_test): Likewise.
14193 (riscv_emit_int_order_test): Likewise.
14194 (riscv_function_arg_boundary): Likewise.
14195 (riscv_pass_mode_in_fpr_p): Likewise.
14196 (riscv_pass_fpr_single): Likewise.
14197 (riscv_pass_fpr_pair): Likewise.
14198 (riscv_get_arg_info): Likewise.
14199 (riscv_function_arg): Likewise.
14200 (riscv_function_arg_advance): Likewise.
14201 (riscv_arg_partial_bytes): Likewise.
14202 (riscv_function_value): Likewise.
14203 (riscv_pass_by_reference): Likewise.
14204 (riscv_setup_incoming_varargs): Likewise.
14205 (riscv_print_operand): Likewise.
14206 (riscv_elf_select_rtx_section): Likewise.
14207 (riscv_save_restore_reg): Likewise.
14208 (riscv_for_each_saved_reg): Likewise.
14209 (riscv_register_move_cost): Likewise.
14210 (riscv_hard_regno_mode_ok_p): Likewise.
14211 (riscv_hard_regno_nregs): Likewise.
14212 (riscv_class_max_nregs): Likewise.
14213 (riscv_memory_move_cost): Likewise.
14214 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
14215 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
14216 (rl78_addr_space_address_mode): Likewise.
14217 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14219 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14220 (rs6000_reassociation_width): Likewise.
14221 (rs6000_invalid_binary_op): Likewise.
14222 (fusion_p9_p): Likewise.
14223 (emit_fusion_p9_load): Likewise.
14224 (emit_fusion_p9_store): Likewise.
14225 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
14226 (ok_for_simple_move_operands): Likewise.
14227 (ok_for_simple_move_strict_operands): Likewise.
14228 (ok_for_simple_arith_logic_operands): Likewise.
14229 (visium_legitimize_reload_address): Likewise.
14230 (visium_select_cc_mode): Likewise.
14231 (output_cbranch): Likewise.
14232 (visium_split_double_move): Likewise.
14233 (visium_expand_copysign): Likewise.
14234 (visium_expand_int_cstore): Likewise.
14235 (visium_expand_fp_cstore): Likewise.
14236 * config/visium/visium.c (visium_pass_by_reference): Likewise.
14237 (visium_function_arg): Likewise.
14238 (visium_function_arg_advance): Likewise.
14239 (visium_libcall_value): Likewise.
14240 (visium_setup_incoming_varargs): Likewise.
14241 (visium_legitimate_constant_p): Likewise.
14242 (visium_legitimate_address_p): Likewise.
14243 (visium_legitimize_address): Likewise.
14244 (visium_secondary_reload): Likewise.
14245 (visium_register_move_cost): Likewise.
14246 (visium_memory_move_cost): Likewise.
14247 (prepare_move_operands): Likewise.
14248 (ok_for_simple_move_operands): Likewise.
14249 (ok_for_simple_move_strict_operands): Likewise.
14250 (ok_for_simple_arith_logic_operands): Likewise.
14251 (visium_function_value_1): Likewise.
14252 (rtx_ok_for_offset_p): Likewise.
14253 (visium_legitimize_reload_address): Likewise.
14254 (visium_split_double_move): Likewise.
14255 (visium_expand_copysign): Likewise.
14256 (visium_expand_int_cstore): Likewise.
14257 (visium_expand_fp_cstore): Likewise.
14258 (visium_split_cstore): Likewise.
14259 (visium_select_cc_mode): Likewise.
14260 (visium_split_cbranch): Likewise.
14261 (output_cbranch): Likewise.
14262 (visium_print_operand_address): Likewise.
14263 * expmed.c (flip_storage_order): Likewise.
14264 * expmed.h (emit_cstore): Likewise.
14265 (flip_storage_order): Likewise.
14266 * genrecog.c (validate_pattern): Likewise.
14267 * hsa-gen.c (gen_hsa_addr): Likewise.
14268 * internal-fn.c (expand_arith_overflow): Likewise.
14269 * ira-color.c (allocno_copy_cost_saving): Likewise.
14270 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14271 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
14272 (process_invariant_for_inheritance): Likewise.
14273 * lra-eliminations.c (move_plus_up): Likewise.
14274 * omp-low.c (lower_oacc_reductions): Likewise.
14275 * simplify-rtx.c (simplify_subreg): Likewise.
14276 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
14277 (TARGET_CHKP_BOUND_MODE): Likewise..
14278 * targhooks.c (default_chkp_bound_mode): Likewise.
14279 (default_setup_incoming_vararg_bounds): Likewise.
14280 * targhooks.h (default_chkp_bound_mode): Likewise.
14281 (default_setup_incoming_vararg_bounds): Likewise.
14282 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
14283 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
14284 (have_whole_vector_shift): Likewise.
14285 * tree-vect-stmts.c (vectorizable_load): Likewise.
14286 * doc/tm.texi: Regenerate.
14288 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14290 Graceful degrade if Binutils PR21472 is not available.
14293 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
14294 .rodata in flash test fails.
14295 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
14296 * confgure: Regenerate.
14297 * config.in: Regenerate.
14298 * config/avr/avr.c (avr_asm_named_section)
14299 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
14300 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
14301 (avr_asm_init_sections): Same.
14303 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14305 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
14306 (fma<VH:mode>4_intrinsic): Likewise.
14307 (*fmsub<VCVTF:mode>4): Likewise.
14308 (*fmsub<VH:mode>4_intrinsic): Likewise.
14310 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14313 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
14314 Don't depend on "optimize > 0".
14315 (out_movhi_r_mr, out_movqi_mr_r): Same.
14316 (out_movhi_mr_r, out_movqi_r_mr): Same.
14317 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
14318 io_address_operand on "optimize > 0".
14320 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14322 * tree-loop-distribution.c: Add general explanantion on the pass.
14323 (generate_loops_for_partition): Mark distributed loop.
14324 (pg_add_dependence_edges): New parameter. Handle alias data
14325 dependence specially and record it in the parameter if asked.
14326 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
14327 (init_partition_graph_vertices, add_partition_graph_edge): New.
14328 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
14329 (free_partition_graph_vdata, build_partition_graph): New.
14330 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
14331 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
14332 (data_ref_segment_size, latch_dominated_by_data_ref): New.
14333 (compute_alias_check_pairs, version_loop_by_alias_check): New.
14334 (version_for_distribution_p, finalize_partitions): New.
14335 (distribute_loop): Handle alias data dependence specially. Factor
14336 out loop fusion code as functions and call these functions.
14338 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14340 * tree-loop-distribution.c (classify_partition): New parameter and
14341 better handle reduction statement.
14342 (rdg_build_partitions): Revise comment.
14343 (distribute_loop): Compute statements in all partitions and pass it
14344 to classify_partition.
14346 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14348 * tree-loop-distribution.c (enum partition_type): New.
14349 (struct partition): New field type.
14350 (partition_merge_into): Add parameter. Update partition type.
14351 (data_dep_in_cycle_p, update_type_for_merge): New functions.
14352 (build_rdg_partition_for_vertex): Compute partition type.
14353 (rdg_build_partitions): Dump partition type.
14354 (distribute_loop): Update calls to partition_merge_into.
14356 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14358 * tree-loop-distribution.c (struct ddr_hasher): New.
14359 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
14361 (classify_partition): Call get_data_dependence.
14362 (pg_add_dependence_edges): Ditto.
14363 (distribute_loop): Release data dependence hash table.
14365 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14367 * tree-loop-distribution.c (ref_base_address): Delete.
14368 (similar_memory_accesses): Rename ...
14369 (share_memory_accesses): ... to this. Check if partitions access
14370 the same memory reference.
14371 (distribute_loop): Call share_memory_accesses.
14373 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14375 * tree-loop-distribution.c (struct partition): New field recording
14376 its data reference.
14377 (partition_alloc, partition_free): Init and release data refs.
14378 (partition_merge_into): Merge data refs.
14379 (build_rdg_partition_for_vertex): Collect data refs for partition.
14380 (pg_add_dependence_edges): Change parameters from vector to bitmap.
14382 (distribute_loop): Remve data refs from vertice data of partition
14385 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14387 * tree-loop-distribution.c (params.h): Include header file.
14388 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
14389 (datarefs_vec): New global var.
14390 (create_rdg_vertices): Use datarefs_vec directly.
14391 (free_rdg): Don't free data references.
14392 (build_rdg): Update use. Don't free data references.
14393 (distribute_loop): Compute global variable for data references.
14394 Bail out if there are too many data references.
14396 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14398 * tree-loop-distribution.c (loop_nest): New global var.
14399 (build_rdg): Use loop directly, rather than loop nest.
14400 (pg_add_dependence_edges): Remove loop nest parameter. Use global
14402 (distribute_loop): Compute global variable loop nest. Update use.
14404 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14406 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
14407 (partition_merge_into): New parameter. Dump reason for fusion.
14408 (distribute_loop): Update use of partition_merge_into.
14410 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14412 * tree-loop-distribution.c (bb_top_order_index): New.
14413 (bb_top_order_index_size, bb_top_order_cmp): New.
14414 (stmts_from_loop): Use topological order.
14415 (pass_loop_distribution::execute): Compute and release topological
14416 order for basic blocks.
14418 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14420 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
14423 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14425 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
14426 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
14427 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
14428 * internal-fn.def (LOOP_DIST_ALIAS): New.
14429 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
14430 (fold_loop_internal_call): ... this.
14431 (vect_loop_dist_alias_call): New function.
14432 (set_uid_loop_bbs): Call fold_loop_internal_call.
14433 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
14436 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14439 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
14440 Require dead FLAGS_REG at the beginning of a peephole.
14442 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14445 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
14446 arguments in the call to __builtin_ia32_sbb_u32.
14447 (_subborrow_u64): Swap _X and _Y arguments in the call to
14448 __builtin_ia32_sbb_u64.
14450 2017-07-04 Jakub Jelinek <jakub@redhat.com>
14453 * tree-vrp.c (compare_assert_loc): Turn into a function template
14454 with stable template parameter. Only test if a->e is NULL,
14455 !a->e == !b->e has been verified already. Use e == NULL or
14456 e != NULL instead of e or ! e tests. If stable is true, don't use
14457 iterative_hash_expr, on the other side allow a or b or both NULL
14458 and sort the NULLs last.
14459 (process_assert_insertions): Sort using compare_assert_loc<false>
14460 instead of compare_assert_loc, later sort using
14461 compare_assert_loc<true> before calling process_assert_insertions_for
14462 in a loop. Use break instead of continue once seen NULL pointer.
14464 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14466 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14467 Cortex-R7 and Cortex-R8 processors.
14469 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14471 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
14472 uninitialized while src is not.
14474 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
14476 * common/config/arm/arm-common.c: Adjust include path for
14478 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
14479 (arm-cpu.h): Create in build directory. Adjust dependency rules.
14480 (arm-cpu-data.h): Likewise.
14481 (arm-cpu-cdata.h): Likewise.
14482 * config/arm/arm-cpu.h: Delete.
14483 * config/arm/arm-cpu-cdata.h: Delete.
14484 * config/arm/arm-cpu-data.h: Delete.
14486 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
14488 * config/arm/arm-cpus.in (cortex-a55): New.
14489 (cortex-a75): Likewise.
14490 (cortex-a75.cortex-a55): Likewise.
14491 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
14493 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
14494 * config/arm/arm-cpu-cdata.h: Regenerate.
14495 * config/arm/arm-cpu-data.h: Regenerate.
14496 * config/arm/arm-cpu.h: Regenerate.
14497 * config/arm/arm-tables.opt: Regenerate.
14498 * config/arm/arm-tune.md: Regenerate.
14500 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14502 * haifa-sched.c (sched_create_recovery_edges): Update profile.
14504 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14506 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
14509 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
14511 PR tree-optimization/81292
14512 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
14513 full_string_p, also call adjust_related_strinfos if the adjustment
14514 is simple, otherwise invalidate related strinfos.
14516 2017-07-04 Martin Liska <mliska@suse.cz>
14519 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
14520 newly created variable as DECL_IGNORED_P.
14522 2017-07-04 Martin Liska <mliska@suse.cz>
14525 * ipa-inline.c (inline_small_functions):
14526 Use xstrdup_for_dump.
14528 2017-07-04 Tom de Vries <tom@codesourcery.com>
14530 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
14532 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14535 * config/darwin.c (darwin_function_switched_text_sections):
14538 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
14540 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
14542 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14544 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
14546 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14548 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
14549 min_profitable_iters, and th as inclusive lower bounds.
14550 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
14551 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
14552 for min_profitable_iters and min_profitable_estimate.
14553 (vect_transform_loop): Treat th as an inclusive lower bound.
14554 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
14556 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14559 * config/darwin.c (darwin_function_switched_text_sections):
14560 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
14561 in two pieces, and suppress the use of buf.
14563 2017-07-03 Nathan Sidwell <nathan@acm.org>
14565 * hash-table.h (hash_table_mod1): Fix indentation.
14567 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14569 PR middle-end/81290
14570 * predict.c (force_edge_cold): Be more careful about propagation
14572 * profile-count.h (profile_probability::guessed,
14573 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
14575 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
14577 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
14579 * doc/invoke.texi (rcpc architecture extension): Document it.
14581 2017-07-03 Richard Biener <rguenther@suse.de>
14583 PR tree-optimization/60510
14584 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
14585 the scalar reduction PHI and use it.
14586 (vectorizable_reduction): Properly guard the single_defuse_cycle
14587 path for non-SLP reduction chains where we cannot use it.
14588 Rework reduc_def/index and vector type deduction. Rework
14589 vector operand gathering during reduction op code-gen.
14590 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
14591 chains dissolve the chain and leave it to non-SLP reduction
14594 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14596 * tree-data-ref.h (dr_alignment): Declare.
14597 * tree-data-ref.c (dr_alignment): New function.
14598 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
14599 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14601 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
14603 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14605 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
14606 and base_misalignment fields.
14607 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
14608 * tree-data-ref.c: Include builtins.h.
14609 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
14610 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
14611 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
14612 * tree-vect-data-refs.c: Include tree-cfg.h.
14613 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
14614 fields instead of calculating an alignment here.
14615 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
14616 innermost_loop_behavior fields.
14618 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14620 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
14622 (DR_STEP_ALIGNMENT): New macro.
14623 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
14624 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
14625 (create_data_ref): Print it.
14626 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
14627 to tell whether the step preserves vector (mis)alignment.
14628 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14629 Move the check for an integer step and generalise to all INTEGER_CST.
14630 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
14631 Print the outer step alignment.
14633 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14635 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
14636 with offset_alignment.
14637 (DR_ALIGNED_TO): Delete.
14638 (DR_OFFSET_ALIGNMENT): New macro.
14639 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
14640 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
14641 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
14642 (create_data_ref): Likewise.
14643 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14644 (vect_analyze_data_refs): Likewise.
14645 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
14646 creating dummy innermost behavior.
14648 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14650 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
14651 with a "innermost_loop_behavior *" and refeence tree.
14652 * tree-data-ref.c (dr_analyze_innermost): Likewise.
14653 (create_data_ref): Update call accordingly.
14654 * tree-predcom.c (find_looparound_phi): Likewise.
14656 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14658 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
14659 fields with dr_wrt_vec_loop.
14660 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
14661 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
14662 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
14663 (vect_dr_behavior): New function.
14664 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14665 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
14666 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
14667 track whether the step preserves the misalignment.
14668 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14669 Use vect_dr_behavior.
14670 (vect_setup_realignment): Update call accordingly.
14671 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
14672 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
14673 call to vect_create_addr_base_for_vector_ref.
14674 (vect_create_cond_for_align_checks): Likewise.
14675 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
14676 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
14677 (vect_recog_mask_conversion_pattern): Likewise.
14678 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
14679 (new_stmt_vec_info): Remove redundant zeroing.
14681 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14683 * common/config/arm/arm-common.c (arm_be8_option): New function.
14684 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
14685 (ISA_ARMv6): Add isa_bit_be8.
14686 * config/arm/arm.h (arm_be8_option): Add prototype.
14687 (BE8_SPEC_FUNCTION): New define.
14688 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
14689 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
14690 (mlittle-endian): Similarly.
14691 (mbe8, mbe32): New options.
14692 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
14693 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
14695 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14697 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
14699 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14701 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
14702 (cleanup_tree_cfg_bb): Use it.
14703 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
14705 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
14707 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14710 * loop-doloop.c (add_test): Update profile.
14712 2017-07-03 Martin Liska <mliska@suse.cz>
14715 * sanopt.c (rewrite_usage_of_param): New function.
14716 (sanitize_rewrite_addressable_params): Likewise.
14717 (pass_sanopt::execute): Call rewrite_usage_of_param.
14719 2017-07-03 Richard Biener <rguenther@suse.de>
14721 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
14722 back to using VIEW_CONVERT_EXPR.
14724 2017-07-03 Martin Liska <mliska@suse.cz>
14727 * doc/extend.texi: Document when a resolver function is
14728 generated for target_clones.
14730 2017-07-03 Martin Liska <mliska@suse.cz>
14732 * asan.c (asan_emit_stack_protection): Unpoison just red zones
14733 and shadow memory of auto variables which are subject of
14734 use-after-scope sanitization.
14735 (asan_expand_mark_ifn): Add do set only when is_poison.
14737 2016-07-03 Richard Biener <rguenther@suse.de>
14739 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
14741 (vect_force_simple_reduction): Record reduction def -> phi mapping.
14742 (vectorizable_reduction): Perform reduction PHI creation when
14743 visiting a reduction PHI and adjust and simplify code generation
14744 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
14745 (vect_transform_loop): Visit reduction PHIs.
14746 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
14747 defs into the SLP tree.
14748 (vect_build_slp_tree): Reduction defs terminate the recursion.
14749 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
14751 (vect_get_vec_defs_for_stmt_copy): Export.
14752 (vect_get_vec_defs): Likewise.
14753 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
14755 (vect_get_vec_defs_for_stmt_copy): Declare.
14756 (vect_get_vec_defs): Likewise.
14758 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14760 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
14761 parameter with a "loop" parameter and use it instead of the
14762 loop containing DR_STMT. Don't check simple_iv when doing
14763 BB analysis. Describe the two analysis modes in the comment.
14765 2017-07-03 Tom de Vries <tom@codesourcery.com>
14767 PR tree-optimization/69468
14768 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
14769 (find_same_succ_bb): Handle ignore_edge_flags.
14771 2017-07-03 Tom de Vries <tom@codesourcery.com>
14773 PR tree-optimization/81192
14774 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
14776 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
14778 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
14780 2017-07-03 Tom de Vries <tom@codesourcery.com>
14782 PR tree-optimization/81192
14783 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
14784 BB_SAME_SUCC (bb) == NULL.
14786 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14788 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
14791 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14793 * dumpfile.c: Include profile-count.h
14794 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
14796 (insert_cond_bb): Update profile.
14797 * tree-cfg.h (insert_cond_bb): Update prototype.
14798 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
14799 * tree-dump.c: Do not include tree-cfg.
14801 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14803 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
14805 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14807 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
14810 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14812 * tree-complex.c (expand_complex_div_wide): update profile.
14814 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14815 Alan Hayward <alan.hayward@arm.com>
14816 David Sherwood <david.sherwood@arm.com>
14818 * Makefile.in (MACHMODE_H): Remove insn-modes.h
14819 (CORETYPES_H): New define.
14820 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
14821 (insn-modes-inline.h, s-modes-inline-h): New rules.
14822 (generated_files): Add insn-modes-inline.h.
14823 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
14824 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
14825 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
14826 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
14827 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
14828 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
14829 (build/gencodes.o, build/genconditions.o): Likewise.
14830 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
14831 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
14832 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
14833 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
14834 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
14835 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
14836 * coretypes.h: Include everything up to real.h for generators.
14837 Include insn-modes.h first. Include wide-int-print.h after
14838 wide-int.h. Include insn-modes-inline.h and then machmode.h.
14839 * machmode.h: Don't include insn-modes.h here.
14840 * function-tests.c: Remove includes of signop.h, machmode.h,
14841 double-int.h and wide-int.h.
14843 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
14845 * optc-save-gen.awk: Likewise.
14846 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
14847 * godump.c: Remove include of wide-int-print.h.
14848 * pretty-print.h: Likewise.
14849 * wide-int-print.cc: Likewise.
14850 * wide-int.cc: Likewise.
14851 * hash-map-tests.c: Remove include of signop.h.
14852 * hash-set-tests.c: Likewise.
14853 * rtl-tests.c: Likewise.
14854 * mkconfig.sh: Remove include of machmode.h.
14855 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
14857 (emit_insn_modes_inline_h): ...this new function. Emit the code
14858 into an insn-modes-inline.h header file, adding appropriate
14859 include guards and end comments.
14860 (emit_insn_modes_c_header): Remove include of machmode.h.
14861 (emit_min_insn_modes_c_header): Include coretypes.h rather than
14863 (main): Handle -i flag and call emit_insn_modes_inline_h when
14866 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14868 * tree-ssa-strlen.c (strinfo): Rename the length field to
14869 nonzero_chars. Add a full_string_p field.
14870 (compare_nonzero_chars, zero_length_string_p): New functions.
14871 (get_addr_stridx): Add an offset_out parameter.
14872 Use compare_nonzero_chars.
14873 (get_stridx): Update accordingly. Use compare_nonzero_chars.
14874 (new_strinfo): Update after above changes to strinfo.
14875 (set_endptr_and_length): Set full_string_p.
14876 (get_string_length): Update after above changes to strinfo.
14877 (unshare_strinfo): Update call to new_strinfo.
14878 (maybe_invalidate): Likewise.
14879 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
14880 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
14881 as a uhwi instead of an shwi. Update after above changes to
14882 strinfo and new_strinfo.
14883 (zero_length_string): Assert that chainsi contains full strings.
14884 Use zero_length_string_p. Update call to new_strinfo.
14885 (adjust_related_strinfos): Update after above changes to strinfo.
14886 Copy full_string_p from origsi.
14887 (adjust_last_stmt): Use zero_length_string_p.
14888 (handle_builtin_strlen): Update after above changes to strinfo and
14889 new_strinfo. Install the lhs as the string length if the previous
14890 entry didn't describe a full string.
14891 (handle_builtin_strchr): Update after above changes to strinfo
14893 (handle_builtin_strcpy): Likewise.
14894 (handle_builtin_strcat): Likewise.
14895 (handle_builtin_malloc): Likewise.
14896 (handle_pointer_plus): Likewise.
14897 (handle_builtin_memcpy): Likewise. Track nonzero characters
14898 that aren't necessarily followed by a nul terminator.
14899 (handle_char_store): Likewise.
14901 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14903 PR tree-optimization/80769
14904 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
14905 for malloc and calloc. Document the new invariant that all related
14906 strinfos have delayed lengths or none do.
14907 (verify_related_strinfos): Move earlier in file.
14908 (set_endptr_and_length): New function, split out from...
14909 (get_string_length): ...here. Also set the lengths of related
14911 (zero_length_string): Assert that chainsi has known (rather than
14913 (adjust_related_strinfos): Likewise.
14915 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14917 PR tree-optimization/81136
14918 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
14919 assert that two references with the same misalignment have the same
14920 compile-time misalignment if those compile-time misalignments
14923 2017-07-01 Andi Kleen <ak@linux.intel.com>
14925 * print-tree.c (print_node): Print all attributes.
14927 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14929 * cfg.c (scale_bbs_frequencies): New function.
14930 * cfg.h (scale_bbs_frequencies): Declare it.
14931 * cfgloopanal.c (single_likely_exit): Cleanup.
14932 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
14934 (scale_loop_profile): Likewise.
14935 (loop_version): Likewise.
14936 (create_empty_loop_on_edge): Update.
14937 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
14938 scale_loop_frequencies, scale_loop_profile, loopify,
14939 loop_version): Update prototypes.
14940 * modulo-sched.c (sms_schedule): Update.
14941 * predict.c (unlikely_executed_edge_p): Also check probability.
14942 (probably_never_executed_edge_p): Fix typo.
14943 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14944 * tree-parloops.c (gen_parallel_loop): Update.
14945 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
14946 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14947 * tree-ssa-loop-split.c (split_loop): Update.
14948 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14949 * tree-vect-loop-manip.c (vect_do_peeling): Update.
14950 (vect_loop_versioning): Update.
14951 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14953 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14955 * trans-mem.c (split_bb_make_tm_edge): Update profile.
14957 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14959 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
14960 to keep profile consistent.
14962 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14964 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
14965 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
14966 * profile-count.h (max_safe_multiplier): Make unsigned.
14967 (profile_count::guessed_zero): New.
14969 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14971 * bb-reorder.c (fix_up_crossing_landing_pad,
14972 fix_crossing_conditional_branches): Use make_single_succ_edge
14973 to keep profile consistent.
14975 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14977 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
14980 2017-07-01 Jakub Jelinek <jakub@redhat.com>
14983 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
14984 the right scopes, make sure cond_jump isn't preserved between multiple
14985 iterations. Search for fallthru edge whenever there are 3+ edges and
14986 use find_fallthru_edge for it.
14988 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14990 Patch by Alexander Monakov <amonakov@ispras.ru>
14991 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
14992 probabilities consistently.
14994 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14996 * pa.c (pa_expand_compare_and_swap_loop): Update call of
14997 emit_cmp_and_jump_insns.
14999 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15002 * tree-inline.c (expand_call_inline): Combine profile statuses.
15004 2017-06-30 Andrew Pinski <apinski@cavium.com>
15006 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
15007 fold_stmt returned true.
15009 2017-06-30 Nathan Sidwell <nathan@acm.org>
15011 * ggc.h (empty_string): Delete.
15012 * cfgexpand.c (expand_asm_stmt): Use plain "".
15013 * optabs.c (expand_asm_memory_barrier): Likewise.
15014 * stringpool.c (empty_string): Delete.
15015 (digit_vector, digit_string): Delete.
15016 (ggc_alloc_string): Use plain "", don't optimize single digit
15017 strings. Use ggc_alloc_atomic.
15019 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
15021 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
15022 comparison set and one other set, use the cost of the non-comparison
15025 2017-06-30 Nathan Sidwell <nathan@acm.org>
15027 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
15030 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
15032 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
15033 loops. Remove now unneeded calls to gimple_switch_set_label() that
15034 just set removed labels to NULL_TREE.
15036 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
15038 * tree-ssanames.c (set_range_info_raw): Abstract from ...
15039 (set_range_info): ...here. Only call set_range_info_raw if domain
15041 (set_nonzero_bits): Call set_range_info_raw.
15042 * tree-ssanames.h (set_range_info_raw): New.
15044 2017-06-30 Jakub Jelinek <jakub@redhat.com>
15047 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
15048 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
15049 of nonimmediate_operand and <store_mask_constraint> instead of m
15050 for the input operand. For V8FI iterator, always split if input
15051 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
15052 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
15053 <store_mask_predicate> instead of register_operand and
15054 <store_mask_constraint> instead of v for the input operand. Make
15055 sure both operands aren't MEMs for if not <mask_applied>.
15057 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
15059 * lto-wrapper.c (copy_file) Close both file descriptors before
15062 2017-06-30 Martin Liska <mliska@suse.cz>
15065 * multiple_target.c (create_dispatcher_calls): Make ifunc
15066 also for function that don't have calls or are not referenced.
15068 2017-06-30 Richard Biener <rguenther@suse.de>
15070 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
15071 analyze the first scalar stmt. Move vector type computation
15072 for the BB case here from ...
15073 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
15074 live operation processing in the SLP case properly.
15076 2017-06-30 Richard Biener <rguenther@suse.de>
15078 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
15080 2017-06-30 Martin Liska <mliska@suse.cz>
15083 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
15084 before BUILT_IN_UNWIND_RESUME when ASAN is used.
15086 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
15088 * doc/invoke.texi (AArch64): Add missing options and remove redundant
15091 2017-06-30 Richard Biener <rguenther@suse.de>
15093 PR tree-optimization/81249
15094 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
15095 condition reduction result to original scalar type.
15097 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15099 * profile-count.h (enum profile_quality): Fix typos and whitespace
15102 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15104 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15105 type for branch probabilities.
15107 2017-06-29 Julian Brown <julian@codesourcery.com>
15108 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15110 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15111 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15112 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15113 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15115 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15117 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15118 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15119 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15120 CC usage from generic code to here.
15121 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15122 CC usage into the target macros.
15124 2017-06-29 Maya Rashish <coypu@sdf.org>
15126 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15129 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15131 * arm/arm-builtins.c: Include profile-count.h
15132 * except.c (sjlj_emit_function_enter): Use
15133 profile_probability::unlikely.
15135 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15137 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15138 and tocrel_offset be pointer args rather than implicitly using
15140 (legitimate_constant_pool_address_p, rs6000_emit_move,
15141 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15142 tocrel_offset and use in toc_relative_expr_p call.
15143 (print_operand, print_operand_address): Use static tocrel_base_oac
15144 and tocrel_offset_oac.
15145 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15148 2017-06-29 Maya Rashish <coypu@sdf.org>
15150 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15152 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15154 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15155 objects, take into account only the alignment of 'op0' and 'mode1' if
15158 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15160 * ccmp.c (ccmp_tree_comparison_p): New function.
15161 (ccmp_candidate_p): Update to use above function.
15162 (get_compare_parts): New function.
15163 (expand_ccmp_next): Update to use new functions.
15164 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15166 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15167 take mode as argument.
15168 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15169 * expr.c (expand_expr_real_1): Pass mode as argument.
15171 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15173 * combine.c (combine_instructions): Print insns to dump_file, together
15176 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15178 * asan.c (asan_emit_stack_protection): Update.
15179 (create_cond_insert_point): Update.
15180 * auto-profile.c (afdo_propagate_circuit): Update.
15181 * basic-block.h (struct edge_def): Turn probability to
15182 profile_probability.
15183 (EDGE_FREQUENCY): Update.
15184 * bb-reorder.c (find_traces_1_round): Update.
15185 (better_edge_p): Update.
15186 (sanitize_hot_paths): Update.
15187 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
15188 (make_single_succ_edge): Update.
15189 (check_bb_profile): Update.
15190 (dump_edge_info): Update.
15191 (update_bb_profile_for_threading): Update.
15192 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
15193 probabilitycount to 0.
15194 * cfgbuild.c (compute_outgoing_frequencies): Update.
15195 * cfgcleanup.c (try_forward_edges): Update.
15196 (outgoing_edges_match): Update.
15197 (try_crossjump_to_edge): Update.
15198 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
15199 (expand_gimple_tailcall): Update.
15200 (construct_init_block): Use make_single_succ_edge.
15201 (construct_exit_block): Use make_single_succ_edge.
15202 * cfghooks.c (verify_flow_info): Update.
15203 (redirect_edge_succ_nodup): Update.
15204 (split_edge): Update.
15205 (account_profile_record): Update.
15206 * cfgloopanal.c (single_likely_exit): Update.
15207 * cfgloopmanip.c (scale_loop_profile): Update.
15208 (set_zero_probability): Remove.
15209 (duplicate_loop_to_header_edge): Update.
15210 * cfgloopmanip.h (loop_version): Update prototype.
15211 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
15212 (force_nonfallthru_and_redirect): Update.
15213 (update_br_prob_note): Update.
15214 (rtl_verify_edges): Update.
15215 (purge_dead_edges): Update.
15216 (rtl_lv_add_condition_to_bb): Update.
15217 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
15218 * cgraphunit.c (init_lowered_empty_function): Update.
15219 (cgraph_node::expand_thunk): Update.
15220 * cilk-common.c: Include profile-count.h
15221 * dojump.c (inv): Remove.
15222 (jumpifnot): Update.
15223 (jumpifnot_1): Update.
15224 (do_jump_1): Update.
15226 (do_jump_by_parts_greater_rtx): Update.
15227 (do_compare_rtx_and_jump): Update.
15228 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
15229 do_jump_1. do_compare_rtx_and_jump): Update prototype.
15230 * dwarf2cfi.c: Include profile-count.h
15231 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
15232 (sjlj_emit_dispatch_table): Likewise.
15233 * explow.c: Include profile-count.h
15234 * expmed.c (emit_store_flag_force): Update.
15235 (do_cmp_and_jump): Update.
15236 * expr.c (compare_by_pieces_d::generate): Update.
15237 (compare_by_pieces_d::finish_mode): Update.
15238 (emit_block_move_via_loop): Update.
15239 (store_expr_with_bounds): Update.
15240 (store_constructor): Update.
15241 (expand_expr_real_2): Update.
15242 (expand_expr_real_1): Update.
15243 * expr.h (try_casesi, try_tablejump): Update prototypes.
15244 * gimple-pretty-print.c (dump_probability): Update.
15245 (dump_profile): New.
15246 (dump_gimple_label): Update.
15247 (dump_gimple_bb_header): Update.
15248 * graph.c (draw_cfg_node_succ_edges): Update.
15249 * hsa-gen.c (convert_switch_statements): Update.
15250 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
15251 (find_if_case_1): Update.
15252 (find_if_case_2): Update.
15253 * internal-fn.c (expand_arith_overflow_result_store): Update.
15254 (expand_addsub_overflow): Update.
15255 (expand_neg_overflow): Update.
15256 (expand_mul_overflow): Update.
15257 (expand_vector_ubsan_overflow): Update.
15258 * ipa-cp.c (good_cloning_opportunity_p): Update.
15259 * ipa-split.c (split_function): Use make_single_succ_edge.
15260 * ipa-utils.c (ipa_merge_profiles): Update.
15261 * loop-doloop.c (add_test): Update.
15262 (doloop_modify): Update.
15263 * loop-unroll.c (compare_and_jump_seq): Update.
15264 (unroll_loop_runtime_iterations): Update.
15265 * lra-constraints.c (lra_inheritance): Update.
15266 * lto-streamer-in.c (input_cfg): Update.
15267 * lto-streamer-out.c (output_cfg): Update.
15268 * mcf.c (adjust_cfg_counts): Update.
15269 * modulo-sched.c (sms_schedule): Update.
15270 * omp-expand.c (expand_omp_for_init_counts): Update.
15271 (extract_omp_for_update_vars): Update.
15272 (expand_omp_ordered_sink): Update.
15273 (expand_omp_for_ordered_loops): Update.
15274 (expand_omp_for_generic): Update.
15275 (expand_omp_for_static_nochunk): Update.
15276 (expand_omp_for_static_chunk): Update.
15277 (expand_cilk_for): Update.
15278 (expand_omp_simd): Update.
15279 (expand_omp_taskloop_for_outer): Update.
15280 (expand_omp_taskloop_for_inner): Update.
15281 * omp-simd-clone.c (simd_clone_adjust): Update.
15282 * optabs.c (expand_doubleword_shift): Update.
15283 (expand_abs): Update.
15284 (emit_cmp_and_jump_insn_1): Update.
15285 (expand_compare_and_swap_loop): Update.
15286 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
15287 * predict.c (predictable_edge_p): Update.
15288 (edge_probability_reliable_p): Update.
15289 (set_even_probabilities): Update.
15290 (combine_predictions_for_insn): Update.
15291 (combine_predictions_for_bb): Update.
15292 (propagate_freq): Update.
15293 (estimate_bb_frequencies): Update.
15294 (force_edge_cold): Update.
15295 * profile-count.c (profile_count::dump): Add missing space into dump.
15296 (profile_count::debug): Add newline.
15297 (profile_count::differs_from_p): Explicitly convert to unsigned.
15298 (profile_count::stream_in): Update.
15299 (profile_probability::dump): New member function.
15300 (profile_probability::debug): New member function.
15301 (profile_probability::differs_from_p): New member function.
15302 (profile_probability::differs_lot_from_p): New member function.
15303 (profile_probability::stream_in): New member function.
15304 (profile_probability::stream_out): New member function.
15305 * profile-count.h (profile_count_quality): Rename to ...
15306 (profile_quality): ... this one.
15307 (profile_probability): New.
15308 (profile_count): Update.
15309 * profile.c (compute_branch_probabilities): Update.
15310 * recog.c (peep2_attempt): Update.
15311 * sched-ebb.c (schedule_ebbs): Update.
15312 * sched-rgn.c (find_single_block_region): Update.
15313 (compute_dom_prob_ps): Update.
15314 (schedule_region): Update.
15315 * sel-sched-ir.c (compute_succs_info): Update.
15316 * stmt.c (struct case_node): Update.
15317 (do_jump_if_equal): Update.
15318 (get_outgoing_edge_probs): Update.
15319 (conditional_probability): Update.
15320 (emit_case_dispatch_table): Update.
15321 (expand_case): Update.
15322 (expand_sjlj_dispatch_table): Update.
15323 (emit_case_nodes): Update.
15324 * targhooks.c: Update.
15325 * tracer.c (better_p): Update.
15326 (find_best_successor): Update.
15327 * trans-mem.c (expand_transaction): Update.
15328 * tree-call-cdce.c: Update.
15329 * tree-cfg.c (gimple_split_edge): Upate.
15330 (move_sese_region_to_fn): Upate.
15331 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
15332 * tree-eh.c (lower_resx): Upate.
15333 (cleanup_empty_eh_move_lp): Upate.
15334 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15335 * tree-inline.c (copy_edges_for_bb): Update.
15336 (copy_cfg_body): Update.
15337 * tree-parloops.c (gen_parallel_loop): Update.
15338 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
15339 (gimple_gen_time_profiler): Update.
15340 * tree-ssa-dce.c (remove_dead_stmt): Update.
15341 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15342 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
15343 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
15344 (unloop_loops): Update.
15345 (try_peel_loop): Update.
15346 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15347 * tree-ssa-loop-split.c (connect_loops): Update.
15348 (split_loop): Update.
15349 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15350 (hoist_guard): Update.
15351 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
15352 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
15353 (value_replacement): Update.
15354 * tree-ssa-reassoc.c (branch_fixup): Update.
15355 * tree-ssa-tail-merge.c (replace_block_by): Update.
15356 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
15357 (create_edge_and_update_destination_phis): Update.
15358 (compute_path_counts): Update.
15359 (recompute_probabilities): Update.
15360 (update_joiner_offpath_counts): Update.
15361 (freqs_to_counts_path): Update.
15362 (duplicate_thread_path): Update.
15363 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
15364 (struct switch_conv_info): Update.
15365 (gen_inbound_check): Update.
15366 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15367 (vect_do_peeling): Update.
15368 (vect_loop_versioning): Update.
15369 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15370 (optimize_mask_stores): Update.
15371 * ubsan.c (ubsan_expand_null_ifn): Update.
15372 * value-prof.c (gimple_divmod_fixed_value): Update.
15373 (gimple_divmod_fixed_value_transform): Update.
15374 (gimple_mod_pow2): Update.
15375 (gimple_mod_pow2_value_transform): Update.
15376 (gimple_mod_subtract): Update.
15377 (gimple_mod_subtract_transform): Update.
15378 (gimple_ic): Update.
15379 (gimple_stringop_fixed_value): Update.
15380 (gimple_stringops_transform): Update.
15381 * value-prof.h: Update.
15383 2017-06-29 Carl Love <cel@us.ibm.com>
15385 * config/rs6000/rs6000-c.c: Add support for built-in functions
15386 vector signed int vec_signed (vector float);
15387 vector signed long long vec_signed (vector double);
15388 vector signed int vec_signed2 (vector double, vector double);
15389 vector signed int vec_signede (vector double);
15390 vector signed int vec_signedo (vector double);
15391 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
15392 instruction generator.
15393 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15394 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
15395 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
15397 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
15398 vunsignede_v2df): Add define_expands.
15399 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
15400 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
15401 VEC_UNSIGNEDO): Add definitions.
15402 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15403 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
15404 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
15405 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
15406 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
15407 * config/rs6000/altivec.h (vec_signed, vec_signed2,
15408 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
15409 vec_unsignede, vec_unsignedo): Add builtin defines.
15410 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
15412 * doc/extend.texi: Update the built-in documentation file for the
15413 new built-in functions.
15415 2017-06-29 Richard Biener <rguenther@suse.de>
15417 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
15418 reduction chains to LOOP_VINFO_REDUCTIONS.
15419 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
15420 SLP reductions after processing reduction chains.
15422 2017-06-29 Nathan Sidwell <nathan@acm.org>
15424 * builtins.c (fold_builtin_FUNCTION): Use
15425 lang_hooks.decl_printable_name.
15427 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
15429 PR middle-end/81194
15430 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
15431 with only one label.
15432 * stmt.c (expand_case): Assert NCASES is greater than one.
15434 2017-06-29 Richard Biener <rguenther@suse.de>
15436 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
15438 (group_case_labels): Likewise.
15439 (find_taken_edge): Push sanity checking on val to workers...
15440 (find_taken_edge_cond_expr): ... here
15441 (find_taken_edge_switch_expr): ... and here, handle cases
15442 with just a default label.
15443 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
15444 (group_case_labels): Likewise.
15445 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
15446 group_case_labels does anything cleanup the CFG again.
15448 2017-06-29 Bin Cheng <bin.cheng@arm.com>
15450 PR tree-optimization/81196
15451 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
15452 exit condition comparing two IVs.
15454 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
15456 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
15457 profile to the dummy entry at the end of the list of architectures.
15458 * config/arm/arm-cpu-cdata.h: Regenerated.
15460 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15461 Michael Collison <michael.collison@arm.com>
15464 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
15466 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
15467 (*aarch64_reg_<mode>3_minus_mask): New pattern.
15468 (*aarch64_<optab>_reg_di3_mask2): New pattern.
15469 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
15470 of shift when the shift amount is masked with constant equal to
15471 the size of the mode.
15472 * config/aarch64/predicates.md (subreg_lowpart_operator): New
15475 2017-06-29 Martin Liska <mliska@suse.cz>
15477 * config/i386/i386.opt: Change range from [1,5] to [0,5].
15479 2017-06-29 Yury Gribov <tetra2005@gmail.com>
15482 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
15485 (edge_growth_cache_entry::edge_growth_cache_entry): New
15487 (reset_edge_growth_cache): Update to use constructor.
15489 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15491 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
15492 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15493 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15495 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
15497 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
15498 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
15500 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
15502 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
15503 (*-linux-uclibc*): Add t-uclibc tmake_file.
15504 * config/t-musl: New.
15505 * config/t-uclibc: New.
15507 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
15509 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
15511 (gen_comm_data): Emit architectural setting of arch_prof.
15512 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
15514 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
15515 (armv8-m.base, armv8-m.main): Likewise.
15516 * arm-protos.h (arm_build_target): Add profile field.
15517 (arch_option): Likewise.
15518 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
15520 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
15521 arm_active_target.profile.
15523 2017-06-28 Richard Biener <rguenther@suse.de>
15525 PR middle-end/81227
15526 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
15527 TYPE_OVERFLOW_WRAPS.
15528 * match.pd (negate_expr_p): Likewise.
15529 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
15530 fold_build2, not fold_binary.
15532 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15534 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
15535 Convert memory address to Pmode.
15536 (aarch64_print_operand): Assert MEM operands are always Pmode.
15538 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15541 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
15542 Remove redundant if.
15543 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
15544 * config/arm/aarch-common-protos.h
15545 (aarch_forward_to_shift_is_not_shifted_re): Remove.
15546 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
15548 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
15551 * multiple_target.c (create_dispatcher_calls): Set the default
15552 clone to be static, not public.
15554 2017-06-28 Richard Biener <rguenther@suse.de>
15556 * tree-vect-loop.c (vectorizable_reduction): Move special
15557 cond reduction IV var creation ...
15558 (vect_create_epilog_for_reduction): ... here. Remove induction_index
15559 parameter. Use STMT_VINFO_VECTYPE.
15560 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
15563 2017-06-28 Martin Liska <mliska@suse.cz>
15566 * ipa-visibility.c (non_local_p): Handle visibility.
15568 2017-06-28 Martin Liska <mliska@suse.cz>
15571 * common.opt: Add IntegerRange to various options.
15572 * opt-functions.awk (integer_range_info): New function.
15573 * optc-gen.awk: Add integer_range_info to cl_options struct.
15574 * opts-common.c (decode_cmdline_option): Handle
15575 CL_ERR_INT_RANGE_ARG.
15576 (cmdline_handle_error): Likewise.
15577 * opts.c (print_filtered_help): Show valid interval in
15578 when --help is provided.
15579 * opts.h (struct cl_option): Add range_min and range_max fields.
15580 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
15582 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
15584 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
15585 (x * C EQ/NE y * C): New transformation.
15587 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
15589 * genmultilib (combination_space): Accept '+' in option names.
15591 2017-06-28 Martin Liska <mliska@suse.cz>
15594 * asan.c (instrument_derefs): Bail out inner references
15595 that are hard register variables.
15597 2017-06-28 Jakub Jelinek <jakub@redhat.com>
15600 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
15601 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
15603 2017-06-28 Richard Biener <rguenther@suse.de>
15605 * tree-vectorizer.h (vect_get_vec_defs): Remove.
15606 (vect_get_slp_defs): Adjust.
15607 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
15609 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
15611 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
15612 get_initial_defs_for_reduction instead of vect_get_vec_defs.
15613 (vectorizable_reduction): Adjust.
15614 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
15616 (vect_get_slp_defs): Likewise.
15617 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
15618 (vectorizable_bswap): Adjust.
15619 (vectorizable_call): Likewise.
15620 (vectorizable_conversion): Likewise.
15621 (vectorizable_assignment): Likewise.
15622 (vectorizable_shift): Likewise.
15623 (vectorizable_operation): Likewise.
15624 (vectorizable_store): Likewise.
15625 (vectorizable_condition): Likewise.
15626 (vectorizable_comparison): Likewise.
15628 2017-06-28 Michael Collison <michael.collison@arm.com>
15631 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
15633 (arm_gen_movmemqi): Removed unused variable 'i'.
15634 Convert 'for' loop into 'while' loop.
15635 (arm_expand_prologue): Remove last unnecessary set of insn.
15636 (thumb_pop): Remove unused variable 'pushed_words'.
15637 (thumb_exit): Remove last unnecessary set of regs_to_pop.
15639 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15641 * config/s390/predicates.md: Use s390_rel_address_ok_p.
15642 * config/s390/s390-protos.h: Add prototype of
15643 s390_rel_address_ok_p.
15644 * config/s390/s390.c (s390_got_symbol): New function.
15645 (s390_rel_address_ok_p): New function.
15646 (legitimize_pic_address): Use s390_rel_address_ok_p.
15647 (s390_load_got): Use s390_got_symbol.
15648 (s390_option_override): Issue error if
15649 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
15650 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
15652 * config/s390/s390.opt: New option mpic-data-is-text-relative.
15654 2017-06-27 Andrew Pinski <apinski@cavium.com>
15656 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
15657 (X * copysign (1.0, X)): New pattern.
15658 (X * copysign (1.0, -X)): New pattern.
15659 (copysign (-1.0, CST)): New pattern.
15661 2017-06-27 Joseph Myers <joseph@codesourcery.com>
15663 * genmultilib (combination_space): Remove variable.
15664 Validate reuse rules against regular expression for any sequence
15665 of multilib options in any order.
15667 2017-06-27 Michael Collison <michael.collison@arm.com>
15669 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
15670 call aarch64_split_simd_combine.
15671 * (aarch64_combine_internal<mode>): Delete pattern.
15672 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
15673 Allow register and subreg operands.
15675 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15677 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
15678 specific need, just fallback on defaults.
15679 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
15681 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15682 Olivier Hainque <hainque@adacore.com>
15684 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
15686 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
15687 targets. Pick a default if no particular attempt applied.
15688 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
15691 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15693 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
15694 (x86_64-wrs-vxworks7): Likewise.
15696 2017-06-27 Marek Polacek <polacek@redhat.com>
15699 * ubsan.c (instrument_null): Check get_base_address's result for null.
15701 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15703 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
15705 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15707 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
15708 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
15709 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
15710 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
15711 New function types.
15712 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
15713 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
15714 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
15715 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
15716 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
15717 BUILT_IN_FEUPDATEENV): New builtins.
15718 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
15719 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
15720 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
15721 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
15723 (builtin_structptr_types): Adjust size.
15724 * tree.c (builtin_structptr_types): Add four entries.
15726 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15727 Olivier Hainque <hainque@adacore.com>
15729 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
15730 (TLS_SYM): New local macro, forcing reference to __tls__ on
15731 link command lines for VxWorks 7 RTPs, triggering initialization
15733 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
15734 OS features TLS support, true for RTPs on VxWorks 7.
15735 * config/vxworks.c (vxworks_override_options): Setup emutls
15738 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15740 * predict.c (test_prediction_value_range): Use -1U instead of -1
15741 to avoid narrowing conversion warning.
15742 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
15743 to avoid narrowing conversion warning.
15744 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
15746 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
15748 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15750 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
15751 64bit configurations.
15752 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
15753 (SIZE_TYPE): Likewise.
15754 * config/vxworks.c (vxworks_emutls_var_fields): Use
15755 long_unsigned_type_node instead of unsigned_type_node as the offset
15756 field type, which is "pointer" mode in emutls.c.
15758 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15761 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
15763 PR middle-end/81207
15764 * gimple-fold.c (replace_call_with_call_and_fold): Handle
15765 gimple_vuse copying separately from gimple_vdef copying.
15767 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15769 * value-prof.c (free_hist): Remove call to memset and the enclosing if
15772 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15773 Olivier Hainque <hainque@adacore.com>
15775 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
15776 for all vxworks7 targets.
15777 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
15778 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
15779 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
15780 variations for VX6/VX7 and 32/64bits later on in ...
15781 (VXWORKS_LIB_SPEC): Leverage new macros.
15782 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
15783 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
15785 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15787 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
15788 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
15790 2017-06-26 Carl Love <cel@us.ibm.com>
15792 * config/rs6000/rs6000-c.c: Add support for built-in functions
15793 vector bool char vec_reve (vector bool char);
15794 vector signed char vec_reve (vector signed char);
15795 vector unsigned char vec_reve (vector unsigned char);
15796 vector bool int vec_reve (vector bool int);
15797 vector signed int vec_reve (vector signed int);
15798 vector unsigned int vec_reve (vector unsigned int);
15799 vector bool long long vec_reve (vector bool long long);
15800 vector signed long long vec_reve (vector signed long long);
15801 vector unsigned long long vec_reve (vector unsigned long long);
15802 vector bool short vec_reve (vector bool short);
15803 vector signed short vec_reve (vector signed short);
15804 vector double vec_reve (vector double);
15805 vector float vec_reve (vector float);
15806 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
15807 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
15808 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
15809 (altivec_vreve): New pattern.
15810 * config/rs6000/altivec.h (vec_reve): New define.
15811 * doc/extend.texi (vec_rev): Update the built-in documentation file
15812 for the new built-in functions.
15814 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15816 PR tree-optimization/71815
15817 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
15819 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
15821 (slsr_process_phi): Likewise.
15822 (replace_uncond_cands_and_profitable_phis): Don't replace a
15823 multiply candidate with a stride of 1 (copy or cast).
15824 (phi_incr_cost): Call uses_consumed_by_stmt rather than
15826 (lowest_cost_path): Likewise.
15827 (total_savings): Likewise.
15829 2017-06-26 Richard Biener <rguenther@suse.de>
15832 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
15833 Use def_builtin_pure for all gather builtins.
15835 2017-06-26 Richard Biener <rguenther@suse.de>
15837 PR tree-optimization/81203
15838 * tree-tailcall.c (find_tail_calls): Do not move stmts into
15839 non-dominating BBs.
15841 2017-06-26 Marek Polacek <polacek@redhat.com>
15844 * doc/invoke.texi: Document -Wmultistatement-macros.
15846 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
15848 * doc/sourcebuild.texi (ARM-specific attributes): Document new
15849 arm_neon_ok_no_float_abi effective target.
15851 2017-06-26 Richard Biener <rguenther@suse.de>
15853 PR tree-optimization/80928
15854 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
15855 (copy_bbs): Set BB_DUPLICATED flag early.
15856 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
15858 (execute_on_shrinking_pred): Likewise.
15859 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
15860 BB_DUPLICATED blocks.
15861 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
15862 iterate over all PHIs considering removal of *gsi.
15864 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
15866 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
15869 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15871 * config/rs6000/rs6000-string.c: (expand_block_clear,
15872 do_load_for_compare, select_block_compare_mode,
15873 compute_current_alignment, expand_block_compare,
15874 expand_strncmp_align_check, expand_strn_compare,
15875 expand_block_move, rs6000_output_load_multiple)
15876 Move functions related to string/block move/compare
15877 to a separate file.
15878 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
15879 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
15880 for this function which is now used in two files.
15881 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
15882 * config.gcc: Add rs6000-string.o to extra_objs for
15883 targets powerpc*-*-* and rs6000*-*-*.
15885 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
15888 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
15889 32-bit, since indexed is not valid for DImode.
15890 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
15891 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
15892 (define_peephole2 for Altivec d-form load): Add 32-bit support.
15893 (define_peephole2 for Altivec d-form store): Likewise.
15896 * multiple_target.c (create_dispatcher_calls): Only create the
15897 dispatcher call if the function is the default clone of a
15898 versioned function.
15900 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
15902 PR middle-end/80902
15903 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
15904 a call, force the call to not be a tail call.
15906 2017-06-23 Jeff Law <law@redhat.com>
15908 * doc/contrib.texi: Add entry for Steven Pemberton's work on
15911 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
15913 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
15914 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
15915 handling for early expansion of vector shifts (sl,sr,sra,rl).
15916 (builtin_function_type): Add vector shift right instructions
15917 to the unsigned argument list.
15919 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
15921 rtl-optimizatoin/79286
15922 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
15923 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
15924 trap. PIC register plus a const unspec without offset can never trap.
15926 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
15928 * tree.h (builtin_structptr_type): New type.
15929 (builtin_structptr_types): Declare new array.
15930 * tree.c (builtin_structptr_types): New array.
15931 (free_lang_data, build_common_tree_nodes): Use it.
15933 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
15936 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
15939 2017-06-22 Matt Turner <mattst88@gmail.com>
15941 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
15942 Lake models to skylake case. Assume skylake for unknown
15943 models with clflushopt.
15945 2017-06-22 Jeff Law <law@redhat.com>
15947 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
15948 frame sizes that do not satisfy aarch64_uimm12_shift.
15950 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
15952 * profile-count.h (apply_probability,
15953 apply_scale, probability_in): Fix checks for zero.
15955 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15957 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
15958 * doc/cppdiropts.texi (-I @var{dir}): Document it.
15960 2016-06-22 Richard Biener <rguenther@suse.de>
15962 * tree-vect-loop.c (vect_model_reduction_cost): Handle
15963 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
15964 REDUC_MAX_EXPR support.
15965 (vectorizable_reduction): Likewise.
15966 (vect_create_epilog_for_reduction): Likewise.
15968 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
15970 * match.pd (A / (1 << B) -> A >> B): New.
15971 * generic-match-head.c: Include optabs-tree.h.
15972 * gimple-match-head.c: Likewise.
15973 * optabs-tree.h (target_supports_op_p): New.
15974 * optabs-tree.c (target_supports_op_p): New.
15976 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15978 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
15979 $gcc_cv_ld --help output.
15980 (gcc_cv_ld_demangle): Likewise.
15981 (gcc_cv_ld_eh_frame_hdr): Likewise.
15982 (gcc_cv_ld_pie): Likewise.
15983 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
15984 (gcc_cv_ld_buildid): Likewise.
15985 (gcc_cv_ld_sysroot): Likewise.
15986 (ld_bndplt_support): Likewise.
15987 (ld_pushpopstate_support): Likewise.
15988 * configure: Regenerate.
15989 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
15991 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15994 * config/i386/sse.md (round<mode>2): Renumber match_dup and
15995 operands indexes to avoid gap between operands and match_dups.
15997 2017-06-21 Andrew Pinski <apinski@cavium.com>
15999 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
16000 Increment Arith_shift and Arith_shift_reg by 1.
16001 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
16003 * config/aarch64/aarch64.c (thunderx_tunings): Enable
16004 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
16005 (aarch64_strip_extend): Add new argument and test for it.
16006 (aarch64_cheap_mult_shift_p): New function.
16007 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
16008 add a cost if it is true.
16009 Update calls to aarch64_strip_extend.
16010 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
16012 2017-06-21 Andrew Pinski <apinski@cavium.com>
16014 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
16016 (thunderxt88): Likewise.
16017 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
16018 (thunderx_prefetch_tune): New variable.
16019 (thunderx2t99_prefetch_tune): Update for the correct values.
16020 (thunderxt88_tunings): New variable.
16021 (thunderx_tunings): Use thunderx_prefetch_tune instead of
16022 generic_prefetch_tune.
16023 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
16025 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16027 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
16028 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
16029 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
16030 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
16031 (aarch64_atomic_cas<mode>, GPI): Likewise.
16033 2017-06-21 Martin Liska <mliska@suse.cz>
16035 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
16036 statements on cold and hot labels.
16037 * predict.c (tree_estimate_probability_bb): Remove the
16038 prediction from this place.
16040 2017-06-21 Martin Liska <mliska@suse.cz>
16042 PR tree-optimization/79489
16043 * gimplify.c (maybe_add_early_return_predict_stmt): New
16045 (gimplify_return_expr): Call the function.
16046 * predict.c (tree_estimate_probability_bb): Remove handling
16048 * predict.def: Update comment about early return predictor.
16049 * gimple-predict.h (is_gimple_predict): New function.
16050 * predict.def: Change default value of early return to 66.
16051 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
16053 * passes.def: Put pass_strip_predict_hints to the beginning of
16056 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
16058 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
16059 FUNCTION_DECL declarations.
16060 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
16062 (dwaf2out_decl): Likewise.
16063 * godump.c (go_early_global_decl): Skip call to the real debug hook
16064 for FUNCTION_DECL declarations.
16065 * passes.c (rest_of_decl_compilation): Skip call to the
16066 early_global_decl debug hook for FUNCTION_DECL declarations, unless
16067 -fdump-go-spec is passed.
16069 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16071 * config/i386/i386.c (struct builtin_isa): New field pure_p.
16072 Reorder for compactness.
16073 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
16074 (def_builtin_pure, def_builtin_pure2): New functions.
16075 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
16077 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16079 * match.pd (nop_convert): New predicate.
16080 ((A +- CST1) +- CST2): Allow some NOP conversions.
16082 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16085 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
16086 with ctors/dtors if GOVD_SHARED is set.
16088 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16090 * config/aarch64/aarch64.md (movti_aarch64):
16091 Emit mov rather than orr.
16092 (movtf_aarch64): Likewise.
16093 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
16094 Emit mov rather than orr.
16096 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16098 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16099 Swap alternatives, make integer dup more expensive.
16101 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16103 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16104 Return true for non-tls symbols.
16106 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16108 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16109 (cortex-a75): Likewise.
16110 (cortex-a75.cortex-a55): Likewise.
16111 * config/aarch64/aarch64-tune.md: Regenerate.
16112 * doc/invoke.texi (-mtune): Document new values for -mtune.
16114 2017-06-21 Tom de Vries <tom@codesourcery.com>
16116 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16117 stack_size feature.
16118 (Effective-Target Keywords, Other attributes): Suggest using
16119 dg-add-options stack_size feature to get stack limit in stack_size
16120 effective target documentation.
16122 2017-06-21 Julian Brown <julian@codesourcery.com>
16123 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16125 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16126 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16127 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16129 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16130 attribute type list for neon_multiply.
16131 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16132 attribute type list for neon_multiply.
16133 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16134 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16135 attribute type list for neon_multiply.
16136 * config/arm/types.md (crypto_pmull): Add.
16137 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16138 attribute type list.
16140 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16142 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16145 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16147 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16148 to make sure not to dereference a NULL cost_classes_ptr pointer.
16150 2017-06-20 Carl Love <cel@us.ibm.com>
16152 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16153 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16154 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16155 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16156 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16157 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16158 VMULOSW): New enum "unspec" values.
16159 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16160 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16161 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16162 altivec_vmulosw): New patterns.
16163 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16164 VMULOSW): Add definitions.
16166 2017-06-20 Julia Koval <julia.koval@intel.com>
16168 * config/i386/i386.c: Fix rounding expand for new pattern.
16169 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16171 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16173 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16174 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
16176 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16178 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
16181 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16183 * config/aarch64/aarch64-cores.def: Rearrange to sort by
16184 architecture, then by implementer ID.
16185 * config/aarch64/aarch64-tune.md: Regenerate.
16187 2017-06-20 Richard Biener <rguenther@suse.de>
16189 PR middle-end/81097
16190 * fold-const.c (split_tree): Fold to type before negating.
16192 2017-06-20 David Malcolm <dmalcolm@redhat.com>
16194 * diagnostic-show-locus.c
16195 (selftest::test_fixit_deletion_affecting_newline): New function.
16196 (selftest::diagnostic_show_locus_c_tests): Call it.
16198 2017-06-20 Andreas Schwab <schwab@suse.de>
16201 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
16204 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16206 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
16207 __ARM_FEATURE_COPROC according to support.
16209 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16211 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16212 Rewritten to avoid overflow for > 32-bit pointers.
16215 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
16216 by removing enum keyword.
16217 (ubsan_type_descriptor): Likewise. Formatting fix.
16220 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
16221 splitter): Require TARGET_SSE2 in the condition.
16223 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
16226 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
16227 for doing vector set of SFmode on ISA 3.0.
16228 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
16229 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
16231 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
16232 SFmode value into a V4SF variable that was extracted from another
16233 V4SF variable without converting the element to double precision
16234 and back to single precision vector format.
16235 (vsx_insert_extract_v4sf_p9_2): Likewise.
16237 2017-06-19 Jakub Jelinek <jakub@redhat.com>
16239 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
16240 in UWHI to avoid undefined overflow.
16243 * ubsan.h (enum ubsan_encode_value_phase): New.
16244 (ubsan_encode_value): Change second argument to
16245 enum ubsan_encode_value_phase with default value of
16246 UBSAN_ENCODE_VALUE_GENERIC.
16247 * ubsan.c (ubsan_encode_value): Change second argument to
16248 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
16249 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
16250 create_tmp_var_raw instead of create_tmp_var and use a
16252 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
16253 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
16254 ubsan_encode_value callers.
16257 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
16258 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
16259 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
16261 2017-06-19 Richard Biener <rguenther@suse.de>
16263 PR middle-end/81118
16264 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
16265 estimates if we changed anything.
16267 2017-06-19 Richard Biener <rguenther@suse.de>
16269 PR tree-optimization/80887
16270 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
16271 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
16272 simplified lookups, then reset mprts_hook.
16273 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
16275 (try_to_simplify): Likewise.
16277 2017-06-19 Martin Liska <mliska@suse.cz>
16280 * gimplify.c (gimplify_switch_expr):
16281 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
16283 2017-06-19 Martin Liska <mliska@suse.cz>
16285 * doc/install.texi: Document that PGO runs in 4 stages.
16287 2017-06-19 Martin Liska <mliska@suse.cz>
16290 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
16291 to dispatcher function name.
16292 * multiple_target.c (replace_function_decl): New function.
16293 (create_dispatcher_calls): Redirect both edges and references.
16295 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
16297 * profile-count.c (profile_count::dump): Dump quality.
16298 (profile_count::differs_from_p): Update for unsigned val.
16299 * profile-count.h (profile_count_quality): New enum.
16300 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
16302 2017-06-19 Richard Biener <rguenther@suse.de>
16304 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
16305 struct function as arg.
16306 (estimate_numbers_of_iterations): Export overload with loop arg.
16307 (free_numbers_of_iterations_estimates_loop): Use an overload of
16308 free_numbers_of_iterations_estimates instead.
16309 * tree-cfg.c (remove_bb): Adjust.
16310 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
16311 * tree-parloops.c (gen_parallel_loop): Likewise.
16312 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
16314 (tree_unroll_loops_completely): Likewise.
16315 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
16316 Use an overload instead and export.
16317 (estimated_loop_iterations): Adjust.
16318 (max_loop_iterations): Likewise.
16319 (likely_max_loop_iterations): Likewise.
16320 (estimate_numbers_of_iterations): Take struct function as arg
16322 (loop_exits_before_overflow): Adjust.
16323 (free_numbers_of_iterations_estimates_loop): Use an overload.
16324 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
16325 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
16327 2017-06-19 Richard Biener <rguenther@suse.de>
16330 * ipa-prop.c (find_constructor_constant_at_offset): Handle
16331 RANGE_EXPR conservatively.
16333 2017-06-16 Carl Love <cel@us.ibm.com>
16335 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16336 definitions for vec_float, vec_float2, vec_floato,
16337 vec_floate built-ins.
16338 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
16339 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
16341 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
16342 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
16343 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
16344 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
16345 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
16346 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
16347 vec_floato): Add builtin defines.
16348 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
16349 Update the built-in documentation file for the new built-in
16352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16354 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
16355 (mthumb): Mark as the negative of -marm.
16357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16359 * doc/invoke.texi (ARM Options, -mcpu): Document supported
16361 (ARM Options, -mtune): Document that this accepts the same
16362 extension options as -mcpu.
16363 (ARM Options, -mfpu): Document addition of -mfpu=auto.
16365 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16367 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
16368 permitted extensions.
16370 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16372 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
16373 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
16374 (armv8-m.main): Add option +nodsp.
16375 * config/arm/arm-cpu-cdata.h: Regenerated.
16377 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16379 * config/arm/t-fuchsia: New file.
16380 * config.gcc (arm*-*-fuchsia*): Use it.
16382 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16384 * config/arm/t-symbian: Rewrite for new option infrastructure.
16386 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16388 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
16389 (MULTILIB_REQUIRED): Likewise.
16391 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16393 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
16394 (MULTILIB_RESUE): Likewise.
16395 (MULTILIB_MATCHES): Likewise.
16396 (MULTLIB_REQUIRED): Likewise.
16398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16400 * config/arm/t-rtems: Rewrite for new option framework.
16402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16404 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
16405 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
16406 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
16407 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
16408 * config/arm/t-multilib: ... here.
16409 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
16410 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
16411 armv7-a and armv8*-a when A-profile libraries have not been built.
16412 * config/arm/t-rmprofile: Rewrite.
16414 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16416 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
16417 with a backslash. Remove the backslash after substituting unescaped
16419 * doc/fragments.texi (MULTILIB_REUSE): Document it.
16421 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16423 * config.gcc: (arm*-*-*): When building a-profile libraries, force
16424 the driver to pass through the default setting of -mfloat-abi.
16425 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
16427 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
16428 (all_feat_combs): New rule.
16429 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
16431 * config/arm/t-aprofile: Rewrite.
16433 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16435 * config/arm/arm.h (FPUTYPE_AUTO): Define.
16436 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
16437 fpu is not specified by the user/command-line.
16438 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
16439 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
16440 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
16441 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
16442 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
16443 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
16445 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16447 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
16448 * config/arm/t-arm-elf: Rewritten.
16450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16452 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
16453 have some floating-point instructions.
16454 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
16455 (TARGET_MAYBE_HARD_FLOAT): New macro.
16456 * config/arm/arm-builtins.c (arm_init_builtins): Use
16457 TARGET_MAYBE_HARD_FLOAT.
16458 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
16460 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16462 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
16463 (configargs.h): Include it.
16464 (arm_print_hint_for_fpu_option): New function.
16465 (arm_parse_fpu_option): New function.
16466 (candidate_extension): New class.
16467 (arm_canon_for_multilib): New function.
16468 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
16469 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
16470 (ARCH_CANONICAL_SPECS): New macro.
16471 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
16473 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16475 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
16476 are set after handling multilib fragments. Set target_cpu_default2
16479 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16481 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
16483 (arm*-*-*): Set target_cpu_default2 to a quoted string.
16484 * config/arm/parsecpu.awk (check_cpu): Validate any extension
16486 (check_arch): Likewise.
16487 * config/arm/arm.c (arm_configure_build_target): Handle
16488 TARGET_CPU_DEFAULT being a string constant. Scan any feature
16489 options in the default.
16491 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16493 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
16494 when an option is an alias of another.
16495 * config/arm/parsecpu.awk (optalias): New parser token.
16496 (gen_comm_data): Mark non-alias options as such. Emit entries
16497 for extension aliases.
16498 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
16499 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
16500 (armv6kz, armv6zk, armv6t2): Likewise.
16501 (armv7): Make vfpv3-d16 an alias.
16502 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
16504 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
16505 Sort in canonical order.
16506 (armv8-a): Sort in canonical order.
16507 (armv8.1-a, armv8.2-a): Likewise.
16508 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
16510 (cortex-a9): Sort in canonical order.
16511 * config/arm/arm.c (selftests.h): Include it.
16512 (arm_test_cpu_arch_data): New function.
16513 (arm_run_self_tests): New function.
16514 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
16515 (targetm): Move declaration to the end of the file.
16516 * arm-cpu-cdata.h: Regenerated.
16518 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16520 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
16521 call to target_mode_check describing the type of option passed.
16522 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
16523 (arm_target_thumb_only): Use arm_parse_arch_option_name or
16524 arm_parse_cpu_option_name to match parameters against list of
16526 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
16527 arm_arch_core_flags data structure.
16528 * config/arm/arm-cpu_cdata.h: Regenerated.
16530 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16532 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
16534 (arm_print_hint_for_cpu_option): Likewise.
16535 (arm_print_hint_for_arch_option): Likewise.
16536 (arm_parse_cpu_option_name): Likewise.
16537 (arm_parse_arch_option_name): Likewise.
16538 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
16539 of entries in the all_fpus list.
16540 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
16541 (arm_parse_cpu_option_name): Declare.
16542 (arm_parse_arch_option_name): Declare.
16543 (arm_parse_option_features): Declare.
16544 (arm_intialize_isa): Declare.
16545 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
16547 (gen_comm_data): ... here. Make definitions non-static.
16548 * config/arm/arm-cpu-data.h: Regenerated.
16549 * config/arm/arm-cpu-cdata.h: Regenerated.
16551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16553 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
16554 (cpu_arch_extension): New structure.
16555 (cpu_arch_option, arch_option, cpu_option): New structures.
16556 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
16557 architecture types.
16558 (gen_data): Generate new format data tables.
16559 * config/arm/arm.c (cpu_tune): New structure.
16560 (cpu_option, processors): Delete.
16561 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
16562 (arm_print_hint_for_cpu_option): ... this and ...
16563 (arm_print_hint_for_arch_option): ... this.
16564 (arm_parse_arch_cpu_name): Delete. Replace with ...
16565 (arm_parse_cpu_option_name): ... this and ...
16566 (arm_parse_arch_option_name): ... this.
16567 (arm_unrecognized_feature): Change type of target parameter to
16569 (arm_parse_arch_cpu_features): Delete. Replace with ...
16570 (arm_parse_option_features): ... this.
16571 (arm_configure_build_target): Rework to use new configuration data
16573 (arm_print_tune_info): Rework for new configuration data tables.
16574 * config/arm/arm-cpu-data.h: Regenerated.
16575 * config/arm/arm-cpu.h: Regenerated.
16577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16579 * Makefile.in (OBJS): Move sbitmap.o from here ...
16580 (OBJS-libcommon): ... to here.
16582 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16584 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
16585 (ISA_ALL_CRYPTO): New macro.
16586 (ISA_ALL_SIMD): New macro
16587 (ISA_ALL_FP): New macro.
16588 * config/arm/arm.c (fpu_bitlist): Update initializer.
16589 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
16591 (arm9e): Add fpu. Add option for nofp
16592 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
16593 (arm926ej-s, arm1026ej-s): Likewise.
16594 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
16595 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
16596 neon-fp16, neon-vfpv4, nofp and nosimd.
16597 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
16598 (cortex-a8): Add fpu. Add option for nofp.
16599 (cortex-a9): Add fpu. Add options for nosimd and nofp.
16600 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
16601 (cortex-r4f): Add fpu.
16602 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
16603 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
16605 (cortex-r8): Likewise.
16606 (cortex-m4): Add fpu. Add option for nofp.
16607 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
16608 (cortex-a17.cortex-a7): Likewise.
16609 (cortex-a32): Add fpu. Add options for crypto and nofp.
16610 (cortex-a35, cortex-a53): Likewise.
16611 (cortex-a57): Add fpu. Add option for crypto.
16612 (cortex-a72, cortex-a73): Likewise.
16613 (exynos-m1): Likewise.
16614 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
16615 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
16616 (cortex-m33): Add fpu. Add option for nofp.
16617 * config/arm/arm-cpu-cdata.h: Regenerated
16618 * config/arm/arm-cpu-data.h: Regenerated.
16620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16622 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
16623 (armv5te, armv5tej): Likewise.
16624 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
16625 (armv7): Add options fp and vfpv3-d16.
16626 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
16627 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
16629 (armv7ve): Likewise.
16630 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
16631 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
16632 (armv8-a): Add nocrypto option.
16633 (armv8.1-a, armv8.2-a): Likewise.
16634 (armv8-m.main): add options fp, fp.dp and nofp.
16636 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16638 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
16640 (armv8-a+crc): Delete.
16641 (armv8.1-a): Add options simd, crypto and nofp.
16642 (armv8.2-a): Add options fp16, simd, crypto and nofp.
16643 (armv8.2-a+fp16): Delete.
16644 (armv8-m.main): Add option dsp.
16645 (armv8-m.main+dsp): Delete.
16646 (cortex-a8): Add fpu. Add nofp option.
16647 (cortex-a9): Add fpu. Add nofp and nosimd options.
16648 * config/arm/parsecpu.awk (gen_data): Generate option tables and
16649 link to main cpu and architecture data structures.
16650 (gen_comm_data): Only put isa attributes from the main architecture
16652 (option): New statement for architecture and CPU entries.
16653 * arm.c (struct cpu_option): New structure.
16654 (struct processors): Add entry for options.
16655 (arm_unrecognized_feature): New function.
16656 (arm_parse_arch_cpu_name): Ignore any characters after the first
16658 (arm_parse_arch_cpu_feature): New function.
16659 (arm_configure_build_target): Separate out any CPU and architecture
16660 features and parse separately. Don't error out if -mfpu=auto is
16661 used with only an architecture string.
16662 (arm_print_asm_arch_directives): New function.
16663 (arm_file_start): Call it.
16664 * config/arm/arm-cpu-cdata.h: Regenerated.
16665 * config/arm/arm-cpu-data.h: Likewise.
16666 * config/arm/arm-tables.opt: Likewise.
16668 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16670 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
16671 assembler when it is not -mfpu=auto.
16673 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16675 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
16676 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
16677 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
16678 (ASM_CPU_SPEC): Rewrite.
16679 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
16680 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
16681 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
16682 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
16683 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
16684 copied string is NUL-terminated. Also strip any characters prefixed
16686 (arm_rewrite_selected_arch): New function.
16687 (arm_rewrite_march): New function.
16689 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16691 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
16692 (x_arm_cpu_string, x_arm_tune_string): Likewise.
16693 (march, mcpu, mtune): Convert to string-based options.
16694 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
16695 (arm_parse_arch_cpu_name): New function.
16696 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
16697 identify selected architecture or CPU.
16698 (arm_option_save): New function.
16699 (TARGET_OPTION_SAVE): Redefine.
16700 (arm_option_restore): Restore string options.
16701 (arm_option_print): Print string options.
16703 2017-06-16 Martin Sebor <msebor@redhat.com>
16705 PR tree-optimization/80933
16706 PR tree-optimization/80934
16707 * builtins.c (fold_builtin_3): Do not handle bcmp here.
16708 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
16709 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
16710 (gimple_fold_builtin): Call them.
16712 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16714 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
16715 as unlikely; update profile.
16717 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16719 * predict.c (force_edge_cold): Handle declaring edges impossible
16722 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16724 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
16726 (try_unroll_loop_completely): Fix reporting.
16728 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16730 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
16732 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
16735 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
16736 if given a non-constant argument for an intrinsic which requires a
16739 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16741 * profile.c (compare_freqs): New function.
16742 (branch_prob): Sort edge list.
16743 (find_spanning_tree): Assume that the list is priority sorted.
16745 2017-06-16 Richard Biener <rguenther@suse.de>
16747 PR tree-optimization/81090
16748 * passes.def (pass_record_bounds): Remove.
16749 * tree-pass.h (make_pass_record_bounds): Likewise.
16750 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
16751 make_pass_record_bounds): Likewise.
16752 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
16753 not free niter estimates at the beginning but at the end.
16754 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
16756 2017-06-16 Richard Biener <rguenther@suse.de>
16758 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
16759 initializer to workaround ICE in host GCC 4.8.
16761 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16763 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
16765 (clone_inlined_nodes): Update.
16767 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16769 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
16770 prefetch settings, and enable prefetching by default at -O3.
16772 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16774 * config/aarch64/aarch64.c (aarch64_override_options_internal):
16775 Set flag_prefetch_loop_arrays according to tuning data.
16777 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16779 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
16780 New tune structure.
16781 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
16782 [Unrelated to main purpose of the patch] Place the pointer field last
16783 to enable type checking errors when tune structure are wrongly merged.
16784 * config/aarch64/aarch64.c (generic_prefetch_tune,)
16785 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
16786 (thunderx2t99_prefetch_tune): New tune constants.
16787 (tune_params *_tunings): Update all tunings (no functional change).
16788 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
16789 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
16790 from tunings structures.
16792 2017-06-16 Jakub Jelinek <jakub@redhat.com>
16795 * ubsan.c (instrument_null): Add T argument, use it instead
16796 of computing it based on IS_LHS.
16797 (instrument_object_size): Likewise.
16798 (pass_ubsan::execute): Adjust instrument_null and
16799 instrument_object_size callers to pass gimple_get_lhs or
16800 gimple_assign_rhs1 result to it. Use instrument_null instead of
16801 calling get_base_address and instrument_mem_ref. Handle
16802 aggregate call arguments for object-size sanitization.
16804 2017-06-16 Yury Gribov <tetra2005@gmail.com>
16806 PR tree-optimization/81089
16807 * tree-vrp.c (is_masked_range_test): Validate operands of
16810 2017-06-15 Martin Sebor <msebor@redhat.com>
16813 * dumpfile.c (dump_register): Avoid calling memset to initialize
16814 a class with a default ctor.
16815 * gcc.c (struct compiler): Remove const qualification.
16816 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
16817 * hash-table.h: Ditto.
16818 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
16820 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
16821 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
16823 * params.h (struct param_info): Make struct members non-const.
16824 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
16825 with default initialization.
16826 * vec.h (vec_copy_construct, vec_default_construct): New helper
16828 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
16829 with vec_copy_construct.
16830 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
16831 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
16832 * doc/invoke.texi (-Wclass-memaccess): Document.
16834 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16836 * emit-rtl.h (is_leaf): Update comment about local
16837 register allocator.
16839 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16842 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
16843 for a variable to have a section before checking if the section has a
16845 Set section to.persistent if persistent attribute is set.
16846 Warn if .persistent attribute is used on an automatic variable.
16848 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
16850 PR rtl-optimization/80474
16851 * reorg.c (update_block): Do not ignore instructions in a delay slot.
16853 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
16855 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
16858 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
16860 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
16861 (casesi): Emit bounds checking as RTL.
16862 (casesi_internal_mips16_<mode>): Remove bounds checking.
16864 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
16866 * config/xtensa/xtensa.c (xtensa_option_override): Append
16867 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
16868 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
16869 xtensa_doloop_hooks): Define unconditionally.
16870 (xtensa_reorg_loops): Only call reorg_loops in the presence of
16872 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
16873 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
16874 for it in xtensa_option_override.
16875 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
16876 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
16878 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
16880 * doc/cppopts.texi: Document '-' special value to -MF.
16882 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
16884 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
16885 (cortex_a53_fconst): Likewise.
16886 (cortex_a53_fpmul): Likewise.
16887 (cortex_a53_f_load_64): Likewise.
16888 (cortex_a53_f_load_many): Likewise.
16889 (cortex_a53_advsimd_alu): Likewise.
16890 (cortex_a53_advsimd_alu_q): Likewise.
16891 (cortex_a53_advsimd_mul): Likewise.
16892 (cortex_a53_advsimd_mul_q): Likewise.
16893 (fpmac bypass): Add new bypass for fpmac-fpmac case.
16894 Add missing fmul, r2f_cvt and fconst cases.
16896 2017-06-14 Richard Biener <rguenther@suse.de>
16898 PR middle-end/81088
16899 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
16901 (fold_binary_loc): When associating do not treat pre-existing
16902 TREE_OVERFLOW on literal constants as a reason to allow
16903 TREE_OVERFLOW on associated literal constants.
16905 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16907 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
16908 (MASK_FEATURES): New macro.
16909 * config/sparc/sparc.c (sparc_option_override): Remove the special
16910 handling of -mfpu and generalize it to all MASK_FEATURES switches.
16912 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16914 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
16915 a division of 0 if non-call exceptions are enabled.
16917 2017-06-14 Andrew Pinski <apinski@cavium.com>
16918 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16921 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
16922 Improve vector initialization code gen for only variable case.
16924 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16926 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
16928 2017-06-14 Richard Biener <rguenther@suse.de>
16930 PR tree-optimization/81083
16931 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
16934 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16936 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
16937 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
16938 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
16939 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
16940 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
16941 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
16943 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16945 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
16946 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
16948 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16950 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
16952 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16954 * config/rs6000/t-rtems: Don't handle SPE.
16956 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16958 * config/rs6000/t-linux: Don't handle SPE.
16960 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16962 * config/rs6000/eabispe.h: Delete file.
16964 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16966 * config/rs6000/t-spe: Delete file.
16968 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16970 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
16971 (rs6000_legitimate_offset_address_p): Return false for anything in
16972 V2SImode or V2SFmode.
16974 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16976 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
16977 except V2SF and V2SI. Rearrange the vector modes, and add comments.
16978 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
16980 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
16981 (rs6000_legitimate_offset_address_p): Ditto.
16982 (rs6000_emit_move): Ditto.
16983 (rs6000_init_builtins): Remove V4HI_type_node.
16985 2017-06-13 Martin Liska <mliska@suse.cz>
16988 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
16989 (gate_asan): Likewise.
16990 * asan.h (asan_no_sanitize_address_p): Remove the function.
16991 (sanitize_flags_p): New function.
16992 * builtins.def: Fix coding style.
16993 * common.opt: Use renamed enum value.
16994 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
16995 * doc/extend.texi: Document no_sanitize attribute.
16996 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
16997 to SANITIZE_UNDEFINED_NONDEFAULT.
16998 * gcc.c (sanitize_spec_function): Use the renamed enum value.
16999 * gimple-fold.c (optimize_atomic_compare_exchange_p):
17000 Use sanitize_flags_p.
17001 * gimplify.c (gimplify_function_tree): Likewise.
17002 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
17003 * opts.c (parse_no_sanitize_attribute): New function.
17004 (common_handle_option): Use renamed enum value.
17005 * opts.h (parse_no_sanitize_attribute): Declare.
17006 * tree.c (sanitize_flags_p): New function.
17007 * tree.h: Declared here.
17008 * tsan.c: Use sanitize_flags_p.
17009 * ubsan.c (ubsan_expand_null_ifn): Likewise.
17010 (instrument_mem_ref): Likewise.
17011 (instrument_bool_enum_load): Likewise.
17012 (do_ubsan_in_current_function): Remove the function.
17013 (pass_ubsan::execute): Use sanitize_flags_p.
17014 * ubsan.h: Remove do_ubsan_in_current_function
17015 * tree-cfg.c (print_no_sanitize_attr_value): New function.
17016 (dump_function_to_file): Use it here.
17018 2017-06-13 Martin Jambor <mjambor@suse.cz>
17020 PR tree-optimization/80803
17021 PR tree-optimization/81063
17022 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
17023 (propagate_subaccesses_across_link): Enqueue subtree whenever
17024 necessary instead of relying on the caller.
17026 2017-06-13 Martin Jambor <mjambor@suse.cz>
17028 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
17029 that have a first_link.
17030 (sort_and_splice_var_accesses): Do not check first_link before
17032 (subtree_mark_written_and_enqueue): Likewise.
17033 (propagate_all_subaccesses): Likewise and do not stop at first
17034 parent with a first_link.
17036 2017-06-13 Martin Jambor <mjambor@suse.cz>
17038 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
17041 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17043 * match.pd: New pattern.
17045 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17047 * tree-vrp.c (is_masked_range_test): New function.
17048 (register_edge_assert_for): Determine ranges for
17051 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17053 PR tree-optimization/67328
17054 * fold-const.c (maskable_range_p): New function.
17055 (build_range_check): Generate bittests if possible.
17057 2017-06-13 Martin Liska <mliska@suse.cz>
17059 * gimple-pretty-print.c (dump_probability): Add new argument.
17060 (dump_edge_probability): Dump both probability and count.
17061 (dump_gimple_label): Likewise.
17062 (dump_gimple_bb_header): Likewise.
17064 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
17067 * config/avr/avr-devices.c: Fix indentation.
17068 * config/avr/gen-avr-mmcu-specs.c: Dito.
17070 2017-06-13 Richard Biener <rguenther@suse.de>
17072 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
17073 instead get vector type from stmt_info.
17074 (vectorizable_reduction): Adjust. Remove dead code.
17076 2017-06-13 Richard Biener <rguenther@suse.de>
17078 PR middle-end/81065
17079 * fold-const.c (extract_muldiv_1): Remove bogus distribution
17080 case of C * (x * C2 + C3).
17081 (fold_addr_of_array_ref_difference): Properly fold index difference.
17083 2017-06-12 David S. Miller <davem@davemloft.net>
17086 * config/sparc/sparc.md (return expander): Emit frame blockage if
17087 function uses alloca.
17089 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17091 * combine.c (make_field_assignment): Check len rather than the mode
17092 precision when calling force_to_mode.
17094 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
17096 Support multilibs and devices that see flash in RAM address range.
17099 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17100 (avr_mcu_t) <flash_pm_offset>: New field.
17101 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17102 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17103 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17104 (AVR_TINY_PM_OFFSET): Remove macro.
17105 * config/avr/avr.opt (-mshort-calls): New option.
17106 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17107 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17108 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17109 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17110 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17111 instead of avr_arch->have_jmp_call.
17112 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17113 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17114 avr_arch->flash_pm_offset to define.
17115 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17116 new field flash_pm_offset. Add entry for avrxmega3.
17117 (avr_texinfo): Add entry for avrxmega3.
17118 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17119 attiny212, attiny214,
17120 attiny412, attiny414, attiny416, attiny417,
17121 attiny814, attiny816, attiny817,
17122 attiny1614, attiny1616, attiny1617,
17123 attiny3214, attiny3216, attiny3217.
17124 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17125 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17126 (avr_print_operand_address) [AVR_TINY]: Same.
17127 (avr_asm_init_sections) <readonly_data_section>: Only patch
17128 callback if avr_arch->flash_pm_offset = 0.
17129 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17130 for rodata if avr_arch->flash_pm_offset != 0.
17131 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17132 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17133 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17134 (m_options): Append opt_rcall.
17135 (m_dirnames): Append dir_rcall.
17136 * config/avr/t-multilib: Regenerate.
17138 * configure.ac [target=avr]: Check whether avrxmega3 default
17139 linker description file works as needed.
17140 * configure: Regenerate.
17141 * doc/avr-mmcu.texi: Regenerate.
17142 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17143 <__AVR_ARCH__>: Document avrxmega3 and 103.
17144 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17145 <__AVR_SHORT_CALLS__>: Document it.
17146 <__AVR_PM_BASE_ADDRESS__>: Document it.
17147 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17148 (AVR Variable Attributes) <progmem>: Document this is
17149 not needed for avrxmega3.
17150 (AVR Named Address Spaces) <__flash>: Dito.
17152 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17154 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17156 2017-06-12 Doug Rupp <rupp@adacore.com>
17158 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17159 Append vxworks-stdint.h to the tm_file list.
17160 * config/vxworks-stdint.h: New file.
17162 2017-06-12 Martin Liska <mliska@suse.cz>
17164 PR tree-optimization/81041
17165 * tree-profile.c (gimple_gen_ic_func_profiler):
17166 Create an extra BB in profile-generate
17167 (gimple_gen_time_profiler): Likewise.
17169 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17171 PR tree-optimization/81003
17172 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17173 (update_range_test): Use it instead of force_gimple_operand_gsi.
17175 2017-06-12 Richard Biener <rguenther@suse.de>
17177 PR tree-optimization/81053
17178 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
17179 with backedge value not defined in loop. Simplify def stmt
17182 2017-06-11 Tom de Vries <tom@codesourcery.com>
17185 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
17187 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
17188 nvptx_cannot_force_const_mem.
17190 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17192 * opts.c (finish_options): Move test for flag_split_stack after
17193 it has been initialized.
17195 2017-06-11 Jason Merrill <jason@redhat.com>
17197 * tree.h (id_equal): New.
17198 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
17199 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
17200 instead of strcmp of IDENTIFIER_POINTER.
17202 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17204 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
17205 (mark_all_inlined_calls_cdtor): Fix formating.
17206 (inline_transform): Rescale profile before inlining.
17208 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17210 * cgraph.h (cgraph_edge::clone): Update prototype.
17211 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
17212 (cgraph_node::create_clone): Update.
17213 (cgraph_node::create_version_clone): Update.
17214 * tree-inline.c (copy_bb): Update.
17215 (expand_call_inline): Update.
17217 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17219 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
17220 factored out from ...
17221 (rs6000_emit_prologue): ... here.
17223 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17225 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
17226 factored out from ...
17227 (rs6000_emit_prologue): ... here.
17229 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17231 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
17233 (handle_missing_profiles): Fix computation of tp_first_run.
17234 (counts_to_freqs): Do not touch freqs when count is 0.
17236 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17238 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
17241 2017-06-10 Tom de Vries <tom@codesourcery.com>
17243 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17244 attributes): Document signal effective target.
17246 2017-06-10 Tom de Vries <tom@codesourcery.com>
17248 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17249 Document effective target stack_size.
17251 2017-06-09 David Malcolm <dmalcolm@redhat.com>
17253 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
17254 to the edit_context if they can be auto-applied.
17256 2017-06-9 Ian Lance Taylor <iant@golang.org>
17258 * opts.c (finish_options): If -fsplit-stack, disable implicit
17259 -forder-blocks-and-partition.
17260 * doc/invoke.texi (Optimize Options): Document that when using
17261 -fsplit-stack -forder-blocks-and-partition is not implicitly
17264 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17266 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
17267 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
17268 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
17269 * builtins.def (abort, trap, unreachable): Declare cold.
17270 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
17271 * tree-core.h (ECF_COLD): New.
17272 * tree.c (set_call_expr_flags): Handle ECF_COLD.
17273 (build_common_builtin_nodes): Mark unreachable and abort as cold.
17275 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17277 * predict.c (unlikely_executed_stmt_p): Cleanup.
17279 2017-06-09 Richard Biener <rguenther@suse.de>
17281 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
17282 model if the ref is always written to.
17284 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17286 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
17288 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17290 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
17293 2017-06-09 Tom de Vries <tom@codesourcery.com>
17296 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
17297 "target cannot support label values" when encountering LABEL_REF.
17299 2017-06-09 Martin Liska <mliska@suse.cz>
17301 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
17302 (gimple_gen_ic_func_profiler): Emit direct comparison
17303 of __gcov_indirect_call_callee with NULL.
17304 (gimple_gen_time_profiler): Change probability from
17305 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
17307 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17309 * profile.c (edge_gcov_counts): Turn to pointer.
17310 (compute_branch_probabilities, compute_branch_probabilities): Update.
17311 (branch_prob): Do not clear edge_gcov_count.
17312 * profile.h (edge_gcov_counts): Turn to pointer.
17313 (edge_gcov_count): Update.
17315 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17317 * gimple.h (gimple_check_failed): Mark cold.
17319 2017-06-09 Richard Biener <rguenther@suse.de>
17321 PR tree-optimization/66623
17322 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
17323 refactor check_reduction into two parts, properly computing
17324 whether we have to check reduction validity for outer loop
17327 2017-06-09 Richard Biener <rguenther@suse.de>
17329 PR tree-optimization/79483
17330 * graphite-scop-detection.c (order): New global.
17331 (get_order): Compute bb to order mapping that satisfies code
17332 generation constraints.
17333 (cmp_pbbs): New helper.
17334 (build_scops): Start domwalk at entry block, sort generated
17337 2017-06-09 Richard Biener <rguenther@suse.de>
17339 PR middle-end/81007
17340 * ipa-polymorphic-call.c
17341 (ipa_polymorphic_call_context::restrict_to_inner_class):
17342 Skip FIELD_DECLs with error_mark_node type.
17343 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
17346 2017-06-09 Martin Liska <mliska@suse.cz>
17348 * predict.c (struct branch_predictor): New struct.
17349 (test_prediction_value_range): New test.
17350 (predict_c_tests): New function.
17351 * selftest-run-tests.c (selftest::run_tests): Run the function.
17352 * selftest.h: Declare new tests.
17354 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
17357 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
17358 gen_add3_insn did not fail.
17359 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
17360 r0, construct that number in a temporary reg and add that reg to r0.
17361 If asked to put the result in r0 as well, fail.
17363 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
17365 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17366 for early expansion of vec_eqv.
17368 2017-06-08 Jakub Jelinek <jakub@redhat.com>
17370 PR middle-end/81005
17371 * ubsan.c (instrument_null): Avoid pointless code temporary.
17372 (pass_ubsan::execute): Instrument aggregate arguments of calls.
17374 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
17378 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
17381 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
17384 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17386 * predict.c (unlikely_executed_edge_p): Move ahead.
17387 (probably_never_executed_edge_p): Use it.
17389 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17391 PR middle-end/79988
17392 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
17393 gimple_call_builtin_p call.
17395 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17397 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
17398 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
17399 rtl_check_failed_type2, rtl_check_failed_code1,
17400 rtl_check_failed_code2, rtl_check_failed_code_mode,
17401 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
17402 rtvec_check_failed_bounds, rtl_check_failed_flag,
17403 _fatal_insn_not_found, _fatal_insn): Likewise.
17404 * tree.h (tree_contains_struct_check_failed,
17405 tree_check_failed, tree_not_check_failed,
17406 tree_class_check_failed, tree_range_check_failed,
17407 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
17408 tree_vec_elt_check_failed, phi_node_elt_check_failed,
17409 tree_operand_check_failed, omp_clause_check_failed,
17410 omp_clause_operand_check_failed, omp_clause_range_check_failed):
17413 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17415 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
17416 flag_branch_probabilities.
17417 * ipa-inline.c (edge_badness): Likewise.
17418 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17419 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
17420 * predict.c (maybe_hot_frequency_p): Likewise.
17421 (probably_never_executed): Likewise.
17422 * sched-ebb.c (schedule_ebbs): Likewise.
17423 * sched-rgn.c (find_single_block_region): Likewise.
17424 * tracer.c (tail_duplicate): Likewise.
17426 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17428 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
17429 longer requires x_flag_profile_use.
17431 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17433 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
17434 instead of flag_reorder_blocks_and_partition.
17435 * dbxout.c (dbxout_function_end): Likewise.
17436 * dwarf2out.c (gen_subprogram_die): Likewise.
17437 * haifa-sched.c (sched_create_recovery_edges): Likewise.
17438 * hw-doloop.c (reorg_loops): Likewise.
17439 * varasm.c (assemble_start_function,
17440 assemble_end_function): Likewise.
17441 (decide_function_section): Do not check for
17442 flag_reorder_blocks_and_partition.
17444 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17446 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17448 (chkp_get_hard_register_fake_addr_expr): Ditto.
17449 (chkp_build_addr_expr): Add check for hard reg case.
17450 (chkp_parse_array_and_component_ref): Ditto.
17451 (chkp_find_bounds_1): Ditto.
17452 (chkp_process_stmt): Don't generate bounds store for
17455 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17457 * predict.c (maybe_hot_bb_p): Do not check profile status.
17458 (maybe_hot_edge_p): Likewise.
17459 (probably_never_executed): Check for zero counts even if profile
17461 (unlikely_executed_edge_p): New function.
17462 (unlikely_executed_stmt_p): New function.
17463 (unlikely_executed_bb_p): New function.
17464 (set_even_probabilities): Use unlikely predicates.
17465 (combine_predictions_for_bb): Likewise.
17466 (predict_paths_for_bb): Likewise.
17467 (predict_paths_leading_to_edge): Likewise.
17468 (determine_unlikely_bbs): New function.
17469 (estimate_bb_frequencies): Use it.
17470 (compute_function_frequency): Use zero counts even if profile is
17472 * profile-count.h: Fix typo.
17474 2017-08-08 Julia Koval <julia.koval@intel.com>
17476 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
17477 _mm512_mask_cvtsepi16_storeu_epi8,
17478 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
17479 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
17480 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
17481 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
17482 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
17483 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
17484 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
17485 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
17486 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
17487 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
17488 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
17489 __builtin_ia32_pmovuswb256mem_mask,
17490 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
17491 __builtin_ia32_pmovwb512mem_mask): New builtins.
17493 2017-08-08 Julia Koval <julia.koval@intel.com>
17495 PR target/73350,80862
17496 * config/i386/subst.md (round): Fix round pattern.
17497 * config/i386/i386.c (ix86_erase_embedded_rounding):
17498 Fix erasing rounding for the fixed pattern.
17500 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17502 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
17504 2017-06-08 Martin Liska <mliska@suse.cz>
17506 PR gcov-profile/80911
17507 * gcov.c (block_info::block_info): New constructor.
17509 2017-06-07 Carl Love <cel@us.ibm.com>
17511 * config/rs6000/rs6000-c: The return type of the following
17512 built-in functions was implemented as int not long long. Fix sign
17513 of return value for the unsigned version of vec_mulo and vec_mule.
17514 vector unsigned long long vec_bperm (vector unsigned long long,
17515 vector unsigned char)
17516 vector signed long long vec_mule (vector signed int,
17518 vector unsigned long long vec_mule (vector unsigned int,
17519 vector unsigned int)
17520 vector signed long long vec_mulo (vector signed int,
17522 vector unsigned long long vec_mulo (vector unsigned int,
17523 vector unsigned int)
17524 * doc/extend.texi: Fix the documentation for the built-in
17527 2017-06-07 Carl Love <cel@us.ibm.com>
17530 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
17533 2017-06-07 Carl Love <cel@us.ibm.com>
17535 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
17536 support, Generate doublehv for signed int/float for BE case only.
17538 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
17540 * doc/invoke.texi (mcx16): Rewrite.
17542 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17544 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
17545 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
17546 *mov<mode>_softfloat, and an anonymous splitter): Use
17547 nonimmediate_operand instead of rs6000_nonimmediate_operand.
17549 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17551 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
17553 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17554 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
17555 (rs6000_debug_reg_global): Adjust.
17556 (rs6000_init_hard_regno_mode_ok): Adjust.
17557 (rs6000_dbx_register_number): Adjust.
17558 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
17559 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17560 Remove SPE_ACC and SPEFSCR.
17561 (REG_ALLOC_ORDER): Ditto.
17562 (FRAME_POINTER_REGNUM): Change to 111.
17563 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
17564 (REG_CLASS_NAMES): Ditto.
17565 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
17566 (REGISTER_NAMES): Ditto.
17567 (ADDITIONAL_REG_NAMES): Ditto.
17568 (rs6000_reg_names): Ditto.
17569 * config/rs6000/rs6000.md: Renumber some register number
17572 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17574 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
17576 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17577 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
17579 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
17580 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17581 Delete the SPE high registers.
17582 (REG_ALLOC_ORDER): Ditto.
17583 (enum reg_class): Remove SPE_HIGH_REGS.
17584 (REG_CLASS_NAMES): Ditto.
17585 (REG_CLASS_CONTENTS): Delete the SPE high registers.
17586 (REGISTER_NAMES): Ditto.
17587 (rs6000_reg_names): Ditto.
17588 * doc/tm.texi.in: Remove SPE as example.
17589 * doc/tm.texi: Regenerate.
17591 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17593 * config/rs6000/8540.md (ppc8540_brinc): Delete.
17594 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
17595 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
17596 * config/rs6000/rs6000.md (type): Remove "brinc".
17598 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17600 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
17601 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
17602 * config/rs6000/linuxspe.h: Delete file.
17603 * config/rs6000/rs6000.md: Don't include spe.md.
17604 * config/rs6000/spe.h: Delete file.
17605 * config/rs6000/spe.md: Delete file.
17606 * config/rs6000/t-rs6000: Remove spe.md.
17608 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17610 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
17611 (reg_or_none500mem_operand): Delete.
17612 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
17613 instead of reg_or_none500mem_operand.
17615 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17617 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17618 handling of SPE flags.
17619 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
17621 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17623 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17625 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
17626 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
17627 paired_divv2sf3): Similar.
17628 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
17629 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
17630 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
17632 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
17633 Rename the paired_* instruction patterns.
17634 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
17636 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
17637 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
17638 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
17639 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
17640 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
17641 PAIRED_VECTOR_MODE.
17642 (struct machine_function): Delete field spe_insn_chain_scanned_p.
17643 (spe_func_has_64bit_regs_p): Delete.
17644 (spe_expand_predicate_builtin): Delete.
17645 (spe_expand_evsel_builtin): Delete.
17646 (TARGET_DWARF_REGISTER_SPAN): Do not define.
17647 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
17648 (invalid_e500_subreg): Delete.
17649 (rs6000_legitimize_address): Always force_reg op2 as well, for
17650 paired single memory accesses.
17651 (rs6000_member_type_forces_blk): Delete.
17652 (rs6000_spe_function_arg): Delete.
17653 (rs6000_expand_unop_builtin): Delete SPE handling.
17654 (rs6000_expand_binop_builtin): Ditto.
17655 (spe_expand_stv_builtin): Delete.
17656 (bdesc_2arg_spe): Delete.
17657 (spe_expand_builtin): Delete.
17658 (spe_expand_predicate_builtin): Delete.
17659 (spe_expand_evsel_builtin): Delete.
17660 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
17661 (spe_init_builtins): Delete.
17662 (spe_func_has_64bit_regs_p): Delete.
17663 (savres_routine_name): Delete "info" parameter. Adjust callers.
17664 (rs6000_emit_stack_reset): Ditto.
17665 (rs6000_dwarf_register_span): Delete.
17666 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
17667 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
17668 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
17670 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
17672 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
17673 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
17674 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
17675 mulv2sf3, divv2sf3): Delete expanders.
17677 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17679 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
17681 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17683 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
17684 * config/rs6000/rs6000.c: Ditto.
17686 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17688 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
17689 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
17690 comparison_operator.
17692 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17694 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
17695 * config/rs6000/rs6000.opt: Ditto.
17696 * config/rs6000/t-rtems: Ditto.
17698 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17700 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
17701 TARGET_E500_SINGLE by 0, simplify.
17702 * config/rs6000/rs6000.c: Ditto.
17703 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
17704 (spe_build_register_parallel): Delete.
17705 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
17706 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
17707 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
17708 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
17709 (E500_CONVERT): Delete.
17710 * config/rs6000/spe.md: Remove many patterns and all define_constants.
17712 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17714 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
17715 * config/rs6000/dfp.md: Ditto.
17716 (negdd2, *negdd2_fpr): Merge.
17717 (absdd2, *absdd2_fpr): Merge.
17718 (negtd2, *negtd2_fpr): Merge.
17719 (abstd2, *abstd2_fpr): Merge.
17720 * config/rs6000/e500.h: Delete file.
17721 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
17722 TARGET_FPRS by 1 and simplify.
17723 * config/rs6000/rs6000-c.c: Ditto.
17724 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
17725 TARGET_DF_SPE by 0.
17726 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
17728 * config/rs6000/rs6000.md: Ditto.
17729 (floatdidf2, *floatdidf2_fpr): Merge.
17730 (move_from_CR_gt_bit): Delete.
17731 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
17732 (E500_CR_IOR_COMPARE): Delete.
17733 (All patterns that require !TARGET_FPRS): Delete.
17734 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
17736 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17738 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
17740 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17742 * graphds.c (add_edge): Intitialize edge's attached data.
17743 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
17744 pointer parameter. Call pointed function on each edge during
17745 graph traversing. Skip traversing the edge when the function
17747 (graphds_dfs, graphds_scc): Ditto.
17748 (for_each_edge): New parameter. Pass the new parameter to callback
17750 * graphds.h (skip_edge_callback): New function pointer type.
17751 (graphds_dfs, graphds_scc): New function pointer parameter.
17752 (graphds_edge_callback, for_each_edge): New parameter.
17754 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17756 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
17757 out code checking if runtime alias check is possible to below ...
17758 Call the new function.
17759 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
17760 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
17762 2017-06-07 Marek Polacek <polacek@redhat.com>
17765 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
17766 TYPE_OVERFLOW_WRAPS check.
17768 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17770 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
17771 if versioning is required.
17772 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
17773 peeling with the check for versioning.
17775 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17777 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
17778 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
17779 Set true to new parameter if new ssa variable is defined.
17780 (vect_gen_vector_loop_niters): Refactor. Set range information
17781 for the new vector loop bound variable.
17782 (vect_do_peeling): Ditto.
17784 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17786 * tree-affine.c (ssa.h): Include header file.
17787 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
17788 has wrapping overflow behavior.
17790 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17792 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
17794 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17796 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
17797 (tree_to_aff_combination): ... here.
17799 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17801 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
17802 reg_pressure model function.
17803 (ivopts_global_cost_for_size): Delete.
17804 (determine_set_costs, iv_ca_recount_cost): Call new model function
17805 ivopts_estimate_reg_pressure.
17807 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17809 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
17810 expensive than udiv. Remove floating point cases from mod.
17812 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17814 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
17815 Increase idiv cost.
17817 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17819 * config/aarch64/aarch64.md
17820 (copysignsf3): Fix mask generation.
17822 2017-06-07 Jakub Jelinek <jakub@redhat.com>
17824 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
17826 (class dump_manager): Add register_dumps method.
17827 * dumpfile.c: Include langhooks.h.
17828 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
17829 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
17830 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
17831 (dump_manager::dump_register): Start with 512 entries instead of 32.
17832 (dump_manager::register_dumps): New method.
17833 * toplev.c (general_init): Instead of invoking register_dumps
17834 langhook, invoke register_dumps method on the dump manager.
17835 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
17838 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
17840 * doc/md.texi: Clarify the restrictions on a define_insn condition.
17841 Say that # requires an associated define_split to exist, and that
17842 the define_split must be suitable for use after register allocation.
17844 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17846 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
17847 (compute_outgoing_frequencies): Also initialize zero counts.
17848 (find_many_sub_basic_blocks): Do not produce uninitialized profile
17849 around loops; preserve more of profile when nothing changes.
17851 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
17853 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
17855 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
17856 * config/arm/arm-cpu-cdata.h: Regenerate.
17857 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
17858 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
17859 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
17860 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
17861 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
17863 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
17865 * config/arm/t-rmprofile: Likewise.
17866 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
17868 2017-06-06 David S. Miller <davem@davemloft.net>
17871 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
17872 blockage if function uses alloca.
17874 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17876 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
17877 New "uid" fields to hold pretty-print IDs of group and ref.
17878 Memory references are now identified as <group_id>:<ref_id>
17879 instead of using [random] addresses.
17880 (dump_mem_details): Simplify, no functional change.
17881 (dump_mem_ref): Simplify and make output more concise.
17882 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
17883 (find_or_create_group): Initialize group uid.
17884 (record_ref): Initialize ref uid. Improve debug output.
17885 (prune_group_by_reuse, should_issue_prefetch_p,)
17886 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
17887 (mark_nontemporal_store, determine_loop_nest_reuse):
17888 Improve debug output.
17890 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17892 * dbgcnt.def (prefetch): New debug counter.
17893 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
17894 (schedule_prefetches): Stop issueing prefetches if debug counter
17897 2017-06-06 Tom de Vries <tom@codesourcery.com>
17899 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
17900 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
17902 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17904 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
17905 Use aarch64_reg_or_zero predicate for operand 4.
17906 (aarch64_compare_and_swap<mode> define_insn_and_split):
17907 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
17908 (aarch64_store_exclusive<mode>): Likewise for operand 2.
17910 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17912 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
17913 (arm_compute_save_core_reg_mask): This.
17914 (thumb1_compute_save_reg_mask): Rename into ...
17915 (thumb1_compute_save_core_reg_mask): This.
17916 (arm_compute_save_reg0_reg12_mask): Adapt comment.
17917 (arm_compute_frame_layout): Likewise.
17919 2017-06-06 Richard Biener <rguenther@suse.de>
17921 PR tree-optimization/80974
17922 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
17923 keep or clear leaders SSA info.
17925 2017-06-06 Tom de Vries <tom@codesourcery.com>
17927 * config/nvptx/nvptx.c (split_mode_p): New function.
17928 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
17930 2017-06-06 Tom de Vries <tom@codesourcery.com>
17932 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
17934 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17937 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
17940 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17942 * shrink-wrap.c (handle_simple_exit): Update profile.
17943 (try_shrink_wrapping): Upate profile.
17945 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17947 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
17948 (tree_guess_outgoing_edge_probabilities): New.
17949 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
17950 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
17952 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17954 * ipa-split.c (split_function): Initialize return bb profile.
17956 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17958 * profile.c (compute_branch_probabilities): Also initialize
17959 EXIT_BLOCK profile.
17961 2017-06-06 Richard Biener <rguenther@suse.de>
17963 PR tree-optimization/80928
17964 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
17965 (vect_analyze_loop_operations): Properly guard analysis for
17967 (vect_transform_loop): Likewise.
17968 (vect_analyze_loop_2): Also reset SLP type on PHIs.
17969 (vect_model_induction_cost): Do not cost for pure SLP.
17970 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
17971 of induction in inner loop vectorization.
17972 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
17973 (vect_get_and_check_slp_defs): Handle vect_induction_def.
17974 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
17976 (vect_analyze_slp_cost_1): Cost induction.
17977 (vect_detect_hybrid_slp_stmts): Handle PHIs.
17978 (vect_get_slp_vect_defs): Likewise.
17979 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
17980 (vect_transform_stmt): Handle SLP reductions.
17981 * tree-vectorizer.h (vectorizable_induction): Adjust.
17983 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17985 * config/rs6000/rs6000.c (make_resolver_func): Update
17986 init_lowered_empty_function call.
17988 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
17990 * doc/invoke.texi: Document the -fprofile-abs-path option.
17991 * common.opt (fprofile-abs-path): New option.
17992 * gcov-io.h (gcov_write_filename): Declare.
17993 * gcov-io.c (gcov_write_filename): New function.
17994 * coverage.c (coverage_begin_function): Use gcov_write_filename.
17995 * profile.c (output_location): Likewise.
17997 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17999 * shring-wrap.c: Revert accidental commit.
18001 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
18003 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
18005 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18007 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
18009 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
18010 profile in callgraph edge.
18011 * profile-count.h (apply_probability): If THIS is 0, then result is 0
18012 (apply_scale): Likewise.
18013 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
18014 Also scale profile when inlining function with zero profile.
18015 (initialize_cfun): Update exit block profile even when it is zero.
18016 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
18017 when profile is read.
18019 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18021 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
18022 (CLONE_*): New constants to define the processors we can generate
18023 code for with the target_clone attribute.
18024 (rs6000_clone_map): New array to identify which clone processors
18025 the current program is running on.
18026 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
18027 target_clone attribute.
18028 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
18029 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
18030 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
18031 (cpu_expand_builtin): Add support for target_clone attribute.
18032 (rs6000_valid_attribute_p): Allow "default" attribute.
18033 (get_decl_name): New debug function to simplify printing the
18034 current function name in debugging statements.
18035 (rs6000_clone_priority): New functions to support the target_clone
18036 attribute, and be able to generate code to switch between ISA 2.05
18037 through ISA 3.0 (power6 through power9).
18038 (rs6000_compare_version_priority): Likewise.
18039 (rs6000_get_function_versions_dispatcher): Likewise.
18040 (make_resolver_func): Likewise.
18041 (add_condition_to_bb): Likewise.
18042 (dispatch_function_versions): Likewise.
18043 (rs6000_generate_version_dispatcher_body): Likewise.
18044 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
18045 (fusion_gpr_load_p): Fix a spacing issue.
18046 * doc/extend.texi (Common Function Attributes): Document that the
18047 PowerPC supports the target_clone attribute.
18049 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18051 * config/arm/arm.h: explain F symbol found in description of ARM
18052 register allocation in its legend.
18054 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18056 * config/mips/frame-header-opt.c: Include profile-count.h.
18057 * config/riscv/riscv.c: Include profile-count.h
18059 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18061 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
18063 (sm_set_flag_if_changed): Add bbs field.
18064 (execute_sm_if_changed_flag_set): Pass BBS.
18065 (execute_sm): Update.
18067 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18069 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
18072 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18074 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
18075 (peephole2): New peephole2 to emit the above.
18076 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
18078 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18080 * config/aarch64/aarch64.c (define_peephole2 above
18081 *sub_<shift>_<mode>): New peephole.
18083 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18085 * config/i386/i386.c (make_resolver_func): Update.
18086 * Makefile.in: Add profile-count.h and profile-count.o
18087 * auto-profile.c (afdo_indirect_call): Update to new API.
18088 (afdo_set_bb_count): Update.
18089 (afdo_propagate_edge): Update.
18090 (afdo_propagate_circuit): Update.
18091 (afdo_calculate_branch_prob): Update.
18092 (afdo_annotate_cfg): Update.
18093 * basic-block.h: Include profile-count.h
18094 (struct edge_def): Turn count to profile_count.
18095 (struct basic_block_def): Likewie.
18096 (REG_BR_PROB_BASE): Move to profile-count.h
18097 (RDIV): Move to profile-count.h
18098 * bb-reorder.c (max_entry_count): Turn to profile_count.
18099 (find_traces): Update.
18100 (rotate_loop):Update.
18101 (connect_traces):Update.
18102 (sanitize_hot_paths):Update.
18103 * bt-load.c (migrate_btr_defs): Update.
18104 * cfg.c (RDIV): Remove.
18105 (init_flow): Use alloc_block.
18106 (alloc_block): Uninitialize count.
18107 (unchecked_make_edge): Uninitialize count.
18108 (check_bb_profile): Update.
18109 (dump_edge_info): Update.
18110 (dump_bb_info): Update.
18111 (update_bb_profile_for_threading): Update.
18112 (scale_bbs_frequencies_int): Update.
18113 (scale_bbs_frequencies_gcov_type): Update.
18114 (scale_bbs_frequencies_profile_count): New.
18115 * cfg.h (update_bb_profile_for_threading): Update.
18116 (scale_bbs_frequencies_profile_count): Declare.
18117 * cfgbuild.c (compute_outgoing_frequencies): Update.
18118 (find_many_sub_basic_blocks): Update.
18119 * cfgcleanup.c (try_forward_edges): Update.
18120 (try_crossjump_to_edge): Update.
18121 * cfgexpand.c (expand_gimple_tailcall): Update.
18122 (construct_exit_block): Update.
18123 * cfghooks.c (verify_flow_info): Update.
18124 (dump_bb_for_graph): Update.
18125 (split_edge): Update.
18126 (make_forwarder_block): Update.
18127 (duplicate_block): Update.
18128 (account_profile_record): Update.
18129 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18130 (get_estimated_loop_iterations): Update.
18131 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18132 (single_likely_exit): Update.
18133 * cfgloopmanip.c (scale_loop_profile): Update.
18135 (set_zero_probability): Update.
18136 (lv_adjust_loop_entry_edge): Update.
18137 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18138 (purge_dead_edges): Update.
18139 (rtl_account_profile_record): Update.
18140 * cgraph.c (cgraph_node::create): Uninitialize count.
18141 (symbol_table::create_edge): Uninitialize count.
18142 (cgraph_update_edges_for_call_stmt_node): Update.
18143 (cgraph_edge::dump_edge_flags): Update.
18144 (cgraph_node::dump): Update.
18145 (cgraph_edge::maybe_hot_p): Update.
18146 * cgraph.h: Include profile-count.h
18147 (create_clone), create_edge, create_indirect_edge): Update.
18148 (cgraph_node): Turn count to profile_count.
18149 (cgraph_edge0: Likewise.
18150 (make_speculative, clone): Update.
18151 (create_edge): Update.
18152 (init_lowered_empty_function): Update.
18153 * cgraphclones.c (cgraph_edge::clone): Update.
18154 (duplicate_thunk_for_node): Update.
18155 (cgraph_node::create_clone): Update.
18156 * cgraphunit.c (cgraph_node::analyze): Update.
18157 (cgraph_node::expand_thunk): Update.
18158 * final.c (dump_basic_block_info): Update.
18159 * gimple-streamer-in.c (input_bb): Update.
18160 * gimple-streamer-out.c (output_bb): Update.
18161 * graphite.c (print_global_statistics): Update.
18162 (print_graphite_scop_statistics): Update.
18163 * hsa-brig.c: Include basic-block.h.
18164 * hsa-dump.c: Include basic-block.h.
18165 * hsa-gen.c (T sum_slice): Update.
18166 (convert_switch_statements):Update.
18167 * hsa-regalloc.c: Include basic-block.h.
18168 * ipa-chkp.c (chkp_produce_thunks): Update.
18169 * ipa-cp.c (struct caller_statistics): Update.
18170 (init_caller_stats): Update.
18171 (gather_caller_stats): Update.
18172 (ipcp_cloning_candidate_p): Update.
18173 (good_cloning_opportunity_p): Update.
18174 (get_info_about_necessary_edges): Update.
18175 (dump_profile_updates): Update.
18176 (update_profiling_info): Update.
18177 (update_specialized_profile): Update.
18178 (perhaps_add_new_callers): Update.
18179 (decide_about_value): Update.
18180 (ipa_cp_c_finalize): Update.
18181 * ipa-devirt.c (struct odr_type_warn_count): Update.
18182 (struct decl_warn_count): Update.
18183 (struct final_warning_record): Update.
18184 (possible_polymorphic_call_targets): Update.
18185 (ipa_devirt): Update.
18186 * ipa-fnsummary.c (redirect_to_unreachable): Update.
18187 * ipa-icf.c (sem_function::merge): Update.
18188 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
18189 * ipa-inline.c (compute_uninlined_call_time): Update.
18190 (compute_inlined_call_time): Update.
18191 (want_inline_small_function_p): Update.
18192 (want_inline_self_recursive_call_p): Update.
18193 (edge_badness): Update.
18194 (lookup_recursive_calls): Update.
18195 (recursive_inlining): Update.
18196 (inline_small_functions): Update.
18197 (dump_overall_stats): Update.
18198 (dump_inline_stats): Update.
18199 * ipa-profile.c (ipa_profile_generate_summary): Update.
18200 (ipa_propagate_frequency): Update.
18201 (ipa_profile): Update.
18202 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18203 * ipa-utils.c (ipa_merge_profiles): Update.
18204 * loop-doloop.c (doloop_modify): Update.
18205 * loop-unroll.c (report_unroll): Update.
18206 (unroll_loop_runtime_iterations): Update.
18207 * lto-cgraph.c (lto_output_edge): Update.
18208 (lto_output_node): Update.
18209 (input_node): Update.
18210 (input_edge): Update.
18211 (merge_profile_summaries): Update.
18212 * lto-streamer-in.c (input_cfg): Update.
18213 * lto-streamer-out.c (output_cfg): Update.
18214 * mcf.c (create_fixup_graph): Update.
18215 (adjust_cfg_counts): Update.
18216 (sum_edge_counts): Update.
18217 * modulo-sched.c (sms_schedule): Update.
18218 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
18219 * predict.c (maybe_hot_count_p): Update.
18220 (probably_never_executed): Update.
18221 (dump_prediction): Update.
18222 (combine_predictions_for_bb): Update.
18223 (propagate_freq): Update.
18224 (handle_missing_profiles): Update.
18225 (counts_to_freqs): Update.
18226 (rebuild_frequencies): Update.
18227 (force_edge_cold): Update.
18228 * predict.h: Include profile-count.h
18229 (maybe_hot_count_p, counts_to_freqs): UPdate.
18230 * print-rtl-function.c: Do not include cfg.h
18231 * print-rtl.c: Include basic-block.h
18232 * profile-count.c: New file.
18233 * profile-count.h: New file.
18234 * profile.c (is_edge_inconsistent): Update.
18235 (correct_negative_edge_counts): Update.
18236 (is_inconsistent): Update.
18237 (set_bb_counts): Update.
18238 (read_profile_edge_counts): Update.
18239 (compute_frequency_overlap): Update.
18240 (compute_branch_probabilities): Update; Initialize and deinitialize
18242 (branch_prob): Update.
18243 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
18244 (edge_gcov_count): New.
18245 (bb_gcov_count): New.
18246 * shrink-wrap.c (try_shrink_wrapping): Update.
18247 * tracer.c (better_p): Update.
18248 * trans-mem.c (expand_transaction): Update.
18249 (ipa_tm_insert_irr_call): Update.
18250 (ipa_tm_insert_gettmclone_call): Update.
18251 * tree-call-cdce.c: Update.
18252 * tree-cfg.c (gimple_duplicate_sese_region): Update.
18253 (gimple_duplicate_sese_tail): Update.
18254 (gimple_account_profile_record): Update.
18255 (execute_fixup_cfg): Update.
18256 * tree-inline.c (copy_bb): Update.
18257 (copy_edges_for_bb): Update.
18258 (initialize_cfun): Update.
18259 (freqs_to_counts): Update.
18260 (copy_cfg_body): Update.
18261 (expand_call_inline): Update.
18262 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18263 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
18264 (try_unroll_loop_completely): Update.
18265 (try_peel_loop): Update.
18266 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18267 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
18268 * tree-ssa-loop-split.c (connect_loops): Update.
18269 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
18270 * tree-ssa-reassoc.c (branch_fixup): Update.
18271 * tree-ssa-tail-merge.c (replace_block_by): Update.
18272 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
18273 (compute_path_counts): Update.
18274 (update_profile): Update.
18275 (recompute_probabilities): Update.
18276 (update_joiner_offpath_counts): Update.
18277 (estimated_freqs_path): Update.
18278 (freqs_to_counts_path): Update.
18279 (clear_counts_path): Update.
18280 (ssa_fix_duplicate_block_edges): Update.
18281 (duplicate_thread_path): Update.
18282 * tree-switch-conversion.c (case_bit_test_cmp): Update.
18283 (struct switch_conv_info): Update.
18284 * tree-tailcall.c (decrease_profile): Update.
18285 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18286 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18287 * value-prof.c (check_counter): Update.
18288 (gimple_divmod_fixed_value): Update.
18289 (gimple_mod_pow2): Update.
18290 (gimple_mod_subtract): Update.
18291 (gimple_ic_transform): Update.
18292 (gimple_stringop_fixed_value): Update.
18293 * value-prof.h (gimple_ic): Update.
18295 2017-06-02 Carl Love <cel@us.ibm.com>
18297 * config/rs6000/rs6000-c: Add support for built-in functions
18298 vector double vec_doublee (vector signed int);
18299 vector double vec_doublee (vector unsigned int);
18300 vector double vec_doublee (vector float);
18301 vector double vec_doubleh (vector signed int);
18302 vector double vec_doubleh (vector unsigned int);
18303 vector double vec_doubleh (vector float);
18304 vector double vec_doublel (vector signed int);
18305 vector double vec_doublel (vector unsigned int);
18306 vector double vec_doublel (vector float);
18307 vector double vec_doubleo (vector signed int);
18308 vector double vec_doubleo (vector unsigned int);
18309 vector double vec_doubleo (vector float);.
18310 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
18311 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
18313 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
18314 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
18315 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
18317 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
18318 vec_doublel, vec_doubleh.
18319 * doc/extend.texi: Update the built-in documentation file for the
18320 new built-in functions.
18322 2017-06-02 David Malcolm <dmalcolm@redhat.com>
18325 * ipa-inline-analysis.c (free_growth_caches): Set
18326 edge_removal_hook_holder to NULL after removing it.
18328 2017-06-02 Sudakshina Das <sudi.das@arm.com>
18330 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
18331 comparision with zero.
18333 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18334 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18335 for early expansion of vec_min and vec_max builtins.
18336 (builtin_function_type): Add min/max unsigned variants to those
18337 identified as having unsigned arguments.
18339 2017-06-02 Olivier Hainque <hainque@adacore.com>
18341 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
18343 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18345 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
18346 Use VALL_F16 iterator rather than VALL.
18348 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18350 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
18351 Emit CBNZ inside loop when doing a strong exchange and comparing
18352 against zero. Generate the CC flags after the loop.
18354 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
18356 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
18357 (dl_section_ref): New.
18358 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
18359 On AIX, append an expression to subtract the size of the
18360 section length to dl_section_ref.
18362 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18364 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18365 for early expansion of vector absolute builtins.
18367 2017-06-02 Richard Biener <rguenther@suse.de>
18369 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
18370 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
18372 2017-06-02 Richard Biener <rguenther@suse.de>
18374 PR tree-optimization/80948
18375 * tree-tailcall.c (find_tail_calls): Track stmts to move in
18376 stmt order as well.
18378 2017-06-02 Richard Biener <rguenther@suse.de>
18380 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
18382 * tree-vect-stmts.c (process_use): Do not mark backedge defs
18383 for inductions as relevant.
18385 2017-06-02 Richard Biener <rguenther@suse.de>
18387 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
18388 (vectorizable_induction): ... this. Remove dead code.
18390 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
18392 * builtins. (expand_builtin_alloca): Remove second parameter and
18393 infer its value from the first parameter instead.
18394 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
18396 2017-06-02 Jakub Jelinek <jakub@redhat.com>
18398 PR rtl-optimization/80903
18399 * loop-doloop.c (add_test): Unshare sequence.
18401 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
18403 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
18405 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
18407 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
18409 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
18410 xlogue_layout::get_instance, logue_layout::xlogue_layout,
18411 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
18412 (xlogue_layout::get_stub_rtx): Make static.
18413 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
18414 (xlogue_layout::compute_stub_managed_regs): Rename to...
18415 (xlogue_layout::count_stub_managed_regs): ...this.
18416 (xlogue_layout::is_stub_managed_reg): New function.
18417 (xlogue_layout::m_stub_names): Rename to...
18418 (xlogue_layout::s_stub_names): ...this, make static.
18419 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
18420 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
18421 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
18422 xlogue_layout::s_stub_names): Instantiate statics.
18423 (stub_managed_regs): Remove.
18424 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
18425 (disable_call_ms2sysv_xlogues): Rename to...
18426 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
18427 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
18429 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
18430 change after reload_completed.
18431 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
18433 (ix86_expand_prologue): Likewise.
18434 (ix86_expand_epilogue): Likewise.
18435 (ix86_expand_split_stack_prologue): Likewise.
18436 (ix86_compute_frame_layout): Remove frame parameter ...
18437 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
18438 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
18440 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
18441 (ix86_frame): Move from here ...
18442 * config/i386/i386.h (ix86_frame): ... to here.
18443 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
18444 complete ix86_frame data structure instead. Remove some_ld_name.
18446 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18448 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
18449 symbols that hold a DECL_VALUE_EXPR.
18451 2017-06-01 Martin Jambor <mjambor@suse.cz>
18453 PR tree-optimization/80898
18454 * tree-sra.c (process_subtree_disqualification): Removed.
18455 (disqualify_candidate): Do not acll
18456 process_subtree_disqualification.
18457 (subtree_mark_written_and_enqueue): New function.
18458 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
18459 RHS has been disqualified and re-queue LHS if necessary. Apart
18460 from that, ignore disqualified RHS.
18462 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18464 * config/s390/s390.c (s390_emit_epilogue): Disable early return
18465 address fetch for z10 or later.
18467 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18469 * config/arc/arc.md (tst_movb): Add guard when splitting.
18471 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18473 * config/arc/arc.c (arc_can_eliminate): Test against
18474 arc_frame_pointer_needed.
18476 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18478 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
18479 to prevent store reordering.
18480 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
18481 (type): Add block type.
18482 (stack_tie): Define special instruction to be used in
18485 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18487 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
18488 constraint. It is not valid for the pattern.
18489 (noncommutative_binary_comparison): Likewise.
18491 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18493 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
18496 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18498 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
18499 be used by the reg-alloc.
18501 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18503 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
18504 reg-alloc when having mul64 or mul32x16 instructions.
18505 (mulsidi3): Likewise.
18506 (umulsidi3): Likewise.
18507 (mulsi32x16): New pattern.
18508 (mulsi64): Likewise.
18509 (mulsidi64): Likewise.
18510 (umulsidi64): Likewise.
18511 (MUL32x16_REG): Define.
18512 (mul64_600): Use MUL32x16_REG.
18513 (mac64_600): Likewise.
18514 (umul64_600): Likewise.
18515 (umac64_600): Likewise.
18517 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18519 * config/arc/arc.md (mulsi3_700): Make it commutative.
18521 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
18523 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
18525 (*sign_extendsidi2_insn): Likewise for movstosw.
18527 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18529 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
18530 the type of the input discriminant value. Convert the
18531 discriminant value of signedness vary.
18533 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
18535 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
18536 Add to -Wall section.
18538 2017-06-01 Richard Biener <rguenther@suse.de>
18540 PR middle-end/66313
18541 * fold-const.c (fold_plusminus_mult_expr): If the factored
18542 factor may be zero use a wrapping type for the inner operation.
18543 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
18544 and handle moved defs.
18545 (process_assignment): Properly guard the unary op case. Return a
18546 tri-state indicating that moving the stmt before the call may allow
18547 to continue. Pass through to_move.
18548 (find_tail_calls): Handle moving unrelated defs before
18551 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
18554 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
18555 splitter result in the canonical way.
18557 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
18559 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
18560 also for 32bit target. Update insn attributes.
18561 (zero-extendsidi2 splitter): Allow all registers for operand 1.
18563 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
18565 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
18566 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
18567 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
18568 (_mm_maskz_min_ss): New intrinsics.
18570 2017-05-31 Martin Liska <mliska@suse.cz>
18572 * tree-vect-loop.c (vect_create_epilog_for_reduction):
18573 Change comment style to one we normally use.
18574 (vectorizable_reduction): Likewise.
18575 (vectorizable_induction): Likewise.
18576 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18577 (vectorizable_call): Likewise.
18578 (vectorizable_simd_clone_call): Likewise.
18579 (vectorizable_conversion): Likewise.
18580 (vectorizable_assignment): Likewise.
18581 (vectorizable_shift): Likewise.
18582 (vectorizable_operation): Likewise.
18583 (vectorizable_store): Likewise.
18584 (vectorizable_load): Likewise.
18585 * tree-vectorizer.h: Likewise.
18587 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
18589 * passes.c (emergency_dump_function): New.
18590 * tree-pass.h (emergency_dump_function): Declare.
18591 * plugin.c (plugins_internal_error_function): Remove.
18592 * plugin.h (plugins_internal_error_function): Remove declaration.
18593 * toplev.c (internal_error_function): New static function. Use it...
18594 (general_init): ...here.
18596 2017-05-31 Graham Markall <graham.markall@embecosm.com>
18598 * config/arc/arc.c (arc_print_operand): Handle constant operands.
18599 (arc_rtx_costs): Add costs for new patterns.
18600 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
18601 * config/arc/predicates.md: Add _1_2_3_operand predicate.
18603 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18605 * tree-ssa-strlen.c (get_next_strinfo): New function.
18606 (get_stridx_plus_constant): Use it.
18607 (zero_length_string): Likewise.
18608 (adjust_related_strinfos): Likewise.
18609 (adjust_last_stmt): Likewise.
18611 2017-05-31 Richard Biener <rguenther@suse.de>
18614 * config/i386/i386.c (ix86_expand_builtin): Remove assert
18615 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
18617 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18619 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
18620 loop_vinfo argument and use of dependence distance vectors.
18621 Check instead whether the two references differ only in their
18622 initial value and assume that they have the same alignment if the
18623 difference is a multiple of the vector alignment.
18624 (vect_analyze_data_refs_alignment): Update call accordingly.
18626 2017-05-31 Martin Liska <mliska@suse.cz>
18629 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
18631 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18633 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
18634 (create_intersect_range_checks): Move from ...
18635 * tree-data-ref.c (create_intersect_range_checks_index)
18636 (create_intersect_range_checks): ... to here.
18637 (create_runtime_alias_checks): New function factored from ...
18638 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
18639 here. Call above function.
18640 * tree-data-ref.h (create_runtime_alias_checks): New function.
18642 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18644 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
18645 segment length for dr_b and compute it in wide_int.
18647 2017-05-31 Richard Biener <rguenther@suse.de>
18649 PR tree-optimization/80906
18650 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
18651 and pass through iv_map.
18652 (copy_bb_and_scalar_dependences): Adjust.
18653 (translate_pending_phi_nodes): Likewise.
18654 (copy_loop_close_phi_args): Handle code-generating IVs instead
18657 2017-05-30 David Malcolm <dmalcolm@redhat.com>
18659 * diagnostic-color.c (color_dict): Add "type-diff".
18660 (parse_gcc_colors): Update comment.
18661 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
18662 -fdiagnostics-show-template-tree and -fno-elide-type.
18663 (GCC_COLORS): Add type-diff to example.
18665 (-fdiagnostics-show-template-tree): New.
18666 (-fno-elide-type): New.
18667 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
18668 the pp_format_decoder callback. Call any m_format_postprocessor's
18670 (pretty_printer::pretty_printer): Initialize
18671 m_format_postprocessor.
18672 (pretty_printer::~pretty_printer): Delete any
18673 m_format_postprocessor.
18674 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
18675 (class format_postprocessor): New class.
18676 (struct pretty_printer::format_decoder): Document the new parameters.
18677 (struct pretty_printer::m_format_postprocessor): New field.
18678 * tree-diagnostic.c (default_tree_printer): Update for new
18679 bool and const char ** params.
18680 * tree-diagnostic.h (default_tree_printer): Likewise.
18682 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
18684 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
18685 (lwa_operand): Delete rs6000_gen_cell_microcode test.
18686 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18687 rs6000_gen_cell_microcode code.
18688 (rs6000_final_prescan_insn): Delete.
18689 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
18690 "warn-cell-microcode" entries.
18691 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
18692 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
18693 throughout. Change cc_reg_not_micro_cr0_operand to
18694 cc_reg_not_cr0_operand throughout.
18695 (*extendhi<mode>2_noload): Delete.
18696 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
18697 (mwarn-cell-microcode): Delete.
18698 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
18699 -mgen-cell-microcode and -mwarn-cell-microcode.
18701 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
18704 * config/i386/constraints.md (Yd): New constraint.
18706 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
18707 and (?Yd, r) alternatives. Update insn attributes.
18708 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
18709 and (?*Yd, r) alternatives. Update insn attributes.
18710 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
18712 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
18714 * gimplify.c (gimplify_modify_expr): Don't create a
18715 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
18718 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
18720 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
18722 2017-05-30 Richard Biener <rguenther@suse.de>
18724 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
18725 and reduc_def fields.
18726 (STMT_VINFO_REDUC_TYPE): New define.
18727 (STMT_VINFO_REDUC_DEF): Likewise.
18728 (vect_force_simple_reduction): Adjust prototype.
18729 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
18730 (vect_is_simple_reduction): Remove check_reduction argument.
18731 (vect_force_simple_reduction): Adjust and set
18732 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18733 (vectorizable_reduction): Do not re-do reduction analysis
18734 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18735 * tree-parloops.c (gather_scalar_reductions): Adjust.
18737 2017-05-30 Richard Biener <rguenther@suse.de>
18739 PR middle-end/80901
18740 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
18743 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
18745 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
18746 Introduce unknown_misalignment parameter and remove vf.
18747 (vect_peeling_hash_get_lowest_cost):
18748 Pass unknown_misalignment parameter.
18749 (vect_enhance_data_refs_alignment):
18750 Fix unsupportable data ref treatment.
18752 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18754 * tree-vect-data-refs.c (vect_get_data_access_cost):
18755 Workaround for SLP handling.
18756 (vect_enhance_data_refs_alignment):
18757 Compute costs for doing no peeling at all, compare to the best
18758 peeling costs so far and avoid peeling if cheaper.
18760 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18762 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
18763 Return peeling info and set costs to zero for unlimited cost
18765 (vect_enhance_data_refs_alignment): Also inspect all datarefs
18766 with unknown misalignment. Compute and costs for unknown
18767 misalignment, compare them to the costs for known misalignment
18768 and choose the cheapest for peeling.
18770 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18772 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
18773 (vect_get_peeling_costs_all_drs): Create function.
18774 (vect_peeling_hash_get_lowest_cost):
18775 Use vect_get_peeling_costs_all_drs.
18776 (vect_peeling_supportable): Create function.
18777 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
18779 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18781 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
18782 DR_HAS_NEGATIVE_STEP.
18783 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
18784 (vect_enhance_data_refs_alignment): Use.
18785 (vect_duplicate_ssa_name_ptr_info): Use.
18786 * tree-vectorizer.h (dr_misalignment): Use.
18787 (known_alignment_for_access_p): Use.
18789 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18792 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
18794 (has_section_name): New function.
18796 2017-05-30 Martin Liska <mliska@suse.cz>
18799 * auto-profile.c (get_function_decl_from_block): Fix
18802 2017-05-30 Richard Biener <rguenther@suse.de>
18804 PR middle-end/80876
18805 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
18807 2017-05-30 Martin Liska <mliska@suse.cz>
18809 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
18810 * dumpfile.h (struct dump_file_info): Remove ctors.
18812 2017-05-30 Martin Liska <mliska@suse.cz>
18814 * predict.def: Fix GNU coding style.
18816 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18818 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18819 Mark 'to' argument with ATTRIBUTE_UNUSED.
18821 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18823 * config/xtensa/xtensa.c (xtensa_emit_call): Use
18824 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
18825 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
18828 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
18830 * doc/install.texi (Options specification): Restore entry of
18831 --enable-sjlj-exceptions.
18833 2017-05-27 Michael Eager <eager@eagercon.com>
18836 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
18838 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
18840 * config/microblaze/microblaze.h
18841 (FIXED_REGISTERS): Update in macro.
18842 (CALL_USED_REGISTERS): Update in macro.
18844 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
18846 * doc/install.texi: Add links to macOS binary distributions.
18848 2017-05-27 Jakub Jelinek <jakub@redhat.com>
18852 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18854 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18856 2017-05-26 Martin Liska <mliska@suse.cz>
18858 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
18860 2017-05-26 Martin Liska <mliska@suse.cz>
18862 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
18863 always leading ';; '.
18864 (dump_bb_info): Likewise.
18865 (brief_dump_cfg): Likewise.
18866 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
18867 * dumpfile.c: Remove usage of TDF_VERBOSE.
18868 * dumpfile.h (enum dump_kind): Likewise.
18869 (dump_gimple_bb_header): Do not use TDF_COMMENT.
18870 * print-tree.c (debug_verbose): Remove.
18871 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
18872 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
18873 * tree-diagnostic.c (default_tree_printer): Replace
18874 TDF_DIAGNOSTIC with TDF_SLIM.
18876 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18878 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
18879 in parameter loop, rather than loop_vinfo.
18880 (create_intersect_range_checks): Ditto.
18881 (vect_create_cond_for_alias_checks): Update call to above functions.
18883 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18885 PR tree-optimization/80815
18886 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
18887 for merging runtime alias checks. Handle negative DR_STEPs.
18889 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18891 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
18893 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
18894 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
18895 out code pruning runtime alias checks.
18896 * tree-data-ref.c (prune_runtime_alias_test_list): New function
18897 factored out from above.
18898 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18900 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18902 (prune_runtime_alias_test_list): New decalaration.
18904 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18906 * tree-vect-data-refs.c (compare_tree): Rename and move ...
18907 * tree-data-ref.c (data_ref_compare_tree): ... to here.
18908 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
18909 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
18910 (operator==, comp_dr_with_seg_len_pair): Ditto.
18911 (vect_prune_runtime_alias_test_list): Ditto.
18913 2017-05-26 Martin Liska <mliska@suse.cz>
18916 * params.def: Bound partial-inlining-entry-probability param.
18918 2017-05-26 Marek Polacek <polacek@redhat.com>
18921 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
18924 2017-05-26 Richard Biener <rguenther@suse.de>
18926 PR tree-optimization/80842
18927 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
18930 2017-05-26 Richard Biener <rguenther@suse.de>
18932 PR tree-optimization/80844
18933 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
18935 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
18937 * doc/md.texi (Machine Constraints): Update x86 family
18938 machine constraints section to match 'config/i386/constraints.md'.
18940 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
18942 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
18944 2017-05-25 Nathan Sidwell <nathan@acm.org>
18946 * doc/invoke.texi (--enable-languages): Update documentation.
18948 2017-05-25 Martin Liska <mliska@suse.cz>
18950 * dumpfile.c: Add TDF_FOLDING.
18951 * dumpfile.h (enum dump_kind): Likewise.
18952 * genmatch.c (dt_simplify::gen_1): Use it.
18954 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18956 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
18958 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18960 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18961 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
18963 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18965 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
18966 * match.pd (X == C): Rewrite it here.
18967 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
18968 with_certain_nonzero_bits2): New predicates.
18969 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
18971 2017-05-24 Nathan Sidwell <nathan@acm.org>
18973 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
18976 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
18979 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18981 * config/powerpcspe: New port. Files are copied from the rs6000
18982 port, with "rs6000" in filenames replaced by "powerpcspe".
18984 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
18986 PR rtl-optimization/80754
18987 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
18989 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
18991 * config/sparc/sparc.md (length): Return the correct value for -mflat
18992 sibcalls to match output_sibcall.
18994 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18998 * config/rs6000/rs6000.c (struct machine_function): Add new field
19000 (rs6000_get_separate_components): Init that field, use it.
19001 (rs6000_components_for_bb): Use the field.
19003 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
19005 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
19007 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
19009 PR middle-end/80823
19010 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
19012 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19015 * config/s390/s390.c (s390_check_qrst_address): Check incoming
19016 address against address_operand predicate.
19017 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
19019 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
19021 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
19022 parameters passed indirectly.
19024 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
19026 * config/i386/i386.md (*movdi_internal): Remove SSE4
19027 alternative 18 (?r, *v). Update insn attributes.
19028 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
19029 Update insn attributes.
19030 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
19031 Update insn attributes.
19032 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
19033 alternative 1 (r, v). Remove isa attribute.
19034 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
19035 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
19036 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19038 2017-05-23 Tom de Vries <tom@codesourcery.com>
19040 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
19043 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19045 * cgraphunit.c (symbol_table::process_new_functions): Update.
19046 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
19047 (inline_generate_summary): Rename to ...
19048 (ipa_fn_summary_generate): ... this one.
19049 (inline_read_summary): Rename to ...
19050 (ipa_fn_summary_read): ... this one.
19051 (inline_write_summary): Rename to ...
19052 (ipa_fn_summary_write): ... this one.
19053 (inline_free_summary): Rename to ...
19054 (ipa_free_fn_summary): ... this one.
19055 (pass_data_local_fn_summary, pass_local_fn_summary,
19056 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
19057 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
19058 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
19059 make_pass_ipa_fn_summary): New.
19060 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
19061 inline_write_summary, inline_free_summary): Remove.
19062 (ipa_free_fn_summary) : New.
19063 * ipa-inline.c (ipa_inline): Update.
19064 (pass_ipa_inline): Do not generate summaries.
19065 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
19067 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
19068 and add pass_ipa_fn_summary.
19069 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
19071 (make_pass_inline_parameters): Remove.
19073 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
19075 * omp-low.c (struct omp_context): Remove "default_kind" member.
19078 * omp-offload.c (execute_oacc_device_lower): Remove the
19079 parallelism dimensions function attributes for unparallelized
19080 OpenACC kernels constructs.
19082 2017-05-23 Martin Liska <mliska@suse.cz>
19084 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
19086 (cgraph_edge::make_speculative): Likewise.
19087 (cgraph_edge::resolve_speculation): Likewise.
19088 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19089 (cgraph_node::dump): Likewise.
19090 * cgraph.h: Likewise.
19091 * cgraphunit.c (analyze_functions): Likewise.
19092 (symbol_table::compile): Likewise.
19093 * ipa-cp.c (print_all_lattices): Likewise.
19094 (determine_versionability): Likewise.
19095 (initialize_node_lattices): Likewise.
19096 (ipcp_verify_propagated_values): Likewise.
19097 (estimate_local_effects): Likewise.
19098 (update_profiling_info): Likewise.
19099 (create_specialized_node): Likewise.
19100 (perhaps_add_new_callers): Likewise.
19101 (decide_about_value): Likewise.
19102 (decide_whether_version_node): Likewise.
19103 (identify_dead_nodes): Likewise.
19104 (ipcp_store_bits_results): Likewise.
19105 * ipa-devirt.c (dump_targets): Likewise.
19106 (ipa_devirt): Likewise.
19107 * ipa-icf.c (sem_item::dump): Likewise.
19108 (sem_function::equals): Likewise.
19109 (sem_variable::equals): Likewise.
19110 (sem_item_optimizer::read_section): Likewise.
19111 (sem_item_optimizer::execute): Likewise.
19112 (congruence_class::dump): Likewise.
19113 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19114 (dump_inline_summary): Likewise.
19115 (estimate_node_size_and_time): Likewise.
19116 (inline_analyze_function): Likewise.
19117 * ipa-inline-transform.c (inline_call): Likewise.
19118 * ipa-inline.c (report_inline_failed_reason): Likewise.
19119 (want_early_inline_function_p): Likewise.
19120 (edge_badness): Likewise.
19121 (update_edge_key): Likewise.
19122 (inline_small_functions): Likewise.
19123 * ipa-profile.c (ipa_profile): Likewise.
19124 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19125 (ipa_make_edge_direct_to_target): Likewise.
19126 (remove_described_reference): Likewise.
19127 (ipa_impossible_devirt_target): Likewise.
19128 (propagate_controlled_uses): Likewise.
19129 (ipa_print_node_params): Likewise.
19130 (ipcp_transform_function): Likewise.
19131 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19132 (propagate_pure_const): Likewise.
19133 * ipa-reference.c (generate_summary): Likewise.
19134 (read_write_all_from_decl): Likewise.
19135 (propagate): Likewise.
19136 (ipa_reference_read_optimization_summary): Likewise.
19137 * ipa-utils.c (ipa_merge_profiles): Likewise.
19138 * ipa.c (walk_polymorphic_call_targets): Likewise.
19139 (symbol_table::remove_unreachable_nodes): Likewise.
19140 (ipa_single_use): Likewise.
19141 * passes.c (execute_todo): Likewise.
19142 * predict.c (drop_profile): Likewise.
19143 * symtab.c (symtab_node::get_dump_name): New function.
19144 (symtab_node::dump_name): Likewise.
19145 (symtab_node::dump_asm_name): Likewise.
19146 (symtab_node::dump_references): Likewise.
19147 (symtab_node::dump_referring): Likewise.
19148 (symtab_node::dump_base): Likewise.
19149 (symtab_node::debug_symtab): Likewise.
19150 * tree-sra.c (convert_callers_for_node): Likewise.
19151 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19152 * value-prof.c (init_node_map): Likewise.
19154 2017-05-23 Martin Liska <mliska@suse.cz>
19156 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19157 and symtab_node::debug_symtab to symbol_table::debug.
19158 * cgraphunit.c (analyze_functions): Use the renamed function.
19159 (symbol_table::compile): Likewise.
19160 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19161 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19162 * passes.c (execute_todo): Likewise.
19163 * symtab.c (symbol_table::dump): New function.
19164 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19166 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19168 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19169 that nonconst implies exec.
19171 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19173 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19174 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19175 (inline_edge_summary_vec): Turn into ...
19176 (ipa_call_summaries): ... this one.
19177 (redirect_to_unreachable, edge_set_predicate,
19178 evaluate_properties_for_edge, inline_summary_alloc,
19179 reset_ipa_call_summary, reset_inline_summary,
19180 inline_summary_t::duplicate): Update.
19181 (inline_edge_duplication_hook): Turn to ...
19182 (ipa_call_summary_t::duplicate): ... this one.
19183 (inline_edge_removal_hook): Turn to ...
19184 (ipa_call_summary_t::remove): ... this one.
19185 (dump_inline_edge_summary): Turn to ...
19186 (dump_ipa_call_summary): ... this one.
19187 (estimate_function_body_sizes): Update.
19188 (inline_update_callee_summaries): Update.
19189 (remap_edge_change_prob): Update.
19190 (remap_edge_summaries): Update.
19191 (inline_merge_summary): Update.
19192 (do_estimate_edge_time): Update.
19193 (inline_generate_summary): Update.
19194 (inline_read_section): Update.
19195 (inline_read_summary): Update.
19196 (inline_free_summary): Update.
19197 * ipa-inline.c (can_inline_edge_p): Update.
19198 (compute_inlined_call_time): Update.
19199 (want_inline_small_function_p): Update.
19200 (edge_badness): Update.
19201 (early_inliner): Update.
19202 * ipa-inline.h (inline_edge_summary): Turn to ...
19203 (ipa_call_summary): ... this one.
19204 (ipa_call_summary_t): New class.
19205 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19206 (ipa_call_summaries): New.
19207 (inline_edge_summary): Remove.
19208 (estimate_edge_growth): Update.
19209 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19210 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19211 * ipa-split.c (execute_split_functions): Update.
19212 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19214 2017-05-23 Tom de Vries <tom@codesourcery.com>
19216 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19217 attributes): Document rdrand effective target.
19219 2017-05-23 Tom de Vries <tom@codesourcery.com>
19221 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19222 attributes): Sort alphabetically.
19224 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
19226 * config/avr/genmultilib.awk: Use gsub instead of gensub.
19228 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19231 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
19232 V2DF/V2DI splat into two separate patterns, one that handles
19233 registers, and the other that only handles memory. Drop support
19234 for splatting from a GPR on ISA 2.07 and then splitting the
19235 splat into direct move and splat.
19236 (vsx_splat_<mode>_reg): Likewise.
19237 (vsx_splat_<mode>_mem): Likewise.
19239 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
19241 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
19243 2017-05-22 Jakub Jelinek <jakub@redhat.com>
19245 PR middle-end/80809
19246 * omp-low.c (finish_taskreg_remap): New function.
19247 (finish_taskreg_scan): If unit size of ctx->record_type
19248 is non-constant, unshare the size expression and replace
19249 decls in it with possible outer var refs.
19251 PR middle-end/80809
19252 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
19253 GOVD_SHARED rather than GOVD_PRIVATE with it.
19254 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
19255 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
19257 PR middle-end/80853
19258 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
19259 as last argument to build_outer_var_ref for pointer bases of array
19260 section reductions.
19262 2017-05-19 Martin Sebor <msebor@redhat.com>
19264 * print-tree.c (print_node): Print DECL_READ_P flag.
19266 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19268 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
19269 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
19270 * cgraph.c: Likewise.
19271 * cgraphunit.c: Likewise.
19272 * gengtype.c: Likewise.
19273 * ipa-cp.c: Likewise.
19274 * ipa-devirt.c: Likewise.
19275 * ipa-icf.c: Likewise.
19276 * ipa-predicate.c: Likewise.
19277 * ipa-profile.c: Likewise.
19278 * ipa-prop.c: Likewise.
19279 * ipa-split.c: Likewise.
19281 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
19282 edge_predicate_pool, dump_inline_hints,
19283 inline_summary::account_size_time, redirect_to_unreachable,
19284 edge_set_predicate, set_hint_predicate,
19285 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
19286 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
19287 inline_summary_t::remove, remap_hint_predicate_after_duplication,
19288 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
19289 ipa_call_summary_t::remove, initialize_growth_caches,
19290 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
19291 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
19292 mark_modified, unmodified_parm_1, unmodified_parm,
19293 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
19294 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
19295 compute_bb_predicates, will_be_nonconstant_expr_predicate,
19296 will_be_nonconstant_predicate, record_modified_bb_info,
19297 get_minimal_bb, record_modified, param_change_prob,
19298 phi_result_unknown_predicate, predicate_for_phi_result,
19299 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
19300 estimate_function_body_sizes, compute_inline_parameters,
19301 compute_inline_parameters_for_curren, pass_data_inline_parameters,
19302 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
19303 inline_update_callee_summaries, remap_edge_change_prob,
19304 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
19305 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
19306 inline_analyze_function, inline_summary_t::insert,
19307 inline_generate_summary, read_ipa_call_summary, inline_read_section,
19308 inline_read_summary, write_ipa_call_summary, inline_write_summary,
19309 inline_free_summary): Move to ipa-fnsummary.h
19310 (predicate_t): Remove.
19311 * ipa-fnsummary.c: New file.
19312 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
19313 (enum inline_hints_vals, inline_hints, agg_position_info,
19314 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
19315 inline_summaries, ipa_call_summary, ipa_call_summary_t,
19316 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
19317 dump_inline_summary, dump_inline_hints, inline_generate_summary,
19318 inline_read_summary, inline_write_summary, inline_free_summary,
19319 inline_analyze_function, initialize_inline_failed,
19320 inline_merge_summary, inline_update_overall_summary,
19321 compute_inline_parameters): Move to ipa-fnsummary.h
19322 * ipa-fnsummary.h: New file.
19323 * ipa-inline-transform.h: Include ipa-inline.h.
19324 * ipa-inline.c: LIkewise.
19326 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19328 * ipa-inline.c (edge_badness): Use inlined_time instead of
19329 inline_summaries->get.
19331 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19333 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
19335 2017-05-22 Nathan Sidwell <nathan@acm.org>
19337 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
19338 (fdump-lang): Document 'raw' option.
19339 * dumpfile.h (TDI_tu): Delete.
19340 * dumpfile.c (dump_files): Remove translation-unit.
19341 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
19343 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
19345 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
19346 command option from $(AWK) call.
19347 * config/avr/genmultilib.awk: Simplify and rewrite so that it
19348 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
19349 [FORMAT]: Remove handling of variable.
19350 * config/avr/t-multilib: Regenerate.
19352 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19354 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
19356 (dump_inline_summary): Do not print self_time.
19357 (estimate_function_body_sizes): Do not set self_time.
19358 (compute_inline_parameters): Likewise.
19359 (inline_read_section, inline_write_summary): Do not stream self_time.
19360 * ipa-inline.h (inline_summary): Drop self_time.
19362 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19364 * ipa-inline-analysis.c (account_size_time): Rename to ...
19365 (inline_summary::account_size_time): ... this one.
19366 (reset_ipa_call_summary): Turn to ...
19367 (ipa_call_summary::reset): ... this one.
19368 (reset_inline_summary): Turn to ...
19369 (inline_summary::reset): ... this one.
19370 (inline_summary_t::remove): Update.
19371 (inline_summary_t::duplicate): Update.
19372 (ipa_call_summary_t::remove): Update.
19373 (dump_inline_summary): Update.
19374 (estimate_function_body_sizes): Update.
19375 (compute_inline_parameters): Update.
19376 (estimate_node_size_and_time): Update.
19377 (inline_merge_summary): Update.
19378 (inline_update_overall_summary): Update.
19379 (inline_read_section): Update.
19380 (inline_write_summary): Update.
19381 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
19382 add account_size_time and reset member functions.
19383 (ipa_call_summary): Add reset function.
19384 * ipa-predicate.h (predicate::operator &): Constify.
19386 2017-05-22 Richard Biener <rguenther@suse.de>
19388 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
19390 2017-05-19 Jason Merrill <jason@redhat.com>
19392 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
19394 2017-05-19 Marek Polacek <polacek@redhat.com>
19397 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
19398 TYPE_OVERFLOW_WRAPS checks.
19400 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
19402 * tree-core.h (enum omp_clause_default_kind): Add
19403 "OMP_CLAUSE_DEFAULT_PRESENT".
19404 * tree-pretty-print.c (dump_omp_clause): Handle it.
19405 * gimplify.c (enum gimplify_omp_var_data): Add
19406 "GOVD_MAP_FORCE_PRESENT".
19407 (gimplify_adjust_omp_clauses_1): Map it to
19408 "GOMP_MAP_FORCE_PRESENT".
19409 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
19411 * gimplify.c (oacc_default_clause): Clarify.
19413 2017-05-19 Nathan Sidwell <nathan@acm.org>
19415 LANG_HOOK_REGISTER_DUMPS
19416 * toplev.c (general_init): Call register dump lang hook.
19417 * doc/invoke.texi: Document -fdump-lang option family.
19418 * dumpfile.c (dump_files): Remove class dump here.
19419 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
19420 * dumpfile.h (tree_dump_index): Remove TDI_class.
19421 * langhooks-def.h (lhd_register_dumps): Declare.
19422 (LANG_HOOKS_REGISTER_DUMPS): Define.
19423 (LANG_HOOKS_INITIALIZER): Add it.
19424 * langhooks.c (lhd_register_dumps): Define.
19425 * langhooks.h (struct lang_hooks): Add register_dumps.
19427 2017-05-19 Nathan Sidwell <nathan@acm.org>
19429 * context.h (context::set_passes): New.
19430 * context.c (context::context): Do not create pass manager.
19431 * toplev.c (general_init): Create pass manager here.
19433 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
19435 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
19436 use this splitter if two add or or instructions would also work for
19437 the constant we want to generate.
19439 2017-05-19 Richard Biener <rguenther@suse.de>
19442 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
19443 predicate evaluation.
19445 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19447 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
19449 * ipa-inline.c (want_inline_small_function_p): Do not cast to
19452 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19454 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19455 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19456 (inline_edge_summary_vec): Turn into ...
19457 (ipa_call_summaries): ... this one.
19458 (redirect_to_unreachable, edge_set_predicate,
19459 evaluate_properties_for_edge, inline_summary_alloc,
19460 reset_ipa_call_summary, reset_inline_summary,
19461 inline_summary_t::duplicate): Update.
19462 (inline_edge_duplication_hook): Turn to ...
19463 (ipa_call_summary_t::duplicate): ... this one.
19464 (inline_edge_removal_hook): Turn to ...
19465 (ipa_call_summary_t::remove): ... this one.
19466 (dump_inline_edge_summary): Turn to ...
19467 (dump_ipa_call_summary): ... this one.
19468 (estimate_function_body_sizes): Update.
19469 (inline_update_callee_summaries): Update.
19470 (remap_edge_change_prob): Update.
19471 (remap_edge_summaries): Update.
19472 (inline_merge_summary): Update.
19473 (do_estimate_edge_time): Update.
19474 (inline_generate_summary): Update.
19475 (inline_read_section): Update.
19476 (inline_read_summary): Update.
19477 (inline_free_summary): Update.
19478 * ipa-inline.c (can_inline_edge_p): Update.
19479 (compute_inlined_call_time): Update.
19480 (want_inline_small_function_p): Update.
19481 (edge_badness): Update.
19482 (early_inliner): Update.
19483 * ipa-inline.h (inline_edge_summary): Turn to ...
19484 (ipa_call_summary): ... this one.
19485 (ipa_call_summary_t): New class.
19486 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19487 (ipa_call_summaries): New.
19488 (inline_edge_summary): Remove.
19489 (estimate_edge_growth): Update.
19490 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19491 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19492 * ipa-split.c (execute_split_functions): Update.
19493 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19495 2017-05-19 Richard Biener <rguenther@suse.de>
19497 PR middle-end/80764
19498 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
19500 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
19502 * config/rs6000/rs6000.c (struct machine_function): Add field
19503 fpr_is_wrapped_separately.
19504 (rs6000_get_separate_components): Use 64 components. Handle the
19505 new FPR components.
19506 (rs6000_components_for_bb): Handle the FPR components.
19507 (rs6000_emit_prologue_components): Handle the FPR components.
19508 (rs6000_emit_epilogue_components): Handle the FPR components.
19509 (rs6000_set_handled_components): Handle the FPR components.
19510 (rs6000_emit_prologue): Don't output prologue code for those FPRs
19511 that are already separately shrink-wrapped.
19512 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
19513 that are already separately shrink-wrapped.
19515 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19518 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
19521 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
19522 (define_peephole2 for Altivec d-form load): Add peepholes to catch
19523 cases where the register allocator uses a move and an offsettable
19524 memory operation to/from a FPR register on ISA 2.06/2.07.
19525 (define_peephole2 for Altivec d-form store): Likewise.
19527 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
19530 * config/i386/mmx.md (*mov<mode>_internal): Enable
19531 alternatives 11, 12, 13 and 14 also for 32bit targets.
19532 Remove alternatives 15, 16, 17 and 18.
19533 * config/i386/sse.md (vec_concatv2di): Change
19534 alternative (!x, *y) to (x, ?!*Yn).
19536 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
19538 * dumpfile.h (enum dump_kind): Remove stray comma.
19540 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19542 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
19543 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
19544 predicate::num_conditions
19545 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
19546 (CHANGED): turn into predicate::changed.
19547 (agg_position_info): Move to ipa-predicate.h
19548 (add_condition, predicate::add_clause, predicate::operator &=,
19549 predicate::or_with, predicate::evaluate, predicate::probability,
19550 dump_condition, dump_clause, predicate::dump,
19551 predicate::remap_after_duplication, predicate::remap_after_inlining,
19552 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
19553 (evaluate_conditions_for_known_args): Update.
19554 (set_cond_stmt_execution_predicate): Update.
19555 * ipa-inline.h: Include ipa-predicate.h
19556 (condition, inline_param_summary, conditions, agg_position_info,
19557 predicate): Move to ipa-predicate.h
19558 * ipa-predicate.c: New file.
19559 * ipa-predicate.h: New file.
19561 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
19563 * final.c (leaf_function_p): Check we are not in a sequence.
19565 2017-05-18 Martin Liska <mliska@suse.cz>
19567 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
19568 * dumpfile.c (dump_register): Use new enum dump_kind.
19569 (get_dump_file_name): Likewise.
19570 (dump_enable_all): Likewise.
19571 (dump_switch_p_1): Likewise.
19572 (enable_rtl_dump_file): Remove usage of TDF_RTL.
19573 * dumpfile.h (enum dump_kind): New enum type.
19574 (struct dump_file_info): Create constructor and
19575 format fields and comments.
19576 * passes.c (pass_manager::register_one_dump_file):
19578 * statistics.c (statistics_early_init): Likewise.
19579 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
19580 TDF_TREE with TDF_SLIM.
19581 (gather_memory_references_ref): Likewise.
19583 2017-05-18 Martin Liska <mliska@suse.cz>
19585 * vec.h (struct vnull): Use it.
19587 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19589 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
19590 (true_predicate, false_predicate, true_predicate_p,
19591 false_predicate_p): Remove.
19592 (single_cond_predicate, not_inlined_predicate): Turn to member function
19594 (add_condition): Update.
19595 (add_clause): Turn to...
19596 (predicate::add_clause): ... this one; update; allow passing NULL
19598 (and_predicates): Turn to ...
19599 (predicate::operator &=): ... this one.
19600 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
19601 (or_predicates): Turn to ...
19602 (predicate::or_with): ... this one.
19603 (evaluate_predicate): Turn to ...
19604 (predicate::evaluate): ... this one.
19605 (predicate_probability): Turn to ...
19606 (predicate::probability): ... this one.
19607 (dump_condition): Update.
19608 (dump_predicate): Turn to ...
19609 (predicate::dump): ... this one.
19610 (account_size_time): Update.
19611 (edge_set_predicate): Update.
19612 (set_hint_predicate): UPdate.
19613 (evaluate_conditions_for_known_args): Update.
19614 (evaluate_properties_for_edge): Update.
19615 (remap_predicate_after_duplication): Turn to...
19616 (predicate::remap_after_duplication): ... this one.
19617 (remap_hint_predicate_after_duplication): Update.
19618 (inline_summary_t::duplicate): UPdate.
19619 (dump_inline_edge_summary): Update.
19620 (dump_inline_summary): Update.
19621 (set_cond_stmt_execution_predicate): Update.
19622 (set_switch_stmt_execution_predicate): Update.
19623 (compute_bb_predicates): Update.
19624 (will_be_nonconstant_expr_predicate): Update.
19625 (will_be_nonconstant_predicate): Update.
19626 (phi_result_unknown_predicate): Update.
19627 (predicate_for_phi_result): Update.
19628 (array_index_predicate): Update.
19629 (estimate_function_body_sizes): Update.
19630 (estimate_node_size_and_time): Update.
19631 (estimate_ipcp_clone_size_and_time): Update.
19632 (remap_predicate): Rename to ...
19633 (predicate::remap_after_inlining): ... this one.
19634 (remap_hint_predicate): Update.
19635 (inline_merge_summary): Update.
19636 (inline_update_overall_summary): Update.
19637 (estimate_size_after_inlining): Update.
19638 (read_predicate): Rename to ...
19639 (predicate::stream_in): ... this one.
19640 (read_inline_edge_summary): Update.
19641 (write_predicate): Rename to ...
19642 (predicate::stream_out): ... this one.
19643 (write_inline_edge_summary): Update.
19644 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
19645 (clause_t): Turn to uint32_t
19646 (predicate): Turn to class; implement constructor and operators
19648 (size_time_entry): Update.
19649 (inline_summary): Update.
19650 (inline_edge_summary): Update.
19652 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
19654 * fold-const.c (fold_binary_loc): Move transformation...
19655 * match.pd (C - X CMP X): ... here.
19657 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
19659 * config/sparc/sparc.c (sparc_option_override): Set function
19660 alignment for -mcpu=niagara7 to 64 to match the I$ line.
19661 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
19663 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
19665 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
19667 2017-05-18 Marek Polacek <polacek@redhat.com>
19670 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
19671 (pass_ubsan::execute): Call gimple_assign_single_p instead of
19672 gimple_assign_load_p.
19674 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
19676 PR middle-end/80692
19677 * real.c (do_compare): Give decimal_do_compare preference over
19678 comparing just the signs.
19680 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
19682 * doc/md.texi (Canonicalization of Instructions): Describe the
19683 canonical form of instructions that inherently set a condition
19686 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
19688 PR middle-end/80775
19689 * tree-cfg.c: Move deletion of unreachable case statements to after
19690 the merging of consecutive case labels.
19692 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
19694 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
19695 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
19696 restoring of callee-saved registers.
19698 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
19700 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
19701 * config/visium/visium.c (single_set_and_flags): Likewise.
19702 * config/visium/visium.md (Substitutions): Likewise.
19704 2017-05-17 Martin Liska <mliska@suse.cz>
19706 * cfg.c: Introduce dump_flags_t type and
19707 use it instead of int type.
19709 * cfghooks.c: Likewise.
19710 * cfghooks.h (struct cfg_hooks): Likewise.
19711 * cfgrtl.c: Likewise.
19712 * cfgrtl.h: Likewise.
19713 * cgraph.c (cgraph_node::get_body): Likewise.
19714 * coretypes.h: Likewise.
19715 * domwalk.c: Likewise.
19716 * domwalk.h: Likewise.
19717 * dumpfile.c (struct dump_option_value_info): Likewise.
19718 (dump_enable_all): Likewise.
19719 (dump_switch_p_1): Likewise.
19720 (opt_info_switch_p): Likewise.
19721 * dumpfile.h (enum tree_dump_index): Likewise.
19722 (struct dump_file_info): Likewise.
19723 * genemit.c: Likewise.
19724 * generic-match-head.c: Likewise.
19725 * gengtype.c (open_base_files): Likewise.
19726 * gimple-pretty-print.c: Likewise.
19727 * gimple-pretty-print.h: Likewise.
19728 * graph.c (print_graph_cfg): Likewise.
19729 * graphite-scop-detection.c (dot_all_sese): Likewise.
19730 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
19731 * loop-unroll.c (report_unroll): Likewise.
19732 * passes.c (pass_manager::register_one_dump_file): Likewise.
19733 * print-tree.c: Likewise.
19734 * statistics.c: Likewise.
19735 * tree-cfg.c: Likewise.
19736 * tree-cfg.h: Likewise.
19737 * tree-dfa.c: Likewise.
19738 * tree-dfa.h: Likewise.
19739 * tree-dump.c (dump_function): Likewise.
19740 * tree-dump.h (struct dump_info): Likewise.
19741 * tree-pretty-print.c: Likewise.
19742 * tree-pretty-print.h: Likewise.
19743 * tree-ssa-live.c: Likewise.
19744 * tree-ssa-live.h: Likewise.
19745 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
19746 * tree-vect-loop.c: Likewise.
19747 * tree-vect-slp.c: Likewise.
19749 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
19750 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19752 PR tree-optimization/80457
19753 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
19754 of all arguments to a statement as scalar_to_vec operations.
19755 (vectorizable_call): Adjust call to vect_model_simple_cost for
19757 (vectorizable_conversion): Likewise.
19758 (vectorizable_assignment): Likewise.
19759 (vectorizable_shift): Likewise.
19760 (vectorizable_operation): Likewise.
19761 (vectorizable_comparison): Likewise.
19762 (vect_is_simple_cond): Record the def types for operands.
19763 (vectorizable_condition): Likewise, call vect_model_simple_cost.
19764 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
19765 for statement argument count.
19767 2017-05-16 Carl Love <cel@us.ibm.com>
19769 * config/rs6000/rs6000-c: Add support for built-in functions
19770 vector unsigned long long vec_bperm (vector unsigned long long,
19771 vector unsigned char)
19772 vector signed long long vec_mule (vector signed int,
19774 vector unsigned long long vec_mule (vector unsigned int,
19775 vector unsigned int)
19776 vector signed long long vec_mulo (vector signed int,
19778 vector unsigned long long vec_mulo (vector unsigned int,
19779 vector unsigned int)
19780 vector signed char vec_sldw (vector signed char,
19781 vector signed char,
19783 vector unsigned char vec_sldw (vector unsigned char,
19784 vector unsigned char,
19786 vector signed short vec_sldw (vector signed short,
19787 vector signed short,
19789 vector unsigned short vec_sldw (vector unsigned short,
19790 vector unsigned short,
19792 vector signed int vec_sldw (vector signed int,
19795 vector unsigned int vec_sldw (vector unsigned int,
19796 vector unsigned int,
19798 vector signed long long vec_sldw (vector signed long long,
19799 vector signed long long,
19801 vector unsigned long long vec_sldw (vector unsigned long long,
19802 vector unsigned long long,
19804 * config/rs6000/rs6000-c: Add support for built-in functions
19805 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
19806 * config/rs6000/altivec.h: Add defintion for vec_sldw.
19807 * doc/extend.texi: Update the built-in documentation for the
19808 new built-in functions.
19810 2017-05-16 Marek Polacek <polacek@redhat.com>
19814 * tree.c (save_expr): Don't fold the expression.
19816 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
19818 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
19819 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
19820 and (?*y,m). Update insn attributes.
19822 2017-05-16 Martin Liska <mliska@suse.cz>
19824 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
19825 flags argument of print_gimple_stmt, print_gimple_expr,
19826 print_generic_stmt and print_generic_expr.
19827 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
19828 * coretypes.h: Likewise.
19829 * except.c (dump_eh_tree): Likewise.
19830 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
19831 * gimple-pretty-print.h: Likewise.
19832 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
19833 (backprop::push_to_worklist): Likewise.
19834 (backprop::pop_from_worklist): Likewise.
19835 (backprop::process_use): Likewise.
19836 (backprop::intersect_uses): Likewise.
19837 (note_replacement): Likewise.
19838 * gimple-ssa-store-merging.c
19839 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
19840 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
19841 (pass_store_merging::execute): Likewise.
19842 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
19843 (ssa_base_cand_dump_callback): Likewise.
19844 (dump_incr_vec): Likewise.
19845 (replace_refs): Likewise.
19846 (replace_mult_candidate): Likewise.
19847 (create_add_on_incoming_edge): Likewise.
19848 (create_phi_basis): Likewise.
19849 (insert_initializers): Likewise.
19850 (all_phi_incrs_profitable): Likewise.
19851 (introduce_cast_before_cand): Likewise.
19852 (replace_one_candidate): Likewise.
19853 * gimplify.c (gimplify_expr): Likewise.
19854 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
19855 (set_rename): Likewise.
19856 (rename_uses): Likewise.
19857 (copy_loop_phi_nodes): Likewise.
19858 (add_close_phis_to_merge_points): Likewise.
19859 (copy_loop_close_phi_args): Likewise.
19860 (copy_cond_phi_args): Likewise.
19861 (graphite_copy_stmts_from_block): Likewise.
19862 (translate_pending_phi_nodes): Likewise.
19863 * graphite-poly.c (print_pdr): Likewise.
19864 (dump_gbb_cases): Likewise.
19865 (dump_gbb_conditions): Likewise.
19866 (print_scop_params): Likewise.
19867 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
19868 (build_cross_bb_scalars_use): Likewise.
19869 (gather_bbs::before_dom_children): Likewise.
19870 * hsa-dump.c (dump_hsa_immed): Likewise.
19871 * ipa-cp.c (print_ipcp_constant_value): Likewise.
19872 (get_replacement_map): Likewise.
19873 * ipa-inline-analysis.c (dump_condition): Likewise.
19874 (estimate_function_body_sizes): Likewise.
19875 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
19876 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
19877 * ipa-prop.c (ipa_dump_param): Likewise.
19878 (ipa_print_node_jump_functions_for_edge): Likewise.
19879 (ipa_modify_call_arguments): Likewise.
19880 (ipa_modify_expr): Likewise.
19881 (ipa_dump_param_adjustments): Likewise.
19882 (ipa_dump_agg_replacement_values): Likewise.
19883 (ipcp_modif_dom_walker::before_dom_children): Likewise.
19884 * ipa-pure-const.c (check_stmt): Likewise.
19885 (pass_nothrow::execute): Likewise.
19886 * ipa-split.c (execute_split_functions): Likewise.
19887 * omp-offload.c (dump_oacc_loop_part): Likewise.
19888 (dump_oacc_loop): Likewise.
19889 * trans-mem.c (tm_log_emit): Likewise.
19890 (tm_memopt_accumulate_memops): Likewise.
19891 (dump_tm_memopt_set): Likewise.
19892 (dump_tm_memopt_transform): Likewise.
19893 * tree-cfg.c (gimple_verify_flow_info): Likewise.
19894 (print_loop): Likewise.
19895 * tree-chkp-opt.c (chkp_print_addr): Likewise.
19896 (chkp_gather_checks_info): Likewise.
19897 (chkp_get_check_result): Likewise.
19898 (chkp_remove_check_if_pass): Likewise.
19899 (chkp_use_outer_bounds_if_possible): Likewise.
19900 (chkp_reduce_bounds_lifetime): Likewise.
19901 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
19902 (chkp_mark_completed_bounds): Likewise.
19903 (chkp_register_incomplete_bounds): Likewise.
19904 (chkp_mark_invalid_bounds): Likewise.
19905 (chkp_maybe_copy_and_register_bounds): Likewise.
19906 (chkp_build_returned_bound): Likewise.
19907 (chkp_get_bound_for_parm): Likewise.
19908 (chkp_build_bndldx): Likewise.
19909 (chkp_get_bounds_by_definition): Likewise.
19910 (chkp_generate_extern_var_bounds): Likewise.
19911 (chkp_get_bounds_for_decl_addr): Likewise.
19912 * tree-chrec.c (chrec_apply): Likewise.
19913 * tree-data-ref.c (dump_data_reference): Likewise.
19914 (dump_subscript): Likewise.
19915 (dump_data_dependence_relation): Likewise.
19916 (analyze_overlapping_iterations): Likewise.
19917 * tree-inline.c (expand_call_inline): Likewise.
19918 (tree_function_versioning): Likewise.
19919 * tree-into-ssa.c (dump_defs_stack): Likewise.
19920 (dump_currdefs): Likewise.
19921 (dump_names_replaced_by): Likewise.
19922 (dump_update_ssa): Likewise.
19923 (update_ssa): Likewise.
19924 * tree-object-size.c (pass_object_sizes::execute): Likewise.
19925 * tree-parloops.c (build_new_reduction): Likewise.
19926 (try_create_reduction_list): Likewise.
19927 (ref_conflicts_with_region): Likewise.
19928 (oacc_entry_exit_ok_1): Likewise.
19929 (oacc_entry_exit_single_gang): Likewise.
19930 * tree-pretty-print.h: Likewise.
19931 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
19932 (get_scalar_evolution): Likewise.
19933 (add_to_evolution): Likewise.
19934 (get_loop_exit_condition): Likewise.
19935 (analyze_evolution_in_loop): Likewise.
19936 (analyze_initial_condition): Likewise.
19937 (analyze_scalar_evolution): Likewise.
19938 (instantiate_scev): Likewise.
19939 (number_of_latch_executions): Likewise.
19940 (gather_chrec_stats): Likewise.
19941 (final_value_replacement_loop): Likewise.
19942 (scev_const_prop): Likewise.
19943 * tree-sra.c (dump_access): Likewise.
19944 (disqualify_candidate): Likewise.
19945 (create_access): Likewise.
19946 (reject): Likewise.
19947 (maybe_add_sra_candidate): Likewise.
19948 (create_access_replacement): Likewise.
19949 (analyze_access_subtree): Likewise.
19950 (analyze_all_variable_accesses): Likewise.
19951 (sra_modify_assign): Likewise.
19952 (initialize_constant_pool_replacements): Likewise.
19953 (find_param_candidates): Likewise.
19954 (decide_one_param_reduction): Likewise.
19955 (replace_removed_params_ssa_names): Likewise.
19956 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
19957 * tree-ssa-copy.c (dump_copy_of): Likewise.
19958 (copy_prop_visit_cond_stmt): Likewise.
19959 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
19960 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
19961 (record_equivalences_from_stmt): Likewise.
19962 * tree-ssa-dse.c (compute_trims): Likewise.
19963 (delete_dead_call): Likewise.
19964 (delete_dead_assignment): Likewise.
19965 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
19966 (forward_propagate_into_cond): Likewise.
19967 (pass_forwprop::execute): Likewise.
19968 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
19969 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
19971 (move_computations_worker): Likewise.
19972 (execute_sm): Likewise.
19973 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
19974 (remove_exits_and_undefined_stmts): Likewise.
19975 (remove_redundant_iv_tests): Likewise.
19976 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
19977 (adjust_iv_update_pos): Likewise.
19978 * tree-ssa-math-opts.c (bswap_replace): Likewise.
19979 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
19980 (value_replacement): Likewise.
19981 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
19982 * tree-ssa-pre.c (print_pre_expr): Likewise.
19983 (get_representative_for): Likewise.
19984 (create_expression_by_pieces): Likewise.
19985 (insert_into_preds_of_block): Likewise.
19986 (eliminate_insert): Likewise.
19987 (eliminate_dom_walker::before_dom_children): Likewise.
19988 (eliminate): Likewise.
19989 (remove_dead_inserted_code): Likewise.
19990 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
19991 * tree-ssa-reassoc.c (get_rank): Likewise.
19992 (eliminate_duplicate_pair): Likewise.
19993 (eliminate_plus_minus_pair): Likewise.
19994 (eliminate_not_pairs): Likewise.
19995 (undistribute_ops_list): Likewise.
19996 (eliminate_redundant_comparison): Likewise.
19997 (update_range_test): Likewise.
19998 (optimize_range_tests_var_bound): Likewise.
19999 (optimize_vec_cond_expr): Likewise.
20000 (rewrite_expr_tree): Likewise.
20001 (rewrite_expr_tree_parallel): Likewise.
20002 (linearize_expr): Likewise.
20003 (break_up_subtract): Likewise.
20004 (linearize_expr_tree): Likewise.
20005 (attempt_builtin_powi): Likewise.
20006 (attempt_builtin_copysign): Likewise.
20007 (transform_stmt_to_copy): Likewise.
20008 (transform_stmt_to_multiply): Likewise.
20009 (dump_ops_vector): Likewise.
20010 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
20011 (print_scc): Likewise.
20012 (set_ssa_val_to): Likewise.
20013 (visit_reference_op_store): Likewise.
20014 (visit_use): Likewise.
20015 (sccvn_dom_walker::before_dom_children): Likewise.
20016 (run_scc_vn): Likewise.
20017 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
20019 (expr_hash_elt::print): Likewise.
20020 (const_and_copies::pop_to_marker): Likewise.
20021 (const_and_copies::record_const_or_copy_raw): Likewise.
20022 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
20023 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
20024 (dump_predicates): Likewise.
20025 (find_uninit_use): Likewise.
20026 (warn_uninitialized_phi): Likewise.
20027 (pass_late_warn_uninitialized::execute): Likewise.
20028 * tree-ssa.c (verify_vssa): Likewise.
20029 (verify_ssa): Likewise.
20030 (maybe_optimize_var): Likewise.
20031 * tree-vrp.c (dump_value_range): Likewise.
20032 (dump_all_value_ranges): Likewise.
20033 (dump_asserts_for): Likewise.
20034 (register_edge_assert_for_2): Likewise.
20035 (vrp_visit_cond_stmt): Likewise.
20036 (vrp_visit_switch_stmt): Likewise.
20037 (vrp_visit_stmt): Likewise.
20038 (vrp_visit_phi_node): Likewise.
20039 (simplify_cond_using_ranges_1): Likewise.
20040 (fold_predicate_in): Likewise.
20041 (evrp_dom_walker::before_dom_children): Likewise.
20042 (evrp_dom_walker::push_value_range): Likewise.
20043 (evrp_dom_walker::pop_value_range): Likewise.
20044 (execute_early_vrp): Likewise.
20046 2017-05-16 Richard Biener <rguenther@suse.de>
20048 * dwarf2out.c (loc_list_from_tree_1): Do not create
20049 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
20051 2017-05-16 Richard Biener <rguenther@suse.de>
20053 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
20055 (note_variable_value_in_expr): If we resolved the decl ref
20056 do not push to the stack.
20058 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
20060 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
20061 operations in fast-math mode.
20062 (vaddq_f16): Likewise.
20063 (vmul_f16): Likewise.
20064 (vmulq_f16): Likewise.
20065 (vsub_f16): Likewise.
20066 (vsubq_f16): Likewise.
20067 * config/arm/neon.md (add<mode>3): New.
20069 (fma:<VH:mode>3): New. Also remove outdated comment.
20072 2017-05-16 Martin Liska <mliska@suse.cz>
20076 * ipa-devirt.c (warn_types_mismatch): Fix typo.
20077 (odr_types_equivalent_p): Likewise.
20079 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
20081 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
20083 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
20086 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
20087 non-interunit SSE move alternatives with '?'.
20088 (zero-extendsidi peephole2): New peephole to skip intermediate
20089 general register in SSE zero-extend sequence.
20091 2017-05-15 Jeff Law <law@redhat.com>
20093 * reorg.c (relax_delay_slots): Create a new variable to hold
20094 the temporary target rather than clobbering TARGET_LABEL.
20096 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20097 missing argument to extract_bit_field call.
20098 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20100 2017-05-15 Martin Liska <mliska@suse.cz>
20103 * gcc.c (process_command): Do not allow empty argument of -o option.
20105 2017-05-15 Renlin Li <renlin.li@arm.com>
20107 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20108 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20109 * config/aarch64/constraints.md (Usf): Add long call check.
20110 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20111 (call_value): Likewise.
20112 (sibcall): Likewise.
20113 (sibcall_value): Likewise.
20115 (call_value_insn): New.
20116 (sibcall_insn): Update rtx pattern.
20117 (sibcall_value_insn): Likewise.
20118 (call_internal): Remove.
20119 (call_value_internal): Likewise.
20120 (sibcall_internal): Likewise.
20121 (sibcall_value_internal): Likewise.
20122 (call_reg): Likewise.
20123 (call_symbol): Likewise.
20124 (call_value_reg): Likewise.
20125 (call_value_symbol): Likewise.
20127 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20130 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20132 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20134 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20135 compatible with CCGOCmode and with CCZmode.
20137 2017-05-14 Martin Sebor <msebor@redhat.com>
20139 PR middle-end/77671
20140 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20141 (gimple_fold_builtin_snprintf): Same.
20142 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20143 (gimple_fold_builtin_snprintf): Same.
20144 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20145 of character types.
20146 (is_call_safe): New function.
20147 (try_substitute_return_value): Call it.
20148 (try_simplify_call): New function.
20149 (pass_sprintf_length::handle_gimple_call): Call it.
20151 2017-05-14 Martin Sebor <msebor@redhat.com>
20153 PR middle-end/80669
20154 * builtins.c (expand_builtin_stpncpy): Simplify.
20156 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20158 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20159 * config/i386/i386.h
20160 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20161 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20162 (struct machine_function): Add new members call_ms2sysv,
20163 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20164 (struct machine_frame_state): New fields sp_realigned and
20165 sp_realigned_offset.
20166 * config/i386/i386.c
20167 (enum xlogue_stub): New enum.
20168 (enum xlogue_stub_sets): New enum.
20169 (class xlogue_layout): New class.
20170 (struct ix86_frame): New fields stack_realign_allocate_offset,
20171 stack_realign_offset and outlined_save_offset. Modify comments to
20172 detail stack layout when using out-of-line stubs.
20173 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20174 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
20175 -mcall-ms2sysv-xlogues.
20176 (stub_managed_regs): New static variable.
20177 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
20178 registers managed by out-of-line stub.
20179 (disable_call_ms2sysv_xlogues): New function.
20180 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
20181 m->call_ms2sysv when appropriate and compute frame layout for
20183 (sp_valid_at, fp_valid_at): New inline functions.
20184 (choose_basereg): New function.
20185 (choose_baseaddr): Add align parameter, use choose_basereg and modify
20187 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
20188 Use align parameter of choose_baseaddr to generated aligned SSE movs
20190 (pro_epilogue_adjust_stack): Modify to track
20191 machine_frame_state::sp_realigned.
20192 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
20193 (ix86_nsaved_sseregs): Likewise.
20194 (ix86_emit_save_regs): Likewise.
20195 (ix86_emit_save_regs_using_mov): Likewise.
20196 (ix86_emit_save_sse_regs_using_mov): Likewise.
20197 (get_scratch_register_on_entry): Likewise.
20198 (gen_frame_set): New function.
20199 (gen_frame_load): Likewise.
20200 (gen_frame_store): Likewise.
20201 (emit_outlined_ms2sysv_save): Likewise.
20202 (emit_outlined_ms2sysv_restore): Likewise.
20203 (ix86_expand_prologue): Modify stack re-alignment code and call
20204 emit_outlined_ms2sysv_save when appropriate.
20205 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
20206 parameter rtx_insn *insn, which allows the function to be used to only
20207 generate the notes.
20208 (ix86_expand_epilogue): Modify validity checks of frame and stack
20209 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
20210 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
20211 * config/i386/predicates.md
20212 (save_multiple): New predicate.
20213 (restore_multiple): Likewise.
20214 * config/i386/sse.md
20215 (save_multiple<mode>): New pattern.
20216 (save_multiple_realign<mode>): Likewise.
20217 (restore_multiple<mode>): Likewise.
20218 (restore_multiple_and_return<mode>): Likewise.
20219 (restore_multiple_leave_return<mode>): Likewise.
20220 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
20222 2017-05-14 Julia Koval <julia.koval@intel.com>
20224 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
20225 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
20226 (__builtin_ia32_xsetbv): New builtins.
20227 * config/i386/i386.c (ix86_expand_special_args_builtin):
20229 (ix86_expand_builtin): Special expand for new intrinsics.
20230 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
20231 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
20232 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
20234 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20236 * cfganal.c (inverted_post_order_compute): Change argument type
20238 * cfganal.h (inverted_post_order_compute): Adjust prototype.
20239 * df-core.c (rest_of_handle_df_initialize): Adjust.
20240 (rest_of_handle_df_finish): Likewise.
20241 (df_analyze_1): Likewise.
20242 (df_analyze): Likewise.
20243 (loop_inverted_post_order_compute): Change argument to be a vec *.
20244 (df_analyze_loop): Adjust.
20245 (df_get_n_blocks): Likewise.
20246 (df_get_postorder): Likewise.
20247 * df.h (struct df_d): Change field to be a vec.
20248 * lcm.c (compute_laterin): Adjust.
20249 (compute_available): Likewise.
20250 * lra-lives.c (lra_create_live_ranges_1): Likewise.
20251 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
20252 * tree-ssa-pre.c (compute_antic): Likewise.
20254 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20256 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
20257 (depth_first_search::depth_first_search): Change structure init
20258 function to this constructor.
20259 (depth_first_search::add_bb): Rename function to this member.
20260 (depth_first_search::execute): Likewise.
20261 (flow_dfs_compute_reverse_finish): Adjust.
20263 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20265 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
20266 (longest_simple_path): Likewise.
20267 * shrink-wrap.c (spread_components): Likewise.
20268 (disqualify_problematic_components): Likewise.
20269 (emit_common_heads_for_components): Likewise.
20270 (emit_common_tails_for_components): Likewise.
20271 (insert_prologue_epilogue_for_components): Likewise.
20273 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20275 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
20278 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20280 * df-core.c (df_set_blocks): Start using auto_bitmap.
20281 (df_compact_blocks): Likewise.
20282 * df-problems.c (df_rd_confluence_n): Likewise.
20283 * df-scan.c (df_insn_rescan_all): Likewise.
20284 (df_process_deferred_rescans): Likewise.
20285 (df_update_entry_block_defs): Likewise.
20286 (df_update_exit_block_uses): Likewise.
20287 (df_entry_block_bitmap_verify): Likewise.
20288 (df_exit_block_bitmap_verify): Likewise.
20289 (df_scan_verify): Likewise.
20290 * lra-constraints.c (lra_constraints): Likewise.
20291 (undo_optional_reloads): Likewise.
20292 (lra_undo_inheritance): Likewise.
20293 * lra-remat.c (calculate_gen_cands): Likewise.
20294 (do_remat): Likewise.
20295 * lra-spills.c (assign_spill_hard_regs): Likewise.
20296 (spill_pseudos): Likewise.
20297 * tree-ssa-pre.c (bitmap_set_and): Likewise.
20298 (bitmap_set_subtract_values): Likewise.
20300 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20302 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
20303 management with auto_bitmap.
20304 (fix_inter_tick): Likewise.
20305 (fix_recovery_deps): Likewise.
20306 * ira.c (add_store_equivs): Likewise.
20307 (find_moveable_pseudos): Likewise.
20308 (split_live_ranges_for_shrink_wrap): Likewise.
20309 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
20310 (rtx_reuse_manager::seen_def_p): Likewise.
20311 (rtx_reuse_manager::set_seen_def): Likewise.
20312 * print-rtl.h (class rtx_reuse_manager): Likewise.
20314 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20316 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
20318 (migrate_btr_def): Likewise.
20319 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
20320 * df-core.c (loop_post_order_compute): Likewise.
20321 (loop_inverted_post_order_compute): Likewise.
20322 * hsa-common.h: Likewise.
20323 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
20324 * init-regs.c (initialize_uninitialized_regs): Likewise.
20325 * ipa-inline.c (resolve_noninline_speculation): Likewise.
20326 (inline_small_functions): Likewise.
20327 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
20328 * ira.c (combine_and_move_insns): Likewise.
20329 (build_insn_chain): Likewise.
20330 * loop-invariant.c (find_invariants): Likewise.
20331 * lower-subreg.c (propagate_pseudo_copies): Likewise.
20332 * predict.c (tree_predict_by_opcode): Likewise.
20333 (predict_paths_leading_to): Likewise.
20334 (predict_paths_leading_to_edge): Likewise.
20335 (estimate_loops_at_level): Likewise.
20336 (estimate_loops): Likewise.
20337 * shrink-wrap.c (try_shrink_wrapping): Likewise.
20338 (spread_components): Likewise.
20339 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
20340 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
20341 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
20342 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
20343 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
20344 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
20345 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
20346 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
20347 (mark_threaded_blocks): Likewise.
20348 (thread_through_all_blocks): Likewise.
20349 * tree-ssa.c (verify_ssa): Likewise.
20350 (execute_update_addresses_taken): Likewise.
20351 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
20353 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20355 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
20357 (post_order_compute): Likewise.
20358 (inverted_post_order_compute): Likewise.
20359 (pre_and_rev_post_order_compute_fn): Likewise.
20361 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20363 * genrecog.c (int_set::int_set): Explicitly construct our
20364 auto_vec base class.
20365 * vec.h (auto_vec::auto_vec): New constructor.
20367 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20369 * bitmap.h (class auto_bitmap): New constructor taking
20370 bitmap_obstack * argument.
20372 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20374 * bitmap.h (class auto_bitmap): Change type of m_bits to
20375 bitmap_head, and adjust ctor / dtor and member operators.
20377 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
20379 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
20380 when returned register mode doesn't match original mode.
20382 2017-05-12 Jeff Law <law@redhat.com>
20383 Jakub Jelinek <jakub@redhat.com>
20385 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
20386 we look for cc setter after the compare-elim changes.
20387 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
20388 within the vector to match what compare-elim now expects.
20389 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
20390 (xorsi3_flags, one_cmplsi2_flags): Likewise.
20392 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
20393 after the compare-elim changes.
20394 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
20395 the vector to match what compare-elim now expects.
20396 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
20397 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
20398 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
20399 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
20400 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
20402 * config/visium/visium.c (single_set_and_flags): Fix where
20403 we look for cc setter after the compare-elim changes.
20404 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
20405 with the vector to match what compare-elim now expects.
20406 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
20407 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
20408 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
20409 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
20410 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
20411 (neg<mode>2_insn_set_overflow): Likewise.
20413 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
20415 PR middle-end/79794
20416 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
20417 maybe_expand_insn call, set ops[0].target. If still set after call,
20418 set alt_rtl. Add extra arg to recursive calls.
20419 (extract_bit_field): Add alt_rtl argument. Pass to
20421 * expmed.h (extract_bit_field): Fix prototype.
20422 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
20423 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
20424 to extract_bit_field_calls.
20425 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
20426 Pass alt_rtl to extract_bit_field calls.
20427 * calls.c (store_unaligned_arguments_into_psuedos)
20428 load_register_parameters): Pass extra NULL to extract_bit_field calls.
20429 * optabs.c (maybe_legitimize_operand): Clear op->target when call
20431 * optabs.h (struct expand_operand): Add target bitfield.
20433 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20435 * compare-elim.c (try_eliminate_compare): Canonicalize
20436 operation with embedded compare to
20437 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
20438 (set (reg) (operation)].
20440 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
20442 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20445 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
20446 cost of adding a carry flag for ADC instruction.
20447 [case MINUS]: Ignore the cost of subtracting a carry flag
20448 for SBB instruction.
20450 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
20452 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
20454 * config/rs6000/bmiintrin.h: New file.
20455 * config/rs6000/bmi2intrin.h: New file.
20456 * config/rs6000/x86intrin.h: New file.
20458 2017-05-12 Jeff Law <law@redhat.com>
20460 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
20463 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
20465 PR middle-end/80707
20466 * tree-cfg.c: Remove cfg edges of unreachable case statements.
20468 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20470 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20471 early expansion of vector divide builtins.
20472 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
20473 builtins identified as having unsigned arguments.
20475 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20477 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
20478 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
20479 expansion of vector logical operations (and, andc, or, xor,
20482 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20484 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
20485 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
20487 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20489 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20490 early GIMPLE expansion of vector multiplies.
20492 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20494 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
20495 TARGET_HAVE_MOVT conditional.
20496 (movt splitter): Likewise.
20498 2017-05-12 Richard Biener <rguenther@suse.de>
20500 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
20501 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20502 Fold all stmts not inplace.
20504 2017-05-12 Richard Biener <rguenther@suse.de>
20506 PR tree-optimization/80713
20507 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
20508 inserted_exprs bit for not removed stmts.
20510 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
20512 PR middle-end/69921
20513 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
20514 parallelized" attribute for parallelized OpenACC kernels.
20515 * omp-offload.c (execute_oacc_device_lower): Use it.
20517 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
20518 Set "oacc kernels" attribute.
20519 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
20520 parameter. Adjust all users.
20521 (oacc_fn_attrib_kernels_p): Remove function.
20522 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
20523 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
20524 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
20525 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
20526 assert "oacc kernels" attribute is set.
20528 2017-05-11 Carl Love <cel@us.ibm.com>
20530 * config/rs6000/rs6000-c: Add support for built-in functions
20531 vector unsigned char vec_popcnt (vector signed char)
20532 vector unsigned char vec_popcnt (vector unsigned char)
20533 vector unsigned short vec_popcnt (vector signed short)
20534 vector unsigned short vec_popcnt (vector unsigned short)
20535 vector unsigned int vec_popcnt (vector signed int)
20536 vector unsigned int vec_popcnt (vector unsigned int)
20537 vector unsigned long long vec_popcnt (vector signed long long)
20538 vector unsigned long long vec_popcnt (vector unsigned long long)
20539 vector signed long long vec_slo (vector signed long long,
20540 vector signed char)
20541 vector signed long long vec_slo (vector signed long long,
20542 vector unsigned char)
20543 vector unsigned long long vec_slo (vector unsigned long long,
20544 vector signed char)
20545 vector unsigned long long vec_slo (vector unsigned long long,
20546 vector unsigned char)
20547 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
20548 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
20549 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
20550 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
20551 * doc/extend.texi: Update the built-in documentation file for the
20552 new built-in functions.
20554 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20556 * attribs.h (sorted_attr_string): Move machine independent
20557 functions for target clone support from the i386 port to common
20558 code. Rename ix86_function_versions to common_function_versions.
20559 Rename make_name to make_unique_name.
20560 (common_function_versions): Likewise.
20561 (make_unique_name): Likewise.
20562 (make_dispatcher_decl): Likewise.
20563 (is_function_default_version): Likewise.
20564 * attribs.c (attr_strcmp): Likewise.
20565 (sorted_attr_string): Likewise.
20566 (common_function_versions): Likewise.
20567 (make_unique_name): Likewise.
20568 (make_dispatcher_decl): Likewise.
20569 (is_function_default_version): Likewise.
20570 * config/i386/i386.c (attr_strcmp): Likewise.
20571 (sorted_attr_string): Likewise.
20572 (ix86_function_versions): Likewise.
20573 (make_name): Likewise.
20574 (make_dispatcher_decl): Likewise.
20575 (is_function_default_version): Likewise.
20576 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
20578 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20581 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20582 Account for direct move costs for vec_construct of integer
20585 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
20588 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
20589 (UNSPEC_STX_ATOMIC): Ditto.
20590 (loaddi_via_sse): New insn.
20591 (storedi_via_sse): Ditto.
20592 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
20593 Update corresponding peephole2 patterns.
20594 (atomic_storedi_fpu): Ditto.
20596 2017-05-11 Julia Koval <julia.koval@intel.com>
20598 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
20599 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
20601 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
20602 (__builtin_ia32_rsqrt14ss_mask): New builtins.
20603 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
20605 2017-05-11 Nathan Sidwell <nathan@acm.org>
20607 * graphite-poly.c: Include dumpfile.h.
20609 * dumpfle.h (dump_function): Declare here ...
20610 * tree-dump.h (dump_function): ... not here.
20611 * dumpfile.c: #include tree-cfg.h.
20612 (dump_function): Move here from ...
20613 * tree-dump.c (dump_function): ... here.
20614 * gimplify.c: #include splay-tree.h, not tree-dump.h.
20615 * graphite-poly.c: Don't include tree-dump.h.
20616 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
20617 * print-tree.c: Likewise.
20618 * stor-layout.c: Likewise.
20619 * tree-nested.c: Likewise.
20621 * dumpfile.c (dump_start): Use TDF_FLAGS.
20622 (dump_enable_all): Fix TDF_KIND check thinko.
20624 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20626 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20627 array entries to represent two legal parameterizations of the
20628 overloaded __builtin_cmpb function, as represented by the
20629 P6_OV_BUILTIN_CMPB constant.
20630 (altivec_resolve_overloaded_builtin): Add special case handling
20631 for the __builtin_cmpb function, as represented by the
20632 P6_OV_BUILTIN_CMPB constant.
20633 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
20634 (BU_P6_64BIT_2): New macro.
20635 (BU_P6_OVERLOAD_2): New macro
20636 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
20637 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
20638 (CMPB): Add overload support to represent both 32-bit and 64-bit
20639 compare-bytes function.
20640 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
20641 support for TARGET_CMPB.
20642 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
20643 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20644 documentation of the __builtin_cmpb overloaded built-in function.
20646 2017-05-11 Richard Biener <rguenther@suse.de>
20648 PR tree-optimization/80705
20649 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
20650 bases are not vectorizable.
20652 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20654 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
20655 when counting register pressure.
20657 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20659 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
20660 (iv_ca_more_deps): Renamed to ...
20661 (iv_ca_compare_deps): ... this.
20662 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
20664 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20666 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
20668 (determine_group_iv_costs): ... here.
20669 (find_inv_vars_cb): Record inv var if it's not recorded before.
20671 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20673 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
20674 (get_shiftadd_cost): Ditto.
20676 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20678 * tree-ssa-address.c: Include header file.
20679 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
20681 (add_to_parts): Refactor.
20682 (addr_to_parts): New parameter. Update use of move_hint_to_base.
20683 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
20686 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20688 PR tree-optimization/53090
20689 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
20691 (extract_cond_operands): Detect condition with IV on both sides
20692 and return COMP_IV_EXPR_2.
20693 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
20694 (rewrite_use_compare): Simplify by removing call to function
20695 extract_cond_operands.
20697 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20699 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
20700 (extract_cond_operands): Detect condition comparing against non-
20701 invariant bound and return appropriate enum value.
20702 (find_interesting_uses_cond): Update use of extract_cond_operands.
20703 Handle its return value accordingly.
20704 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
20706 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20708 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
20709 nonlinear iv_use computation in loop invariant sensitive way.
20711 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20713 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
20714 (find_iv_candidates): Call relate_compare_use_with_all_cands.
20716 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20718 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
20719 (dump_cand): Support iv_cand.inv_exprs.
20720 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
20722 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
20725 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20727 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
20729 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
20730 as local function. Include necessary header files.
20731 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
20733 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20735 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
20737 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20739 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
20740 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
20741 RSHIFT_EXPR and BIT_NOT_EXPR.
20743 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20745 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
20746 (adjust_setup_cost): New parameter supporting round up adjustment.
20747 (struct address_cost_data): Delete.
20748 (force_expr_to_var_cost): Don't bound cost with spill_cost.
20749 (split_address_cost, ptr_difference_cost): Delete.
20750 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
20751 (struct ainc_cost_data): New struct.
20752 (get_address_cost_ainc): New function.
20753 (get_address_cost, get_computation_cost): Reimplement.
20754 (determine_group_iv_cost_address): Record inv_expr for all uses of
20756 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
20757 (iv_ca_has_deps): Reimplemented to ...
20758 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
20760 (iv_ca_extend): Call iv_ca_more_deps.
20762 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20764 * tree-ssa-address.c (struct mem_address): Move to header file.
20765 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
20766 * tree-ssa-address.h (struct mem_address): Move from C file.
20767 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
20769 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20771 * tree-affine.h (aff_combination_type): New interface.
20772 (aff_combination_zero_p): Remove static.
20773 (aff_combination_const_p): New interface.
20774 (aff_combination_singleton_var_p): New interfaces.
20776 2017-05-11 Richard Biener <rguenther@suse.de>
20778 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20779 Skip unreachable blocks and destinations.
20780 (eliminate): Move stmt removal and fixup ...
20781 (fini_eliminate): ... here. Skip inserted exprs.
20782 (pass_pre::execute): Move fini_pre after fini_eliminate.
20783 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
20784 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
20785 PRE to get rid of dead code that has invalid SSA form and
20786 split critical edges again.
20788 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20790 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20792 2017-05-11 Richard Biener <rguenther@suse.de>
20794 * passes.c (execute_function_todo): Verify loops if they are
20795 said to be up-to-date.
20796 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
20797 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
20799 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
20802 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
20803 handle calling assemble_external ourself.
20806 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
20807 modes with zero size. Enhance comment.
20809 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20811 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
20812 built-ins for vec_xl and vec_xst with short and char pointer
20815 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20817 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
20818 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
20819 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
20820 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
20821 (_mm_maskz_min_round_ss): New intrinsics.
20822 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
20823 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
20824 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
20825 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
20826 (__builtin_ia32_minss_mask_round): New builtins.
20827 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20828 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20829 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
20831 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
20832 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
20834 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
20837 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20839 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
20840 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
20841 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
20842 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
20843 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
20844 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
20845 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
20846 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20847 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20848 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
20849 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
20850 (__builtin_ia32_mulss_mask_round): New builtins.
20851 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20852 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20853 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
20855 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
20856 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20858 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20861 2017-05-10 Julia Koval <julia.koval@intel.com>
20863 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
20864 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
20865 (_mm256_setr_m128i): New intrinsics.
20867 2017-05-10 Julia Koval <julia.koval@intel.com>
20869 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
20870 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
20871 (_mm_maskz_rcp14_ss): New intrinsics.
20872 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
20873 (__builtin_ia32_rcp14ss_mask): New builtins.
20874 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
20876 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
20878 PR tree-optimization/51513
20879 * tree-cfg.c (gimple_seq_unreachable_p): New function.
20880 (assert_unreachable_fallthru_edge_p): Use it.
20881 (group_case_labels_stmt): Likewise.
20882 * tree-cfg.h: Prototype it.
20883 * stmt.c: Include cfghooks.h and tree-cfg.h.
20884 (emit_case_dispatch_table) <gap_label>: New local variable.
20885 Use it to fill dispatch table gaps.
20886 Test for default_label before updating probabilities.
20887 (expand_case) <default_label>: Remove unneeded initialization.
20888 Test for unreachable default case statement and remove its edge.
20889 Set default_label accordingly.
20890 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
20892 2017-05-10 Carl Love <cel@us.ibm.com>
20894 * config/rs6000/rs6000-c: Add support for built-in functions
20895 vector signed char vec_neg (vector signed char)
20896 vector signed short int vec_neg (vector short int)
20897 vector signed int vec_neg (vector signed int)
20898 vector signed long long vec_neg (vector signed long long)
20899 vector float vec_neg (vector float)
20900 vector double vec_neg (vector double)
20901 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
20903 * config/rs6000/altivec.h: Add define for vec_neg
20904 * doc/extend.texi: Update the built-in documentation for the
20905 new built-in functions.
20907 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20909 PR tree-optimization/77644
20910 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
20912 2017-05-10 Nathan Sidwell <nathan@acm.org>
20914 * dumpfile.h (TDI_lang_all): New.
20915 (TDF_KIND): New. Renumber others
20916 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
20918 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
20920 (get_dump_file_name): Adjust suffix generation.
20921 (dump_enable_all): Use TDF_KIND.
20922 * doc/invoke.texi (-fdump-lang-all): Document.
20924 * dumpfile.h: Tabify.
20926 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
20929 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
20930 Move member access before delete.
20932 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
20934 * tree-inline.c (expand_call_inline): Split block at stmt
20937 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
20940 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
20941 are now unused after splitting mov{sf,sd}_hardfloat.
20942 (f32_lr2): Likewise.
20943 (f32_lm): Likewise.
20944 (f32_lm2): Likewise.
20945 (f32_li): Likewise.
20946 (f32_li2): Likewise.
20947 (f32_lv): Likewise.
20948 (f32_sr): Likewise.
20949 (f32_sr2): Likewise.
20950 (f32_sm): Likewise.
20951 (f32_sm2): Likewise.
20952 (f32_si): Likewise.
20953 (f32_si2): Likewise.
20954 (f32_sv): Likewise.
20955 (f32_dm): Likewise.
20956 (f32_vsx): Likewise.
20957 (f32_av): Likewise.
20958 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
20959 For movsf, order stores so the VSX stores occur before the GPR
20960 store which encourages the register allocator to use a traditional
20961 FPR instead of a GPR. For movsd, order the stores so that the GPR
20962 store comes before the VSX stores to allow the power6 to work.
20963 This is due to the power6 not having a 32-bit integer store
20964 instruction from a FPR.
20965 (movsf_hardfloat): Likewise.
20966 (movsd_hardfloat): Likewise.
20968 2017-05-09 Martin Sebor <msebor@redhat.com>
20970 PR translation/80280
20971 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
20974 PR translation/80280
20975 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
20976 data member added in r247778.
20977 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
20979 2017-05-09 Nathan Sidwell <nathan@acm.org>
20981 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
20983 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
20986 2017-05-09 Marek Polacek <polacek@redhat.com>
20988 * doc/invoke.texi: Fix typo.
20990 2017-05-09 Richard Biener <rguenther@suse.de>
20992 * tree-vrp.c (vrp_val_is_max): Adjust comment.
20993 (vrp_val_is_min): Likewise.
20994 (set_value_range_to_value): Likewise.
20995 (set_value_range_to_nonnegative): Likewise.
20996 (gimple_assign_nonzero_p): Likewise.
20997 (gimple_stmt_nonzero_p): Likewise.
20998 (vrp_int_const_binop): Likewise. Remove unreachable case.
20999 (adjust_range_with_scev): Adjust comments.
21000 (compare_range_with_value): Likewise.
21001 (extract_range_from_phi_node): Likewise.
21002 (test_for_singularity): Likewise.
21004 2017-05-09 Richard Biener <rguenther@suse.de>
21006 * tree-vrp.c (get_single_symbol): Add assert that we don't
21007 get overflowed constants as invariant part.
21008 (compare_values_warnv): Add comment before the TREE_NO_WARNING
21009 checks. Use wi::cmp instead of recursing for integer constants.
21010 (compare_values): Just ignore whether we assumed undefined
21011 overflow instead of failing the compare.
21012 (extract_range_for_var_from_comparison_expr): Add comment before the
21013 TREE_NO_WARNING sets.
21014 (test_for_singularity): Likewise.
21015 (extract_range_from_comparison): Do not disable optimization
21016 when we assumed undefined overflow.
21017 (extract_range_basic): Remove init of unused var.
21019 2017-05-09 Richard Biener <rguenther@suse.de>
21021 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
21022 (extract_range_from_multiplicative_op_1): Adjust.
21023 (extract_range_from_binary_expr_1): Use int_const_binop.
21025 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21028 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
21029 rs6000_store_data_bypass_p in seven define_bypass directives and
21030 in several comments.
21031 * config/rs6000/rs6000-protos.h: Add prototype for
21032 rs6000_store_data_bypass_p function.
21033 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
21034 function implements slightly different (rs6000-specific) semantics
21035 than store_data_bypass_p, returning false rather than aborting
21036 with assertion error when arguments do not satisfy the
21037 requirements of store data bypass.
21038 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
21039 rs6000_store_data_bypass_p.
21041 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
21043 * config/xtensa/xtensa-protos.h
21044 (xtensa_initial_elimination_offset): New declaration.
21045 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21046 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
21047 macro definition, add case for FRAME_POINTER_REGNUM when
21048 FRAME_GROWS_DOWNWARD.
21049 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
21050 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
21051 xtensa_initial_elimination_offset.
21053 2017-05-08 Nathan Sidwell <nathan@acm.org>
21055 * doc/invoke.texi: Alphabetize -fdump options.
21057 2017-05-08 Martin Sebor <msebor@redhat.com>
21059 PR translation/80280
21060 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
21062 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
21064 * target.def (compute_frame_layout): New optional target hook.
21065 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
21066 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
21067 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
21069 * reload1.c (verify_initial_elim_offsets): Likewise.
21070 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
21071 (use_simple_return_p): Call arm_compute_frame_layout if needed.
21072 (arm_get_frame_offsets): Split up into this ...
21073 (arm_compute_frame_layout): ... and this function.
21075 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
21077 * config/aarch64/constraints.md (Usa): New constraint.
21078 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
21080 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
21082 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
21083 with_multilib_list after it has been checked.
21085 2017-05-08 Richard Biener <rguenther@suse.de>
21087 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
21088 (bitmap_set_subtract_values): Likewise.
21090 2017-05-08 Richard Biener <rguenther@suse.de>
21092 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
21093 (gimple_assign_nonzero): ... this and remove strict_overflow_p
21095 (gimple_stmt_nonzero_warnv_p): Rename to ...
21096 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21098 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21099 (extract_range_basic): Adjust, do not disable propagation on
21100 strict overflow sensitive simplification.
21101 (vrp_visit_cond_stmt): Likewise.
21103 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21105 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21106 body size unconditionally.
21108 2017-05-07 Jeff Law <law@redhat.com>
21111 2017-05-06 Jeff Law <law@redhat.com>
21112 PR tree-optimization/78496
21113 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21116 PR tree-optimization/78496
21117 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21118 (simplify_stmt_using_ranges): Call it.
21119 (vrp_dom_walker::before_dom_children): Extract equivalences
21120 from an ASSERT_EXPR with an equality comparison against a
21123 2017-05-06 Jeff Law <law@redhat.com>
21125 PR tree-optimization/78496
21126 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21129 PR tree-optimization/78496
21130 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21131 (simplify_stmt_using_ranges): Call it.
21132 (vrp_dom_walker::before_dom_children): Extract equivalences
21133 from an ASSERT_EXPR with an equality comparison against a
21136 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21138 * lra-constraints.c (lra_copy_reg_equiv): New function.
21139 (split_reg): Use it to copy equivalence information from the
21140 original register to the spill register.
21142 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21144 PR rtl-optimization/75964
21145 * simplify-rtx.c (simplify_const_relational_operation): Remove
21146 invalid handling of comparisons of integer ABS.
21148 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21150 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21151 initialize to zero.
21152 (init_regs): Remove declaration.
21153 (function_arg_advance_32): Initialize error_p as boolean variable.
21155 2017-05-05 Nathan Sidwell <nathan@acm.org>
21157 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21158 lines. Use for (;;).
21160 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21162 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21163 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21164 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21165 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21166 VF=2 that require versioning.
21168 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21170 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21173 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21175 * diagnostic.h (diagnostic_override_option_index): Convert from
21176 macro to inline function.
21178 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21180 * diagnostic.c (last_module_changed_p): New function.
21181 (set_last_module): New function.
21182 (diagnostic_report_current_module): Convert macro usage to
21183 the above functions.
21184 * diagnostic.h (diagnostic_context::last_module): Strengthen
21185 from const line_map * to const line_map_ordinary *.
21186 (diagnostic_last_module_changed): Delete macro.
21187 (diagnostic_set_last_module): Delete macro.
21189 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21191 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
21192 with diagnostic_report_diagnostic.
21193 (diagnostic_n_impl_richloc): Likewise.
21194 * diagnostic.h (report_diagnostic): Delete macro.
21195 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
21196 with diagnostic_report_diagnostic.
21197 * substring-locations.c (format_warning_va): Likewise.
21199 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21201 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
21202 save/restor of format_spec. Move option-printing code to...
21203 (print_option_information): ...this new function, and
21204 reimplement by simply printing to the pretty_printer,
21205 rather than appending to the format string.
21207 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21209 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
21210 handling logic into...
21211 (update_effective_level_from_pragmas): ...this new function.
21213 2017-05-04 Andrew Waterman <andrew@sifive.com>
21215 * config/riscv/riscv.opt (mstrict-align): New option.
21216 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
21217 (SLOW_UNALIGNED_ACCESS): Define.
21218 (riscv_slow_unaligned_access): Declare.
21219 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
21221 (riscv_slow_unaligned_access): New variable.
21222 (rocket_tune_info): Set slow_unaligned_access to true.
21223 (optimize_size_tune_info): Set slow_unaligned_access to false.
21224 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
21225 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
21226 (riscv_option_override): Set riscv_slow_unaligned_access.
21227 * doc/invoke.texi: Add -mstrict-align to RISC-V.
21229 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
21231 * config/riscv/riscv.md: Unify indentation.
21233 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21238 * config/rs6000/rs6000.md (u code attribute): Add FIX and
21240 (extendsi<mode>2): Add support for doing sign extension via
21241 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
21242 don't have ISA 3.0 instructions.
21243 (extendsi<mode>2 splitter): Likewise.
21244 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
21245 generate the normal insns since SImode can now go in vector
21246 registers. Disallow the special UNSPECs needed for previous
21247 machines to hide SImode being used. Add new insns
21248 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
21249 (fix_trunc<mode>si2_stfiwx): Likewise.
21250 (fix_trunc<mode>si2_internal): Likewise.
21251 (fixuns_trunc<mode>si2): Likewise.
21252 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21253 (fctiw<u>z_<mode>_smallint): Likewise.
21254 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
21255 of floating point to 32-bit integer from doing a direct move to
21256 the GPR registers to do a store.
21257 (fctiwz_<mode>): Break long line.
21259 2017-05-05 Bin Cheng <bin.cheng@arm.com>
21261 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
21262 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
21263 (addr_list, addr_offset_valid_p): New.
21264 (split_address_groups): Check offset validity with above function.
21265 (gt-tree-ssa-loop-ivopts.h): Include header file.
21267 2017-05-05 Nathan Sidwell <nathan@acm.org>
21269 * config.gcc (arm*-*-*): Add missing 'fi'.
21271 2017-05-05 Steve Ellcey <sellcey@cavium.com>
21273 * doc/invoke.texi (-fopt-info): Explicitly say order of options
21274 included in -fopt-info does not matter.
21275 * doc/optinfo.texi (-fopt-info): Fix description of default
21276 behavour. Explicitly say order of options included in -fopt-info
21279 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21281 * config.gcc: Allow combinations of aprofile and rmprofile values for
21282 --with-multilib-list.
21283 * config/arm/t-multilib: New file.
21284 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
21285 variables. Remove setting of ISA and floating-point ABI in
21286 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
21287 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
21288 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
21289 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
21291 * config/arm/t-rmprofile: Likewise except for the matches changes.
21292 * doc/install.texi (--with-multilib-list): Document the combination of
21293 aprofile and rmprofile values and warn about pitfalls in doing that.
21295 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21297 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
21298 (movdi_aarch64): Likewise.
21300 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21302 PR tree-optimization/80632
21303 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
21305 (build_arrays): Initialize it for virtual phis.
21306 (fix_phi_nodes): Use it for virtual phis.
21308 PR tree-optimization/80558
21309 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
21310 [x, y] op z into [x op, y op z] for op & or | if conditions
21313 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
21314 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21317 * config/arm/arm.md (use_literal_pool): Remove.
21318 (64-bit immediate split): No longer takes cost into consideration
21319 if arm_disable_literal_pool is enabled.
21320 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
21321 used when arm_disable_literal_pool is enabled.
21322 (arm_max_const_double_inline_cost): Remove use of
21323 arm_disable_literal_pool.
21324 (push_minipool_fix): Add assert.
21325 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
21326 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
21327 (no_literal_pool_sf_immediate): New.
21329 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21331 PR tree-optimization/80613
21332 * tree-ssa-dce.c (propagate_necessity): Remove cases for
21333 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
21335 2017-05-05 Richard Biener <rguenther@suse.de>
21337 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
21339 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
21341 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
21342 of this flag from insn conditions due to removal from r247495.
21344 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21346 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
21348 (arm_early_store_addr_dep_ptr): Likewise.
21349 * config/arm/aarch-common-protos.h
21350 (arm_early_load_addr_dep_ptr): Add prototype.
21351 (arm_early_store_addr_dep_ptr): Likewise.
21352 * config/arm/cortex-a53.md: Add new bypasses.
21354 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21356 * tree.c (next_type_uid): Change type to unsigned.
21357 (type_hash_canon): Decrement back next_type_uid if
21358 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
21359 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
21362 2017-05-04 Martin Sebor <msebor@redhat.com>
21364 * builtins.c: Fix a trivial typo in a comment.
21366 PR middle-end/79234
21367 * builtins.c (check_sizes): Adjust to handle reading past the end.
21368 Avoid printing excessive upper bound of ranges. Use %E to print
21369 tree nodes instead of converting them to %wu.
21370 (expand_builtin_memchr): New function.
21371 (compute_dest_size): Rename...
21372 (compute_objsize): ...to this.
21373 (expand_builtin_memcpy): Adjust.
21374 (expand_builtin_mempcpy): Adjust.
21375 (expand_builtin_strcat): Adjust.
21376 (expand_builtin_strcpy): Adjust.
21377 (check_strncat_sizes): Adjust.
21378 (expand_builtin_strncat): Adjust.
21379 (expand_builtin_strncpy): Adjust and simplify.
21380 (expand_builtin_memset): Adjust.
21381 (expand_builtin_bzero): Adjust.
21382 (expand_builtin_memcmp): Adjust.
21383 (expand_builtin): Handle memcmp.
21384 (maybe_emit_chk_warning): Check strncat just once.
21386 2017-05-04 Martin Sebor <msebor@redhat.com>
21388 PR preprocessor/79214
21389 PR middle-end/79222
21390 PR middle-end/79223
21391 * builtins.c (check_sizes): Add inlining context and issue
21392 warnings even when -Wno-system-headers is set.
21393 (check_strncat_sizes): Same.
21394 (expand_builtin_strncat): Same.
21395 (expand_builtin_memmove): New function.
21396 (expand_builtin_stpncpy): Same.
21397 (expand_builtin): Handle memmove and stpncpy.
21399 2017-05-04 Bin Cheng <bin.cheng@arm.com>
21401 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
21402 which is not used any more.
21404 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21406 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
21408 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21410 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
21411 (cortexa53_tunings): Likewise.
21412 (cortexa57_tunings): Likewise.
21413 (cortexa72_tunings): Likewise.
21414 (cortexa73_tunings): Likewise.
21416 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21418 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
21419 Set loop alignment to 8.
21421 2017-05-04 Martin Sebor <msebor@redhat.com>
21423 PR translation/80280
21424 * builtins.c (expand_builtin_object_size): Add missing quoting to
21425 %D and like directives.
21426 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
21427 (hsa_type_for_tree_type): Same.
21428 (verify_function_arguments): Same.
21429 * symtab.c (symbol_table::change_decl_assembler_name): Same.
21430 * varasm.c (get_section): Same.
21433 2017-05-04 Martin Sebor <msebor@redhat.com>
21435 PR translation/80280
21436 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
21438 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21440 * config/aarch64/aarch64.c (generic_addrcost_table):
21441 Change HI/TI mode setting.
21443 2017-05-04 Martin Jambor <mjambor@suse.cz>
21445 PR tree-optimization/80622
21446 * tree-sra.c (comes_initialized_p): New function.
21447 (build_accesses_from_assign): Only set write lazily when
21448 comes_initialized_p is false.
21449 (analyze_access_subtree): Use comes_initialized_p.
21450 (propagate_subaccesses_across_link): Assert !comes_initialized_p
21451 instead of testing for PARM_DECL.
21453 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21455 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
21456 constraint on operand 0 to allow more general addressing modes.
21457 Adjust output template.
21458 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
21460 * config/aarch64/aarch64-protos.h
21461 (aarch64_address_valid_for_prefetch_p): Declare prototype.
21462 * config/aarch64/constraints.md (Dp): New address constraint.
21463 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
21466 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
21468 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
21469 update use of estimate_ipcp_clone_size_and_time.
21470 (estimate_local_effects): Update use of
21471 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
21472 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
21473 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
21474 Return nonspecialized time.
21476 2017-05-04 Richard Biener <rguenther@suse.de>
21478 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
21479 for the last VUSE which def dominates the PHI. Directly call
21481 (get_continuation_for_phi_1): Remove.
21483 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
21485 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
21486 to explain the use of truncating division. Cap the number of
21487 iterations to the maximum given by nb_iterations_upper_bound,
21490 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21492 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
21493 * configure: Regenerate.
21494 * config.in: Regenerate.
21495 * config/i386/driver-mingw32.c: new file.
21496 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
21497 * config.host: Link driver-mingw32.o on MinGW host.
21498 * doc/install.texi: Document new --enable-mingw-wildcard configure
21501 2017-05-04 Marek Polacek <polacek@redhat.com>
21503 PR tree-optimization/80612
21504 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
21506 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21507 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
21509 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
21510 (movt splitter): Likewise.
21511 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
21512 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
21513 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
21514 block for Thumb-1 with MOVT.
21515 (thumb2_legitimate_address_p): Move code block ...
21516 (can_avoid_literal_pool_for_label_p): ... into this new function.
21517 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
21519 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
21520 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
21521 "M-profile targets with the MOVT instruction".
21523 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21525 * config/arm/arm-builtins.c (arm_init_builtins): Rename
21526 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
21527 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
21529 2017-05-04 Martin Liska <mliska@suse.cz>
21531 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
21532 variable cond_code.
21534 2017-05-04 Richard Biener <rguenther@suse.de>
21536 * tree.c (array_at_struct_end_p): Handle arrays at struct
21537 end with flexarrays more conservatively. Refactor and treat
21538 arrays of arrays or aggregates more strict. Fix
21539 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
21540 * tree.c (array_at_struct_end_p): Adjust prototype.
21541 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
21542 * gimple-fold.c (get_range_strlen): Likewise.
21543 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
21545 2017-05-04 Richard Biener <rguenther@suse.de>
21547 PR tree-optimization/31130
21548 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
21550 (supports_overflow_infinity): Likewise.
21551 (is_negative_overflow_infinity): Likewise.
21552 (is_positive_overflow_infinity): Likewise.
21553 (is_overflow_infinity): Likewise.
21554 (stmt_overflow_infinity): Likewise.
21555 (overflow_infinity_range_p): Likewise.
21556 (usable_range_p): Remove as always returning true.
21557 (make_overflow_infinity): Remove.
21558 (negative_overflow_infinity): Likewise.
21559 (positive_overflow_infinity): Likewise.
21560 (avoid_overflow_infinity): Likewise.
21561 (set_value_range): Adjust accordingly.
21562 (set_value_range_to_nonnegative): Likewise, remove now unused
21563 overflow_infinity arg.
21564 (vrp_operand_equal_p): Adjust.
21565 (update_value_range): Likewise.
21566 (range_int_cst_singleton_p): Likewise.
21567 (operand_less_p): Likewise.
21568 (compare_values_warnv): Likewise.
21569 (extract_range_for_var_from_comparison_expr): Likewise.
21570 (vrp_int_const_binop): Likewise.
21571 (zero_nonzero_bits_from_vr): Likewise.
21572 (extract_range_from_multiplicative_op_1): Likewise.
21573 (extract_range_from_binary_expr_1): Likewise.
21574 (extract_range_from_unary_expr): Likewise.
21575 (extract_range_from_comparison): Likewise.
21576 (extract_range_basic): Likewise.
21577 (adjust_range_with_scev): Likewise.
21578 (compare_ranges): Likewise.
21579 (compare_range_with_value): Likewise.
21580 (dump_value_range): Likewise.
21581 (test_for_singularity): Likewise, remove strict_overflow_p parameter
21583 (simplify_cond_using_ranges): Adjust.
21585 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21587 * brig-builtins.def: Added a builtin for class_f64.
21588 * builtin-types.def: Added a builtin type needed by class_f64.
21590 2017-05-03 Jason Merrill <jason@redhat.com>
21592 * timevar.def: Add TV_CONSTEXPR.
21594 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21596 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
21598 2017-05-03 Martin Jambor <mjambor@suse.cz>
21600 * ipa-prop.c (ipa_update_after_lto_read): Removed.
21601 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
21602 * ipa-cp.c (ipcp_propagate_stage): Do not call
21603 ipa_update_after_lto_read.
21604 * ipa-inline.c (ipa_inline): Likewise.
21606 2017-05-03 Martin Jambor <mjambor@suse.cz>
21608 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
21609 tag. Added a default constructor and a destructor.
21610 (ipa_edge_args_sum_t): New class;
21611 (ipa_edge_args_sum): Declare.
21612 (ipa_edge_args_vector): Remove declaration.
21613 (IPA_EDGE_REF): Use ipa_edge_args_sum.
21614 (ipa_free_edge_args_substructures): Remove declaration.
21615 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
21616 (ipa_edge_args_info_available_for_edge_p): Likewise.
21617 * ipa-prop.c (ipa_edge_args_vector): Removed.
21618 (edge_removal_hook_holder): Likewise.
21619 (edge_duplication_hook_holder): Likewise.
21620 (ipa_edge_args_sum): New variable.
21621 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
21622 ipa_edge_args_vector.
21623 (ipa_free_edge_args_substructures): Likewise.
21624 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
21625 ipa_edge_args_vector.
21626 (ipa_edge_removal_hook): Turned into method
21627 ipa_edge_args_sum_t::remove.
21628 (ipa_edge_duplication_hook): Turned into method
21629 ipa_edge_args_sum_t::duplicate.
21630 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
21631 registering edge hooks.
21632 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
21633 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
21634 ipa_edge_args_sum instead of ipa_edge_args_vector.
21635 * ipa-profile.c (ipa_profile): Likewise.
21637 2017-05-03 Martin Jambor <mjambor@suse.cz>
21639 * symbol-summary.h (function_summary): New method exists.
21640 (function_summary::symtab_removal): Deallocate through release.
21641 (call_summary): New class.
21642 (gt_ggc_mx): New overload.
21643 (gt_pch_nx): Likewise.
21644 (gt_pch_nx): Likewise.
21646 2017-05-03 Jeff Law <law@redhat.com>
21648 PR tree-optimization/78496
21649 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
21650 from simplify_cond_using_ranges. Split off code to walk
21651 backwards through casts into ...
21652 (simplify_cond_using_ranges_2): New function.
21653 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
21654 (execute_vrp): After identifying jump threads, call
21655 simplify_cond_using_ranges_2.
21657 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
21660 * ipa-inline.h (inline_summary): Add ctor.
21661 (create_ggc): Do not use ggc_cleared_alloc.
21663 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
21664 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21666 * gcc.c (handle_braces): Support escaping in switch matching
21668 * doc/invoke.texi (Spec Files): Document it.
21669 Remove superfluous @code markup in items.
21671 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21673 * diagnostic-show-locus.c (struct column_range): New struct.
21674 (get_affected_columns): New function.
21675 (get_printed_columns): New function.
21676 (struct correction): New struct.
21677 (correction::ensure_capacity): New function.
21678 (correction::ensure_terminated): New function.
21679 (struct line_corrections): New struct.
21680 (line_corrections::~line_corrections): New dtor.
21681 (line_corrections::add_hint): New function.
21682 (layout::print_trailing_fixits): Reimplement in terms of the new
21684 (selftest::test_overlapped_fixit_printing): New function.
21685 (selftest::diagnostic_show_locus_c_tests): Call it.
21687 2017-05-03 Nathan Sidwell <nathan@acm.org>
21689 Canonicalize canonical type hashing
21690 * tree.h (type_hash_canon_hash): Declare.
21691 * tree.c (type_hash_list, attribute_hash_list): Move into
21692 type_hash_canon_hash.
21693 (build_type_attribute_qual_variant): Break out hash code calc into
21694 type_hash_canon_hash.
21695 (type_hash_canon_hash): New. Generic type hash computation.
21696 (build_range_type_1, build_array_type_1, build_function_type,
21697 build_method_type_directly, build_offset_type, build_complex_type,
21698 make_vector_type): Call it.
21700 2017-05-03 Richard Biener <rguenther@suse.de>
21702 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
21703 When all DRs have unknown misaligned do not always peel
21704 when there is a store but apply the same costing model as if
21705 there were only loads.
21707 2017-05-03 Richard Biener <rguenther@suse.de>
21710 PR tree-optimization/80492
21711 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21712 compare_base_decls returning dont-know properly.
21714 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
21716 * config/arm/iterators.md (CCSI): New mode iterator.
21717 (arch): New mode attribute.
21718 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
21719 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
21720 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
21721 code iterator for success result mode.
21722 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
21723 the corresponding new insn generators.
21725 2017-05-03 Bin Cheng <bin.cheng@arm.com>
21728 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21729 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21731 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
21733 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
21734 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
21735 (DDR_A): Wrap DDR argument in brackets.
21736 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
21737 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
21738 (DDR_REVERSED_P): Likewise.
21740 2017-05-03 Jakub Jelinek <jakub@redhat.com>
21742 PR tree-optimization/79472
21743 * tree-switch-conversion.c (struct switch_conv_info): Add
21744 contiguous_range and default_case_nonstandard fields.
21745 (collect_switch_conv_info): Compute contiguous_range and
21746 default_case_nonstandard fields, don't clear final_bb if
21747 contiguous_range and only the default case doesn't have the required
21749 (check_all_empty_except_final): Set default_case_nonstandard instead
21750 of failing if contiguous_range and the default case doesn't have empty
21752 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
21753 and only the default case doesn't have the required constants. Skip
21755 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
21756 if default_case_nonstandard.
21757 (build_constructors): Build constant 1 just once. Assert that default
21758 values aren't inserted in between cases if contiguous_range. Skip
21760 (build_arrays): Skip virtual phis.
21761 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
21762 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
21763 Handle virtual phis.
21764 (gen_inbound_check): Handle default_case_nonstandard case.
21765 (process_switch): Adjust check_final_bb caller. Call
21766 gather_default_values with the first non-default case instead of
21767 default case if default_case_nonstandard.
21769 2017-05-02 Nathan Sidwell <nathan@acm.org>
21771 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
21772 check. Fix formatting.
21774 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
21776 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
21777 errors when comparing specialized and unspecialized times.
21779 2017-05-02 David Malcolm <dmalcolm@redhat.com>
21781 * diagnostic-show-locus.c
21782 (layout::should_print_annotation_line_p): Make private.
21783 (layout::print_annotation_line): Make private.
21784 (layout::annotation_line_showed_range_p): Make private.
21785 (layout::show_ruler): Make private.
21786 (layout::print_source_line): Make private. Pass in line and
21787 line_width, rather than calling location_get_source_line. Drop
21789 (layout::print_leading_fixits): New method.
21790 (layout::print_any_fixits): Rename to...
21791 (layout::print_trailing_fixits): ...this, and make private.
21792 Don't print newline fixits.
21793 (diagnostic_show_locus): Move logic for printing one row into...
21794 (layout::print_line): ...this new function. Move the
21795 location_get_source_line call and error-handling from
21796 print_source_line to here. Call print_leading_fixits, and rename
21797 print_any_fixits to print_trailing_fixits.
21798 (selftest::test_fixit_insert_containing_newline): Update now that
21799 newlines are partially supported.
21800 (selftest::test_fixit_insert_containing_newline_2): New test.
21801 (selftest::test_fixit_replace_containing_newline): Update comments.
21802 (selftest::diagnostic_show_locus_c_tests): Call the new test.
21803 * edit-context.c (class added_line): New class.
21804 (class edited_line): Describe newline handling in comment.
21805 (edited_line::actually_edited_p): New method.
21806 (edited_line::print_content): Delete redundant decl.
21807 (edited_line::m_predecessors): New field.
21808 (edited_file::print_content): Call edited_line::print_content.
21809 (edited_file::print_diff): Update to support newlines.
21810 (edited_file::print_diff_hunk): Likewise.
21811 (edited_file::print_run_of_changed_lines): New function.
21812 (edited_file::print_diff_line): Convert to...
21813 (print_diff_line): ...this.
21814 (edited_file::get_effective_line_count): New function.
21815 (edited_line::edited_line): Initialize new field m_predecessors.
21816 (edited_line::~edited_line): Clean up m_predecessors.
21817 (edited_line::apply_fixit): Handle newlines.
21818 (edited_line::get_effective_line_count): New function.
21819 (edited_line::print_content): New function.
21820 (edited_line::print_diff_lines): New function.
21821 (selftest::test_applying_fixits_insert_containing_newline): New
21823 (selftest::test_applying_fixits_replace_containing_newline): New
21825 (selftest::insert_line): New function.
21826 (selftest::test_applying_fixits_multiple_lines): Add example of
21828 (selftest::edit_context_c_tests): Call the new tests.
21830 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21832 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
21833 parameter cand. Update dump information.
21834 (get_computation_cost): Update uses.
21836 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21838 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
21839 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
21840 (get_computation_at, rewrite_use_address): Update use of
21841 get_computation_aff.
21843 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21845 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
21846 (get_computation): Delete.
21847 (get_computation_cost): Implement like get_computation_cost_at.
21848 Use get_computation_at.
21849 (get_computation_cost_at): Delete.
21850 (rewrite_use_nonlinear_expr): Use get_computation_at.
21851 (rewrite_use_compare, remove_unused_ivs): Ditto.
21853 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21855 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
21857 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21859 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
21860 (ivopts_global_cost_for_size): Rename parameter and update uses.
21861 (iv_ca_recount_cost): Update uses.
21862 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21863 candidates seperately in n_invs and n_cands.
21864 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
21866 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21868 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
21869 (find_inv_vars_cb): New.
21870 (find_depends): Renamed to ...
21871 (find_inv_vars): ... this.
21872 (add_candidate_1, force_var_cost): Call find_inv_vars.
21873 (split_address_cost, determine_group_iv_cost_cond): Ditto.
21875 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21877 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
21878 inv_vars. Add inv_exprs.
21879 (struct iv_cand): Rename depends_on to inv_vars.
21880 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
21881 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
21882 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
21883 (dump_cand): Dump inv_vars.
21884 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
21885 (record_invariant, find_depends, add_candidate_1): Ditto.
21886 (set_group_iv_cost, force_var_cost): Ditto.
21887 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
21888 (get_computation_cost_at, get_computation_cost): Ditto.
21889 (determine_group_iv_cost_generic): Ditto.
21890 (determine_group_iv_cost_address): Ditto.
21891 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
21892 (determine_group_iv_costs): Ditto.
21893 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
21894 (iv_ca_set_remove_invariants): Renamed to ...
21895 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
21896 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
21897 (iv_ca_set_add_invariants): Renamed to ...
21898 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
21899 (iv_ca_set_cp): Use iv_ca_set_add_invs.
21900 (iv_ca_has_deps): Support inv_vars and inv_exprs.
21901 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
21902 (create_new_ivs): Remove useless dump.
21904 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21906 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
21908 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
21909 (iv_ca_set_no_cp, create_new_iv): Ditto.
21911 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21913 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21915 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21917 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
21918 function tree_check2.
21920 2017-05-02 Martin Liska <mliska@suse.cz>
21922 * doc/gcov.texi: Add missing preposition.
21923 * gcov.c (function_info::function_info): Properly fill up
21924 all member variables.
21926 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21928 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
21930 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21932 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
21934 2017-05-02 Martin Liska <mliska@suse.cz>
21937 * lto-streamer-in.c (lto_read_tree_1): Remove
21938 LTO_STREAMER_DEBUG.
21939 * lto-streamer.c (struct tree_hash_entry): Likewise.
21940 (struct tree_entry_hasher): Likewise.
21941 (tree_entry_hasher::hash): Likewise.
21942 (tree_entry_hasher::equal): Likewise.
21943 (lto_streamer_init): Likewise.
21944 (lto_orig_address_map): Likewise.
21945 (lto_orig_address_get): Likewise.
21946 (lto_orig_address_remove): Likewise.
21947 * lto-streamer.h: Likewise.
21948 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
21949 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
21951 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
21953 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
21954 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
21955 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
21956 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
21957 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
21958 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
21959 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
21960 (mm_maskz_sub_ss): New intrinsics.
21961 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21962 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21963 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
21964 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
21965 (__builtin_ia32_subss_mask_round): New builtins.
21966 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21967 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21968 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
21970 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
21971 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21973 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21976 2017-05-02 Martin Jambor <mjambor@suse.cz>
21978 PR tree-optimization/78687
21979 * tree-sra.c (access): New field parent.
21980 (process_subtree_disqualification): New function.
21981 (disqualify_candidate): Call it.
21982 (build_accesses_from_assign): Reset write flag if creating an
21984 (build_access_subtree): Fill in parent field and also prpagate
21985 down grp_write flag.
21986 (create_artificial_child_access): New parameter set_grp_write, set
21987 grp_write to its value.
21988 (propagate_subaccesses_across_link): Also propagate grp_write flag
21990 (propagate_all_subaccesses): Push the closest parent back to work
21991 queue if add_access_to_work_queue returned true.
21993 2017-05-02 Richard Biener <rguenther@suse.de>
21995 * common.opt (fstrict-overflow): Alias negative to fwrapv.
21996 * doc/invoke.texi (fstrict-overflow): Remove all traces of
21997 -fstrict-overflow documentation.
21998 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
21999 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
22000 flag_strict_overflow.
22001 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
22002 * lto-opts.c (lto_write_options): Do not stream it.
22003 * lto-wrapper.c (merge_and_complain): Do not handle it.
22004 * opts.c (default_options_table): Do not set -fstrict-overflow.
22005 (finish_options): Likewise do not clear it when sanitizing.
22006 * simplify-rtx.c (simplify_const_relational_operation): Do not
22007 test flag_strict_overflow.
22009 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22011 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
22012 using enabled attribute.
22013 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
22014 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
22015 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
22016 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
22017 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
22018 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
22019 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
22020 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
22021 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
22022 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
22024 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22026 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
22028 2017-05-02 Richard Biener <rguenther@suse.de>
22030 PR tree-optimization/80591
22032 2017-04-10 Richard Biener <rguenther@suse.de>
22034 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22037 2017-05-02 Richard Biener <rguenther@suse.de>
22039 PR tree-optimization/80549
22040 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
22041 (cleanup_tree_cfg_noloop): Create forwarders to known loop
22042 headers if they do not have a preheader.
22044 2017-05-02 Martin Liska <mliska@suse.cz>
22047 * common.opt: Fix typo.
22048 * doc/invoke.texi: Likewise.
22050 2017-05-01 Jan Beulich <jbeulich@suse.com>
22052 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
22053 swapping, add (x,x,m,x,n) alternative.
22055 2017-05-01 Nathan Sidwell <nathan@acm.org>
22057 * calls.c (combine_pending_stack_adjustment_and_call): Remove
22058 unnecessary unadjusted_alignment check.
22060 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22063 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
22065 * gimplify.c (gimplify_cilk_detach): New function.
22066 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
22067 * tree-core.h: Document EXPR_CILK_SPAWN.
22068 * tree.h (EXPR_CILK_SPAWN): Define.
22070 2017-05-01 David Malcolm <dmalcolm@redhat.com>
22072 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
22073 to use new fixit_hint representation, using the "replace" logic.
22074 (get_line_span_for_fixit_hint): Likewise.
22075 (layout::print_any_fixits): Likewise.
22076 (selftest::test_one_liner_many_fixits): Rename to...
22077 (selftest::test_one_liner_many_fixits_1): ...this, and update
22078 comment and expected output to reflect that the multiple fix-it
22079 hints are now consolidated into one insertion.
22080 (selftest::test_one_liner_many_fixits_2): New test.
22081 (selftest::test_diagnostic_show_locus_one_liner): Update for
22083 (selftest::test_fixit_consolidation): Update for fix-it API
22085 * diagnostic.c (print_parseable_fixits): Likewise.
22086 * edit-context.c (edited_line::m_line_events): Convert from
22087 auto_vec <line_event *> to auto_vec <line_event>.
22088 (class line_event): Convert from abstract base class to a concrete
22089 class, taking over the role of replace_event.
22090 (class insert_event): Delete.
22091 (class replace_event): Rename to class line_event. Convert to
22093 (edit_context::add_fixits): Reimplement.
22094 (edit_context::apply_insert): Delete.
22095 (edit_context::apply_replace): Rename to...
22096 (edit_context::apply_fixit): ...this. Convert to half-open range.
22097 (edited_file::apply_insert): Delete.
22098 (edited_file::apply_replace): Rename to...
22099 (edited_file::apply_fixit): ...this.
22100 (edited_line::~edited_line): Drop deletion of events.
22101 (edited_line::apply_insert): Delete.
22102 (edited_line::apply_replace): Rename to...
22103 (edited_line::apply_fixit): ...this. Convert to half-open range.
22104 Update for change to type of m_line_events.
22105 * edit-context.h (edit_context::apply_insert): Delete.
22106 (edit_context::apply_replace): Rename to...
22107 (edit_context::apply_fixit): ...this.
22109 2017-05-01 Martin Sebor <msebor@redhat.com>
22111 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22114 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22117 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22118 __get_cpuid_max returns 0.
22119 (__get_cpuid_count): Ditto.
22121 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22123 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22124 replacement expression is another instance of one of its arguments.
22126 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22129 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22130 check for stack push/pop autoinc.
22131 * config/i386/i386.c (ix86_agi_dependent): Return false
22132 if the only reason why modified_in_p returned true is that
22133 addr is SP based and set_insn is a push or pop.
22135 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22137 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22140 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22143 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22144 (account_size_time): Use two predicates - exec_pred and
22146 (evaluate_conditions_for_known_args): Compute both clause and
22148 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22149 (inline_summary_t::duplicate): Update.
22150 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22152 (compute_inline_parameters): Likewise.
22153 (estimate_edge_size_and_time): Update caluclation of time.
22154 (estimate_node_size_and_time): Compute both time and nonspecialized
22156 (estimate_ipcp_clone_size_and_time): Update.
22157 (inline_merge_summary): Update.
22158 (do_estimate_edge_time): Update.
22159 (do_estimate_edge_size): Update.
22160 (do_estimate_edge_hints): Update.
22161 (inline_read_section, inline_write_summary): Stream both new predicates.
22162 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22164 (compute_inlined_call_time): Cleanup.
22165 (big_speedup_p): Update.
22166 (edge_badness): Update.
22167 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22168 (size_time_entry): Replace predicate by exec_predicate and
22169 nonconst_predicate.
22170 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22171 (estimate_edge_time): Return also nonspec_time.
22172 (reset_edge_growth_cache): Update.
22174 2017-04-29 Jakub Jelinek <jakub@redhat.com>
22176 PR rtl-optimization/80491
22177 * ifcvt.c (noce_process_if_block): When looking for x setter
22178 with missing else_bb, don't check only the insn right before
22179 cond_earliest, but look for the last insn that x is modified in
22180 within the same bb.
22182 PR rtl-optimization/80491
22183 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
22185 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
22187 PR tree-optimization/80487
22188 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
22190 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22192 PR tree-optimization/79697
22193 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
22194 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
22195 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
22197 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
22198 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
22200 2017-04-28 Martin Sebor <msebor@redhat.com>
22202 PR tree-optimization/80523
22203 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
22204 (init_target_to_host_charmap, target_to_host, target_strtol10): New
22206 (maybe_warn, format_directive, parse_directive): Use new functions.
22207 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
22209 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
22211 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
22213 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
22215 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
22216 target_header_dir): Set correctly.
22217 * configure: Regenerated.
22218 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22219 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22220 instead of SYSTEM_HEADER_DIR.
22222 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
22224 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
22225 (estimate_local_effects): Likewise.
22226 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
22227 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
22228 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
22229 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
22230 do_estimate_edge_time, estimate_edge_time): Likewise.
22231 * ipa-inline-analysis.c (estimate_node_size_and_time,
22232 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
22233 (estimate_time_after_inlining): Remove.
22235 2017-04-28 Martin Liska <mliska@suse.cz>
22237 * doc/gcov.texi: Enhance documentation of gcov.
22239 2017-04-28 Martin Liska <mliska@suse.cz>
22241 * doc/gcov.texi: Sort options in alphabetic order.
22242 * doc/gcov-dump.texi: Likewise.
22243 * doc/gcov-tool.texi: Likewise.
22244 * gcov.c (print_usage): Likewise.
22245 * gcov-dump.c (print_usage): Likewise.
22246 * gcov-tool.c (print_merge_usage_message): Likewise.
22247 (print_rewrite_usage_message): Likewise.
22248 (print_overlap_usage_message): Likewise.
22250 2017-04-28 Martin Liska <mliska@suse.cz>
22252 PR gcov-profile/53915
22253 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
22255 2017-04-28 Martin Liska <mliska@suse.cz>
22257 PR gcov-profile/79891
22258 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
22259 is marked by compiler as living on a line.
22260 (get_cycles_count): Remove usage of the union.
22261 (output_intermediate_file): Likewise.
22262 (find_source): Fix GNU coding style.
22263 (accumulate_line_counts): Remove old non-all block mode.
22264 (output_lines): Remove usage of the union.
22265 * profile.c (output_location): Include all BBs, even if
22266 belonging to a same line (and file) as a previous BB.
22268 2017-04-28 Martin Liska <mliska@suse.cz>
22270 * gcov.c (process_args): Handle new argument 'w'.
22271 (read_graph_file): Assign ID to BBs.
22272 (output_branch_count): Display BB # if verbose flag is set.
22273 (output_lines): Likewise for arcs.
22274 (print_usage): Add '--verbose' option help.
22275 * doc/gcov.texi: Document --verbose (-w) option.
22277 2017-04-28 Martin Liska <mliska@suse.cz>
22279 * gcov.c (struct block_location_info): New struct.
22280 (process_file): Fill up the new structure.
22281 (read_graph_file): Replace usage of encoding by the newly added
22283 (add_line_counts): Likewise.
22284 (accumulate_line_counts): Remove usage of the union.
22285 (function_info::function_info): New function.
22286 (function_info::~function_info): Likewise.
22287 (process_file): Call delete instead of release_function.
22288 (release_function): Release the function.
22289 (release_structures): Call delete instead of release_function.
22290 (solve_flow_graph): Replace usage of num_blocks.
22291 (find_exception_blocks): Likewise.
22292 (output_lines): Fix GNU coding style.
22294 2017-04-28 Martin Liska <mliska@suse.cz>
22297 * coverage.c (coverage_remove_note_file): New function.
22298 * coverage.h: Declare the function.
22299 * toplev.c (finalize): Clean if an error has been seen.
22301 2017-04-28 Martin Liska <mliska@suse.cz>
22303 PR gcov-profile/80031
22304 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
22305 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
22306 * gcov.c (read_graph_file): Read just number of blocks.
22307 * profile.c (branch_prob): Do not stream 0 flags per a basic
22310 2017-04-28 Martin Liska <mliska@suse.cz>
22312 * gcov-dump.c (tag_*): Add new argument to declarations.
22313 (dump_gcov_file): Likewise.
22314 (tag_blocks): Add and use new argument depth.
22315 (tag_arcs): Likewise.
22316 (tag_lines): Likewise.
22317 (tag_counters): Likewise.
22318 (tag_summary): Likewise.
22319 (dump_working_sets): Use depth to do a proper indentation.
22321 2017-04-28 Jakub Jelinek <jakub@redhat.com>
22324 * cgraph.h (symtab_node::debug_symtab): No longer inline.
22325 * symtab.c (symtab_node::debug_symtab): Move definition here.
22327 2017-04-28 Richard Biener <rguenther@suse.de>
22329 * lto-streamer.h (LTO_major_version): Bump to 7.
22331 2017-04-28 Richard Biener <rguenther@suse.de>
22333 * tree-vrp.c (assert_info): New struct.
22334 (add_assert_info): New helper.
22335 (register_edge_assert_for_2): Refactor to add asserts to a vector
22337 (register_edge_assert_for_1): Likewise.
22338 (register_edge_assert_for): Likewise.
22339 (finish_register_edge_assert_for): New helper actually registering
22340 asserts where live on edge.
22341 (find_conditional_asserts): Adjust.
22342 (find_switch_asserts): Likewise.
22343 (evrp_dom_walker::try_find_new_range): Generalize.
22344 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
22346 2017-04-27 Marek Polacek <polacek@redhat.com>
22349 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
22350 arg10 and arg11 to itype.
22352 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
22354 * doc/extend.texi (Object Size Checking): Improve grammar.
22356 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
22359 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
22360 that the logic for permitting reciprocal estimates matches that
22363 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22366 * tree.c (type_cache_hasher::equal): Only compare
22367 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
22368 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
22369 non-aggregate element types.
22370 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
22371 about the flag on ARRAY_TYPEs in the comment, formatting fix.
22373 2017-04-27 Richard Biener <rguenther@suse.de>
22375 PR middle-end/80533
22376 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
22377 stripping ARRAY_REFs from MEM_EXPR make sure we're not
22378 keeping a reference to a trailing array.
22380 2017-04-27 Richard Biener <rguenther@suse.de>
22382 PR middle-end/80539
22383 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
22384 being in loop-closed SSA form conservatively.
22385 (chrec_fold_multiply_poly_poly): Likewise.
22387 2017-04-27 Tamar Christina <tamar.christina@arm.com>
22389 PR middle-end/79665
22390 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
22391 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
22393 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22396 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
22397 (aarch64_function_arg_alignment): Return unsigned int again, but still
22398 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
22399 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22400 Don't emit -Wpsabi note.
22401 (aarch64_function_arg_boundary): Likewise.
22402 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22405 2017-04-26 Nathan Sidwell <nathan@acm.org>
22407 * tree.h (crc32_unsigned_n): Declare.
22408 (crc32_unsigned, crc32_unsigned): Make inline.
22409 * tree.c (crc32_unsigned_bits): Replace with ...
22410 (crc32_unsigned_n): ... this.
22411 (crc32_unsigned, crc32_byte): Remove.
22412 (crc32_string): Remove unnecessary braces.
22414 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22416 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
22417 * ipa-inline-analysis.c (MAX_TIME): Remove.
22418 (account_size_time): Use sreal for time.
22419 (dump_inline_summary): Update.
22420 (estimate_function_body_sizes): Update.
22421 (estimate_edge_size_and_time): Update.
22422 (estimate_calls_size_and_time): Update.
22423 (estimate_node_size_and_time): Update.
22424 (inline_merge_summary): Update.
22425 (inline_update_overall_summary): Update.
22426 (estimate_time_after_inlining): Update.
22427 (inline_read_section): Update.
22428 (inline_write_summary): Update.
22429 * ipa-inline.c (compute_uninlined_call_time): Update.
22430 (compute_inlined_call_time): Update.
22431 (recursive_inlining): Update.
22432 (inline_small_functions): Update.
22433 (dump_overall_stats): Update.
22434 * ipa-inline.h: Include sreal.h.
22435 (size_time_entry): Turn time to sreal.
22436 (inline_summary): Turn self_time nad time to sreal.
22438 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22440 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
22442 (sreal::stream_out, sreal::stream_in): New.
22443 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
22445 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22447 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
22450 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
22453 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
22455 (dimode_scalar_chain::compute_convert_gain): Ditto.
22456 (dimode_scalar_chain::make_vector_copies): Ditto.
22457 (dimode_scalar_chain::convert_reg): Ditto.
22458 (dimode_scalar_chain::convert_insn): Ditto.
22459 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
22460 (VI248_AVX512BW_1): New mode iterator.
22461 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
22462 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
22465 2017-04-25 Martin Sebor <msebor@redhat.com>
22467 PR tree-optimization/80497
22468 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
22469 constants are representable in HOST_WIDE_INT.
22470 (parse_directive): Ditto.
22472 2017-04-25 Martin Sebor <msebor@redhat.com>
22475 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
22476 (new_zero_array): Adjust signature.
22477 (dom_info::dom_init): Used unsigned rather that size_t.
22478 (dom_info::dom_info): Same.
22480 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22481 Jakub Jelinek <jakub@redhat.com>
22484 * config/arm/arm.c: Include gimple.h.
22485 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22486 returns negative, increment ncrn only if it returned positive.
22487 (arm_needs_doubleword_align): Return int instead of bool,
22488 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
22489 members, but if there is any such non-FIELD_DECL
22490 > PARM_BOUNDARY aligned decl, return -1 instead of false.
22491 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22492 returns negative, increment nregs only if it returned positive.
22493 (arm_setup_incoming_varargs): Likewise.
22494 (arm_function_arg_boundary): Emit -Wpsabi note if
22495 arm_needs_doubleword_align returns negative, return
22496 DOUBLEWORD_ALIGNMENT only if it returned positive.
22498 2017-04-25 Marek Polacek <polacek@redhat.com>
22501 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22502 first argument to type.
22504 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
22507 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
22508 type checks to test for compatibility instead of equality.
22510 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22511 Jakub Jelinek <jakub@redhat.com>
22514 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
22516 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
22517 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
22518 the alignment computation, but return their maximum in warn_alignment.
22519 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22520 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
22522 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
22523 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22526 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22528 * config/arc/simdext.md (dmpyh): Fix typo.
22530 2017-04-25 Richard Biener <rguenther@suse.de>
22532 PR tree-optimization/80492
22533 * alias.c (compare_base_decls): Handle registers with asm
22534 specification conservatively.
22535 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22536 compare_base_decls returning dont-know properly.
22538 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22540 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
22541 (legitimate_offset_address_p): New function.
22542 (arc_legitimate_address_p): Use above function.
22544 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22546 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
22548 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22550 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
22551 ACCH registers whenever they are available.
22553 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22555 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
22556 double regs fix when not used.
22558 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22560 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
22562 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
22563 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
22565 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22567 * config/arc/arc.c (arc_output_addsi): Check for h-register class
22568 when emitting short ADD instructions.
22570 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22572 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
22574 (cmpsi_cc_c_insn): Likewise.
22575 (cbranchsi4_scratch): Compute proper instruction length using
22576 compact_hreg_operand.
22577 * config/arc/predicates.md (compact_hreg_operand): New predicate.
22579 2017-04-25 Richard Biener <rguenther@suse.de>
22581 PR middle-end/80509
22582 * passes.c (pass_manager::pass_manager): Initialize
22583 m_name_to_pass_map.
22585 2017-04-25 Richard Biener <rguenther@suse.de>
22587 PR tree-optimization/79201
22588 * tree-ssa-sink.c (statement_sink_location): Handle calls.
22590 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22593 * config/s390/vector.md: Split MEM->GPR vector moves for
22594 non-s_operand addresses.
22596 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22599 * config/s390/predicates.md (reload_const_wide_int_operand): New
22601 * config/s390/s390.md ("movti"): Remove d/P alternative.
22602 ("movti_bigconst"): New pattern definition.
22604 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22607 * s390-protos.h (s390_expand_cs_hqi): Removed.
22608 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
22609 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
22610 modes as well as CCZ1mode and CCZmode.
22611 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
22612 signature of s390_emit_compare_and_swap.
22613 (s390_expand_cs_hqi): Likewise, make static.
22614 (s390_expand_cs_tdsi): Generate an explicit compare before trying
22615 compare-and-swap, in some cases.
22616 (s390_expand_cs): Wrapper function.
22617 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
22619 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
22620 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
22621 patterns for small and large integers. Forbid symref memory operands.
22622 Move expander to s390.c. Require cc register.
22623 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
22624 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
22625 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
22626 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
22627 symref memory operands. Remove CC mode and call s390_match_ccmode
22629 ("atomic_exchange<mode>"): Allow and implement all integer modes.
22631 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22633 * config/s390/s390.md (define_peephole2): New peephole to help
22634 combining the load-and-test pattern with volatile memory.
22636 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22638 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
22639 with CCZmode for TARGET_Z196.
22641 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22643 PR rtl-optimization/80501
22644 * combine.c (make_compound_operation_int): Set subreg_code to SET
22645 even for AND with mask of the sign bit of mode.
22647 PR rtl-optimization/80500
22648 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
22649 sum's initial value.
22651 2017-04-25 Julian Brown <julian@codesourcery.com>
22652 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22654 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
22656 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
22658 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
22660 2017-04-25 Julian Brown <julian@codesourcery.com>
22661 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22663 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
22664 (thunderx2t99_sha): New Reservation.
22666 2017-04-25 Julian Brown <julian@codesourcery.com>
22667 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22669 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
22670 type for 1-element load.
22672 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
22674 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
22676 2017-04-24 Martin Jambor <mjambor@suse.cz>
22678 PR tree-optimization/80293
22679 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
22680 char arrays not totally scalarizable if it is false.
22681 (analyze_all_variable_accesses): Pass correct value in the new
22682 parameter. Add a statistics counter.
22684 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
22686 PR middle-end/79931
22687 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
22689 2017-04-24 Richard Biener <rguenther@suse.de>
22691 PR tree-optimization/80494
22692 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
22693 out for complex types.
22695 2017-04-24 Richard Biener <rguenther@suse.de>
22697 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
22698 * tree-ssa-sccvn.c (print_scc): Print SCC size.
22699 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
22700 (DFS): Adjust and never fail.
22701 (sccvn_dom_walker::fail): Remove.
22702 (sccvn_dom_walker::before_dom_children): Adjust.
22703 (run_scc_vn): Likewise and never fail.
22704 * tree-ssa-pre.c (pass_pre::execute): Adjust.
22705 (pass_fre::execute): Likewise.
22707 2017-04-24 Richard Biener <rguenther@suse.de>
22709 PR tree-optimization/79725
22710 * tree-ssa-sink.c (statement_sink_location): Return whether
22711 failure reason was zero uses. Move that check later.
22712 (sink_code_in_bb): Deal with zero uses by removing the stmt
22715 2017-04-24 Richard Biener <rguenther@suse.de>
22718 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
22719 pointer-based references.
22721 2017-04-24 Richard Biener <rguenther@suse.de>
22724 * pass_manager.h (pass_manager::operator new): Remove.
22725 (pass_manager::operator delete): Likewise.
22726 * passes.c (pass_manager::operator new): Remove.
22727 (pass_manager::operator delete): Likewise.
22728 (pass_manager::pass_manager): Zero individual pass members.
22730 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
22733 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
22734 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
22735 Check "XEXP (src, 1)" operand here.
22736 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
22737 Check "XEXP (src, 1)" operand here.
22738 (dimode_scalar_chain::make_vector_copies): Detect count register
22739 of a shift instruction. Zero extend count register from QImode
22740 to DImode to satisfy vector shift pattern count operand predicate.
22741 Substitute vector shift count operand with a DImode copy.
22742 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
22745 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
22747 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
22748 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22749 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
22750 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22751 (UNSPEC_NOREX_MEM): Remove definition.
22753 2017-04-21 Richard Biener <rguenther@suse.de>
22755 PR tree-optimization/79547
22756 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22757 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
22758 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
22759 without any constraints.
22761 2017-04-21 Richard Biener <rguenther@suse.de>
22763 PR tree-optimization/78847
22764 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
22766 2017-04-21 Richard Biener <rguenther@suse.de>
22768 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
22769 (build_distinct_type_copy): Likewise.
22770 (build_variant_type_copy): Likewise.
22771 * tree.c (build_qualified_type): Pass down mem-stat info.
22772 (build_distinct_type_copy): Likewise.
22773 (build_variant_type_copy): Likewise.
22775 2017-04-21 Richard Biener <rguenther@suse.de>
22777 PR tree-optimization/80237
22778 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
22780 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
22781 for a simplified result.
22783 2016-04-21 Richard Biener <rguenther@suse.de>
22785 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
22786 sth as strict as a simple_iv but a chrec without symbols and an
22787 operand defined in the loop we are peeling (and not some subloop).
22788 (propagate_constants_for_unrolling): Propagate all constants.
22790 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22793 * config/i386/i386.c (print_reg): Remove assert for disalowed
22794 regno values, call output_operand_lossage instead.
22796 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22799 * config/i386/constraints.md (Yc): New register constraint.
22800 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
22801 Use Yc constraint for alternative 2 of operand 0. Remove
22802 preferred_for_speed attribute.
22804 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22806 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
22807 lastprivate clauses in SIMT case.
22809 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
22811 * doc/invoke.texi (-Wextra-semi): Document new warning option.
22813 2017-04-20 Richard Biener <rguenther@suse.de>
22815 PR tree-optimization/57796
22816 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
22817 as N scalar stores.
22818 (vect_model_load_cost): Cost gathers as N scalar loads.
22820 2017-04-20 Richard Biener <rguenther@suse.de>
22822 * ggc-page.c (ggc_allocated_p): Rename to ...
22823 (safe_lookup_page_table_entry): ... this and return the lookup
22825 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
22827 2017-04-20 Richard Biener <rguenther@suse.de>
22829 PR tree-optimization/80453
22830 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
22831 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
22832 from the conditions.
22833 (vn_phi_eq): Pass them down.
22834 (vn_phi_lookup): Record them.
22835 (vn_phi_insert): Likewise.
22837 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
22839 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
22840 uninitialized variable warning to avoid buffer overrun.
22842 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22845 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
22846 is suppressed for '{ 0 }' in C.
22848 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22850 * BASE-VER: Set to 8.0.0.
22852 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
22854 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
22855 priority .init_array and .fini_array section with SECTION_NOTYPE
22858 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22860 PR middle-end/80423
22861 * tree.h (build_array_type): Add typeless_storage default argument.
22862 * tree.c (type_cache_hasher::equal): Also compare
22863 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
22864 (build_array_type): Add typeless_storage argument, set
22865 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
22867 (build_nonshared_array_type): Adjust build_array_type_1 caller.
22868 (build_array_type): Likewise. Add typeless_storage argument.
22870 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22871 Jakub Jelinek <jakub@redhat.com>
22873 PR tree-optimization/80426
22874 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
22875 operation on symbolic operands, also compute the overflow for the
22876 invariant part when the operation degenerates into a negation.
22878 2017-04-19 Jakub Jelinek <jakub@redhat.com>
22881 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
22882 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
22885 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
22887 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
22890 * config/avr/avr.c (tree.h): Include it.
22891 (cgraph.h): Include it.
22892 (avr_encode_section_info): Don't warn for uninitialized progmem
22893 variable if it's just an alias.
22895 2017-04-19 Richard Biener <rguenther@suse.de>
22898 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
22899 when needed by AutoPGO.
22901 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
22904 * doc/lto.texi: Remove an extra 'that'.
22906 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
22908 PR rtl-optimization/80429
22909 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
22910 are only used in debug insns.
22912 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22913 Vladimir Makarov <vmakarov@redhat.com>
22915 * config/sparc/predicates.md (input_operand): Add comment. Return
22916 true for any memory operand when LRA is in progress.
22917 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
22919 2017-04-18 Jeff Law <law@redhat.com>
22922 * mips.md ({return,simple_return}_internal): Do not overwrite
22925 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22927 PR tree-optimization/80443
22928 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
22929 instead of adding 1, subtract -1 and similarly instead of subtracting
22932 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
22934 PR rtl-optimization/80357
22935 * haifa-sched.c (tmp_bitmap): New variable.
22936 (model_recompute): Handle duplicate use records.
22937 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
22938 (free_global_sched_pressure_data): Free it.
22940 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
22943 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22944 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22945 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22946 instead of SYSTEM_HEADER_DIR.
22948 2017-04-18 Jeff Law <law@redhat.com>
22950 PR middle-end/80422
22951 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
22952 predecessors after walking up the insn chain.
22954 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22957 * dwarf2out.c (modified_type_die): Try harder not to emit internal
22958 sizetype type into debug info.
22960 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22963 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
22964 unneeded test for TARGET_UPPER_REGS_SF.
22965 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
22967 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22970 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
22971 instead of gsi_after_labels.
22973 2017-04-18 Jeff Law <law@redhat.com>
22975 * regcprop.c (maybe_mode_change): Avoid creating copies of the
22979 2017-04-13 Jeff Law <law@redhat.com>
22980 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22981 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22983 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
22986 * config/avr/avr.c (intl.h): Include it.
22987 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
22989 2017-04-18 Martin Liska <mliska@suse.cz>
22991 PR gcov-profile/78783
22992 * gcov-tool.c (gcov_output_files): Validate that destination
22993 file is either removed by the tool or by a user.
22995 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22996 Guy Benyei <guybe@mellanox.com>
22998 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
22999 block, and do not negate it, the stored id is already negative.
23001 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23003 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
23005 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23008 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
23009 masks of options that should be turned off if the VSX vector
23010 options are turned off.
23011 (OTHER_P8_VECTOR_MASKS): Likewise.
23012 (OTHER_VSX_VECTOR_MASKS): Likewise.
23013 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
23014 rs6000_disable_incompatible_switches to validate no type switches
23016 (rs6000_incompatible_switch): New function to disallow turning on
23017 other vector options if -mno-vsx, -mno-power8-vector, or
23018 -mno-power9-vector are specified.
23020 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23022 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
23024 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23026 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
23027 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
23028 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
23029 (ARG_POINTER_CFA_OFFSET): Likewise.
23031 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23033 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
23034 conditions to take advantage of various optimizations.
23036 2017-04-13 Jeff Law <law@redhat.com>
23038 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23039 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23040 (zero_extendsidi2_dext): Likewise.
23042 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23045 * fold-const.c (fold_ternary_loc): Revert
23046 use op0 instead of fold_convert_loc (loc, type, arg0) part of
23049 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
23051 PR rtl-optimization/80343
23052 * lra-remat.c (update_scratch_ops): Assign original hard reg to
23053 new scratch pseudo.
23055 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
23058 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
23059 to ubsan_encode_value.
23061 2017-04-13 Jeff Law <law@redhat.com>
23063 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23064 appearing in DEBUG_INSNs.
23066 2017-04-13 Martin Liska <mliska@suse.cz>
23068 PR gcov-profile/80413
23069 * gcov-io.c (gcov_write_string): Copy to buffer just when
23070 allocated size is greater than zero.
23072 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23075 * dwarf2out.c (decls_for_scope): Ignore declarations of
23076 current_function_decl in BLOCK_NONLOCALIZED_VARS.
23078 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
23081 * ipa-visibility.c (non_local_p): Fix typos.
23082 (localize_node): When localizing symbol in same comdat group,
23083 dissolve the group only when we know external symbols are going
23085 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
23087 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23089 PR tree-optimization/79390
23090 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
23091 order does not result in usable sequence, retry with reversed operand
23097 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23098 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23099 op0 instead of fold_convert_loc (loc, type, arg0).
23101 2017-04-12 Jeff Law <law@redhat.com>
23103 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23104 has a delay slot in the generated code.
23106 * config/cris/cris.md (cris_preferred_reload_class): Return
23107 GENNONACR_REGS rather than GENERAL_REGS.
23109 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23112 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23113 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23114 signedness of the result type.
23116 2017-04-12 Richard Biener <rguenther@suse.de>
23117 Jeff Law <law@redhat.com>
23119 PR tree-optimization/80359
23120 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23121 trim stores to TARGET_MEM_REFs.
23123 2017-04-12 Richard Biener <rguenther@suse.de>
23125 PR tree-optimization/79390
23126 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23127 threading case even more.
23129 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23132 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23133 for quad_address_p for TImode, instead of just not indexed_address.
23135 2017-04-12 Richard Biener <rguenther@suse.de>
23136 Bernd Edlinger <bernd.edlinger@hotmail.de>
23138 PR middle-end/79671
23139 * alias.c (component_uses_parent_alias_set_from): Handle
23140 TYPE_TYPELESS_STORAGE.
23141 (get_alias_set): Likewise.
23142 * tree-core.h (tree_type_common): Add typeless_storage flag.
23143 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23144 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23145 for types containing members with TYPE_TYPELESS_STORAGE.
23146 (place_field): Likewise.
23147 (layout_type): Likewise for ARRAY_TYPE.
23148 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23149 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23150 TYPE_TYPELESS_STORAGE.
23151 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23153 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23156 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23157 first argument to type.
23159 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23163 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23164 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23165 (rs6000_expand_binop_builtin): Likewise.
23166 (rs6000_expand_ternop_builtin): Likewise; also add missing
23167 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23168 vshasigma built-ins.
23169 * doc/extend.texi: Document that vec_xxpermdi's third argument
23170 must be a constant.
23172 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23174 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
23175 Use shift_const cost parameter when calculating gain of STV shifts.
23177 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
23179 PR rtl-optimization/70478
23180 * lra-constraints.c (process_alt_operands): Check memory for
23181 disfavoring memory insn operand.
23183 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23185 PR middle-end/80100
23186 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
23187 left shift in unsigned HOST_WIDE_INT type.
23189 PR rtl-optimization/80385
23190 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
23191 (not (neg X)) into (plus X -1) for complex or non-integral modes.
23194 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
23195 if they have any depend clauses.
23197 2017-04-11 Martin Liska <mliska@suse.cz>
23200 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
23201 * ipa-split.c (split_function): Create a local comdat symbol
23202 if caller is in a comdat group.
23204 2017-04-11 Martin Liska <mliska@suse.cz>
23207 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
23210 2017-04-11 Martin Sebor <msebor@redhat.com>
23212 PR middle-end/80364
23213 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
23214 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
23216 (directive::set_width, directive::set_precision, format_character):
23218 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
23221 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
23224 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
23225 conflict, set target->arch_name instead of target->cpu_name.
23227 2017-04-11 Richard Biener <rguenther@suse.de>
23229 PR tree-optimization/80374
23230 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
23231 build_zero_cst, remove fold_convertible_p check again.
23233 2017-04-11 Martin Liska <mliska@suse.cz>
23236 * ubsan.c (instrument_object_size): Do not instrument register
23239 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23242 * config/i386/i386-builtin-types.def
23243 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
23244 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
23245 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
23246 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
23247 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
23248 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
23249 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
23250 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
23251 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
23252 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
23253 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
23254 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
23255 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
23256 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
23257 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
23258 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
23259 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
23260 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
23261 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
23262 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
23263 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
23264 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
23265 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
23266 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
23267 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
23268 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
23269 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
23270 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
23271 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
23272 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
23273 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
23274 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
23275 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
23276 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
23277 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
23278 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
23279 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
23280 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
23281 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
23282 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
23283 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
23284 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
23285 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
23286 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
23287 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
23288 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
23290 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
23291 flag to second_arg_count, handle 4 argument function type _COUNT
23292 aliases, handle second_arg_count on second argument rather than last.
23294 2017-04-10 Jeff Law <law@redhat.com>
23296 PR tree-optimization/80374
23297 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
23298 record anything if we can not convert integer_zero_node to the
23301 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23304 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23305 Enhance special handling given to the TARGET_P9_MINMAX option in
23306 relation to certain other options.
23308 2017-04-10 Bin Cheng <bin.cheng@arm.com>
23310 PR tree-optimization/80153
23311 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
23312 remove POINTER_PLUS_EXPR's base part directly, rather than through
23315 2017-04-10 Richard Biener <rguenther@suse.de>
23316 Bin Cheng <bin.cheng@arm.com>
23318 PR tree-optimization/80153
23319 * tree-affine.c (aff_combination_to_tree): Get base pointer from
23320 the first element of pointer type aff_tree. Build result expr in
23322 (add_elt_to_tree): Convert to type unconditionally. Remove other
23323 fold_convert calls.
23324 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
23325 (rewrite_use_nonlinear_expr): Check invariant using iv information.
23327 2017-04-10 Richard Biener <rguenther@suse.de>
23329 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23332 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
23334 PR rtl-optimization/70478
23335 * lra-constraints.c (curr_small_class_check): New.
23336 (update_and_check_small_class_inputs): New.
23337 (process_alt_operands): Update curr_small_class_check. Disfavor
23338 alternative insn memory operands. Check available regs for small
23341 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23344 * config/mips/mips.opt (-mvirt): Update description.
23345 * doc/invoke.texi (-mvirt): Likewise.
23347 2017-04-10 Richard Biener <rguenther@suse.de>
23349 PR middle-end/80362
23350 * fold-const.c (fold_binary_loc): Look at unstripped ops when
23351 looking for NEGATE_EXPR in -A / -B to A / B folding.
23353 2017-04-10 Martin Liska <mliska@suse.cz>
23355 PR gcov-profile/80224
23356 * gcov.c (print_usage): Fix usage string.
23357 (get_gcov_intermediate_filename): Remove.
23358 (output_gcov_file): Use both for normal and intermediate format.
23359 (generate_results): Do not initialize special file for
23360 intermediate format.
23362 2017-04-10 Richard Biener <rguenther@suse.de>
23364 PR tree-optimization/80304
23365 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
23368 2017-04-10 Nathan Sidwell <nathan@acm.org>
23371 * config/rs6000/rs6000.c (rs6000_vector_type): New.
23372 (rs6000_init_builtins): Use it.
23374 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23376 * config/arm/arm.md (<mrc>): Add mode to SET source.
23377 (<mrrc>): Likewise.
23379 2017-04-10 Richard Biener <rguenther@suse.de>
23381 PR middle-end/80344
23382 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
23384 2017-04-10 Jakub Jelinek <jakub@redhat.com>
23387 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
23388 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
23389 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
23390 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
23391 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
23392 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
23393 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
23394 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
23395 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
23396 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
23397 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
23398 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
23399 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
23400 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
23401 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
23402 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
23403 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
23404 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
23405 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
23406 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
23407 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
23408 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
23409 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
23411 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
23413 PR rtl-optimization/70478
23414 * lra-constraints.c: Reverse the last patch.
23416 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
23418 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
23419 Add comment for WCHAR_T.
23421 2017-04-08 Martin Liska <mliska@suse.cz>
23424 2017-04-07 Martin Liska <mliska@suse.cz>
23427 * ipa-split.c (split_function): Add function part to a same comdat
23430 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23433 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
23435 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
23437 * rs6000/rs6000.c (vec_load_pendulum): Rename...
23438 (vec_pairing): ...to this.
23439 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
23440 (rs6000_sched_init): Adjust for name change.
23441 (struct rs6000_sched_context): Likewise.
23442 (rs6000_init_sched_context): Likewise.
23443 (rs6000_set_sched_context): Likewise.
23445 2017-04-07 Jakub Jelinek <jakub@redhat.com>
23451 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
23453 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
23454 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
23455 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
23457 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
23459 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
23461 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23463 PR rtl-optimization/70703
23464 * ira-color.c (update_conflict_hard_regno_costs): Use
23465 int64_t instead of HOST_WIDE_INT.
23467 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23469 PR rtl-optimization/70478
23470 * lra-constraints.c (process_alt_operands): Disfavor alternative
23471 insn memory operands.
23473 2017-04-07 Jeff Law <law@redhat.com>
23475 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
23476 CALL and NOTE_INSN_CALL_ARG_LOCATION.
23478 2017-04-07 Martin Liska <mliska@suse.cz>
23481 * config/aarch64/aarch64.c (aarch64_process_target_attr):
23482 Show error message instead of an ICE.
23484 2017-04-07 Martin Liska <mliska@suse.cz>
23487 * ipa-split.c (split_function): Add function part to a same comdat
23490 2017-04-07 Richard Biener <rguenther@suse.de>
23492 PR middle-end/80341
23493 * tree.c (get_unwidened): Also handle ! for_type case for
23495 * convert.c (do_narrow): Split out from ...
23496 (convert_to_integer_1): ... here. Do not pass final truncation
23497 type to get_unwidened for TRUNC_DIV_EXPR.
23499 2017-04-07 Richard Biener <rguenther@suse.de>
23501 * tree-affine.c (wide_int_ext_for_comb): Take type rather
23503 (aff_combination_const): Adjust.
23504 (aff_combination_scale): Likewise.
23505 (aff_combination_add_elt): Likewise.
23506 (aff_combination_add_cst): Likewise.
23507 (aff_combination_convert): Likewise.
23508 (add_elt_to_tree): Likewise. Remove unused argument.
23509 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
23511 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
23513 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
23515 * config/arm/arm.c (arm_default_short_enums): Use
23516 ARM_DEFAULT_SHORT_ENUMS.
23517 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
23519 2017-04-06 Jakub Jelinek <jakub@redhat.com>
23522 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
23523 members with redundant out-of-class redeclaration.
23525 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23528 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
23529 * config/i386/i386.md (*zero_extendsidi2):
23530 Add (?*x,*x) and (?*v,*v) alternatives.
23532 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23535 * config/i386/i386.c (ix86_expand_builtin)
23536 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
23537 mode from insn data. Convert operands to insn operand mode.
23538 Copy operands that don't satisfy insn predicate to a register.
23540 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
23542 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
23545 2017-04-06 Richard Biener <rguenther@suse.de>
23547 PR tree-optimization/80334
23548 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
23549 preserve alignment of accesses.
23551 2017-04-06 Richard Biener <rguenther@suse.de>
23553 PR tree-optimization/80262
23554 * tree-sra.c (build_ref_for_offset): Preserve address-space
23556 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
23557 Drop useless address-space information on MEM_REF offsets.
23559 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
23561 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
23563 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23565 PR rtl-optimization/70703
23566 * ira-color.c (update_conflict_hard_regno_costs): Use
23567 HOST_WIDE_INT instead of long.
23569 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
23572 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
23573 not defined for x86_64 target. Add -mmmx target option when __SSE2__
23575 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
23576 for x86_64 target. Handle -m3dnowa option.
23578 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23580 PR rtl-optimization/70703
23581 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
23582 (update_conflict_hard_regno_costs): Use long instead of unsigned
23583 arithmetic for cost calculation.
23585 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23586 Bernd Edlinger <bernd.edlinger@hotmail.de>
23589 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
23592 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
23595 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
23596 ptr_mode with Pmode throughout.
23597 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
23598 into probe_stack_range and use DImode.
23600 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
23603 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
23604 call_eh_return is true.
23606 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23608 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
23609 Initialize last_match_fntype_index.
23611 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23614 * tree-nvr.c: Include internal-fn.h.
23615 (pass_return_slot::execute): Ignore internal calls without
23618 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23619 Richard Biener <rguenther@suse.de>
23622 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
23623 captures used multiple times, except for the last use.
23624 * generic-match-head.c: Include gimplify.h.
23626 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23628 PR tree-optimization/79390
23629 * target.h (struct noce_if_info): Declare.
23630 * targhooks.h (default_noce_conversion_profitable_p): Declare.
23631 * target.def (noce_conversion_profitable_p): New target hook.
23632 * ifcvt.h (struct noce_if_info): New type, moved from ...
23633 * ifcvt.c (struct noce_if_info): ... here.
23634 (noce_conversion_profitable_p): Renamed to ...
23635 (default_noce_conversion_profitable_p): ... this. No longer
23637 (noce_try_store_flag_constants, noce_try_addcc,
23638 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
23639 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
23640 instead of noce_conversion_profitable_p.
23641 * config/i386/i386.c: Include ifcvt.h.
23642 (ix86_option_override_internal): Don't override
23643 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
23644 (ix86_noce_conversion_profitable_p): New function.
23645 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
23646 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
23647 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
23648 * doc/tm.texi: Regenerated.
23650 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23652 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
23655 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23658 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
23659 instructions for small multiply cores.
23661 2017-04-04 Jeff Law <law@redhat.com>
23663 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
23665 (mips_expand_vec_perm_const): Initialize elements in orig_perm
23666 that are not set by the loop over the elements.
23668 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23671 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
23672 int mode, convert_modes it to mode as unsigned, otherwise use
23673 lowpart_subreg to mode rather than SImode.
23674 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
23675 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
23676 Use DImode instead of SImode for the shift count operand.
23677 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
23680 2017-04-04 Richard Biener <rguenther@suse.de>
23682 PR middle-end/80281
23683 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
23684 arithmetic done for the negate or the plus. Simplify.
23685 (A - (-B) -> A + B): Likewise.
23686 * fold-const.c (split_tree): Make sure to not negate pointers.
23688 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
23690 PR rtl-optimization/60818
23691 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
23692 a compare of comparisons with the thing compared if this results
23693 in a different machine mode.
23695 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
23697 * alias.c (base_alias_check): Fix typo in comment.
23698 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
23699 * cgraphunit.c (symbol_table::compile): Likewise.
23700 * collect2.c (maybe_run_lto_and_relink): Likewise.
23701 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
23702 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
23703 * config/avr/avr.c (avr_map_op_t): Likewise.
23704 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
23705 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
23706 * config/epiphany/epiphany.md (movcc): Likewise.
23707 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
23708 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
23710 * config/mips/mips.c (mips_save_restore_reg): Likewise.
23711 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
23712 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
23713 * config/sh/sh.c (sh_rtx_costs): Likewise.
23714 * fold-const.c (fold_truth_andor): Likewise.
23715 * genautomata.c (collapse_flag): Likewise.
23716 * gengtype.h (struct type::u::s): Likewise.
23717 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
23718 * input.c (FORMAT_AMOUNT): Likewise.
23719 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
23720 (known_aggs_to_agg_replacement_list): Likewise.
23721 * ipa-inline-analysis.c: Likewise.
23722 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
23723 * ipa-polymorphic-call.c
23724 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
23725 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
23726 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
23728 * modulo-sched.c (apply_reg_moves): Likewise.
23729 * omp-expand.c (build_omp_regions_1): Likewise.
23730 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
23731 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
23732 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
23733 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
23734 * value-prof.c: Likewise.
23735 * var-tracking.c (val_reset): Likewise.
23737 2017-04-03 Richard Biener <rguenther@suse.de>
23739 PR tree-optimization/80275
23740 * fold-const.c (split_address_to_core_and_offset): Handle
23743 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
23745 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
23746 descriptors is at least equal to that of functions.
23748 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23750 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
23752 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23755 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
23756 (mov<IMOD4:mode>): New expander.
23757 (*mov<IMOD4:mode>_internal): New insn and split pattern.
23759 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
23761 PR rtl-optimization/79405
23762 * fwprop.c (propagations_left): New variable.
23763 (forward_propagate_into): Decrement it.
23764 (fwprop_init): Initialize it.
23765 (fw_prop): If the variable has reached zero, stop propagating.
23766 (fwprop_addr): Ditto.
23768 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23771 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
23772 a FUNCTION_DECL, pass it as decl instead of origin to
23775 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
23777 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
23780 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
23783 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
23784 TARGET_VSX_SMALL_INTEGER.
23786 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23788 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23789 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
23791 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23793 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
23794 extraction from odd-numbered MSA register.
23796 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23798 PR middle-end/80173
23799 * expmed.c (store_bit_field_1): Don't attempt to create
23800 a word subreg out of hard registers wider than word if they
23801 have HARD_REGNO_NREGS of 1 for their mode.
23803 PR middle-end/80163
23804 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
23805 conversions to integer types wider than word and pointer.
23808 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
23809 (rtx_equal_for_cselib_p): Pass 0 to it.
23810 * cselib.c (cselib_hasher::equal): Likewise.
23811 (rtx_equal_for_cselib_1): Add depth argument. If depth
23812 is 128, don't look up VALUE locs and punt. Increment
23813 depth in recursive calls when walking VALUE locs.
23815 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
23817 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
23818 (make_gcov_file_name): Use the canonical path name for generating
23820 (read_line): Fix handling of files with ascii null bytes.
23822 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
23824 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
23825 to initialise a vector register instead
23826 of using a const_int.
23828 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23830 PR translation/80189
23831 * gimplify.c (omp_default_clause): Use %qs instead of %s in
23832 diagnostic messages.
23834 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
23837 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
23838 (dfp_diex_<mode>): Update mode of operand 1.
23839 * doc/extend.texi (dxex, dxexq): Document change to return type.
23840 (diex, diexq): Document change to argument type.
23842 2017-03-30 Martin Jambor <mjambor@suse.cz>
23845 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
23846 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
23847 it reflects the signature changes performed at the callee side.
23848 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
23849 to cgraph_build_function_type_skip_args.
23850 (build_function_decl_skip_args): Adjust call to the above function.
23852 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23855 * config/i386/sse.md
23856 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
23857 register as dest whenever it is a MEM not rtx_equal_p to the
23858 corresponding dup operand, and when forcing into reg move the
23859 reg into the memory afterwards.
23860 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
23861 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
23862 for the force_reg mode.
23863 (avx512vl_vextractf128<mode>): Use register as dest either
23864 always when a MEM, or when it is a MEM not rtx_equal_p to the
23865 corresponding dup operand, or even not when it is a CONST_VECTOR
23866 depending on the mode and lo vs. hi.
23867 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
23869 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
23870 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
23871 Likewise. Require that operands[2] is even.
23872 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
23873 Remove extraneous parens. Require that operands[2] is a multiple
23875 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
23876 operands[0] is a MEM if <mask_applied>, the predicates/constraints
23877 disallow memory then.
23879 2017-03-30 Richard Biener <rguenther@suse.de>
23881 PR tree-optimization/77498
23882 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
23883 to non-constants over backedges.
23885 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
23887 PR rtl-optimization/80233
23888 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
23889 as last_combined_insn. Do not test for BARRIER_P separately.
23891 2017-03-29 Andreas Schwab <schwab@suse.de>
23894 * calls.c (prepare_call_address): Convert funexp to Pmode before
23895 copying to temp reg.
23897 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23899 PR tree-optimization/80158
23900 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
23901 Handle possible future case of more than one alternate
23903 (replace_rhs_if_not_dup): Likewise.
23904 (replace_one_candidate): Likewise.
23906 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
23908 PR rtl-optimization/80193
23909 * ira.c (ira): Do not check allocation for LRA.
23911 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
23913 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
23914 (nvptx_output_simt_exit): Declare.
23915 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
23916 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
23917 (init_softstack_frame): Move initialization of crtl->is_leaf to...
23918 (nvptx_declare_function_name): ...here. Emit declaration of local
23919 memory space buffer for omp_simt_enter insn.
23920 (nvptx_output_unisimt_switch): New.
23921 (nvptx_output_softstack_switch): New.
23922 (nvptx_output_simt_enter): New.
23923 (nvptx_output_simt_exit): New.
23924 * config/nvptx/nvptx.h (struct machine_function): New fields
23925 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
23926 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
23927 (UNSPECV_SIMT_EXIT): Ditto.
23928 (omp_simt_enter_insn): New insn.
23929 (omp_simt_enter): New expansion.
23930 (omp_simt_exit): New insn.
23931 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
23933 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
23934 (expand_GOMP_SIMT_ENTER_ALLOC): New.
23935 (expand_GOMP_SIMT_EXIT): New.
23936 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
23937 (GOMP_SIMT_ENTER_ALLOC): Ditto.
23938 (GOMP_SIMT_EXIT): Ditto.
23939 * target-insns.def (omp_simt_enter): New insn.
23940 (omp_simt_exit): Ditto.
23941 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
23943 (lower_rec_simd_input_clauses): Implement SIMT privatization.
23944 (lower_rec_input_clauses): Likewise.
23945 (lower_lastprivate_clauses): Handle SIMT privatization.
23947 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
23948 (ompdevlow_adjust_simt_enter): New.
23949 (find_simtpriv_var_op): New.
23950 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
23951 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
23953 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
23954 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
23955 (copy_decl_for_dup_finish): Ditto.
23957 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
23959 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
23962 * config/i386/i386.c (ix86_option_override_internal): Always
23963 allow -mpreferred-stack-boundary=3 for 64-bit targets.
23965 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23967 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
23969 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23971 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
23972 mark new edge's irreducible flag accordign to it.
23973 (vect_do_peeling): Check loop preheader edge's irreducible flag
23974 and pass it to function slpeel_add_loop_guard.
23976 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
23978 PR tree-optimization/80218
23979 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23980 Update block frequencies and counts.
23982 2017-03-28 Richard Biener <rguenther@suse.de>
23984 PR tree-optimization/78644
23985 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
23986 of a simplification result we may not use it at all.
23988 2017-03-28 Richard Biener <rguenther@suse.de>
23991 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
23992 without arguments, generate default definition of a SSA name.
23994 2017-03-28 Richard Biener <rguenther@suse.de>
23996 PR middle-end/80222
23997 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
23998 TYPE_REF_CAN_ALIAS_ALL references.
23999 * fold-const.c (fold_indirect_ref_1): Likewise.
24001 2017-03-28 Martin Liska <mliska@suse.cz>
24004 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
24005 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
24007 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24008 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
24010 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
24011 (EXTRA_SPECS): Define.
24012 (SUBTARGET_EXTRA_SPECS): Likewise.
24013 (SUBTARGET_CPP_SPEC): Likewise.
24014 * config/arc/elf.h (EXTRA_SPECS): Renamed to
24015 SUBTARGET_EXTRA_SPECS.
24016 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
24018 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24020 * config/arc/simdext.md (vst64_insn): Update pattern.
24021 (vld32wh_insn): Likewise.
24022 (vld32wl_insn): Likewise.
24023 (vld64_insn): Likewise.
24024 (vld32_insn): Likewise.
24026 2017-03-28 Marek Polacek <polacek@redhat.com>
24029 * fold-const.c (fold_comparison): Use protected_set_expr_location
24030 instead of SET_EXPR_LOCATION.
24032 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
24034 * tree.c (add_expr): Avoid name lookup warning.
24036 2017-03-27 Jeff Law <law@redhat.com>
24038 PR tree-optimization/80216
24039 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
24040 function name. Limit recursion depth.
24041 (record_temporary_equivalences): Corresponding changes.
24043 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
24045 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
24048 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24051 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
24053 * cfgcleanup.c (reg_note_cfa_p): New array.
24054 (insns_have_identical_cfa_notes): New function.
24055 (old_insns_match_p): Don't cross-jump in between /f
24056 and non-/f instructions. If both i1 and i2 are frame related,
24057 verify all CFA notes, their order and content.
24059 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24062 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
24063 HImode and SImode with zero extend to DImode to one insn.
24064 (bswap<mode>2_extenddi): Likewise.
24065 (bswapsi2_extenddi): Likewise.
24066 (bswaphi2_extendsi): Likewise.
24067 (bswaphi2): Combine bswap HImode and SImode into one insn.
24068 Separate memory insns from swapping register.
24069 (bswapsi2): Likewise.
24070 (bswap<mode>2): Likewise.
24071 (bswaphi2_internal): Delete, no longer used.
24072 (bswapsi2_internal): Likewise.
24073 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
24074 store, and gpr<-gpr swap insns.
24075 (bswap<mode>2_store): Likewise.
24076 (bswaphi2_reg): Register only splitter, combine with the splitter.
24077 (bswaphi2 splitter): Likewise.
24078 (bswapsi2_reg): Likewise.
24079 (bswapsi2 splitter): Likewise.
24080 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
24081 the insns into load, store, and register/register insns.
24082 (bswapdi2_ldbrx): Likewise.
24083 (bswapdi2_load): Likewise.
24084 (bswapdi2_store): Likewise.
24085 (bswapdi2_reg): Likewise.
24087 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
24089 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
24090 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
24092 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
24095 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24097 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24098 special handling for target option conflicts between dform
24099 options (-mpower9-dform, -mpower9-dform-vector,
24100 -mpower9-dform-scalar) and -mno-direct-move.
24102 2017-03-27 Richard Biener <rguenther@suse.de>
24104 PR tree-optimization/80181
24105 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24107 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24109 * config/arc/predicates.md (move_double_src_operand): Replace the
24110 call to move_double_src_operand with a call to address_operand.
24112 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24114 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24115 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24116 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24118 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24120 * config/arc/predicates.md (long_immediate_loadstore_operand):
24121 Consider scaled addresses cases.
24123 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24125 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24126 restored when in interrupt.
24127 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24128 doesn't have delay slot.
24130 2017-03-27 Richard Biener <rguenther@suse.de>
24133 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24134 inlined thunk clones.
24136 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24139 * asan.c (instrument_derefs): Copy over last operand from
24140 original COMPONENT_REF to the new COMPONENT_REF with
24141 DECL_BIT_FIELD_REPRESENTATIVE.
24142 * ubsan.c (instrument_object_size): Likewise.
24144 2017-03-27 Richard Biener <rguenther@suse.de>
24146 PR tree-optimization/80170
24147 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24148 sure DR/SCEV didnt fold in constants we do not see when looking
24149 at the reference base alignment.
24151 2017-03-27 Richard Biener <rguenther@suse.de>
24153 PR middle-end/80171
24154 * gimple-fold.c (fold_ctor_reference): Properly guard against
24155 NULL return value from canonicalize_constructor_val.
24157 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24160 * config/i386/i386.c (ix86_expand_builtin)
24161 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24162 flags reg setting and flags reg using instructions.
24163 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24164 clobbering instructions to zero extend op2.
24166 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24168 * doc/install.texi (Configuration) <--with-aix-soname>:
24169 Update link to AIX ld.
24171 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24173 PR rtl-optimization/80160
24174 PR rtl-optimization/80159
24175 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
24176 reg_alternate_class into account.
24178 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
24181 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
24182 to consider in curr_insn_transform.
24184 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24186 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
24187 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
24188 and emit_mode_inner.
24190 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24192 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
24193 argument to the overloaded builtin variants. Use the new flag to
24194 deprecate certain builtin variants.
24195 * config/s390/s390-builtin-types.def: Add new builtin types.
24196 * config/s390/s390-builtins.h: Support new flags field for
24197 overloaded builtins.
24198 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
24199 (s390_macro_to_expand): Enable vector float data type.
24200 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24201 builtins by incrementing the __VEC__ version number.
24202 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
24204 (s390_resolve_overloaded_builtin): Emit error messages depending
24205 on the builtin flags.
24206 * config/s390/s390.c (s390_expand_builtin): Support additional
24207 flags argument. Change error message to match the messages
24208 emitted in s390-c.c.
24209 * config/s390/s390.md: New UNSPEC_* constants.
24210 (op_type): Add new instruction types.
24211 * config/s390/vecintrin.h: Add new builtins and test data class
24213 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
24214 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
24215 (VEC_INEXACT, VEC_NOINEXACT): New constants.
24216 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
24217 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
24218 ("vec_mergel<mode>"): V_HW -> VEC_HW.
24220 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
24221 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
24222 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
24223 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
24225 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
24226 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
24227 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
24228 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
24230 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
24231 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
24232 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
24233 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
24234 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
24235 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
24236 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
24238 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
24239 ("vec_scatter_element<V_HW_4:mode>_DI")
24240 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
24241 ("vec_fpint<mode>", "vflls")
24242 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
24243 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
24244 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
24245 ("*vec_cmphe<mode>_cc"): ... these.
24247 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
24248 mode constant instead of magic value.
24250 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24252 * config/s390/s390.c (s390_expand_vec_compare): Support other
24253 vector floating point modes than just V2DF.
24254 (s390_expand_vcond): Likewise.
24255 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
24256 (s390_cannot_change_mode_class): Prevent mode changes between TF
24257 and V1TF in vector registers.
24258 * config/s390/s390.md (DF, SF): New mode attributes.
24259 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
24260 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
24261 SFmode support for VRs.
24262 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
24264 (VFT, VF_HW): New mode iterators.
24265 (vw, sdx): New mode attributes.
24266 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
24267 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
24268 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
24269 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
24270 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
24271 also the new vector floating point modes. Renaming to ...
24273 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
24274 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
24275 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
24276 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
24277 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
24278 ("vec_unordered<mode>"): ... these.
24280 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
24281 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
24282 ("*vec_extendv2df"): New insn definitions.
24284 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24286 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
24287 ("mulditi3_2", "*muldi3_sign"): New patterns.
24288 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
24289 rename the pattern definition.
24291 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24293 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
24295 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
24297 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24299 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
24300 instruction if possible.
24301 * config/s390/vector.md (vec_halfnumelts): New mode
24303 ("*vec_vllezlf<mode>"): New pattern.
24305 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24307 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
24308 ("popcountv4si2", "popcountv2di2"): Rename to ...
24309 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
24310 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
24312 ("popcount<mode>2_vxe"): New pattern.
24314 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24316 * common/config/s390/s390-common.c (processor_flags_table): Add
24318 * config.gcc: Add arch12.
24319 * config/s390/driver-native.c (s390_host_detect_local_cpu):
24320 Default to arch12 for unknown CPU model numbers.
24321 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
24322 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
24323 PROCESSOR_max sanity check.
24324 * config/s390/s390-opts.h (enum processor_type): Add
24326 * config/s390/s390.c (processor_table): Add arch12.
24327 (s390_expand_builtin): Add check for B_VXE flag.
24328 (s390_issue_rate): Add PROCESSOR_ARCH12.
24329 (s390_get_sched_attrmask): Likewise.
24330 (s390_get_unit_mask): Likewise.
24331 (s390_sched_score): Enable z13 scheduling for arch12.
24332 (s390_sched_reorder): Likewise.
24333 (s390_sched_variable_issue): Likewise.
24334 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
24336 (s390_tune_attr): Use z13 scheduling also for arch12.
24337 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
24338 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
24339 (TARGET_VXE_P): New macros.
24340 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
24341 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
24342 * config/s390/s390.opt: Add arch12 as processor_type.
24344 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24346 * config/s390/s390.md
24347 ("fixuns_truncdddi2", "fixuns_trunctddi2")
24348 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
24349 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
24351 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
24352 Rename expanders to ...
24354 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
24355 ("fixuns_truncdddi2_emu"): ... these.
24357 ("fixuns_trunc<mode>si2_emu"): New expander.
24359 ("*fixuns_truncdfdi2_z13"): Rename to ...
24360 ("*fixuns_truncdfdi2_vx"): ... this.
24362 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24364 * config/s390/2964.md: Remove the single element vector compare
24365 instructions which are no longer used.
24366 * config/s390/s390.c (s390_select_ccmode): Remove handling of
24368 (s390_canonicalize_comparison): Remove handling of DFmode
24370 (s390_expand_vec_compare_scalar): Remove function.
24371 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
24372 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
24374 ("*cmp<mode>_ccs"): Add wfcdb instruction.
24376 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24378 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
24380 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
24381 will anyway by matched by mov<mode>_64dfp.
24383 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24385 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
24386 vlef/vstef. Add missing operand to vleif.
24388 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24390 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
24391 pair for all vector types with 64 bit elements.
24392 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
24393 * config/s390/vector.md (V_HW_64): ... here.
24394 (V_128_NOSINGLE): New mode iterator.
24395 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
24396 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
24397 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
24398 ("*vec_load_pairv2di"): Change to ...
24399 ("*vec_load_pair<mode>"): ... this one.
24401 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24403 * config/s390/constraints.md: Add comments.
24404 (jKK): Reject element sizes > 8 bytes.
24405 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
24407 * config/s390/s390.md: Add the s_operand checks formerly in
24408 s390_split_ok_p to various splitters where they are still
24410 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
24411 for 128 bit vectors. Plus two splitters.
24413 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24415 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
24418 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24421 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
24422 error if the boundary argument is not constant.
24424 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24426 PR rtl-optimization/80112
24427 * loop-doloop.c (doloop_condition_get): Don't check condition
24428 if cmp isn't SET with IF_THEN_ELSE src.
24430 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24432 PR tree-optimization/80158
24433 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
24434 replacing a candidate statement, also replace it for the
24435 candidate's alternate interpretation.
24436 (replace_rhs_if_not_dup): Likewise.
24437 (replace_one_candidate): Likewise.
24439 2017-03-24 Richard Biener <rguenther@suse.de>
24441 PR tree-optimization/80167
24442 * graphite-isl-ast-to-gimple.c
24443 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
24445 (translate_isl_ast_to_gimple::get_rename): Likewise.
24447 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24449 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
24450 handling of certain combinations of target options, including the
24451 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
24452 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
24454 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24457 * config/arm/arm.md (*load_multiple): Add reload_completed to
24458 matching condition.
24460 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24461 Richard Biener <rguenth@suse.de>
24463 PR tree-optimization/79908
24464 PR tree-optimization/80136
24465 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24466 been cast away, gimplify_and_add suffices.
24468 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
24470 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
24472 2017-03-23 Richard Biener <rguenther@suse.de>
24474 PR tree-optimization/80032
24475 * gimplify.c (gimple_push_cleanup): Forced unconditional
24476 cleanups still have to go to the conditional_cleanups
24479 2017-03-22 Jakub Jelinek <jakub@redhat.com>
24481 PR tree-optimization/80072
24482 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
24484 (next_operand_entry_id): Change type to unsigned int.
24485 (sort_by_operand_rank): Make sure to return the right return value
24486 even if unsigned fields are bigger than INT_MAX.
24487 (struct oecount): Change cnt and id type to unsigned int.
24488 (oecount_hasher::equal): Formatting fix.
24489 (oecount_cmp): Make sure to return the right return value
24490 even if unsigned fields are bigger than INT_MAX.
24491 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
24494 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
24495 TREE_READONLY on result if writing it more than once.
24498 * doc/invoke.texi (-fsanitize=thread): Document that with
24499 -fnon-call-exceptions atomics are not able to throw
24503 * tsan.c: Include tree-eh.h.
24504 (instrument_builtin_call): Call maybe_clean_eh_stmt or
24505 maybe_clean_or_replace_eh_stmt where needed.
24506 (instrument_memory_accesses): Add cfg_changed argument.
24507 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
24508 if it returned true.
24509 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
24511 PR rtl-optimization/63191
24512 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
24513 wrapper function, moved the whole old content into ...
24514 (ix86_delegitimize_address_1): ... this. New inline function.
24515 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
24516 true as last argument instead of ix86_delegitimize_address.
24518 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24520 * config/aarch64/aarch64.c (generic_branch_cost): Copy
24521 cortexa57_branch_cost.
24523 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24525 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
24527 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24530 * doc/md.texi (Constraints): Document wA constraint.
24531 * config/rs6000/constraints.md (wA): New.
24532 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
24533 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
24534 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
24535 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
24537 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
24540 * gimplify.c (is_oacc_declared): New function.
24541 (oacc_default_clause): Use it to set default flags for acc declared
24542 variables inside parallel regions.
24543 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
24544 declared variables.
24545 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
24546 declare attribute to any decl as necessary.
24548 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24551 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
24552 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
24553 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
24554 (arm_arch_lpae): This.
24555 * config/arm/arm.c (arm_arch7ve): Rename into ...
24556 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
24557 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
24560 2017-03-22 Martin Liska <mliska@suse.cz>
24563 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
24564 error message instead of an ICE.
24566 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24568 * doc/extend.texi (6.11 Additional Floating Types): Revise.
24570 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
24572 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24574 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24577 2017-03-21 Martin Sebor <msebor@redhat.com>
24579 * doc/extend.texi: Use "cannot" instead of "can't."
24580 * doc/hostconfig.texi: Same.
24581 * doc/install.texi: Same.
24582 * doc/invoke.texi: Same.
24583 * doc/loop.texi: Same.
24584 * doc/md.texi: Same.
24585 * doc/objc.texi: Same.
24586 * doc/rtl.texi: Same.
24587 * doc/tm.texi: Same.
24588 * doc/tm.texi.in: Same.
24589 * doc/trouble.texi: Same.
24591 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
24594 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
24595 (collect_checksum_attributes): Set it.
24596 (die_checksum_ordered): Use it.
24598 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24600 PR tree-optimization/79908
24601 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
24602 change: For a VA_ARG whose LHS has been cast away, use
24603 force_gimple_operand to construct the side effects.
24605 2017-03-21 David Malcolm <dmalcolm@redhat.com>
24607 PR translation/80001
24608 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
24609 more amenable to translation.
24610 (oacc_loop_auto_partitions): Likewise.
24612 2017-03-21 Marek Polacek <polacek@redhat.com>
24613 Martin Sebor <msebor@redhat.com>
24615 PR tree-optimization/80109
24616 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
24617 on INTEGRAL_TYPE_P.
24619 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24620 Segher Boessenkool <segher@kernel.crashing.org>
24623 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
24624 check reg_used_between_p between insn and one of succ or succ2
24625 depending on if succ is artificial insn not inserted into insn
24628 2017-03-21 Martin Liska <mliska@suse.cz>
24630 PR gcov-profile/80081
24631 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
24632 * doc/gcc.texi: Include gcov-dump stuff.
24633 * doc/gcov-dump.texi: New file.
24635 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
24637 PR rtl-optimization/79150
24638 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
24639 conditional jump, if the jump is the last insn of the loop.
24641 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24642 Richard Biener <rguenth@suse.de>
24644 PR tree-optimization/79908
24645 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24646 been cast away, use force_gimple_operand to construct the side
24649 2017-03-21 Martin Liska <mliska@suse.cz>
24651 PR libfortran/79956
24652 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
24655 2017-03-21 Brad Spengler <spender@grsecurity.net>
24658 * plugin.c (htab_hash_plugin): New function.
24659 (add_new_plugin): Use it and adjust.
24660 (parse_plugin_arg_opt): Adjust.
24661 (init_one_plugin): Likewise.
24663 2017-03-21 Richard Biener <rguenther@suse.de>
24665 PR tree-optimization/80032
24666 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
24667 if set force the cleanup to happen unconditionally.
24668 (gimplify_target_expr): Push inserted clobbers with force_uncond
24669 to avoid them being removed by control-dependent DCE.
24671 2017-03-21 Richard Biener <rguenther@suse.de>
24673 PR tree-optimization/80122
24674 * tree-inline.c (copy_bb): Do not expans va-arg packs or
24675 va_arg_pack_len when the inlined call stmt requires pack
24677 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
24679 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24682 * tsan.c (instrument_builtin_call): If the memory model argument
24683 is not a constant, assume it is valid.
24686 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
24689 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24691 PR rtl-optimization/79910
24692 * combine.c (can_combine_p): Do not allow combining an I0 or I1
24693 if its dest is used by an insn before I2 (other than the combined
24694 insns themselves, which are properly handled already).
24696 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24699 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24701 * combine.c (record_used_regs): New static function.
24702 (try_combine): Handle situations where there is an additional
24703 instruction between I2 and I3 which needs to have a LOG_LINK
24707 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24709 * combine.c (try_combine): Delete redundant i1 test. Call
24710 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24712 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24715 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
24716 alternatives 13/14.
24718 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24720 PR tree-optimization/80054
24721 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
24722 the optimization if a PHI or any of its arguments is not dominated
24723 by the candidate's basis. Use gphi* rather than gimple* as
24725 (replace_profitable_candidates): Clean up a gimple* variable that
24726 should be a gphi* variable.
24728 2017-03-20 Martin Sebor <msebor@redhat.com>
24731 * doc/extend.texi (attribute constructor): Document present limitation.
24733 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
24736 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
24737 __POWER9_VECTOR__ #ifdef control, change template definition to
24738 use Power9-specific built-in function.
24739 (vec_any_eq): Likewise.
24740 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
24741 to control outcomes from this test.
24742 (vector_ae_<mode>p): For VEC_F modes, likewise.
24744 2017-03-20 Ian Lance Taylor <iant@google.com>
24746 * config/i386/i386.c (ix86_function_regparm): Save an extra
24747 register for -fsplit-stack with DECL_STATIC_CHAIN.
24749 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24752 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
24753 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
24755 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24757 * config/riscv/riscv.c (riscv_print_operand): Use "fence
24759 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
24762 2017-03-20 Marek Polacek <polacek@redhat.com>
24765 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
24767 2017-03-20 Richard Biener <rguenther@suse.de>
24769 PR tree-optimization/80113
24770 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
24771 allocate extra SSA name for PHI def.
24772 (add_close_phis_to_outer_loops): Likewise.
24773 (add_close_phis_to_merge_points): Likewise.
24774 (copy_loop_close_phi_args): Likewise.
24775 (copy_cond_phi_nodes): Likewise.
24777 2017-03-20 Martin Liska <mliska@suse.cz>
24779 PR middle-end/79753
24780 * tree-chkp.c (chkp_build_returned_bound): Do not build
24781 returned bounds for a LHS that's not a BOUNDED_P type.
24783 2017-03-20 Martin Liska <mliska@suse.cz>
24787 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
24788 COMPLEX_CST and VECTOR_CST.
24790 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24793 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
24794 target operand. A new splitter adds the clobber statement in case
24795 the target operand is dead anyway.
24797 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
24799 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
24800 to age-old versions of binutils and glibc.
24802 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
24804 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
24806 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24808 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
24810 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24812 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
24813 requirement for binutils 2.13.
24815 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24817 * combine.c (try_combine): Delete redundant i1 test. Call
24818 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24820 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
24822 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
24823 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
24825 <riscv64-*-elf>: Re-arrange section
24826 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
24827 <riscv32-*-linux>: Likewise.
24828 <riscv64-*-elf>: Likewise
24829 <riscv64-*-linux>: Likewise.
24831 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
24834 * aarch64.opt(verbose-cost-dump): Fix typo.
24836 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
24839 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
24840 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
24842 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24844 * reload.c (find_reloads): When reloading a nonoffsettable address,
24845 use RELOAD_OTHER for it and its address reloads.
24847 PR rtl-optimization/79910
24848 * combine.c (record_used_regs): New static function.
24849 (try_combine): Handle situations where there is an additional
24850 instruction between I2 and I3 which needs to have a LOG_LINK
24853 2017-03-17 Jeff Law <law@redhat.com>
24855 PR tree-optimization/71437
24856 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
24857 conditional in the hash table first.
24858 (vrp_dom_walker::before_dom_children): Extract condition from
24859 ASSERT_EXPR. Record condition, its inverion and any implied
24860 conditions as well.
24862 2017-03-17 Marek Polacek <polacek@redhat.com>
24863 Markus Trippelsdorf <markus@trippelsdorf.de>
24865 PR tree-optimization/80079
24866 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
24869 2017-03-17 Richard Biener <rguenther@suse.de>
24871 PR middle-end/80075
24872 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
24873 Properly verify the LHS before the RHS possibly claims to be
24875 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
24878 2017-03-17 Martin Jambor <mjambor@suse.cz>
24880 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
24881 (List of -O2 options): Likewise.
24882 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
24885 2017-03-17 Tom de Vries <tom@codesourcery.com>
24887 * gcov-dump.c (print_usage): Print bug_report_url.
24889 2017-03-17 Richard Biener <rguenther@suse.de>
24891 PR middle-end/80050
24892 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
24893 (parser::peek): Likewise.
24895 2017-03-17 Richard Biener <rguenther@suse.de>
24897 PR tree-optimization/80048
24898 * sese.c (free_sese_info): Properly release rename_map and
24899 copied_bb_map elements.
24901 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
24903 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
24904 Add linked-list forward and backlinks. Insert on
24905 construction, remove on destruction.
24906 (class pass_store_merging): Add m_stores_head field.
24907 (pass_store_merging::terminate_and_process_all_chains):
24908 Iterate over m_stores_head list.
24909 (pass_store_merging::terminate_all_aliasing_chains):
24911 (pass_store_merging::execute): Check for debug stmts first.
24912 Push new chains onto the m_stores_head stack.
24914 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24917 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
24918 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
24919 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
24921 2017-03-16 Jeff Law <law@redhat.com>
24923 PR tree-optimization/71437
24924 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
24925 member function. Implementation moved into after_dom_children
24926 member function and into the threader's thread_outgoing_edges
24928 (dom_opt_dom_walker::after_dom_children): Simplify by moving
24929 some code into new thread_outgoing_edges.
24930 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
24931 definition. Simplify marker handling (do it here). Assume we always
24932 have the available expression and the const/copies tables.
24933 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
24935 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
24936 * tree-vrp.c (equiv_stack): No longer file scoped.
24937 (vrp_dom_walker): New class.
24938 (vrp_dom_walker::before_dom_children): New member function.
24939 (vrp_dom_walker::after_dom_children): Likewise.
24940 (identify_jump_threads): Setup domwalker. Use it rather than
24941 walking edges in a random order by hand. Simplify setup/finalization.
24942 (finalize_jump_threads): Remove.
24943 (vrp_finalize): Do not call identify_jump_threads here.
24944 (execute_vrp): Do it here instead and call thread_through_all_blocks
24947 PR tree-optimization/71437
24948 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
24950 (simplify_stmt_for_jump_threading): Add basic_block argument. All
24952 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
24953 (dom_opt_dom_walker::thread_across_edge): Remove
24954 handle_dominating_asserts argument. All callers changed.
24955 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
24956 changes. Remove calls to lhs_of_dominating_assert. Other
24957 uses of handle_dominating_asserts turn into unconditional code
24958 (simplify_control_stmt_condition_1): Likewise.
24959 (simplify_control_stmt_condition): Likewise.
24960 (thread_through_normal_block, thread_across_edge): Likewise.
24961 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
24962 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
24963 object if it is not an SSA_NAME.
24964 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
24965 before calling into the VRP specific simplifiers.
24966 (identify_jump_threads): Remove handle_dominating_asserts
24969 2017-03-16 Jakub Jelinek <jakub@redhat.com>
24972 * tree-diagnostic.c (default_tree_printer): No longer static.
24973 * tree-diagnostic.h (default_tree_printer): New prototype.
24975 2017-03-16 Tamar Christina <tamar.christina@arm.com>
24977 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
24978 Change ins into fmov.
24980 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24982 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
24983 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
24984 Use h_con constraint for operand 1.
24985 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
24986 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
24988 2017-03-15 Jeff Law <law@redhat.com>
24990 PR tree-optimization/71437
24991 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
24992 (record_temporary_equivalences): Use it.
24994 PR tree-optimization/71437
24995 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
24996 tree-ssa-scopedtables.
24997 (lookup_avail_expr, build_and_record_new_cond): Likewise.
24998 (record_conditions, record_cond, vuse_eq): Likewise.
24999 (record_edge_info): Adjust to API tweak of record_conditions.
25000 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
25001 (record_temporary_equivalences, optimize_stmt): Likewise.
25002 (eliminate_redundant_computations): Likewise.
25003 (record_equivalences_from_stmt): Likewise.
25004 * tree-ssa-scopedtables.c: Include options.h and params.h.
25005 (vuse_eq): New function, moved from tree-ssa-dom.c
25006 (build_and_record_new_cond): Likewise.
25007 (record_conditions): Likewise. Accept vector of conditions rather
25008 than edge_equivalence structure for first argument.
25009 for the first argument.
25010 (avail_exprs_stack::lookup_avail_expr): New member function, moved
25011 from tree-ssa-dom.c.
25012 (avail_exprs_stack::record_cond): Likewise.
25013 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
25014 from tree-ssa-dom.c.
25015 (avail_exprs_stack): Add new member functions lookup_avail_expr
25017 (record_conditions): Declare.
25019 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
25022 * lra-constraints.c (process_alt_operands): Increase reject for
25023 reloading an input/output operand.
25025 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
25028 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
25029 insns to convert from signed/unsigned char/short to IEEE 128-bit
25031 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
25033 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
25036 * config/i386/i386.c (ix86_vector_duplicate_value): Create
25037 subreg of inner mode for values already in registers.
25039 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
25041 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
25042 iteration reg is used after the loop.
25044 2017-03-14 Martin Sebor <msebor@redhat.com>
25046 PR tree-optimization/79800
25047 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
25048 precision in negative-positive range.
25049 (format_floating): Call non-const overload with adjusted precision.
25051 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25054 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
25057 2017-03-14 Martin Sebor <msebor@redhat.com>
25059 PR middle-end/80020
25060 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
25061 * builtins.def (aligned_alloc): Use it.
25064 * Makefile.in (GTFILES): Add calls.c.
25065 * calls.c: Include "gt-calls.h".
25067 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
25069 PR rtl-optimization/79728
25070 * regs.h (struct target_regs): New field
25071 x_contains_allocatable_regs_of_mode.
25072 (contains_allocatable_regs_of_mode): New macro.
25073 * reginfo.c (init_reg_sets_1): Initialize it, and change
25074 contains_reg_of_mode so it includes global regs as well.
25075 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
25076 rather than contains_regs_of_mode.
25078 2017-03-14 Martin Liska <mliska@suse.cz>
25080 * doc/invoke.texi: Document options that can't be combined with
25081 -fcheck-pointer-bounds.
25083 2017-03-14 Martin Liska <mliska@suse.cz>
25085 PR middle-end/79831
25086 * doc/invoke.texi (-Wchkp): Document the option.
25088 2017-03-14 Martin Liska <mliska@suse.cz>
25090 * Makefile.in: Install gcov-dump.
25092 2017-03-14 Martin Liska <mliska@suse.cz>
25094 * multiple_target.c (expand_target_clones): Bail out for
25095 an invalid attribute.
25097 2017-03-14 Richard Biener <rguenther@suse.de>
25099 * alias.c (struct alias_set_entry): Pack properly.
25100 * cfgloop.h (struct loop): Likewise.
25101 * cse.c (struct set): Likewise.
25102 * ipa-utils.c (struct searchc_env): Likewise.
25103 * loop-invariant.c (struct invariant): Likewise.
25104 * lra-remat.c (struct cand): Likewise.
25105 * recog.c (struct change_t): Likewise.
25106 * rtl.h (struct address_info): Likewise.
25107 * symbol-summary.h (function_summary): Likewise.
25108 * tree-loop-distribution.c (struct partition): Likewise.
25109 * tree-object-size.c (struct object_size_info): Likewise.
25110 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25111 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25112 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25113 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25114 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25115 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25116 (struct _stmt_vec_info): Likewise.
25118 2017-03-14 Martin Liska <mliska@suse.cz>
25121 * multiple_target.c (create_dispatcher_calls): Check that
25122 a target can create a function dispatcher.
25124 2017-03-14 Martin Liska <mliska@suse.cz>
25127 * multiple_target.c (expand_target_clones): Drop local.local
25128 flag for default implementation.
25130 2017-03-14 Richard Biener <rguenther@suse.de>
25132 PR tree-optimization/80030
25133 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25135 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25137 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25138 gcc_fallthrough() instead of __attribute__((fallthrough));
25140 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25142 * doc/gcc.texi: Remove "up" link to (DIR).
25143 * doc/gccint.texi: Ditto.
25145 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25147 * doc/install.texi (Specific) <avr>: Remove reference to
25150 2017-03-13 Jeff Law <law@redhat.com>
25152 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25153 attribute rather than comments.
25155 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25156 match_scratch operand is highest.
25158 2017-03-13 Martin Liska <mliska@suse.cz>
25160 PR middle-end/78339
25161 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25162 is a CHKP clone, use original declaration.
25164 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25166 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25167 (arc_conditional_register_usage): Use a different allocation order
25168 when optimizing for size.
25169 * common/config/arc/arc-common.c (arc_option_optimization_table):
25170 Section anchors default on when optimizing for size.
25172 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25174 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
25176 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25178 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
25179 * config/arc/arc.md (cpu_facility): Add cd variant.
25180 (*movqi_insn): Add code density variant.
25181 (*movhi_insn): Likewise.
25182 (*movqi_insn): Likewise.
25183 (*addsi3_mixed): Likewise.
25184 (subsi3_insn): Likewise.
25186 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25188 * config/arc/arc.md (movsi_cond_exec): Update constraint.
25190 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25192 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
25193 expressions with MINUS and UNARY ops.
25195 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25198 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
25200 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
25201 between vec_select and vector argument.
25202 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
25203 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
25204 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
25205 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
25206 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
25207 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
25209 2017-03-13 Richard Biener <rguenther@suse.de>
25212 * params.def (vect-max-peeling-for-alignment): Fix typo.
25214 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25216 * doc/install.texi (Specific) <mips-*-*>: Remove description of
25217 issue that only occurred with binutils below 2.18.
25219 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25221 * doc/install.texi (Specific) <cris-axis-elf>: No longer
25222 refer to binutils 2.11/2.12 minimum.
25224 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25226 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
25227 ftp.kernel.org and simplify binutils requirement.
25229 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
25231 * doc/invoke.texi (Warning Options): Fix spelling of link-time
25233 (Optimize Options): Ditto. Also remove redundancy.
25235 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25237 PR translation/79848
25238 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
25240 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
25241 to G_ to avoid double translation.
25243 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25245 PR translation/79923
25246 * auto-profile.c (get_combined_location): Convert leading
25247 character of diagnostics to lower case and remove trailing period.
25248 (read_profile): Likewise for various diagnostics.
25249 * config/arm/arm.c (arm_option_override): Remove trailing period
25250 from various diagnostics.
25251 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
25252 (msp430_expand_delay_cycles): Likewise.
25254 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25257 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
25258 full command-line argument, rather than just "str".
25259 (aarch64_validate_march): Likewise.
25260 (aarch64_validate_mtune): Likewise.
25262 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
25264 PR rtl-optimization/78911
25265 * lra-assigns.c (must_not_spill_p): New function.
25266 (spill_for): Use it.
25268 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25270 PR tree-optimization/79981
25271 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
25272 ATOMIC_COMPARE_EXCHANGE ifn result.
25273 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
25274 IFN_ATOMIC_COMPARE_EXCHANGE.
25276 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25279 * opts.c (parse_sanitizer_options): Add missing question mark to
25280 "did you mean" message.
25282 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25284 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
25286 (VMULEUH_UNS): Likewise.
25287 (VMULOUB_UNS): Likewise.
25288 (VMULOUH_UNS): Likewise.
25289 * config/rs6000/rs6000.c (builtin_function_type): Remove
25290 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
25292 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25295 * read-rtl-function.c (function_reader::read_rtx_operand): Update
25296 x with result of extra_parsing_for_operand_code_0.
25297 (function_reader::extra_parsing_for_operand_code_0): Convert
25298 return type from void to rtx, returning x. When reading
25299 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
25300 larger size containing struct block_symbol.
25302 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
25304 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
25305 -mfloat128-hardware without -m64.
25307 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
25310 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
25311 entries to the case statement that marks unsigned arguments to
25312 overloaded functions.
25314 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25316 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
25317 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
25319 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
25322 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
25323 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
25325 2017-03-10 Martin Liska <mliska@suse.cz>
25329 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
25330 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
25333 2017-03-10 Olivier Hainque <hainque@adacore.com>
25335 * tree-switch-conversion (array_value_type): Start by resetting
25336 candidate type to it's main variant.
25338 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25340 PR rtl-optimization/79909
25341 * combine.c (try_combine): Use simplify_replace_rtx on individual
25342 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
25343 of the whole CALL_INSN_FUNCTION_USAGE.
25345 PR tree-optimization/79972
25346 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
25347 get_range_info on SSA_NAMEs. Formatting fixes.
25349 2017-03-10 Richard Biener <rguenther@suse.de>
25350 Jakub Jelinek <jakub@redhat.com>
25352 PR tree-optimization/77975
25353 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
25354 edge to be constant.
25355 (get_val_for): For constant x return it. Formatting fix.
25356 (loop_niter_by_eval): Avoid pointless looping if the next iteration
25357 would use the same bases as the current one.
25359 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25361 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
25362 instead of vec_select for V1TImode.
25363 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
25365 (VSX_LE_128): Add V1TI to this mode iterator.
25366 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
25367 (*vsx_le_perm_store_<mode>): Likewise.
25368 (pre-reload splitter for VSX stores): Likewise.
25369 (post-reload splitter for VSX stores): Likewise.
25370 (*vsx_xxpermdi2_le_<mode>): Likewise.
25371 (*vsx_lxvd2x2_le_<mode>): Likewise.
25372 (*vsx_stxvd2x2_le_<mode>): Likewise.
25374 2017-03-09 Michael Eager <eager@eagercon.com>
25376 Correct failures with --enable-checking=yes,rtl.
25378 * config/microblaze/microblaze.c (microblaze_expand_shift):
25379 Replace GET_CODE test with CONST_INT_P and INTVAL test with
25380 test for const0_rtx.
25381 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
25382 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
25384 2017-03-09 Richard Biener <rguenther@suse.de>
25386 PR tree-optimization/79977
25387 * graphite-scop-detection.c (scop_detection::merge_sese):
25388 Handle the case of extra exits to blocks dominating the entry.
25390 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
25392 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25395 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
25397 PR rtl-optimization/79949
25398 * lra-constraints.c (process_alt_operands): Check memory when
25399 trying to predict a cycle. Print about the overall increase.
25401 2017-03-09 Richard Biener <rguenther@suse.de>
25403 PR middle-end/79971
25404 * gimple-expr.c (useless_type_conversion_p): Preserve
25405 TYPE_SATURATING for fixed-point types.
25407 2017-03-09 Richard Biener <rguenther@suse.de>
25410 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
25411 alignment of BLKmode params.
25413 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25416 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
25417 (VALL_NO_V2Q): Likewise.
25419 * config/aarch64/aarch64-simd.md
25420 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
25422 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
25423 VALL_NO_V2Q mode iterator.
25424 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
25426 2017-03-09 Martin Liska <mliska@suse.cz>
25428 PR tree-optimization/79631
25429 * tree-chkp-opt.c (chkp_is_constant_addr): Call
25430 tree_int_cst_sign_bit just for INTEGER constants.
25432 2017-03-09 Martin Liska <mliska@suse.cz>
25436 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
25439 2017-03-09 Marek Polacek <polacek@redhat.com>
25442 * tree.c (inchash::add_expr): Handle TREE_VEC.
25444 2017-03-09 Martin Liska <mliska@suse.cz>
25447 (chkp_narrow_size_and_offset): New function.
25448 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
25449 (void chkp_parse_bit_field_ref): New function.
25450 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
25451 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
25453 2017-03-09 Martin Liska <mliska@suse.cz>
25456 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
25457 (chkp_find_bounds_1): Remove gcc_unreachable.
25459 2017-03-09 Jakub Jelinek <jakub@redhat.com>
25462 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
25463 BUILT_IN_SYNC*, determine the access type from the size suffix and
25464 always build a MEM_REF with that type. Handle forgotten
25465 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
25468 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
25469 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
25470 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
25471 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
25472 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
25473 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
25474 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
25475 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
25476 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
25477 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
25478 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
25479 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
25480 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
25481 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
25482 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
25483 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
25484 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
25485 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
25486 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
25487 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
25488 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
25489 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
25490 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
25491 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
25492 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
25493 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
25494 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
25495 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
25496 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
25497 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
25498 definitions outside of __OPTIMIZE__ guarded section.
25501 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
25502 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
25503 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
25504 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
25507 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25509 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
25510 ("vfenez<mode>"): Add missing constraints.
25512 2017-03-08 Martin Sebor <msebor@redhat.com>
25515 * config/nds32/nds32.c (nds32_option_override):
25516 Fix misspelled diagnostic.
25518 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25521 * gimplify.c (gimplify_omp_for): Replace index var in outer
25522 taskloop statement with an artificial variable and add
25523 OMP_CLAUSE_PRIVATE clause for it.
25525 2017-03-08 Richard Biener <rguenther@suse.de>
25527 PR tree-optimization/79955
25528 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
25529 for accesses that are completely outside of the variable.
25531 2017-03-08 Andrew Haley <aph@redhat.com>
25533 PR tree-optimization/79943
25534 * tree-ssa-loop-split.c (compute_new_first_bound): When
25535 calculating the new upper bound, (END-BEG) should be added, not
25538 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25540 * config/avr/avr.md (setmemhi): Make sure match_dup
25541 operand number comes before match_scratch.
25543 2017-03-08 Richard Biener <rguenther@suse.de>
25545 PR tree-optimization/79920
25546 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
25547 with ncopies == 1 to ...
25548 (vect_transform_slp_perm_load): ... here. Properly compute
25549 all element loads by iterating VF times over the group. Do
25550 not handle ncopies (computed in a broken way) in
25551 vect_create_mask_and_perm.
25553 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25556 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
25557 is a uniform vector, use uniform_vector_p return value instead of
25558 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
25560 2017-03-07 Marek Polacek <polacek@redhat.com>
25562 PR middle-end/79809
25563 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
25564 (alloca_call_type): Likewise.
25566 2017-03-07 Martin Liska <mliska@suse.cz>
25568 * gcov.c (process_args): Put comment to correct location.
25570 2017-03-07 Martin Liska <mliska@suse.cz>
25572 PR middle-end/68270
25573 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
25574 Use array_at_struct_end_p instead of DECL_CHAIN (field).
25575 (chkp_narrow_bounds_for_field): Likewise.
25576 (chkp_parse_array_and_component_ref): Pass one more argument to
25579 2017-03-07 Richard Biener <rguenther@suse.de>
25581 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
25584 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25586 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
25587 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
25589 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25592 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
25593 to end of description.
25594 (PARAM_MAX_STORES_TO_MERGE): Likewise.
25596 2017-03-07 Jakub Jelinek <jakub@redhat.com>
25598 PR rtl-optimization/79901
25599 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
25601 (*avx512f_<code><mode>3<mask_name>): ... this.
25602 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
25603 iterator instead of VI8_AVX2_AVX512BW.
25605 PR rtl-optimization/79901
25606 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
25607 min/max expander, expand it using expand_vec_cond_expr.
25610 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
25613 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25616 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
25617 to void * for PCH reasons.
25618 * dwarf2out.c (output_loc_operands, output_die): Cast
25619 v.val_vec.array to unsigned char *.
25621 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
25624 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
25627 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
25629 PR rtl-optimization/79571
25630 * lra-constraints.c (process_alt_operands): Calculate static
25631 reject and subtract it from overall when only addresses will be
25634 2017-03-06 Julia Koval <julia.koval@intel.com>
25637 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
25638 incoming stack boundary to 128 for 64-bit targets.
25640 2017-03-06 Richard Biener <rguenther@suse.de>
25642 PR tree-optimization/79894
25643 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
25644 to NULL after folding it.
25646 2017-03-06 Richard Biener <rguenther@suse.de>
25648 PR tree-optimization/79824
25649 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
25650 check disabling peeling for gaps.
25652 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
25654 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25655 attributes): Document gettimeofday.
25657 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25659 * config/s390/s390.c (s390_option_override_internal): Set
25660 PARAM_MIN_VECT_LOOP_BOUND
25662 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25664 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
25665 * config/s390/s390.md: Likewise.
25667 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25670 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
25671 (<avx2_avx512>_perm<mode>): Rename to ...
25672 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
25674 (<avx512>_perm<mode>_mask): Rename to ...
25675 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
25677 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
25678 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
25679 instead of VI8F_256_512.
25680 (avx512f_perm<mode>): New define_expand.
25681 (avx512f_perm<mode>_mask): Likewise.
25682 (avx512f_perm<mode>_1<mask_name>): New define_insn.
25683 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
25685 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25687 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
25688 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
25690 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
25692 2017-03-06 Martin Liska <mliska@suse.cz>
25695 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
25696 when having a SSA NAME w/o VAR_DECL assigned to it.
25698 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25700 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
25701 msa_dpsub_<su>_d): Fix MODE for vec_select.
25703 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25705 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
25707 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
25709 2017-03-06 Richard Biener <rguenther@suse.de>
25711 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
25712 * plugin.c (register_plugin_info): Likewise.
25713 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
25715 2017-03-05 Jakub Jelinek <jakub@redhat.com>
25717 * config/i386/sse.md (sse_storehps, sse_storelps,
25718 avx_<castmode><avxsizesuffix>_<castmode>,
25719 avx512f_<castmode><avxsizesuffix>_<castmode>,
25720 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
25721 in condition that at least one operand is not a MEM.
25723 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25725 PR middle-end/79805
25726 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
25727 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
25729 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
25730 gimple_call_nothrow_p flag based on whether original builtin can throw.
25731 If it can, emit following stmts on the fallthrough edge.
25732 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
25733 don't create new bb if inserting just debug stmts on the edge, try to
25734 insert them on the fallthru bb or just reset debug stmts.
25736 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
25739 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
25740 restore recog_data (including the operand rtxes inside it) around
25741 the call to get_insn_template.
25743 2017-03-03 Martin Sebor <msebor@redhat.com>
25745 PR tree-optimization/79699
25746 * context.c (context::~context): Free MPFR caches to avoid
25747 a memory leak on program exit.
25749 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25751 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
25752 Use wide_int::ulow () instead of .elt (0).
25754 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25756 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
25757 (*pushxf): Limit oF constraint to 32bit targets and add oC
25758 constraint for 64bit targets.
25759 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
25760 (*pushdf): Change rmF constraint to rmC.
25762 2017-03-03 Martin Liska <mliska@suse.cz>
25764 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
25765 Remove unused variable.
25767 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25770 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
25771 is a memory operand, increase num_memory.
25772 (ix86_expand_args_builtin): Likewise.
25774 2017-03-03 Jan Hubicka <jh@suse.cz>
25777 * ipa-devirt.c (maybe_record_node): Properly handle
25778 __cxa_pure_virtual visibility.
25780 2017-03-03 Martin Liska <mliska@suse.cz>
25782 PR tree-optimization/79803
25783 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
25785 (pass_loop_prefetch::execute): Disabled optimization if an
25786 assumption about L1 cache size is not met.
25788 2017-03-03 Martin Liska <mliska@suse.cz>
25790 PR rtl-optimization/79574
25791 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
25792 (hash_scan_set): Likewise.
25793 (dump_hash_table): Likewise.
25794 (hoist_code): Likewise.
25796 2017-03-03 Richard Biener <rguenther@suse.de>
25798 * fixed-value.c (fixed_from_string): Restore use of elt (1)
25799 in place of uhigh ().
25800 (fixed_convert_from_real): Likewise.
25802 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25805 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
25807 2017-03-03 Richard Biener <rguenther@suse.de>
25809 PR middle-end/79818
25810 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
25811 TYPE_OVERFLOW_UNDEFINED check.
25813 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25815 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
25817 (vector_ae_<mode>_p): Likewise.
25818 (vector_nez_<mode>_p): Likewise.
25819 (vector_ne_v2di_p): Likewise.
25820 (vector_ae_v2di_p): Likewise.
25821 (vector_ne_<mode>_p): Likewise.
25822 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
25824 (vsx_tsqrt<mode>2_fe): Likewise.
25826 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
25829 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
25831 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25833 PR rtl-optimization/79780
25834 * cprop.c (one_cprop_pass): When second and further conditional trap
25835 in a single basic block is turned into an unconditional trap, turn it
25836 into a deleted note to avoid RTL verification failures.
25838 2017-03-02 Richard Biener <rguenther@suse.de>
25840 * fold-const.c (const_binop): Use ulow () instead of elt (0).
25842 2017-03-02 Richard Biener <rguenther@suse.de>
25844 PR tree-optimization/79345
25846 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
25847 param and abort the walk, returning -1 if it is hit.
25848 (walk_aliased_vdefs): Take a limit param and pass it on.
25849 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
25850 defaulting to 0 and return a signed int.
25851 * tree-ssa-uninit.c (struct check_defs_data): New struct.
25852 (check_defs): New helper.
25853 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
25854 about uninitialized memory.
25855 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
25856 bogus uninitialized warning.
25857 (fixed_convert_from_real): Likewise.
25859 2017-03-02 Bin Cheng <bin.cheng@arm.com>
25861 PR tree-optimization/66768
25862 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
25863 iv_use if base object can't be determined.
25865 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25867 PR tree-optimization/79345
25868 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
25869 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
25870 (get_pattern_stats): Initialize it.
25871 * genemit.c (gen_expand): Verify match_scratch numbers come after
25872 match_operand/match_dup numbers.
25873 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
25874 match_scratch numbers.
25875 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
25877 * config/s390/s390.md (trunctdsd2): Likewise.
25879 2017-03-02 Richard Biener <rguenther@suse.de>
25881 * wide-int.h (wide_int_storage::operator=): Implement in terms
25884 2017-03-02 Richard Biener <rguenther@suse.de>
25886 PR tree-optimization/79777
25887 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
25888 the to insert expression to sth existing.
25890 2017-03-01 Martin Sebor <msebor@redhat.com>
25892 PR middle-end/79692
25893 * gimple-ssa-sprintf.c
25894 (directive::known_width_and_precision): New function.
25895 (format_integer): Use it.
25896 (get_mpfr_format_length): Consider the full range of precision
25897 when computing %g output with the # flag. Set the likely byte
25898 count to 3 rather than 1 when precision is indeterminate.
25899 (format_floating): Correct the lower bound of precision.
25901 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25903 * doc/invoke.texi: Document default code model for 64-bit Linux.
25905 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25908 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
25909 udiv rather than div since input pattern is unsigned.
25911 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
25913 * config/i386/i386.c (print_reg): Warn for values of
25914 unsupported size in integer register.
25916 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
25919 * config/rs6000/predicates.md (current_file_function_operand): Do
25920 not allow self calls to be local if the function is replaceable.
25922 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
25925 * config/rs6000/altivec.h (vec_ctz and others): Change the
25926 preprocessor macro that controls conditional compilation from
25927 _ARCH_PWR9 to __POWER9_VECTOR__.
25928 (vec_all_ne): Change parameterization of __altivec_scalar_pred
25929 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
25930 control (instead of _ARCH_PWR9 control) so that template
25931 definition uses power9-specific function.
25932 (vec_any_eq): Likewise.
25933 (vec_all_ne): Change macro definition to use a power9-specific
25934 expansion under #ifdef __POWER9_VECTOR__ control (instead of
25935 _ARCH_PWR9 control).
25936 (vec_any_eq) Likewise.
25937 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
25938 expansion for CMPNEF to remove support for xvcmpnesp instruction.
25939 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
25940 support for xvcmpnedp instruction.
25941 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
25942 macro expansion so that Power9 implementation of vec_all_ne does
25943 not use the AltiVec predicate framework.
25944 (VCMPNEH_P): Likewise.
25945 (VCMPNEW_P): Likewise.
25946 (VCMPNED_P): Likewise.
25947 (VCMPNEFP_P): Likewise.
25948 (VCMPNEDP_P): Likewise.
25949 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
25950 implementation of vec_any_eq to not use AltiVec predicate
25952 (VCMPAEH_P): Likewise.
25953 (VCMPAEW_P): Likewise.
25954 (VCMPAED_P): Likewise.
25955 (VCMPAEFP_P): Likewise.
25956 (VCMPAEDP_P): Likewise.
25957 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
25958 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
25959 not use the AltiVec predicate framework.
25960 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
25961 of vec_any_eq to not use AltiVec predicate framework.
25962 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25963 support for predefined __POWER9_VECTOR__ macro to indicate that
25964 Power9 instruction selection is enabled.
25965 (altivec_overloaded_builtins): Remove extraneous
25966 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
25967 function argument types RS6000_BTI_bool_V16QI and
25968 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
25969 entry for overloaded function argument types RS6000_BTI_bool_V4SI
25970 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
25971 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
25972 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
25973 Power9 for implementations of vec_cmpne. Change the signature for
25974 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
25975 (representing vec_all_ne) to remove the previously described first
25976 argument of type RS6000_BTI_INTSI, as this was an artifact of
25977 reliance on the AltiVec predicate framework, which is no longer
25978 used in the implementation of these functions. Add
25979 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
25980 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
25981 since, unlike the AltiVec predicate framework implementation, we
25982 do not share function descriptors between vec_alle and vec_anyeq.
25983 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
25984 set of modes that receive special treatment even when
25985 TARGET_P9_VECTOR is true. The special treatment emits code that
25986 does not depend on Power9 instructions.
25987 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
25988 define_expand to not rely on AltiVec predicate framework.
25989 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25991 (vector_ne_v2di_p): Change this define_expand to not rely on
25992 AltiVec predicate framework.
25993 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
25995 (vector_ne_<mode>_p): Change this define_expand to not rely on
25996 AltiVec predicate framework.
25997 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25999 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
26000 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
26001 define_insn pattern.
26002 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
26003 define_insn pattern because the xvcmpne<VSs>. instruction is not
26005 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
26006 instruction is not supported.
26008 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26010 * config/nvptx/nvptx.c: Include intl.h.
26012 2017-03-01 Martin Jambor <mjambor@suse.cz>
26015 * ipa-prop.h (ipa_bits): Removed field known.
26016 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
26017 to pointers. Adjusted their comments to warn about their sharing.
26018 (ipcp_transformation_summary): Change bits to a vector of pointers.
26019 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
26020 (ipa_get_ipa_bits_for_value): Declare.
26021 * tree-vrp.h (value_range): Mark as GTY((for_user)).
26022 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
26023 (ipa_bits_hash_table): Likewise.
26024 (ipa_vr_ggc_hash_traits): Likewise.
26025 (ipa_vr_hash_table): Likewise.
26026 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
26027 being pointers and vr_known being removed.
26028 (ipa_set_jf_unknown): Likewise.
26029 (ipa_get_ipa_bits_for_value): New function.
26030 (ipa_set_jfunc_bits): Likewise.
26031 (ipa_get_value_range): New overloaded functions.
26032 (ipa_set_jfunc_vr): Likewise.
26033 (ipa_compute_jump_functions_for_edge): Use the above functions to
26034 construct bits and vr parts of jump functions.
26035 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
26036 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26038 (ipcp_grow_transformations_if_necessary): Also allocate
26039 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26041 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
26042 them. Fix too long lines.
26043 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
26044 vr_known being removed.
26045 (ipa_read_jump_function): Use new setter functions to construct bits
26046 and vr parts of jump functions or set them to NULL.
26047 (write_ipcp_transformation_info): Adjust for bits being pointers.
26048 (read_ipcp_transformation_info): Likewise.
26049 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
26051 Include gt-ipa-prop.h.
26052 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
26054 (ipcp_store_bits_results): Likewise.
26055 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
26056 Do not write to existing jump functions but use a temporary instead.
26058 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26061 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
26062 attempt to use its first operand as BIT_FIELD_REF base.
26064 2017-03-01 Richard Biener <rguenther@suse.de>
26066 PR middle-end/79721
26067 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
26068 interpolating formula in wrapping arithmetic.
26069 (chrec_apply): Convert chrec_evaluate return value to wanted type.
26071 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26073 PR tree-optimization/79734
26074 * tree-vect-generic.c (expand_vector_condition): Optimize
26075 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
26076 Handle VEC_COND_EXPR where comparison has different inner width from
26077 type's inner width.
26079 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
26081 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
26082 markup, and similar issues. Remove @opindex entries for things
26083 that aren't options. Add missing -mmpy-option entries.
26085 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26087 PR tree-optimization/79737
26088 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
26089 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
26090 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
26091 instead of byte_size. Formatting fix.
26092 (shift_bytes_in_array_right): Formatting fix.
26094 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26097 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26098 condition on optimize for the leaf function test.
26100 2017-02-28 Martin Liska <mliska@suse.cz>
26103 * read-rtl-function.c (function_reader::handle_unknown_directive):
26104 Bail out when one uses -flto.
26106 2017-02-28 Martin Liska <mliska@suse.cz>
26108 * common.opt: Replace space with tabular for options of <number>
26110 * config/i386/i386.opt: Show <number> value for
26111 -mlarge-data-threshold.
26112 * opts.c (print_filtered_help): Do not display number in hexadecimal
26115 2017-02-28 Martin Liska <mliska@suse.cz>
26117 * common.opt: Fix --help=option -Q for options which are of
26120 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26122 * config/i386/i386.c (print_reg): Error out for values
26123 of 8-bit size in invalid integer register.
26125 2017-02-28 Martin Sebor <msebor@redhat.com>
26127 PR tree-optimization/79691
26128 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26130 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26133 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26134 gcc_unreachable with output_operand_lossage.
26136 2017-02-28 Richard Biener <rguenther@suse.de>
26138 PR tree-optimization/79740
26139 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26141 (visit_nary_op): Insert the nary into the hashtable if we
26142 pattern-matched sth.
26143 * tree-ssa-pre.c (eliminate_insert): Robustify.
26145 2017-02-28 Richard Biener <rguenther@suse.de>
26147 PR middle-end/79731
26148 * fold-const.c (decode_field_reference): Reject out-of-bound
26151 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26153 * config/i386/i386.c: Include intl.h.
26154 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26155 instead of just cond ? "..." : "...".
26156 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26157 * coverage.c (read_counts_file): Likewise.
26158 * omp-offload.c: Include intl.h.
26159 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26160 of just cond ? "..." : "...".
26161 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26162 of just cond ? "..." : "...".
26164 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26167 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26169 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26170 'tune for' CPU name.
26171 * config/arm/arm-cpu-data.h: Regenerated.
26173 2017-02-28 Richard Biener <rguenther@suse.de>
26175 PR tree-optimization/79732
26176 * tree-inline.c (expand_call_inline): Do not shadow var.
26178 2017-02-28 Richard Biener <rguenther@suse.de>
26180 PR tree-optimization/79723
26181 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
26182 address-space properly.
26184 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
26186 * doc/optinfo.texi (Optimization groups): Fix option used for
26188 * doc/invoke.texi (-fopt-info): Document "omp".
26189 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
26190 (OPTGROUP_ALL): Add OPTGROUP_OMP.
26191 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
26192 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
26193 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
26195 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
26197 * dumpfile.c (optgroup_options): Instead of "openmp", associate
26198 OPTGROUP_OMP with "omp".
26200 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
26203 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
26204 for arithmetic shift of unsigned V2DI.
26206 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
26208 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
26209 arc/linux.h headers.
26210 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
26211 (LINK_SPEC): Likewise.
26212 (ARC_TLS_EXTRA_START_SPEC): Likewise.
26213 (EXTRA_SPECS): Likewise.
26214 (STARTFILE_SPEC): Likewise.
26215 (ENDFILE_SPEC): Likewise.
26216 (LIB_SPEC): Likewise.
26217 (TARGET_SDATA_DEFAULT): Likewise.
26218 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
26219 (MULTILIB_DEFAULTS): Likewise.
26220 (DWARF2_UNWIND_INFO): Likewise.
26221 * config/arc/big.h: New file.
26222 * config/arc/elf.h: Likewise.
26223 * config/arc/linux.h: Likewise.
26224 * config/arc/t-uClibc: Remove.
26226 2017-02-27 Bin Cheng <bin.cheng@arm.com>
26228 PR tree-optimization/77536
26229 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
26230 (tree_transform_and_unroll_loop): Use above function to compute the
26231 estimated niter of unrolled loop and use it when scaling profile.
26232 Also use count info rather than frequency if it's non-zero.
26233 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
26234 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
26235 (vect_transform_loop): Call above function.
26237 2017-02-27 Richard Biener <rguenther@suse.de>
26239 PR tree-optimization/45397
26240 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
26241 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
26242 (visit_nary_op): Add pattern matching for CSEing sign-changed
26243 or truncated operations with wider ones.
26245 2017-02-27 Richard Biener <rguenther@suse.de>
26247 PR tree-optimization/79690
26248 * tree-vect-stmts.c (vectorizable_store): Use vector type
26249 built from the DR with address-space.
26251 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
26253 * doc/invoke.texi (Optimize Options): Refine the description
26254 of asan-use-after-return.
26256 2017-02-25 Alan Modra <amodra@gmail.com>
26258 PR rtl-optimization/79584
26259 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
26260 base, not ad->base_term, the reg within base. Remove assertion
26261 that ad->base == ad->base_term. Replace gen_int_mode using
26262 bogus mode with const0_rtx.
26264 2017-02-25 Jakub Jelinek <jakub@redhat.com>
26266 PR middle-end/79396
26267 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
26268 FMA_EXPR like tcc_binary or tcc_unary.
26270 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
26273 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
26275 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
26276 (output_loc_operands): Handle DW_OP_call_ref and
26277 DW_OP_GNU_variable_value.
26278 (struct variable_value_struct): New type.
26279 (struct variable_value_hasher): Likewise.
26280 (variable_value_hash): New variable.
26281 (string_types): Remove.
26282 (copy_loc_descr): New function.
26283 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
26284 (prepend_loc_descr_to_each): New function.
26285 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
26286 instead of add_loc_descr_to_each if the first argument is single
26287 location list and the second has multiple.
26288 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
26289 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
26290 when looking for variable value which doesn't have other location info.
26291 (loc_list_from_tree): Formatting fix.
26292 (gen_array_type_die): Simplify DW_AT_string_length handling.
26293 (adjust_string_types): Remove.
26294 (gen_subprogram_die): Don't call adjust_string_types nor test/set
26295 string_types. Call resolve_variable_values.
26296 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
26297 (resolve_addr_in_expr): Likewise. Add A argument.
26298 (copy_deref_exprloc): Remove deref argument. Adjust for the
26299 original expression being DW_OP_GNU_variable_value with optionally
26300 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
26301 optionally after it.
26302 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
26303 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
26304 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
26305 (variable_value_hasher::hash, variable_value_hasher::equal): New
26307 (resolve_variable_value_in_expr, resolve_variable_value,
26308 resolve_variable_values, note_variable_value_in_expr,
26309 note_variable_value): New functions.
26310 (dwarf2out_early_finish): Call note_variable_value on all toplevel
26313 2017-02-24 Jakub Jelinek <jakub@redhat.com>
26316 * opts.h (handle_generated_option): Add GENERATED_P argument.
26317 * opts-common.c (handle_option): Adjust function comment.
26318 (handle_generated_option): Add GENERATED_P argument, pass it to
26320 (control_warning_option): Pass false to handle_generated_option
26322 * opts.c (maybe_default_option): Pass true to handle_generated_option
26324 * optc-gen.awk: Likewise.
26326 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26328 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
26329 a REG, look at the REG it is a SUBREG of.
26330 (splitter for cmpeqsi_t): Ditto.
26332 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26334 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
26335 the special USEs with the pattern of the insn, not the insn itself.
26337 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
26340 * doc/invoke.texi: Document -mload-store-pairs.
26342 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26343 Sandra Loosemore <sandra@codesourcery.com>
26345 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
26346 argument isn't a CONST_INT.
26347 (nios2_alternate_compare_const): Assert op is a CONST_INT.
26348 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
26349 (nios2_validate_compare): Bypass alternate compare logic if *op2
26350 is not a CONST_INT.
26351 (ldstwm_operation_p): Return false if first_base is not a REG or
26352 if first_offset is not a CONST_INT.
26354 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26356 * config/cris/cris.md: Use correct operand in a define_peephole2.
26358 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26360 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
26362 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26364 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
26365 this_insn if it is an INSN or JUMP_INSN.
26366 (force_offsettable): Look at base, not at addr.
26367 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
26368 on things that aren't necessarily CONST_INTs.
26370 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
26372 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
26373 -mfpmath=sse is the default also for x86-32 targets with SSE2
26374 instruction set when @option{-ffast-math} is enabled
26376 2017-02-24 Jeff Law <law@redhat.com>
26378 PR rtl-optimizatoin/79286
26379 * ira.c (update_equiv_regs): Drop may_trap_p exception to
26382 2017-02-24 Richard Biener <rguenther@suse.de>
26384 PR tree-optimization/79389
26385 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
26388 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
26390 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
26391 function comment to reflect reality.
26392 (loop_exits_before_overflow): Fix typo in function description.
26394 2017-02-24 Richard Biener <rguenther@suse.de>
26396 PR tree-optimization/79389
26397 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
26398 properly that a threading opportunity exists. Detect conditional
26399 copy/constant propagation opportunities.
26401 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
26403 * config/visium/visium.md (type): Add trap.
26404 (b): New mode attribute.
26405 (*btst): Rename into...
26406 (*btst<mode>): ...this and adjust.
26407 (*cbranchsi4_btst_insn): Rename into...
26408 (*cbranch<mode>4_btst_insn): ...this and adjust.
26409 (trap): New define_insn.
26411 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26413 PR tree-optimization/79389
26414 * ifcvt.c (struct noce_if_info): Add rev_cond field.
26415 (noce_reversed_cond_code): New function.
26416 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
26417 reversed_comparison_code. Formatting fix.
26418 (noce_try_store_flag): Test rev_cond != NULL in addition to
26419 reversed_comparison_code.
26420 (noce_try_store_flag_constants): Likewise.
26421 (noce_try_store_flag_mask): Likewise.
26422 (noce_try_addcc): Use rev_cond if non-NULL instead of
26423 reversed_comparison_code.
26424 (noce_try_cmove_arith): Likewise. Formatting fixes.
26425 (noce_try_minmax, noce_try_abs): Clear rev_cond.
26426 (noce_find_if_block): Initialize rev_cond.
26427 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
26428 instead of false as last argument never attempt to reverse it
26431 2017-02-23 Bin Cheng <bin.cheng@arm.com>
26433 PR tree-optimization/79663
26434 * tree-predcom.c (combine_chains): Process refs in reverse order
26435 only for ZERO length chains, and add explaining comment.
26437 2017-02-23 Jeff Law <law@redhat.com>
26439 PR tree-optimization/79578
26440 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
26441 in call to operand_equal_p.
26443 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
26446 * config/i386/cpuid.h: Fix another undefined behavior.
26448 2017-02-23 Richard Biener <rguenther@suse.de>
26450 PR tree-optimization/79683
26451 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
26452 vector types for data-refs.
26454 2017-02-23 Martin Liska <mliska@suse.cz>
26456 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
26458 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26460 PR middle-end/79665
26461 * internal-fn.c (get_range_pos_neg): Moved to ...
26462 * tree.c (get_range_pos_neg): ... here. No longer static.
26463 * tree.h (get_range_pos_neg): New prototype.
26464 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
26465 are known to be in between 0 and signed maximum inclusive, try to
26466 expand both unsigned and signed divmod and use the cheaper one from
26469 2017-02-22 Jeff Law <law@redhat.com>
26471 PR tree-optimization/79578
26472 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
26473 to compare base operands.
26475 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
26478 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
26479 gpc_reg_operand instead of fpr_reg_operand.
26481 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
26483 * config/mips/mips.c (mips_return_in_memory): Force FP
26484 vector types to be returned in memory for o32 ABI.
26486 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26488 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
26489 instead of DW_TAG_member for static data member declarations and don't
26490 set no_linkage_name for static inline data members.
26491 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
26494 2017-02-22 Martin Liska <mliska@suse.cz>
26496 * doc/invoke.texi: Replace inequality signs with square brackets
26499 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26501 PR tree-optimization/68644
26502 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
26504 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
26507 * lra-constraints.c (simplify_operand_subreg): Handle
26508 WORD_REGISTER_OPERATIONS targets.
26510 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26513 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
26514 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
26515 elimination by swapping fld*.
26517 2017-02-22 Richard Biener <rguenther@suse.de>
26519 PR tree-optimization/79673
26520 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
26521 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
26522 irrelevant address-space qualifiers and avoiding a
26523 ADDR_SPACE_CONVERT_EXPR from fold_convert.
26525 2017-02-22 Richard Biener <rguenther@suse.de>
26527 PR tree-optimization/79666
26528 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
26529 to not symbolically negate if that may introduce undefined
26532 2017-02-22 Martin Liska <mliska@suse.cz>
26535 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
26536 * data-streamer-out.c (streamer_write_gcov_count_stream):
26538 * value-prof.c (stream_out_histogram_value): Make assert more
26539 precise based on type of counter.
26541 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
26544 * config/i386/i386.md (standard_x87sse_constant_load splitter):
26545 Use nonimmediate_operand instead of memory_operand for operand 1.
26546 (float-extend standard_x87sse_constant_load splitter): Ditto.
26548 2017-02-21 Jeff Law <law@redhat.com>
26550 PR tree-optimization/79621
26551 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
26552 blocks with edges to themselves.
26554 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26557 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
26558 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
26559 Use gimple_call_builtin_p.
26562 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
26563 on temporarily removed DEBUG_INSNs.
26565 PR tree-optimization/79649
26566 * tree-loop-distribution.c (classify_partition): Give up on
26567 non-generic address space loads/stores.
26569 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
26571 * doc/loop.texi (Loop manipulation): Remove nonexistent
26572 tree_ssa_loop_version from the documentation.
26573 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
26575 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26578 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
26579 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
26580 * config/rs6000/rs6000.c: Include except.h.
26581 (rs6000_expand_split_stack_prologue): Call
26582 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
26584 2017-02-21 Martin Jambor <mjambor@suse.cz>
26587 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
26588 have been analyzed.
26590 2017-02-21 Martin Jambor <mjambor@suse.cz>
26592 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
26593 for backward compatibility only.
26594 * doc/invoke.texi (Option Summary): Remove all references to
26595 -fipa-cp-alignment.
26597 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
26601 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26603 * lra-constraints.c (curr_insn_transform): Handle
26604 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26606 2017-02-21 Martin Liska <mliska@suse.cz>
26608 * config/i386/i386.opt: Replace -masm-dialect with -masm.
26610 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
26612 PR translation/79638
26613 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
26615 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
26618 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
26619 (arm_function_ok_for_sibcall): Return false for an indirect call by
26620 descriptor if all the argument registers are used.
26621 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
26622 alignment of the function.
26624 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26626 PR tree-optimization/61441
26627 * simplify-rtx.c (simplify_const_unary_operation): For
26628 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
26629 the sNaN unmodified.
26631 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26633 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26634 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26635 instead of SYSTEM_HEADER_DIR.
26637 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
26638 Martin Liška <mliska@suse.cz>
26640 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
26641 Fix typos and grammar, use active voice, and clarify.
26643 2017-02-20 Marek Polacek <polacek@redhat.com>
26645 PR middle-end/79537
26646 * gimplify.c (gimplify_expr): Handle unused *&&L;.
26649 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
26651 2017-02-20 Jakub Jelinek <jakub@redhat.com>
26654 * config/i386/i386.c (ix86_expand_builtin): Handle
26655 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
26656 ix86_builtins_isa[fcode].isa as a requirement of those
26657 flags and any other flag in the bitmask.
26658 (ix86_init_mmx_sse_builtins): Use 0 instead of
26659 ~OPTION_MASK_ISA_64BIT as mask.
26660 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
26661 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
26662 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
26663 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
26665 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26668 * lra-constraints.c (split_reg): Check requested split mode
26669 is supported by the register.
26671 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26673 * lra-constraints.c (simplify_operand_subreg): Remove early
26676 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26679 * lra-constraints.c (curr_insn_transform): Tighten condition
26680 for converting SUBREG reloads from OP_OUT to OP_INOUT.
26682 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26685 * lra-constraints.c (curr_insn_transform): Handle
26686 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26688 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
26691 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
26693 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
26695 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
26698 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
26701 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26703 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
26706 2017-02-18 Jakub Jelinek <jakub@redhat.com>
26708 * final.c (last_columnnum, override_columnnum): New variables.
26709 (final_start_function): Set last_columnnum, pass it to begin_prologue
26710 hook and pass 0 to dwarf2out_begin_prologue.
26711 (final_scan_insn): Update override_columnnum. Pass last_columnnum
26712 to source_line debug hook.
26713 (notice_source_line): Compute last_columnnum and for debug_column_info
26714 return true on column changes.
26715 * debug.h (struct gcc_debug_hooks): Add column argument to
26716 source_line and begin_prologue hooks.
26717 (debug_nothing_int_charstar_int_bool): Remove prototype.
26718 (debug_nothing_int_int_charstar,
26719 debug_nothing_int_int_charstar_int_bool): New prototypes.
26720 (dwarf2out_begin_prologue): Add column argument.
26721 * debug.c (do_nothing_debug_hooks): Adjust source_line and
26722 begin_prologue hooks.
26723 (debug_nothing_int_charstar_int_bool): Remove.
26724 (debug_nothing_int_int_charstar,
26725 debug_nothing_int_int_charstar_int_bool): New functions.
26726 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
26727 through to dwarf2out_source_line.
26728 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
26729 (dwarf2out_source_line): Add column argument, emit it if requested.
26730 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
26732 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26733 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26734 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
26735 through to dwarf2out_begin_prologue.
26736 (vmsdbgout_source_line): Add column argument, pass it through to
26737 dwarf2out_source_line.
26738 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
26739 dbxout_source_line caller.
26740 (dbxout_source_line): Add column argument.
26742 * common.opt (gno-column-info, gcolumn-info): New options.
26743 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
26744 (check_die): Also test for multiple DW_AT_decl_column attributes.
26745 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
26746 DW_AT_decl_column if requested.
26747 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
26749 (gen_variable_die): Likewise.
26750 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
26751 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
26754 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
26755 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
26756 (ix86_handle_option): Handle OPT_m3dnowa.
26757 * doc/invoke.texi (-m3dnowa): Document.
26758 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
26759 -m3dnowa instead of -m3dnow -march=athlon.
26762 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
26763 instead of gcc_assert for K, r and R code checks. Formatting fixes.
26765 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26768 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
26769 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
26770 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
26771 generator for vsx_xxpermdi_<mode>_be.
26772 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
26773 force big-endian semantics.
26774 (vsx_xxpermdi_<mode>_be): New define_expand with same
26775 implementation as previous version of vsx_xxpermdi_<mode>.
26777 2017-02-17 Jakub Jelinek <jakub@redhat.com>
26779 PR tree-optimization/79327
26780 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
26781 variable, its initialization and use.
26783 2017-02-17 Julia Koval <julia.koval@intel.com>
26785 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
26786 (OPTION_MASK_ISA_PKU_UNSET): New.
26787 (ix86_handle_option): Handle -mrdpid.
26788 * config/i386/cpuid.h (bit_RDPID): New.
26789 * config/i386/driver-i386.c (host_detect_local_cpu):
26790 Detect RDPID feature.
26791 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
26792 * config/i386/i386-c.c (ix86_target_macros_internal):
26794 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
26795 (ix86_valid_target_attribute_inner_p): Add "rdpid".
26796 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26797 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
26798 * config/i386/i386.md (define_insn "rdpid"): New.
26799 * config/i386/i386.opt Add -mrdpid.
26800 * config/i386/immintrin.h (_rdpid_u32): New.
26802 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
26804 PR rtl-optimization/79541
26805 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
26806 instead of transforming it into USE.
26808 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
26810 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
26811 If HONOR_SNANS (SFmode) force the input to a register.
26812 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
26813 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
26814 an frsp or similar insn.
26816 2017-02-17 Martin Liska <mliska@suse.cz>
26818 PR rtl-optimization/79577
26819 * params.def (selsched-max-sched-times): Increase minimum to 1.
26821 2017-02-17 Martin Liska <mliska@suse.cz>
26823 PR rtl-optimization/79574
26824 * gcse.c (want_to_gcse_p): Prevent integer overflow.
26826 2017-02-17 Martin Liska <mliska@suse.cz>
26828 PR tree-optimization/79529
26829 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
26830 ssa_defined_default_def_p to handle cases which are implicitly
26832 * tree-ssa.c (ssa_defined_default_def_p): New function.
26833 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
26834 which are implicitly defined.
26835 * tree-ssa.h (ssa_defined_default_def_p): Declare.
26837 2017-02-17 Richard Biener <rguenther@suse.de>
26839 PR middle-end/79576
26840 * params.def (max-ssa-name-query-depth): Limit to 10.
26842 2017-02-17 Richard Biener <rguenther@suse.de>
26844 PR tree-optimization/79552
26845 * tree-ssa-structalias.c (visit_loadstore): Properly verify
26848 2017-02-17 Richard Biener <rguenther@suse.de>
26851 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
26853 2017-02-17 Marek Polacek <polacek@redhat.com>
26855 PR middle-end/79536
26856 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
26857 (fold_negate_expr): New wrapper.
26859 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
26861 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
26862 Correct terminology and de-emphasize pre-standard behavior.
26864 2017-02-16 Alan Modra <amodra@gmail.com>
26866 PR rtl-optimization/79286
26867 * ira.c (def_dominates_uses): New function.
26868 (update_equiv_regs): Don't create an equivalence for insns that
26869 may trap where the register def does not dominate the use.
26871 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
26873 PR rtl-optimization/78127
26874 * lra.c (lra): Call lra_eliminate before finish the loop after
26877 2017-02-16 Richard Biener <rguenther@suse.de>
26879 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
26881 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
26882 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
26883 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
26884 (isl_val_int_from_wi): New function.
26885 (extract_affine_gmp): Rename to ...
26886 (extract_affine_wi): ... this, take a widest_int.
26887 (extract_affine_int): Just wrap extract_affine_wi.
26888 (add_param_constraints): Use isl_val_int_from_wi.
26889 (add_loop_constraints): Likewise, and extract_affine_wi.
26891 2017-02-15 Jeff Law <law@redhat.com>
26893 PR middle-end/79521
26894 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
26895 ira_init_register_move_cost_if_necessary.
26897 2017-02-15 Martin Sebor <msebor@redhat.com>
26899 PR middle-end/32003
26900 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
26901 removed in a prior commit.
26903 2017-02-15 Bin Cheng <bin.cheng@arm.com>
26905 PR tree-optimization/79347
26906 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
26907 counters during peeling.
26909 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
26911 * Makefile.in (site.exp): Remove "set ISLVER".
26913 2017-02-15 Jakub Jelinek <jakub@redhat.com>
26916 * real.c (real_from_integer): Call real_convert even for decimal.
26918 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
26921 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
26923 2017-02-14 Andrew Pinski <apinski@cavium.com>
26925 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
26926 cores and change the partno/implementer to be correct.
26927 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
26928 the 'B" as the implementer.
26929 * config/aarch64/aarch64-tune.md: Regenerate.
26931 2017-02-14 Carl Love <cel@us.ibm.com>
26933 * config/rs6000/rs6000.c: Add case statement entry to make the
26934 xvcvuxdsp built-in argument unsigned.
26935 * config/rs6000/vsx.md: Fix the source and return operand types so they
26936 match the instruction definitions from the ISA document. Fix typo
26937 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
26940 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
26943 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
26944 member early_clobber_alts.
26945 * lra-lives.c (reg_early_clobber_p): New.
26946 (process_bb_lives): Use it.
26947 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
26948 (debug_operand_data): Initialize early_clobber_alts.
26949 (setup_operand_alternative): Set up early_clobber_alts.
26950 (collect_non_operand_hard_regs): Ditto. Pass early clobber
26951 alternatives to new_insn_reg.
26952 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
26954 (lra_update_insn_regno_info): Pass the new arg.
26956 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26958 PR middle-end/79505
26959 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
26960 (new_oacc_loop_raw): Don't clear already cleared fields.
26963 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
26964 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
26965 _mm512_prefetch_i64gather_ps): New inline functions and macros.
26967 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
26970 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
26972 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
26975 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
26976 the extra instruction to the right place to store 128-bit constant
26979 2017-02-14 Martin Sebor <msebor@redhat.com>
26981 PR middle-end/79448
26982 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
26983 warning for strings of unknown length.
26985 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26987 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
26989 2017-02-14 Jeff Law <law@redhat.com>
26992 * ira-costs.c (scan_one_insn): Initialize register move costs
26993 for pseudos seen in USE/CLOBBER insns.
26995 PR tree-optimization/79095
26996 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
26997 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
26998 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
26999 if the operands are known to be not equal, then the resulting range
27001 (intersect_ranges): If the new range is ~[0,0] and the old range is
27002 wide, then prefer ~[0,0].
27003 * tree-vrp.c (overflow_comparison_p_1): New function.
27004 (overflow_comparison_p): New function.
27005 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
27006 if NAME is used in an overflow test.
27007 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
27008 overflow check that can be expressed as an equality test, then adjust
27009 ops to be that equality test.
27011 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27013 * config/s390/s390-builtin-types.def: Remove flags argument.
27014 * config/s390/s390.c (s390_init_builtins): Likewise.
27016 2017-02-14 Martin Liska <mliska@suse.cz>
27018 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
27019 vector. Fix trailing white spaces.
27021 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
27023 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
27026 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27028 PR rtl-optimization/68664
27029 * config/arm/arm.c (arm_sched_can_speculate_insn):
27030 New function. Declare prototype.
27031 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27033 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27035 PR rtl-optimization/68664
27036 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
27038 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27040 2017-02-14 Amit Pawar <amit.pawar@amd.com>
27042 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
27043 max skip bytes for function, loop and jump.
27045 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27047 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
27048 ABS_EXPR for gimple dump.
27050 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27053 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
27055 PR tree-optimization/79408
27056 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
27057 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
27058 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
27059 also if rhs1 is INTEGER_CST.
27061 2017-02-14 Richard Biener <rguenther@suse.de>
27063 PR middle-end/79432
27064 * tree-into-ssa.c (insert_phi_nodes): When the function can
27065 have abnormal edges rewrite SSA names with broken use-def
27066 dominance out of SSA and register them for PHI insertion.
27068 2017-02-13 Martin Sebor <msebor@redhat.com>
27070 PR middle-end/79496
27071 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
27072 clearing info.nowrite flag when snprintf size argument is a range.
27074 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27076 * cprop.c (cprop_jump): Add missing space in string literal.
27077 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
27078 (get_constraint_for_component_ref): Likewise.
27079 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
27080 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
27081 * lra-constraints.c (process_alt_operands): Likewise.
27082 * ipa-inline.c (inline_small_functions): Likewise.
27083 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
27084 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27085 * trans-mem.c (diagnose_tm_1_op): Likewise.
27086 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
27087 (grid_parallel_clauses_gridifiable): Likewise.
27089 * config/nvptx/mkoffload.c (process): Add space in between
27092 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
27093 "MOD4_SSE_REGS" and "ALL_REGS".
27095 * spellcheck.c (test_data): Add , in between "foo" and "food".
27097 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27100 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27101 boundary crossing check and subsequent code generation agree.
27103 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27105 * config/aarch64/aarch64.c (has_memory_op): Delete.
27106 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27109 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27111 PR rtl-optimization/79388
27112 PR rtl-optimization/79450
27113 * combine.c (distribute_notes): When removing TEM_INSN for which
27114 corresponding dest has last value recorded, invalidate that last
27117 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27119 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27120 of explicit '@'. Add missing assembly comment marker on branch costs
27123 2017-02-13 Nathan Sidwell <nathan@acm.org>
27125 * gengtype-lex.l (<in_struct>): Add '/'.
27127 2017-02-13 Martin Liska <mliska@suse.cz>
27130 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27132 2017-02-13 Richard Biener <rguenther@suse.de>
27134 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27136 * configure: Re-generate.
27137 * config.in: Likewise.
27138 * graphite-dependences.c: Simplify as if
27139 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27140 * graphite-isl-ast-to-gimple.c: Likewise.
27141 * graphite-optimize-isl.c: Likewise.
27142 * graphite-poly.c: Likewise.
27143 * graphite-sese-to-poly.c: Likewise.
27144 * graphite.h: Likewise.
27145 * toplev.c: Include isl/version.h and use isl_version () for
27146 printing the ISL version.
27147 * doc/install.texi: Update ISL requirement.
27149 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27151 * doc/standards.texi (Standards): Update reference to
27154 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27156 * doc/extend.texi (Named Address Spaces): sourceware.org now
27158 * doc/install.texi (Binaries): Ditto.
27161 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27163 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27164 standard terminology "stringize"/"stringizing" throughout.
27165 * doc/cppinternals.texi: Likewise.
27167 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27169 * doc/extend.texi: Fix some spelling mistakes and typos.
27170 * doc/invoke.texi: Likewise.
27172 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27175 * params.def (inline-min-speedup) Change from 10 to 8.
27177 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27179 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
27182 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27185 * ipa-inline-analysis.c (get_minimal_bb): New function.
27186 (record_modified): Use it.
27187 (remap_edge_change_prob): Handle also ancestor functions.
27189 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
27191 * doc/contrib.texi (Contributors): Remove broken link into
27192 the Mauve CVS repository.
27194 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27196 PR middle-end/79454
27197 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
27198 result computation whenever lhs doesn't have vector mode, not
27199 just when it has BLKmode.
27201 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27203 * doc/makefile.texi (profiledbootstrap): Refer to the
27204 installation instructions only in textual form.
27206 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27209 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
27211 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27213 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
27214 (Specific): Update mingw-w64 reference.
27216 (Specific): Remove broken link to Renesas RX processor.
27218 2017-02-10 Richard Biener <rguenther@suse.de>
27220 * toplev.c (process_options): Do not mention obsolete graphite
27221 options when printing sorry message about missing graphite support.
27222 Mention -floop-nest-optimize.
27224 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
27226 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
27227 (vtst_p16): Likewise.
27228 (vtstq_p8): Likewise.
27229 (vtstq_p16): Likewise.
27231 (vtstq_p64): Likewise.
27232 * config/arm/arm_neon.h (vgetq_lane_p64): New.
27233 (vset_lane_p64): New.
27234 (vsetq_lane_p64): New.
27236 2017-02-10 Jakub Jelinek <jakub@redhat.com>
27238 PR tree-optimization/79411
27239 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
27240 stmt operands are SSA_NAMEs used in abnormal phis.
27241 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
27244 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27247 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
27250 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27252 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
27254 2017-02-09 Jakub Jelinek <jakub@redhat.com>
27256 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
27260 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
27261 not arbitrary TREE_CONSTANT.
27264 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
27265 "omp declare target link" attribute unless is_global_var.
27266 * omp-offload.c (find_link_var_op): Likewise.
27268 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
27269 Chung-Lin Tang <cltang@codesourcery.com>
27271 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
27273 (gimplify_adjust_omp_clauses): Don't delete TILE.
27274 (gimplify_omp_for): Deal with TILE.
27275 * internal-fn.c (expand_GOACC_TILE): New function.
27276 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
27278 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
27279 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
27281 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
27282 avoid DIV for outermost collapse var.
27283 (expand_oacc_for): Insert tile element loop as needed. Adjust.
27284 Remove out of date comments, fix whitespace.
27285 * omp-general.c (omp_extract_for_data): Deal with tiling.
27286 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
27287 adjust OLF_DIM_BASE value.
27288 (struct omp_for_data): Add tiling field.
27289 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
27290 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
27291 for auto loops. Remove default auto determining, moved to
27292 oacc_loop_fixed_partitions.
27293 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
27294 stmts, add e_mask field.
27295 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
27296 (oacc_thread_numbers): Use oacc_dim_call.
27297 (oacc_xform_tile): New.
27298 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
27299 (finish_oacc_loop): Adjust for ifns vector.
27300 (oacc_loop_discover_walk): Append loop abstraction sites to list,
27301 add case for GOACC_TILE fns.
27302 (oacc_loop_xform_loop): Delete.
27303 (oacc_loop_process): Iterate over call list directly, and add
27304 handling for GOACC_TILE fns.
27305 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
27307 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
27308 vector partitioning to outer loops. Assign 2 partitions to loops
27309 when available. Add TILE handling.
27310 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
27311 (execite_oacc_device_lower): Process GOACC_TILE fns,
27312 ignore unknown specs.
27313 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
27314 * tree.c (omp_clause_num_ops): Adjust TILE ops.
27315 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
27317 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
27319 * configure.ac (ACX_BUGURL): Update.
27320 * configure: Regenerate.
27322 2017-02-09 Richard Biener <rguenther@suse.de>
27324 PR tree-optimization/69823
27325 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27326 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
27328 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27330 * config/arc/arc-c.def: Add __NPS400__ definition.
27331 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
27332 (TARGET_NPS400): Define.
27334 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27336 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
27338 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
27339 pointer, arch_info.
27340 (arc_cpu_types): Fill the arch_info field with a pointer into the
27341 arc_arch_types table.
27342 (arc_selected_cpu): Declare.
27343 * config/arc/arc.c (arc_selected_cpu): Make global.
27344 (arc_selected_arch): Delete.
27345 (arc_base_cpu): Delete.
27346 (arc_override_options): Remove references to deleted variables,
27347 update access to arch information.
27348 (ARC_OPT): Update access to arch information.
27349 (ARC_OPTX): Likewise.
27350 * config/arc/arc.h (arc_base_cpu): Remove declaration.
27351 (TARGET_ARC600): Update access to arch information.
27352 (TARGET_ARC601): Likewise.
27353 (TARGET_ARC700): Likewise.
27354 (TARGET_EM): Likewise.
27355 (TARGET_HS): Likewise.
27356 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
27359 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
27362 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
27363 condition/operands for integer GE/LE/GEU/LEU operations.
27365 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
27367 PR translation/79397
27368 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
27371 2017-02-08 Martin Jambor <mjambor@suse.cz>
27374 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
27375 whether allocation happened.
27376 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
27377 nothing was allocated.
27379 2017-02-08 Jakub Jelinek <jakub@redhat.com>
27381 PR tree-optimization/79408
27382 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
27383 constant, but SSA_NAME with a known integer range, use the minimum
27384 of that range instead of op1 to determine if modulo can be replaced
27385 with its first operand.
27387 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27389 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
27391 2017-02-08 Richard Biener <rguenther@suse.de>
27393 PR tree-optimization/71824
27394 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27395 Check all loops contained in the merged region.
27397 2017-02-07 Andrew Pinski <apinski@cavium.com>
27399 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
27401 2017-02-07 Andrew Pinski <apinski@cavium.com>
27403 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
27404 (thunderxt88): Likewise.
27405 (thunderxt81): Disable LSE and change v8.1 to v8.
27406 (thunderxt83): Likewise.
27408 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27409 Richard Biener <rguenther@suse.de>
27411 PR middle-end/79399
27412 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
27413 type from int to size_t.
27414 * ira-costs.c (struct_costs_size): Change type from int to size_t.
27416 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27418 PR rtl-optimization/79386
27419 * cprop.c (bypass_conditional_jumps): Initialize
27420 bypass_last_basic_block already before splitting bbs after
27421 unconditional traps...
27422 (bypass_conditional_jumps): ... rather than here.
27425 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
27426 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
27427 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
27428 fix -masm=intel patterns.
27430 2017-02-07 Richard Biener <rguenther@suse.de>
27432 PR tree-optimization/79256
27433 PR middle-end/79278
27434 * builtins.c (get_object_alignment_2): Use min_align_of_type
27435 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
27436 and ADJUST_FIELD_ALIGN.
27438 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
27440 * doc/tm.texi: Regenerate.
27441 * stor-layout.c (layout_decl): Adjust.
27442 (update_alignment_for_field): Likewise.
27443 (place_field): Likewise.
27444 (min_align_of_type): Likewise.
27445 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
27446 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
27447 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
27448 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
27449 * config/frv/frv.c (frv_adjust_field_align): Likewise.
27450 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
27451 * config/i386/i386.c (x86_field_alignment): Likewise.
27452 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
27453 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
27454 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
27455 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
27456 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
27457 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
27461 2017-01-30 Richard Biener <rguenther@suse.de>
27463 PR tree-optimization/79256
27464 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27465 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27468 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
27470 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
27471 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
27472 builtins to SImode and emit a zero-extend, if necessary.
27474 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27476 * docs/invoke.texi (RISC-V Options): Alphabetize.
27478 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27480 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
27483 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27485 * config/riscv/riscv.c: New file.
27486 * common/config/riscv/riscv-common.c: Likewise.
27487 * config.gcc: Likewise.
27488 * config/riscv/constraints.md: Likewise.
27489 * config/riscv/elf.h: Likewise.
27490 * config/riscv/generic.md: Likewise.
27491 * config/riscv/linux.h: Likewise.
27492 * config/riscv/multilib-generator: Likewise.
27493 * config/riscv/peephole.md: Likewise.
27494 * config/riscv/pic.md: Likewise.
27495 * config/riscv/predicates.md: Likewise.
27496 * config/riscv/riscv-builtins.c: Likewise.
27497 * config/riscv/riscv-c.c: Likewise.
27498 * config/riscv/riscv-ftypes.def: Likewise.
27499 * config/riscv/riscv-modes.def: Likewise.
27500 * config/riscv/riscv-opts.h: Likewise.
27501 * config/riscv/riscv-protos.h: Likewise.
27502 * config/riscv/riscv.h: Likewise.
27503 * config/riscv/riscv.md: Likewise.
27504 * config/riscv/riscv.opt: Likewise.
27505 * config/riscv/sync.md: Likewise.
27506 * config/riscv/t-elf-multilib: Likewise.
27507 * config/riscv/t-linux: Likewise.
27508 * config/riscv/t-linux-multilib: Likewise.
27509 * config/riscv/t-riscv: Likewise.
27510 * configure.ac: Likewise.
27511 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
27512 Waterman as RISC-V maintainers.
27513 * doc/install.texi: Add RISC-V entries.
27514 * doc/invoke.texi: Add RISC-V options section.
27515 * doc/md.texi: Add RISC-V constraints section.
27516 * configure: Regenerated.
27518 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27521 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
27522 false values to be constant vectors with all 0 or all 1 bits set.
27523 (vcondu<mode><mode>): Likewise.
27524 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
27526 (fpmask_comparison_operator): Update comment.
27527 (vecint_comparison_operator): New predicate.
27528 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
27529 vector conditionals when the true and false values are constant
27530 vectors with all 0 bits or all 1 bits set.
27532 2017-02-06 Martin Sebor <msebor@redhat.com>
27534 PR tree-optimization/79376
27535 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
27537 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
27539 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
27540 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
27541 to simplify split condition.
27543 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27545 * omp-expand.c (oxpand_omp_atomic_fetch_op,
27546 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
27549 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
27551 PR rtl-optimization/68664
27552 * target.def (can_speculate_insn): New hook.
27553 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
27554 * doc/tm.texi: Regenerate.
27555 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
27556 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
27557 (rs6000_sched_can_speculate_insn): New function.
27559 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27561 PR tree-optimization/79284
27562 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
27563 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
27564 vectorizable_mask_load_store, vectorizable_operation,
27565 vect_is_simple_cond, get_same_sized_vectype): Use it instead
27566 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
27567 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
27568 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
27569 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
27570 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27571 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
27572 is_gimple_assign (stmt). Replace another such test with
27573 is_gimple_assign (stmt).
27575 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27578 * config/avr/avr.c (rtl-iter.h): Include it.
27579 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
27580 (avr_legitimate_combined_insn): ...and implementation.
27582 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
27584 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
27585 * config/s390/s390.c (s390_const_operand_ok)
27586 (s390_canonicalize_comparison, s390_extract_part)
27587 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
27588 (s390_contiguous_bitmask_p, s390_rtx_costs)
27589 (legitimize_pic_address): Likewise.
27590 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
27591 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
27592 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
27593 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
27594 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
27596 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27598 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
27599 REGNO($0) == REGNO($1).
27601 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27603 * config/s390/linux.h(SIZE_TYPE): Add comment.
27605 2017-02-06 Julian Brown <julian@codesourcery.com>
27606 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27607 Virendra Pathak <virendra.pathak@broadcom.com>
27609 * config/aarch64/aarch64-cores.def: Change the scheduler
27611 * config/aarch64/aarch64.md: Include thunderx2t99.md.
27612 * config/aarch64/thunderx2t99.md: New file.
27614 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27616 * doc/standards.texi (Go Language): Update link to language
27619 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27621 * tree-eh.c (lower_resx): Sanitize profile.
27622 (cleanup_empty_eh_move_lp): Likewise.
27624 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27627 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
27629 * cfgloopmanip.h (loop_version): Update prototype.
27630 * modulo-sched.c (sms_schedule): Update call of loop_version.
27631 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
27632 * tree-parloops.c (gen_parallel_loop): Likewise.
27633 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
27634 * tree-ssa-loop-split.c (split_loop): Likewise.
27635 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
27636 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
27638 2017-02-05 Martin Liska <mliska@suse.cz>
27641 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
27643 (print_operand_address): Initialize a struct to zero.
27645 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27647 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
27648 garbage collector only in textual form.
27650 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27652 * doc/extend.texi (x86 specific memory model extensions for
27653 transactional memory): Simplify a phrase.
27655 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
27658 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
27659 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
27660 (atomic_storedi_1): Likewise.
27662 2017-02-04 Jakub Jelinek <jakub@redhat.com>
27664 PR tree-optimization/79338
27665 * tree-parloops.c (gather_scalar_reductions): Don't call
27666 vect_analyze_loop_form for loop->inner before destroying loop's
27669 2017-02-03 Martin Sebor <msebor@redhat.com>
27671 PR tree-optimization/79327
27672 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
27673 when precision has resulted in leading zeros.
27674 (format_integer): Adjust the likely counter to assume an unknown
27675 argument that may be zero is non-zero.
27677 2017-02-03 Jason Merrill <jason@redhat.com>
27680 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
27681 avoid copying non-taken branch.
27683 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27685 PR tree-optimization/79340
27686 * tree-vect-loop.c (vectorizable_reduction): Release
27687 vec_defs elements after safe_splicing them into other vectors.
27690 PR tree-optimization/79327
27691 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
27692 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
27694 (format_integer): Use wide_int_to_tree instead of build_int_cst
27695 + to_?hwi. If argmin is NULL, just set argmin and argmax to
27696 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
27697 of shortest and longest sequence.
27699 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
27701 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
27702 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
27704 2017-02-03 Walter Lee <walt@tilera.com>
27707 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
27708 after initial stackframe link reg save.
27709 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
27711 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27714 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
27715 wu for stxssp alternative.
27717 2017-02-03 Martin Sebor <msebor@redhat.com>
27719 PR tree-optimization/79352
27720 * gimple-fold.c (get_range_strlen): Add argument.
27721 (get_range_strlen): Change return type to bool.
27722 (get_maxval_strlen): Pass in a dummy argument.
27723 * gimple-fold.h (get_range_strlen): Change return type to bool.
27724 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
27725 * tree.h (array_at_struct_end_p): Add argument.
27726 * tree.c (array_at_struct_end_p): Handle it.
27728 2017-02-03 Martin Liska <mliska@suse.cz>
27731 * multiple_target.c (create_dispatcher_calls): Redirect edge
27732 from a caller of a dispatcher.
27733 (expand_target_clones): Make the clones local.
27734 (ipa_target_clone): Do both target clones and resolvers.
27735 (ipa_dispatcher_calls): Remove the pass.
27736 (pass_dispatcher_calls::gate): Likewise.
27737 (make_pass_dispatcher_calls): Likewise.
27738 * passes.def (pass_target_clone): Put as very first IPA early
27741 2017-02-03 Martin Liska <mliska@suse.cz>
27743 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
27744 in case of a function with ifunc attribute.
27746 2017-02-03 Martin Liska <mliska@suse.cz>
27748 * cgraph.c (cgraph_node::dump): Dump function version info.
27749 * symtab.c (symtab_node::dump_base): Add missing new line.
27751 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27753 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
27754 (ifcombine_ifandif): Use it.
27756 2017-02-03 Martin Liska <mliska@suse.cz>
27758 * doc/invoke.texi: Document default value for
27759 use-after-scope-direct-emission-threshold.
27761 2017-02-03 Martin Liska <mliska@suse.cz>
27763 PR tree-optimization/79339
27764 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
27765 (format_floating): Likewise.
27767 2017-02-03 Martin Liska <mliska@suse.cz>
27770 * ipa-prop.c (ipa_node_params_t::insert): Remove current
27772 (ipa_node_params_t::remove): Likewise.
27773 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
27774 initialization from removed ipa_node_params_t::insert.
27775 (ipa_node_params::~ipa_node_params): Move from removed
27776 ipa_node_params_t::release.
27777 * symbol-summary.h (symbol_summary::m_released): New member.
27778 Do not release a summary twice. Do not allow to call finalizer
27779 for types of a summary that live in GGC memory.
27781 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27783 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
27786 2017-02-02 Martin Sebor <msebor@redhat.com>
27788 PR middle-end/79275
27789 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
27790 (format_string): Tighten up the range of output for non-constant
27791 strings and correct the expected range for wide non-constant strings.
27793 2017-02-02 Martin Sebor <msebor@redhat.com>
27795 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
27797 PR middle-end/32003
27798 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
27800 (-fdump-tree-@var): Add to index and document how to come up
27801 with pass-specific option and dump file names.
27802 (-fdump-passes): Clarify where to look for output.
27804 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27806 PR middle-end/77445
27807 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
27808 statistics of the analyzed path; allow threading for speed when
27809 any of BBs along the path are optimized for speed.
27811 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
27813 PR middle-end/78468
27814 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
27815 settings of the virtual registers.
27818 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
27820 * explow.c (get_dynamic_stack_size): Take known alignment of stack
27821 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
27824 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27826 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
27827 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
27829 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27831 * config/s390/s390.md: Add missing comments with the expanded
27833 * config/s390/vector.md: Likewise.
27834 * config/s390/vx-builtins.md: Likewise.
27836 2017-02-02 Jakub Jelinek <jakub@redhat.com>
27839 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
27840 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
27841 conditions on a single line.
27843 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27845 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27846 __S390_VX__ to __VX__.
27848 2017-02-01 Andrew Pinski <apinski@cavium.com>
27850 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
27851 stmt_info to record_stmt_cost.
27852 (vect_get_known_peeling_cost): Pass stmt_info if known to
27854 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
27855 cpu_vector_cost field into
27856 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
27857 field into vec_int_stmt_cost and vec_fp_stmt_cost.
27858 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
27859 splitting of scalar_stmt_cost and vec_stmt_cost.
27860 (thunderx_vector_cost): Likewise.
27861 (cortexa57_vector_cost): LIkewise.
27862 (exynosm1_vector_cost): Likewise.
27863 (xgene1_vector_cost): Likewise.
27864 (thunderx2t99_vector_cost): Improve after the splitting of the two
27866 (aarch64_builtin_vectorization_cost): Update for the splitting of
27867 scalar_stmt_cost and vec_stmt_cost.
27869 2017-02-01 Torvald Riegel <triegel@redhat.com>
27870 Richard Henderson <rth@redhat.com>
27872 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
27873 conditional on existance of a fast atomic load.
27874 * optabs-query.c (can_atomic_load_p): New function.
27875 * optabs-query.h (can_atomic_load_p): Declare it.
27876 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
27877 no fast atomic load is available for the particular size of access.
27878 (expand_atomic_compare_and_swap): Likewise.
27879 (expand_atomic_load): Likewise.
27880 (expand_atomic_store): Likewise.
27881 (expand_atomic_fetch_op): Likewise.
27882 * testsuite/lib/target-supports.exp
27883 (check_effective_target_sync_int_128): Remove x86 because it provides
27884 no fast atomic load.
27885 (check_effective_target_sync_int_128_runtime): Likewise.
27887 2017-02-01 Richard Biener <rguenther@suse.de>
27889 * graphite.c: Include tree-vectorizer.h for find_loop_location.
27890 (graphite_transform_loops): Provide opt-info for optimized nests.
27891 * tree-parloop.c (parallelize_loops): Provide opt-info for
27892 parallelized loops.
27894 2017-02-01 Richard Biener <rguenther@suse.de>
27896 PR middle-end/79315
27897 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
27898 was not set before.
27900 2017-02-01 Richard Biener <rguenther@suse.de>
27902 PR tree-optimization/71824
27903 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27904 Verify the loops are valid in the merged SESE region.
27905 (scop_detection::can_represent_loop_1): Check analyzing the
27906 evolution of the number of iterations in the region succeeds.
27908 2017-01-31 Ian Lance Taylor <iant@golang.org>
27910 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
27911 REG_ARGS_SIZE note to 32-bit push insns and call insn.
27913 2017-01-31 David Malcolm <dmalcolm@redhat.com>
27915 PR preprocessor/79210
27916 * input.c (get_substring_ranges_for_loc): Replace line_width
27917 assertion with error-handling.
27919 2017-01-31 Richard Biener <rguenther@suse.de>
27921 PR tree-optimization/77318
27922 * graphite-sese-to-poly.c (extract_affine): Fix assert.
27923 (create_pw_aff_from_tree): Take loop parameter.
27924 (add_condition_to_pbb): Pass loop of the condition to
27925 create_pw_aff_from_tree.
27927 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27929 * config/s390/s390.c (s390_asan_shadow_offset): New function.
27930 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
27932 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
27936 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
27938 (convert_int_to_float128): Likewise.
27939 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
27940 (convert_int_to_float128): Likewise.
27941 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
27942 (UNSPEC_IEEE128_CONVERT): Likewise.
27943 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
27944 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
27945 Use local variables for IBM extended format.
27946 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
27947 (fix_trunc<mode>si2_fprs): Likewise.
27948 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
27949 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
27950 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
27951 to know that we can now have integers of all sizes in vector
27953 (fix<uns>_<mode>di2_hw): Likewise.
27954 (float<uns>_<mode>si2_hw): Likewise.
27955 (fix_<mode>si2_hw): Likewise.
27956 (fixuns_<mode>si2_hw): Likewise.
27957 (float<uns>_<mode>di2_hw): Likewise.
27958 (float_<mode>di2_hw): Likewise.
27959 (float_<mode>si2_hw): Likewise.
27960 (floatuns_<mode>di2_hw): Likewise.
27961 (floatuns_<mode>si2_hw): Likewise.
27962 (xscvqp<su>wz_<mode>): Delete, no longer used.
27963 (xscvqp<su>dz_<mode>): Likewise.
27964 (xscv<su>dqp_<mode>): Likewise.
27965 (ieee128_mfvsrd_64bit): Likewise.
27966 (ieee128_mfvsrd_32bit): Likewise.
27967 (ieee128_mfvsrwz): Likewise.
27968 (ieee128_mtvsrw): Likewise.
27969 (ieee128_mtvsrd_64bit): Likewise.
27970 (ieee128_mtvsrd_32bit): Likewise.
27972 2017-01-31 Martin Liska <mliska@suse.cz>
27975 * ipa-prop.c (ipa_free_all_node_params): Call release method
27976 instead of ~sumbol_summary to not to trigger double times
27979 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
27981 PR tree-optimization/71691
27982 * bitmap.h (class auto_bitmap): New.
27983 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
27984 is_maybe_undefined instead of ssa_undefined_value_p.
27986 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27988 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27989 __S390_ARCH_LEVEL__ to __ARCH__.
27991 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27993 PR tree-optimization/79267
27994 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
27995 if should_remove_lhs_p is true.
27997 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
28000 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
28001 (add_alignment_attribute): New.
28002 (base_type_die): Add alignment attribute.
28003 (subrange_type_die): Likewise.
28004 (modified_type_die): Likewise.
28005 (gen_array_type_die): Likewise.
28006 (gen_descr_array_type_die: Likewise.
28007 (gen_enumeration_type_die): Likewise.
28008 (gen_subprogram_die): Likewise.
28009 (gen_variable_die): Likewise.
28010 (gen_field_die): Likewise.
28011 (gen_ptr_to_mbr_type_die): Likewise.
28012 (gen_struct_or_union_type_die): Likewise.
28013 (gen_subroutine_type_die): Likewise.
28014 (gen_typedef_die): Likewise.
28015 (base_type_cmp): Compare alignment attribute.
28017 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28020 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
28021 (setb_unsigned) New pattern for setb with CCUNS.
28022 * config/rs6000/rs6000.c (expand_block_compare): Use a different
28023 subfc./subfe sequence to avoid overflow problems. Generate a
28024 shorter sequence with cmpld/setb for power9.
28025 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
28026 for generating subfc. instruction.
28027 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
28028 now uses this instruction.
28030 2017-01-30 Ian Lance Taylor <iant@google.com>
28033 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
28034 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
28036 2017-01-30 Martin Sebor <msebor@redhat.com>
28038 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
28039 Move constant to the right of a relational operator.
28040 (get_mpfr_format_length, format_character, format_string): Ditto.
28041 (should_warn_p, maybe_warn): Same.
28043 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
28045 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
28048 * asan.c (get_translation_unit_decl): Remove function.
28049 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
28051 2017-01-30 Martin Liska <mliska@suse.cz>
28053 PR gcov-profile/79259
28054 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
28055 -fprofile-generate.
28057 2017-01-30 Martin Liska <mliska@suse.cz>
28060 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
28061 Initialize variables with NULL value.
28063 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
28066 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
28068 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
28070 2017-01-30 Richard Biener <rguenther@suse.de>
28072 PR tree-optimization/79276
28073 * tree-vrp.c (process_assert_insertions): Properly adjust common
28074 when removing a duplicate.
28076 * gcc.dg/torture/pr79276.c: New testcase.
28078 2017-01-30 Richard Biener <rguenther@suse.de>
28080 PR tree-optimization/79256
28081 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28082 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28084 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
28086 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
28089 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
28090 ("*r<noxa>sbg_<mode>_sll_bitmask")
28091 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
28092 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
28093 Use contiguous_bitmask_nowrap_operand.
28095 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28098 * config/rs6000/altivec.h (vec_xl): Revise #define.
28099 (vec_xst): Likewise.
28101 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28103 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28105 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28107 PR rtl-optimization/79194
28108 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28109 traps before call to bypass_conditional_jumps.
28111 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28113 PR tree-optimization/71374
28114 * lra-constraints.c (check_conflict_input_operands): New.
28115 (match_reload): Use it.
28117 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28120 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28121 account to calculate conflict_set.
28123 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28125 PR rtl-optimization/78559
28126 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28127 other_insn in combine.
28129 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28131 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28132 uint16_type_node for BT_UINT16.
28134 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28136 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28137 "RTL Tests" to menu.
28138 (GIMPLE Tests): New node.
28139 (RTL Tests): New node.
28141 2017-01-27 Richard Biener <rguenther@suse.de>
28143 PR tree-optimization/79245
28144 * tree-loop-distribution.c (distribute_loop): Apply cost
28145 modeling also to detected patterns.
28147 2017-01-27 Richard Biener <rguenther@suse.de>
28149 PR tree-optimization/71433
28150 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28151 (compare_assert_loc): New function.
28152 (process_assert_insertions): Sort and optimize assert locations
28153 to remove duplicates and push down identical assertions on
28154 edges to their destination block.
28156 2017-01-27 Richard Biener <rguenther@suse.de>
28158 PR tree-optimization/79244
28159 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28160 out SSA names even if abnormal.
28162 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28164 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28165 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28166 instead of MPFR_RNDN.
28168 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28171 * arm.c (arm_option_override): Don't call build_target_option_node
28172 until after doing all option overrides.
28173 (arm_valid_target_attribute_tree): Likewise.
28175 2017-01-27 Martin Liska <mliska@suse.cz>
28177 * doc/invoke.texi (-fprofile-arcs): Document profiling support
28178 for {cd}tors and C++ {cd}tors.
28180 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
28182 * config/s390/s390.md ("*setmem_long_and")
28183 ("*setmem_long_and_31z"): Use zero_extend instead of and.
28185 2017-01-26 Martin Sebor <msebor@redhat.com>
28187 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
28190 2017-01-26 Martin Sebor <msebor@redhat.com>
28192 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
28193 HAVE_DFmode before using XFmode or DFmode.
28194 (parse_directive): Avoid using the z length modifier to avoid
28195 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
28197 PR middle-end/78703
28198 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
28199 to accept adjustment as an array.
28200 (get_int_range): New function.
28201 (struct directive): Make width and prec arrays.
28202 (directive::set_width, directive::set_precision): Call get_int_range.
28203 (format_integer, format_floating): Handle width and precision ranges.
28204 (format_string, parse_directive): Same.
28206 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28209 * dwarf2out.c (generate_skeleton_bottom_up): For children with
28210 comdat_type_p set, just clone them, but keep the children in the
28214 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
28215 which have direct callers with -fvar-tracking-assignments enabled
28217 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
28218 inside of type units.
28220 2017-01-26 Martin Sebor <msebor@redhat.com>
28222 PR middle-end/78703
28223 * gimple-ssa-sprintf.c (struct result_range): Add likely and
28225 (struct format_result): Replace number_chars, number_chars_min,
28226 and number_chars_max with a single member of struct result_range.
28228 (format_result::operator+=): Adjust.
28229 (struct fmtresult): Remove bounded. Handle likely and unlikely
28231 (fmtresult::adjust_for_width_or_precision): New function.
28232 (fmtresult:type_max_digits): New function.
28233 (bytes_remaining): Handle likely and unlikely counters.
28234 (min_bytes_remaining): Remove.
28235 (format_percent): Simplify.
28236 (format_integer, format_floating): Set likely and unlikely counters.
28237 (get_string_length, format_character, format_string): Same.
28238 (format_plain, should_warn_p): New function.
28239 (maybe_warn): Call should_warn_p. Update diagnostic messages
28240 and handle those for all directives, including plain strings.
28241 (format_directive): Handle likely and unlikely counters.
28242 Remove unnecessary quoting from diagnostics. Add an informational
28244 (add_bytes): Remove.
28245 (pass_sprintf_length::compute_format_length): Simplify.
28246 (try_substitute_return_value): Handle likely and unlikely counters.
28248 2017-01-26 Carl Love <cel@us.ibm.com>
28250 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
28251 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
28253 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
28256 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
28257 endianess for subregs into account.
28258 * lra-constraints.c (lra_constraints): Do risky transformations
28259 always on the first iteration.
28260 * lra-lives.c (check_pseudos_live_through_calls): Add arg
28261 last_call_used_reg_set.
28262 (process_bb_lives): Define and use last_call_used_reg_set.
28263 * lra.c (lra): Always continue after lra_constraints on the first
28266 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
28268 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
28270 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
28272 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28274 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
28275 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
28276 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
28277 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
28278 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
28279 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
28280 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
28281 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
28282 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
28284 2017-01-26 Marek Polacek <polacek@redhat.com>
28287 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
28288 for the third operand.
28290 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28292 PR middle-end/79236
28293 * omp-low.c (struct omp_context): Add simt_stmt field.
28294 (scan_omp_for): Return omp_context *.
28295 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
28296 context to the _simt_ SIMD stmt.
28297 (lower_omp_for): For combined SIMD with sibling _simt_
28298 SIMD, make sure to use the same decls in _looptemp_
28299 clauses as in the sibling.
28301 2017-01-26 David Sherwood <david.sherwood@arm.com>
28303 PR middle-end/79212
28304 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
28307 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28310 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
28311 emit fld b; fld a; if possible.
28313 * brig-builtins.def: Update copyright years.
28314 * config/arm/arm_acle_builtins.def: Update copyright years.
28316 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
28319 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
28320 constraint instead of o for the stxsd instruction.
28322 2017-01-25 Carl Love <cel@us.ibm.com>
28324 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
28325 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
28327 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
28329 * doc/invoke.texi (C++ Dialect Options): Fix typo.
28331 2017-01-25 Richard Biener <rguenther@suse.de>
28333 PR tree-optimization/69264
28334 * target.def (vector_alignment_reachable): Improve documentation.
28335 * doc/tm.texi: Regenerate.
28336 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
28338 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
28339 earlier changes with respect to TYPE_USER_ALIGN.
28340 (vector_alignment_reachable_p): Likewise. Improve dumping.
28342 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28345 * config/arm/arm.md (xordi3): Force constant operand into a register
28348 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28350 * doc/invoke.texi (-fstore-merging): Correct default optimization
28351 levels at which it is enabled.
28352 (-O): Move -fstore-merging from list to...
28355 2017-01-25 Richard Biener <rguenther@suse.de>
28358 * omp-expand.c: Include debug.h.
28359 (expand_omp_taskreg): Make sure to generate early debug before
28360 outlining anything from a function.
28361 (expand_omp_target): Likewise.
28362 (grid_expand_target_grid_body): Likewise.
28364 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
28367 * asan.c (get_translation_unit_decl): New function.
28368 (asan_add_global): Extract modules file name from globals
28369 TRANSLATION_UNIT_DECL name.
28371 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
28374 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
28375 for long calls with APCS frame and VFP.
28377 2017-01-24 David Malcolm <dmalcolm@redhat.com>
28379 * cfg.c (original_copy_tables_initialized_p): New function.
28380 * cfg.h (original_copy_tables_initialized_p): New decl.
28381 * cfgrtl.c (relink_block_chain): Guard the call to
28382 free_original_copy_tables with a call to
28383 original_copy_tables_initialized_p.
28384 * cgraph.h (symtab_node::native_rtl_p): New decl.
28385 * cgraphunit.c (symtab_node::native_rtl_p): New function.
28386 (symtab_node::needed_p): Don't assert for early assembly output
28387 for __RTL functions.
28388 (cgraph_node::finalize_function): Set "force_output" for __RTL
28390 (cgraph_node::analyze): Bail out early for __RTL functions.
28391 (analyze_functions): Update assertion to support __RTL functions.
28392 (cgraph_node::expand): Bail out early for __RTL functions.
28393 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
28395 * function.h (struct function): Update comment for field
28397 * gimple-expr.c: Include "tree-pass.h".
28398 (gimple_has_body_p): Return false for __RTL functions.
28399 * Makefile.in (OBJS): Add run-rtl-passes.o.
28400 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
28402 (gcc::pass_manager::get_clean_slate): New accessor.
28403 * passes.c: Include "insn-addr.h".
28404 (should_skip_pass_p): Add logging. Update logic for running
28405 "expand" to be compatible with both __GIMPLE and __RTL. Guard
28406 property-provider override so it is only done for gimple passes.
28408 (skip_pass): New function.
28409 (execute_one_pass): Call skip_pass when skipping passes.
28410 * read-md.c (md_reader::read_char): Support filtering
28411 the input to a subset of line numbers.
28412 (md_reader::md_reader): Initialize fields
28413 m_first_line and m_last_line.
28414 (md_reader::read_file_fragment): New function.
28415 * read-md.h (md_reader::read_file_fragment): New decl.
28416 (md_reader::m_first_line): New field.
28417 (md_reader::m_last_line): New field.
28418 * read-rtl-function.c (function_reader::create_function): Only
28419 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
28420 curr_properties. Set DECL_INITIAL to a dummy block.
28421 (read_rtl_function_body_from_file_range): New function.
28422 * read-rtl-function.h (read_rtl_function_body_from_file_range):
28424 * run-rtl-passes.c: New file.
28425 * run-rtl-passes.h: New file.
28427 2017-01-24 Jeff Law <law@redhat.com>
28429 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
28432 2017-01-24 Bin Cheng <bin.cheng@arm.com>
28434 PR tree-optimization/79159
28435 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
28436 (record_nonwrapping_iv): Improve boundary using above function if no
28437 value range information.
28439 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
28440 Martin Jambor <mjambor@suse.cz>
28442 * brig-builtins.def: New file.
28443 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
28444 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
28445 (DEF_HSAIL_SAT_BUILTIN): Likewise.
28446 (DEF_HSAIL_INTR_BUILTIN): Likewise.
28447 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
28448 * builtin-types.def (BT_INT8): New.
28449 (BT_INT16): Likewise.
28450 (BT_UINT8): Likewise.
28451 (BT_UINT16): Likewise.
28452 (BT_FN_ULONG): Likewise.
28453 (BT_FN_UINT_INT): Likewise.
28454 (BT_FN_UINT_ULONG): Likewise.
28455 (BT_FN_UINT_LONG): Likewise.
28456 (BT_FN_UINT_PTR): Likewise.
28457 (BT_FN_ULONG_PTR): Likewise.
28458 (BT_FN_INT8_FLOAT): Likewise.
28459 (BT_FN_INT16_FLOAT): Likewise.
28460 (BT_FN_UINT32_FLOAT): Likewise.
28461 (BT_FN_UINT16_FLOAT): Likewise.
28462 (BT_FN_UINT8_FLOAT): Likewise.
28463 (BT_FN_UINT64_FLOAT): Likewise.
28464 (BT_FN_UINT16_UINT32): Likewise.
28465 (BT_FN_UINT32_UINT16): Likewise.
28466 (BT_FN_UINT16_UINT16_UINT16): Likewise.
28467 (BT_FN_INT_PTR_INT): Likewise.
28468 (BT_FN_UINT_PTR_UINT): Likewise.
28469 (BT_FN_LONG_PTR_LONG): Likewise.
28470 (BT_FN_ULONG_PTR_ULONG): Likewise.
28471 (BT_FN_VOID_UINT64_UINT64): Likewise.
28472 (BT_FN_UINT8_UINT8_UINT8): Likewise.
28473 (BT_FN_INT8_INT8_INT8): Likewise.
28474 (BT_FN_INT16_INT16_INT16): Likewise.
28475 (BT_FN_INT_INT_INT): Likewise.
28476 (BT_FN_UINT_FLOAT_UINT): Likewise.
28477 (BT_FN_FLOAT_UINT_UINT): Likewise.
28478 (BT_FN_ULONG_UINT_UINT): Likewise.
28479 (BT_FN_ULONG_UINT_PTR): Likewise.
28480 (BT_FN_ULONG_ULONG_ULONG): Likewise.
28481 (BT_FN_UINT_UINT_UINT): Likewise.
28482 (BT_FN_VOID_UINT_PTR): Likewise.
28483 (BT_FN_UINT_UINT_PTR: Likewise.
28484 (BT_FN_UINT32_UINT64_PTR): Likewise.
28485 (BT_FN_INT_INT_UINT_UINT): Likewise.
28486 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
28487 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
28488 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
28489 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
28490 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
28491 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
28492 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
28493 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
28494 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
28495 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
28496 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
28497 * doc/frontends.texi: List BRIG FE.
28498 * doc/install.texi (Testing): Add BRIG tesring requirements.
28499 * doc/invoke.texi (Overall Options): Mention BRIG.
28500 * doc/standards.texi (Standards): Doucment BRIG HSA version.
28502 2017-01-24 Richard Biener <rguenther@suse.de>
28504 PR translation/79208
28505 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
28507 2017-01-24 Martin Jambor <mjambor@suse.cz>
28510 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
28511 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
28512 and known_contexts.
28514 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
28516 PR middle-end/79123
28517 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
28518 casts from signed to unsigned really don't have a range.
28520 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
28522 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
28523 GMP_RNDx for compatiblity.
28525 2017-01-24 Martin Liska <mliska@suse.cz>
28528 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
28529 that would prevent us to call alloca with -1 as argument.
28531 2017-01-24 Jakub Jelinek <jakub@redhat.com>
28533 * dwarf2out.c (output_compilation_unit_header, output_file_names):
28534 Avoid -Wformat-security warning.
28536 2017-01-23 Andrew Pinski <apinski@cavium.com>
28538 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
28541 2017-01-23 Martin Sebor <msebor@redhat.com>
28543 PR middle-end/78703
28544 * gimple-ssa-sprintf.c (warn_level): New global.
28545 (format_integer): Use it here and throughout the rest of the file.
28546 Use the same switch to compute sign as base.
28547 (maybe_warn): New function.
28548 (format_directive): Factor out warnings into maybe_warn.
28549 Add debugging output. Use warn_level.
28550 (add_bytes): Use warn_level.
28551 (pass_sprintf_length::compute_format_length): Add debugging output.
28552 (try_substitute_return_value): Same.
28553 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
28555 PR middle-end/78703
28556 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
28557 (struct fmtresult, format_integer, format_floating): Adjust.
28558 (fmtresult::fmtresult): Set max correctly in two argument ctor.
28559 (get_string_length, format_string,format_directive): Same.
28560 (pass_sprintf_length::compute_format_length): Same.
28561 (try_substitute_return_value): Simplify slightly.
28563 PR middle-end/78703
28564 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
28565 (fmtresult::operator+=): Outlined.
28566 (struct fmtresult): Add ctors.
28567 (struct conversion_spec): Rename...
28568 (struct directive): ...to this. Add and remove data members.
28569 (directive::set_width, directive::set_precision): New functions.
28570 (format_percent): Use fmtresult ctor.
28571 (get_width_and_precision): Remove.
28572 (format_integer): Make naming changes. Avoid computing width and
28574 (format_floating): Same. Adjust indentation.
28575 (format_character, format_none): New functions.
28576 (format_string): Moved character handling to format_character.
28577 (format_directive): Remove arguments, change return type.
28578 (parse_directive): New function.
28579 (pass_sprintf_length::compute_format_length): Move directive
28580 parsing to parse_directive.
28582 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28584 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
28585 (assign_assembler_name_if_needed): ... this.
28586 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
28587 (assign_assembler_name_if_needed): ... this.
28588 (free_lang_data_in_cgraph): Adjust callers.
28589 * cgraphunit.c (cgraph_node::analyze): Likewise.
28590 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
28593 2017-01-23 Richard Biener <rguenther@suse.de>
28595 PR tree-optimization/79088
28596 PR tree-optimization/79188
28597 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
28598 resetting loop bounds after last path deletion. Reset loop
28599 bounds of the target loop, make code match the comments.
28600 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
28601 Make sure loops need no fixups.
28603 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28605 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
28606 exponent support with double type for first argument.
28607 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
28608 type returned by __builtin_vec_extract_sig,
28609 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
28610 functions from "vector int" to "vector unsigned int" or from
28611 "vector long long int" to "vector unsigned long long int".
28612 Changed type returned by __builtin_vec_extract_exp,
28613 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
28614 functions from "vector int" to "vector unsigned int" or from
28615 "vector long long int" to "vector unsigned long long int".
28616 Changed return type of __builtin_vec_test_data_class,
28617 __builtin_vec_test_data_class_sp, and
28618 __builtin_vec_test_data_class_dp from "vector int" to
28619 "vector bool int" or from "vector long long int" to "vector bool
28620 long long int" and changed second argument type from "unsigned
28621 int" to "int". Added new overloaded function forms "vector float
28622 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
28623 "vector float __builtin_vec_insert_exp_sp (vector float, vector
28624 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
28625 double, vector unsigned long long int)" and "vector double
28626 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
28627 long int)". Changed return type of
28628 __builtin_scalar_test_data_class and
28629 __builtin_scalar_test_data_class_sp and
28630 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
28631 int" and changed second argument from "unsigned int" to "int".
28632 Changed type returned by __builtin_scalar_test_neg,
28633 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
28634 from "int" to "bool int". Added new overloaded function form
28635 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
28636 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
28637 exponent double-precision with floating point first argument.
28638 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
28639 documentation of scalar_test_data_class, scalar_test_neg,
28640 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
28641 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
28642 vec_test_data_class built-in functions to reflect refinements in
28643 their type signatures.
28645 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
28647 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
28649 (aarch64_elf_asm_destructor): Likewise.
28651 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
28653 PR rtl-optimization/78634
28654 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
28655 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
28656 * ifcvt.c (noce_try_cmove): Add missing cost check.
28658 PR rtl-optimization/71724
28659 * combine.c (if_then_else_cond): Look for situations where it is
28660 beneficial to undo the work of one of the recursive calls.
28662 2017-01-23 Bin Cheng <bin.cheng@arm.com>
28664 PR tree-optimization/70754
28665 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
28666 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
28667 combined stmt before it if not NULL.
28668 (combine_chains): Process refs reversely and compute dominance point
28671 2017-01-23 Martin Liska <mliska@suse.cz>
28673 PR tree-optimization/79196
28674 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
28675 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
28677 (strlen_optimize_stmt): Call the renamed function.
28679 2017-01-23 Michael Matz <matz@suse.de>
28681 PR tree-optimization/78384
28682 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
28684 2017-01-23 Richard Biener <rguenther@suse.de>
28686 PR tree-optimization/79186
28687 * tree-vrp.c (register_new_assert_for): Make sure we've seen
28688 both incoming edges before moving an assert.
28690 2017-01-23 Martin Jambor <mjambor@suse.cz>
28692 * ipa-prop.c (load_from_param_1): Removed.
28693 (load_from_unmodified_param): Bits from load_from_param_1 put back
28695 (load_from_param): Removed.
28696 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
28697 with stmt. Reverted back to use of load_from_unmodified_param.
28699 2017-01-23 Martin Jambor <mjambor@suse.cz>
28702 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
28703 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
28704 field a pointer to garbage collected vector, mark lattices and
28705 ipcp_orig_node with GTY((skip)).
28706 (ipa_get_param_count): Adjust to descriptors being a pointer.
28707 (ipa_get_param): Likewise.
28708 (ipa_get_type): Likewise.
28709 (ipa_get_param_move_cost): Likewise.
28710 (ipa_set_param_used): Likewise.
28711 (ipa_get_controlled_uses): Likewise.
28712 (ipa_set_controlled_uses): Likewise.
28713 (ipa_is_param_used): Likewise.
28714 (ipa_node_params_t): Move into garbage collector. New methods insert
28716 (ipa_node_params_sum): Annotate wth GTY(()).
28717 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
28719 (ipa_load_from_parm_agg): Adjust declaration.
28720 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
28721 * ipa-profile.c (ipa_profile): Likewise.
28722 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
28723 (ipa_populate_param_decls): Make descriptors parameter garbage
28725 (ipa_dump_param): Adjust to descriptors being a pointer.
28726 (ipa_alloc_node_params): Likewise.
28727 (ipa_initialize_node_params): Likewise.
28728 (load_from_param_1): Make descriptors parameter garbage collected.
28729 (load_from_unmodified_param): Likewise.
28730 (load_from_param): Likewise.
28731 (ipa_load_from_parm_agg): Likewise.
28732 (ipa_node_params::~ipa_node_params): Removed.
28733 (ipa_free_all_node_params): Remove call to delete operator.
28734 (ipa_node_params_t::insert): New.
28735 (ipa_node_params_t::remove): Likewise.
28736 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
28737 copy known_csts and known_contexts vectors.
28738 (ipa_read_node_info): Adjust to descriptors being a pointer.
28739 (ipcp_modif_dom_walker): Make m_descriptors field garbage
28741 (ipcp_transform_function): Make descriptors variable garbage
28744 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
28746 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
28747 * config/i386/avx512dqintrin.h: Ditto.
28748 * config/i386/avx512fintrin.h: Ditto.
28749 * config/i386/i386.c: Handle new builtins.
28750 * config/i386/i386-builtin.def: Add new builtins.
28751 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
28752 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
28754 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28755 Martin Liska <mliska@suse.cz>
28757 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
28758 * asan.c (asan_expand_poison_ifn): Support stores and use
28759 appropriate ASAN report function.
28760 * internal-fn.c (expand_ASAN_POISON_USE): New function.
28761 * internal-fn.def (ASAN_POISON_USE): Declare.
28762 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
28763 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
28764 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
28765 ASAN_POISON calls w/o LHS.
28766 * tree-ssa.c (execute_update_addresses_taken): Create clobber
28767 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
28768 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
28769 * gimplify.c (asan_poison_variables): Add attribute
28770 use_after_scope_memory to variables that really needs to live
28772 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
28773 having the attribute.
28775 2017-01-23 Martin Liska <mliska@suse.cz>
28777 * asan.c (create_asan_shadow_var): New function.
28778 (asan_expand_poison_ifn): Likewise.
28779 * asan.h (asan_expand_poison_ifn): New declaration.
28780 * internal-fn.c (expand_ASAN_POISON): Likewise.
28781 * internal-fn.def (ASAN_POISON): New builtin.
28782 * sanopt.c (pass_sanopt::execute): Expand
28783 asan_expand_poison_ifn.
28784 * tree-inline.c (copy_decl_for_dup_finish): Make function
28786 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
28787 * tree-ssa.c (is_asan_mark_p): New function.
28788 (execute_update_addresses_taken): Rewrite local variables
28789 (identified just by use-after-scope as addressable) into SSA.
28791 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28793 * doc/install.texi (Specific): opensource.apple.com uses https
28794 now. Remove trailing slash.
28796 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28798 * README.Portability: Remove note on an Irix compatibility issue.
28800 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
28802 * gcov.c (INCLUDE_ALGORITHM): Define.
28803 (INCLUDE_VECTOR): Define.
28804 No longer include <vector> and <algorithm> directly.
28806 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
28808 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
28810 * doc/invoke.texi (Code Gen Options): Ditto.
28812 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
28815 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
28817 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
28819 rtl-optimization/79125
28820 * cprop.c (local_cprop_pass): Handle cases where we make an
28821 unconditional trap.
28823 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
28827 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
28828 read from, for big endian.
28830 2017-01-20 Jiong Wang <jiong.wang@arm.com>
28832 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
28833 register pauth builtins for LP64 only.
28835 2017-01-20 Marek Polacek <polacek@redhat.com>
28838 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
28841 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28843 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
28845 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
28846 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
28847 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
28849 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28852 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
28853 in big-endian mode when they are not a single duplicated value.
28855 2017-01-20 Richard Biener <rguenther@suse.de>
28857 * BASE-VER: Bump to 7.0.1.
28859 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28861 * omp-low.c (omplow_simd_context): New struct. Use it...
28862 (lower_rec_simd_input_clauses): ...here and...
28863 (lower_rec_input_clauses): ...here to hold common data. Adjust all
28864 references to idx, lane, max_vf, is_simt.
28866 2017-01-20 Graham Markall <graham.markall@embecosm.com>
28868 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
28871 2017-01-20 Martin Jambor <mjambor@suse.cz>
28873 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
28874 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
28876 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
28877 (GTFILES): Rename hsa.c to hsa-common.c.
28878 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
28879 * hsa-dump.c: Likewise.
28880 * hsa-gen.c: Likewise.
28881 * hsa-regalloc.c: Likewise.
28882 * ipa-hsa.c: Likewise.
28883 * omp-expand.c: Likewise.
28884 * omp-low.c: Likewise.
28885 * toplev.c: Likewise.
28887 2017-01-20 Marek Polacek <polacek@redhat.com>
28890 * doc/invoke.texi: Document -Wduplicated-branches.
28891 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
28892 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
28893 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
28894 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
28895 return 0 only when not OEP_LEXICOGRAPHIC.
28896 (fold_build_cleanup_point_expr): Use the expression
28897 location when building CLEANUP_POINT_EXPR.
28898 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
28899 * tree.c (add_expr): Handle error_mark_node.
28901 2017-01-20 Martin Liska <mliska@suse.cz>
28904 * tree-profile.c (init_ic_make_global_vars): Do not call
28906 (gimple_init_gcov_profiler): Likewise.
28908 2017-01-20 Martin Liska <mliska@suse.cz>
28911 * cgraph.h (maybe_create_reference): Remove argument and
28913 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
28915 * ipa-cp.c (create_specialized_node): Likewise.
28916 * symtab.c (symtab_node::maybe_create_reference): Handle
28917 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
28919 2017-01-20 Martin Liska <mliska@suse.cz>
28921 * read-rtl-function.c (function_reader::create_function): Use
28922 build_decl instread of build_decl_stat.
28924 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
28926 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
28927 * config/i386/avx512dqintrin.h: Ditto.
28928 * config/i386/avx512fintrin.h: Ditto.
28929 * config/i386/i386-builtin-types.def: Add new types.
28930 * config/i386/i386.c: Handle new types.
28931 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
28932 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
28933 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
28934 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
28935 (__builtin_ia32_kshiftridi): New.
28936 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
28938 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
28942 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
28943 define to rs6000_init_stack_protect_guard.
28944 (rs6000_init_stack_protect_guard): New function.
28946 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28947 Yunqiang Su <yunqiang.su@imgtec.com>
28949 * config.gcc (supported_defaults): Add madd4.
28950 (with_madd4): Add validation.
28951 (all_defaults): Add madd4.
28952 * config/mips/mips.opt (mmadd4): New option.
28953 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28955 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28957 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
28958 (ISA_HAS_FUSED_MADD4): Likewise.
28959 * doc/invoke.texi (-mmadd4): Document the new option.
28960 * doc/install.texi (--with-madd4): Document the new option.
28962 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28964 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
28965 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
28966 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
28967 (aarch64_init_pauth_hint_builtins): New.
28968 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
28969 (aarch64_expand_builtin): Expand new builtins.
28971 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28973 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
28974 * combine-stack-adj.c (no_unhandled_cfa): Handle
28975 REG_CFA_TOGGLE_RA_MANGLE.
28976 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
28977 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
28978 info for return address signing.
28979 (aarch64_expand_epilogue): Likewise.
28981 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28983 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
28984 * config/aarch64/aarch64-protos.h
28985 (aarch64_return_address_signing_enabled): New declaration.
28986 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
28988 (aarch64_expand_prologue): Sign return address before it's pushed onto
28990 (aarch64_expand_epilogue): Authenticate return address fetched from
28992 (aarch64_override_options): Sanity check for ILP32 and ISA level.
28993 (aarch64_attributes): New function attributes for "sign-return-address".
28994 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
28995 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
28996 ("*do_return"): Generate combined instructions according to key index.
28997 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
28998 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
29000 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
29001 * config/aarch64/aarch64.opt (msign-return-address=): New.
29002 * doc/extend.texi (AArch64 Function Attributes): Documents
29003 "sign-return-address=".
29004 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
29006 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29008 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
29009 overall option summary.
29011 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29013 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
29014 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
29015 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
29016 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
29018 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
29020 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
29021 -mpower9-minmax by default for -mcpu=power9.
29022 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
29023 128-bit floating point.
29025 2017-01-20 Alan Modra <amodra@gmail.com>
29027 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
29028 optimizing for size.
29030 2017-01-20 Alan Modra <amodra@gmail.com>
29033 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
29034 for strcmp and strncmp from corresponding builtin decl.
29036 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29038 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
29039 instead of i386/rtems-64.h.
29040 * config/i386/rtems-64.h: Remove.
29042 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29046 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
29048 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
29050 2017-01-19 Tamar Christina <tamar.christina@arm.com>
29052 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
29053 Change int to HOST_WIDE_INT.
29054 * config/aarch64/aarch64-protos.h
29055 (aarch64_simd_gen_const_vector_dup): Likewise.
29056 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
29058 2017-01-19 David Malcolm <dmalcolm@redhat.com>
29060 * langhooks-def.h (lhd_type_for_size): New decl.
29061 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
29062 * langhooks.c (lhd_type_for_size): New function, taken from
29065 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
29067 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
29068 define_bypass for CR latency.
29069 (power9-cracked-alu): Update bypass latency and remove power9-branch.
29070 (power9-alu2): Add define_bypass for CR latency.
29072 (power9-mul): Update insn latency.
29073 (power9-mul-compare): Update insn latency, bypass latency and remove
29076 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29078 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
29080 * config/aarch64/aarch64.md
29081 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
29082 aarch64_nopcrelative_literal_loads.
29083 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
29085 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
29087 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
29088 TARGET_LOONGSON_3A.
29089 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
29091 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
29094 * config.gcc (supported_defaults): Add lxc1-sxc1.
29095 (with_lxc1_sxc1): Add validation.
29096 (all_defaults): Add lxc1-sxc1.
29097 * config/mips/mips.opt (mlxc1-sxc1): New option.
29098 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29100 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29101 __mips_no_lxc1_sxc1.
29102 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29103 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29104 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29106 2017-01-19 Richard Biener <rguenther@suse.de>
29108 PR tree-optimization/72488
29109 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29110 sure to restore SSA info.
29111 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29113 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29115 PR rtl-optimization/79121
29116 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29117 of the inner type when shifting an extended value.
29119 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29122 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29123 interposable aliases.
29125 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29128 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29129 Use the evmergelohi instruction.
29130 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29131 (mov_sitf_e500_subreg8_2_be): Likewise.
29132 (mov_sitf_e500_subreg12_2_le): Likewise.
29133 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29134 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29135 (mov_sitf_e500_subreg8_2_le): Likewise.
29136 (mov_sitf_e500_subreg12_2_be): Likewise.
29138 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29140 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29141 attribute from vecsimple to vecperm.
29142 (altivec_vbpermq2): Likewise.
29144 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29147 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29149 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29150 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29151 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29152 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29153 case where N arg is SIZE_MAX.
29154 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29155 (cmpstrsi): Add pattern.
29157 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29159 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29160 __builtin_vec_revb builtins.
29161 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29162 built-in functions to support generation of the ISA 3.0 XXBR<x>
29163 vector byte reverse instructions.
29164 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29165 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29166 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29167 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29168 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29169 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29170 (P9V_BUILTIN_VEC_REVB): Likewise.
29171 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29172 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29173 (p9_xxbrq_v16qi): Likewise.
29174 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
29175 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
29176 (p9_xxbrh_v8hi): Likewise.
29177 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
29178 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
29179 vec_revb built-in functions.
29181 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
29183 PR rtl-optimization/78952
29184 * config/i386/i386.md (any_extract): New code iterator.
29185 (*insvqi_2): Use any_extract for source operand.
29186 (*insvqi_3): Use any_shiftrt for source operand.
29188 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29190 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
29192 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
29194 2017-01-18 Matthias Klose <doko@ubuntu.com>
29196 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
29198 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29200 * config/rs6000/altivec.h (vec_bperm): Change #define.
29201 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
29202 (altivec_vbpermq2): New define_insn.
29203 (altivec_vbpermd): Likewise.
29204 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
29205 function interface.
29206 (VBPERMD): Likewise.
29207 (VBPERM): New polymorphic function interface.
29208 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
29209 Add entries for P9V_BUILTIN_VEC_VBPERM.
29210 * doc/extend.texi: Add interfaces for vec_bperm.
29212 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29214 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
29215 first letter of error messages.
29216 (s390_resolve_overloaded_builtin): Likewise.
29217 * config/s390/s390.c (s390_expand_builtin): Likewise.
29218 (s390_invalid_arg_for_unprototyped_fn): Likewise.
29219 (s390_valid_target_attribute_inner_p): Likewise.
29220 * config/s390/s390.md ("tabort"): Likewise.
29222 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
29224 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
29225 (ISA_AVOID_DIV_HILO): New macro.
29226 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
29227 (ISA_HAS_DDIV): Likewise.
29229 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29231 * doc/invoke.texi (fabi-version): Correct number of occurrences.
29233 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29235 * doc/invoke.texi (fabi-version): Spelling fix.
29237 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29240 * doc/invoke.texi (fabi-version): Mention mangling fix for
29243 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29246 * doc/invoke.texi (fabi-version): Document discriminator mangling.
29248 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
29251 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
29252 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
29254 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
29256 (stack_protect_test): Ditto.
29257 * config/rs6000/rs6000.opt (mstack-protector-guard=,
29258 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
29260 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
29261 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
29262 -mstack-protector-guard-offset=.
29263 (RS/6000 and PowerPC Options): Ditto.
29265 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29267 * config/i386/i386.h (MASK_CLASS_P): New define.
29268 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
29269 there are no registers from different register sets also when
29270 mask registers are used. Update function comment.
29271 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
29272 to (*k/*r) and (*k/*km) alternatives.
29274 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29276 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
29277 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
29278 (EH_RETURN_HANDLER_RTX): New define.
29279 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
29280 Force frame pointer in EH return functions.
29281 (aarch64_expand_epilogue): Add barrier for eh_return.
29282 (aarch64_final_eh_return_addr): Remove.
29283 (aarch64_eh_return_handler_rtx): New function.
29284 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
29286 (aarch64_eh_return_handler_rtx): New prototype.
29288 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29290 * config/rs6000/altivec.h (vec_rlmi): New #define.
29291 (vec_vrlnm): Likewise.
29292 (vec_rlnm): Likewise.
29293 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
29294 (UNSPEC_VRLNM): Likewise.
29295 (VIlong): New mode iterator.
29296 (altivec_vrl<VI_char>mi): New define_insn.
29297 (altivec_vrl<VI_char>nm): Likewise.
29298 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
29300 (VRLDNM): Likewise.
29301 (RLNM): New polymorphic function entry.
29302 (VRLWMI): New monomorphic function entry.
29303 (VRLDMI): Likewise.
29304 (RLMI): New polymorphic function entry.
29305 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
29306 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
29307 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
29310 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29313 * dwarf2out.c (field_byte_offset): Restore the
29314 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
29315 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
29316 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
29319 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29322 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
29324 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29327 * dwarf2out.c (add_data_member_location_attribute): For constant
29328 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
29329 instead of DW_AT_data_member_location, DW_AT_bit_offset and
29330 DW_AT_byte_size attributes.
29332 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29334 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
29335 after forcing to constant memory when the code model is medium.
29337 2017-01-17 Julia Koval <julia.koval@intel.com>
29340 * config/i386/avx512fintrin.h
29341 (_mm512_i32gather_ps): Change __addr type to void const*.
29342 (_mm512_mask_i32gather_ps): Ditto.
29343 (_mm512_i32gather_pd): Ditto.
29344 (_mm512_mask_i32gather_pd): Ditto.
29345 (_mm512_i64gather_ps): Ditto.
29346 (_mm512_mask_i64gather_ps): Ditto.
29347 (_mm512_i64gather_pd): Ditto.
29348 (_mm512_mask_i64gather_pd): Ditto.
29349 (_mm512_i32gather_epi32): Ditto.
29350 (_mm512_mask_i32gather_epi32): Ditto.
29351 (_mm512_i32gather_epi64): Ditto.
29352 (_mm512_mask_i32gather_epi64): Ditto.
29353 (_mm512_i64gather_epi32): Ditto.
29354 (_mm512_mask_i64gather_epi32): Ditto.
29355 (_mm512_i64gather_epi64): Ditto.
29356 (_mm512_mask_i64gather_epi64): Ditto.
29357 (_mm512_i32scatter_ps): Change __addr type to void*.
29358 (_mm512_mask_i32scatter_ps): Ditto.
29359 (_mm512_i32scatter_pd): Ditto.
29360 (_mm512_mask_i32scatter_pd): Ditto.
29361 (_mm512_i64scatter_ps): Ditto.
29362 (_mm512_mask_i64scatter_ps): Ditto.
29363 (_mm512_i64scatter_pd): Ditto.
29364 (_mm512_mask_i64scatter_pd): Ditto.
29365 (_mm512_i32scatter_epi32): Ditto.
29366 (_mm512_mask_i32scatter_epi32): Ditto.
29367 (_mm512_i32scatter_epi64): Ditto.
29368 (_mm512_mask_i32scatter_epi64): Ditto.
29369 (_mm512_i64scatter_epi32): Ditto.
29370 (_mm512_mask_i64scatter_epi32): Ditto.
29371 (_mm512_i64scatter_epi64): Ditto.
29372 (_mm512_mask_i64scatter_epi64): Ditto.
29373 * config/i386/avx512pfintrin.h
29374 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
29375 (_mm512_mask_prefetch_i32gather_ps): Ditto.
29376 (_mm512_mask_prefetch_i64gather_pd): Ditto.
29377 (_mm512_mask_prefetch_i64gather_ps): Ditto.
29378 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
29379 (_mm512_prefetch_i32scatter_ps): Ditto.
29380 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
29381 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
29382 (_mm512_prefetch_i64scatter_pd): Ditto.
29383 (_mm512_prefetch_i64scatter_ps): Ditto.
29384 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
29385 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
29386 * config/i386/avx512vlintrin.h
29387 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
29388 (_mm_mmask_i32gather_ps): Ditto.
29389 (_mm256_mmask_i32gather_pd): Ditto.
29390 (_mm_mmask_i32gather_pd): Ditto.
29391 (_mm256_mmask_i64gather_ps): Ditto.
29392 (_mm_mmask_i64gather_ps): Ditto.
29393 (_mm256_mmask_i64gather_pd): Ditto.
29394 (_mm_mmask_i64gather_pd): Ditto.
29395 (_mm256_mmask_i32gather_epi32): Ditto.
29396 (_mm_mmask_i32gather_epi32): Ditto.
29397 (_mm256_mmask_i32gather_epi64): Ditto.
29398 (_mm_mmask_i32gather_epi64): Ditto.
29399 (_mm256_mmask_i64gather_epi32): Ditto.
29400 (_mm_mmask_i64gather_epi32): Ditto.
29401 (_mm256_mmask_i64gather_epi64): Ditto.
29402 (_mm_mmask_i64gather_epi64): Ditto.
29403 (_mm256_i32scatter_ps): Change __addr type to void*.
29404 (_mm256_mask_i32scatter_ps): Ditto.
29405 (_mm_i32scatter_ps): Ditto.
29406 (_mm_mask_i32scatter_ps): Ditto.
29407 (_mm256_i32scatter_pd): Ditto.
29408 (_mm256_mask_i32scatter_pd): Ditto.
29409 (_mm_i32scatter_pd): Ditto.
29410 (_mm_mask_i32scatter_pd): Ditto.
29411 (_mm256_i64scatter_ps): Ditto.
29412 (_mm256_mask_i64scatter_ps): Ditto.
29413 (_mm_i64scatter_ps): Ditto.
29414 (_mm_mask_i64scatter_ps): Ditto.
29415 (_mm256_i64scatter_pd): Ditto.
29416 (_mm256_mask_i64scatter_pd): Ditto.
29417 (_mm_i64scatter_pd): Ditto.
29418 (_mm_mask_i64scatter_pd): Ditto.
29419 (_mm256_i32scatter_epi32): Ditto.
29420 (_mm256_mask_i32scatter_epi32): Ditto.
29421 (_mm_i32scatter_epi32): Ditto.
29422 (_mm_mask_i32scatter_epi32): Ditto.
29423 (_mm256_i32scatter_epi64): Ditto.
29424 (_mm256_mask_i32scatter_epi64): Ditto.
29425 (_mm_i32scatter_epi64): Ditto.
29426 (_mm_mask_i32scatter_epi64): Ditto.
29427 (_mm256_i64scatter_epi32): Ditto.
29428 (_mm256_mask_i64scatter_epi32): Ditto.
29429 (_mm_i64scatter_epi32): Ditto.
29430 (_mm_mask_i64scatter_epi32): Ditto.
29431 (_mm256_i64scatter_epi64): Ditto.
29432 (_mm256_mask_i64scatter_epi64): Ditto.
29433 (_mm_i64scatter_epi64): Ditto.
29434 (_mm_mask_i64scatter_epi64): Ditto.
29435 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
29436 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
29437 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
29438 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
29439 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
29440 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
29441 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
29442 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
29443 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
29444 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
29445 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
29446 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
29447 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
29448 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
29449 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
29450 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
29451 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
29452 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
29453 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
29454 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
29455 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
29456 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
29457 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
29458 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
29459 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
29460 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
29461 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
29462 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
29463 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
29464 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
29465 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
29466 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
29467 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
29468 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
29469 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
29470 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
29471 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
29472 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
29473 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
29474 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
29475 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
29476 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
29477 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
29478 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
29479 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
29480 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
29481 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
29482 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
29483 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
29484 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
29485 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
29486 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
29487 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
29488 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
29489 definitions accordingly.
29491 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
29492 Kuan-Lin Chen <kuanlinchentw@gmail.com>
29495 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
29498 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
29501 * ira-conflicts.c (ira_build_conflicts): Update total conflict
29502 hard regs for inner regno.
29504 2017-01-17 Martin Liska <mliska@suse.cz>
29507 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
29508 assumption and add comment.
29510 2017-01-17 Nathan Sidwell <nathan@acm.org>
29512 * ipa-visibility.c (localize_node): New function, broken out of ...
29513 (function_and_variable_visibility): ... here. Call it.
29515 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29517 PR middle-end/77445
29518 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
29519 correctly set frequency of oudgoing edge.
29520 (duplicate_thread_path): Fix profile updating.
29522 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29525 * configure.ac: Add GCC_BASE_VER.
29526 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
29527 version from BASE-VER file.
29528 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
29529 (gcc.o): Depend on $(BASEVER).
29530 * common.opt (dumpfullversion): New option.
29531 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
29532 * doc/invoke.texi: Document -dumpfullversion.
29533 * doc/install.texi: Document --with-gcc-major-version-only.
29534 * configure: Regenerated.
29536 2017-01-17 Richard Biener <rguenther@suse.de>
29538 PR tree-optimization/71433
29539 * tree-vrp.c (register_new_assert_for): Merge same asserts
29540 on all incoming edges.
29541 (process_assert_insertions_for): Handle insertions at the
29544 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
29546 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
29547 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
29549 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
29552 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
29555 2017-01-17 Alan Modra <amodra@gmail.com>
29558 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
29559 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
29560 symbolic stack limit when pic.
29562 2017-01-16 Martin Sebor <msebor@redhat.com>
29564 PR tree-optimization/78608
29565 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
29567 2017-01-16 Jeff Law <law@redhat.com>
29570 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
29571 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
29572 for several include directories that may be relative to sysroot.
29573 * config/i386/x-mingw32 (gplus_includedir): Define.
29574 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
29575 (native_system_includedir): Likewise.
29576 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
29577 override if TARGET_SYSTEM_ROOT is defined.
29578 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
29580 PR tree-optimization/79090
29581 PR tree-optimization/33562
29582 PR tree-optimization/61912
29583 PR tree-optimization/77485
29584 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
29585 and computed trims into the dump file.
29587 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29589 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
29591 2017-01-16 Jakub Jelinek <jakub@redhat.com>
29594 * gimplify.c (gimplify_init_constructor): If want_value and
29595 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
29599 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
29600 sequence. Formatting fixes.
29601 (doloop_optimize): Formatting fixes.
29604 * gcc.c (debug_level_greater_than_spec_func): New function.
29605 (static_spec_functions): Add debug-level-gt spec function.
29606 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
29608 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29609 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29610 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
29611 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
29612 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
29613 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
29615 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
29617 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
29618 QImode fixups to general and mask registers only.
29620 2017-01-16 Carl Love <cel@us.ibm.com>
29622 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
29623 for built-in functions
29624 vector signed char vec_nabs (vector signed char)
29625 vector signed short vec_nabs (vector signed short)
29626 vector signed int vec_nabs (vector signed int)
29627 vector signed long long vec_nabs (vector signed long long)
29628 vector float vec_nabs (vector float)
29629 vector double vec_nabs (vector double)
29630 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
29632 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
29633 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
29634 * doc/extend.texi: Update the documentation file for the new built-in
29637 2017-01-16 Martin Sebor <msebor@redhat.com>
29639 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
29642 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29644 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
29645 UNSPEC_VSX__XXSPLTD to require special splat handling.
29647 2017-01-16 David Malcolm <dmalcolm@redhat.com>
29650 * system.h: Poison strndup.
29652 2017-01-16 Alan Modra <amodra@gmail.com>
29655 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
29658 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
29660 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
29662 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
29664 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
29665 call recog here. Assert that INSN_CODE (insn) is non-negative.
29667 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29670 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
29672 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
29673 in the currently scheduled RTL fragment.
29675 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29677 PR rtl-optimization/78751
29678 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
29681 2017-01-14 Jeff Law <law@redhat.com>
29683 PR tree-optimization/79090
29684 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
29685 variable length stores.
29686 (compute_trims): Delete dead assignment to *trim_tail.
29687 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
29690 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
29692 PR rtl-optimization/78626
29693 PR rtl-optimization/78727
29694 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
29695 of a block, and split such blocks after everything else is finished.
29697 2017-01-14 Alan Modra <amodra@gmail.com>
29700 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
29701 target legitimate_combined_insn.
29702 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
29703 (rs6000_legitimate_combined_insn): New function.
29704 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
29706 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
29707 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
29708 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
29710 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
29712 * doc/frontends.texi (G++ and GCC): Remove references to Java.
29714 2017-01-13 Jeff Law <law@redhat.com>
29716 PR tree-optimization/33562
29717 PR tree-optimization/61912
29718 PR tree-optimization/77485
29719 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
29721 (delete_dead_assignment): Likewise.
29722 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
29723 statement to delete_dead_call and delete_dead_assignment.
29725 2017-01-13 David Malcolm <dmalcolm@redhat.com>
29728 * substring-locations.c (format_warning_va): Strengthen case 1 so
29729 that both endpoints of the substring must be within the format
29730 range for just the substring to be printed.
29732 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
29734 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
29735 * config/i386/i386.c (ix86_target_string): Add missing options
29736 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
29737 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
29738 flags_other and ix86_target_other to flags2_other. Display unknown
29740 (ix86_valid_target_attribute_inner_p): Add missing options and
29741 reorder options by implied ISAs, as in ix86_target_string.
29743 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29745 * hash-table.h (hash_table::too_empty_p): New function.
29746 (hash_table::expand): Use it.
29747 (hash_table::traverse): Likewise.
29748 (hash_table::empty_slot): Use sizeof (value_type) instead of
29749 sizeof (PTR) to convert bytes to elements. Shrink the table
29750 if the current size is excessive for the current number of
29753 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29755 * ira-costs.c (record_reg_classes): Break from the inner loop
29756 early once alt_fail is known to be true. Update outer loop
29757 handling accordingly.
29759 2017-01-13 Jeff Law <law@redhat.com>
29761 * tree-ssa-dse.c (decrement_count): New function.
29762 (increment_start_addr, maybe_trim_memstar_call): Likewise.
29763 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
29764 when we know the partially dead statement is a mem* function.
29766 PR tree-optimization/61912
29767 PR tree-optimization/77485
29768 * tree-ssa-dse.c: Include expr.h.
29769 (maybe_trim_constructor_store): New function.
29770 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
29772 PR tree-optimization/33562
29773 PR tree-optimization/61912
29774 PR tree-optimization/77485
29775 * doc/invoke.texi: Document new dse-max-object-size param.
29776 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
29777 * tree-ssa-dse.c: Include params.h.
29778 (dse_store_status): New enum.
29779 (initialize_ao_ref_for_dse): New, partially extracted from
29781 (valid_ao_ref_for_dse, normalize_ref): New.
29782 (setup_live_bytes_from_ref, compute_trims): Likewise.
29783 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
29784 (maybe_trim_partially_dead_store): Likewise.
29785 (maybe_trim_complex_store): Likewise.
29786 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
29787 Track what bytes live from the original store. Return tri-state
29788 for dead, partially dead or live.
29789 (dse_dom_walker): Add constructor, destructor and new private members.
29790 (delete_dead_call, delete_dead_assignment): New extracted from
29792 (dse_optimize_stmt): Make a member of dse_dom_walker.
29793 Use initialize_ao_ref_for_dse.
29795 PR tree-optimization/33562
29796 PR tree-optimization/61912
29797 PR tree-optimization/77485
29798 * sbitmap.h (bitmap_count_bits): Prototype.
29799 (bitmap_clear_range, bitmap_set_range): Likewise.
29800 * sbitmap.c (bitmap_clear_range): New function.
29801 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
29803 2017-01-13 Martin Liska <mliska@suse.cz>
29806 * function.c (set_cfun): Add new argument force.
29807 * function.h (set_cfun): Likewise.
29808 * ipa-inline-transform.c (inline_call): Use the function when
29809 strict alising from is dropped for function we inline to.
29811 2017-01-13 Richard Biener <rguenther@suse.de>
29813 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
29814 for dumping GIMPLE INTEGER_CSTs.
29816 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29818 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
29819 to 201112L since C++17.
29821 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
29824 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
29825 if -fsanitize=kernel-address is present.
29827 2017-01-13 Richard Biener <rguenther@suse.de>
29829 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
29830 as _Literal ( type ) number in case usual suffixes do not
29831 preserve all information.
29833 2017-01-13 Richard Biener <rguenther@suse.de>
29835 PR tree-optimization/77283
29836 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
29837 and ssa-iterators.h.
29838 (is_feasible_trace): Implement a cost model based on joiner
29841 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
29844 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
29845 char or short to __float128/_Float128 directly.
29847 2017-01-12 Martin Sebor <msebor@redhat.com>
29849 to -Wformat-overflow.
29850 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
29851 (min_bytes_remaining): Same.
29852 (get_string_length): Same.
29853 (format_string): Same.
29854 (format_directive): Same.
29856 (pass_sprintf_length::handle_gimple_call): Same.
29858 2017-01-12 Jakub Jelinek <jakub@redhat.com>
29860 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
29861 info.nowrite calls with no lhs that can't throw. Return bool
29862 whether gsi_remove has been called or not.
29863 (pass_sprintf_length::handle_gimple_call): Return bool whether
29864 try_substitute_return_value called gsi_remove. Formatting fix.
29865 (pass_sprintf_length::execute): Don't use gsi_remove if
29866 handle_gimple_call returned true.
29869 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
29870 be removed due to side-effects, don't remove following barrier nor
29871 turn the successor edge into fallthru edge.
29873 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29876 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
29877 element-reversing loads and stores as not swappable.
29879 2017-01-12 Nathan Sidwell <nathan@acm.org>
29880 Nicolai Stange <nicstange@gmail.com>
29882 * combine.c (try_combine): Don't ignore result of overlap checking
29883 loop. Combine overlap & asm check into single loop.
29885 2017-01-12 Richard Biener <rguenther@suse.de>
29887 * tree-pretty-print.c (dump_generic_node): Provide -gimple
29888 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
29890 2017-01-12 Richard Biener <rguenther@suse.de>
29892 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
29893 and TS_TARGET_OPTION directly derive from TS_BASE.
29894 * tree-core.h (tree_optimization_option): Derive from tree_base.
29895 (tree_target_option): Likewise.
29897 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
29899 * config/i386/i386.c (memory_address_length): Increase len
29900 only when rip_relative_addr_p returns false.
29902 2017-01-11 Julia Koval <julia.koval@intel.com>
29904 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
29905 (OPTION_MASK_ISA_SGX_SET): New.
29906 (ix86_handle_option): Handle OPT_msgx.
29907 * config.gcc: Added sgxintrin.h.
29908 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
29909 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
29910 * config/i386/i386.c (ix86_target_string): Add -msgx.
29912 (ix86_option_override_internal): Handle new options.
29913 (ix86_valid_target_attribute_inner_p): Add sgx.
29914 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
29915 * config/i386/i386.opt: Add msgx.
29916 * config/i386/sgxintrin.h: New file.
29917 * config/i386/x86intrin.h: Add sgxintrin.h.
29919 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29922 * fold-const.c (maybe_nonzero_address): Return 1 for function
29924 (tree_single_nonzero_warnv_p): Don't handle function local objects
29928 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
29931 2017-01-11 David Malcolm <dmalcolm@redhat.com>
29934 * opts.c: Include "spellcheck.h"
29935 (struct string_fragment): New struct.
29936 (struct edit_distance_traits<const string_fragment &>): New
29938 (get_closest_sanitizer_option): New function.
29939 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
29941 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29943 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
29945 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
29946 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
29947 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
29948 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
29949 for initial die_offset if dwarf_split_debug_info.
29950 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
29951 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
29953 (output_skeleton_debug_sections): Formatting fix. Use
29954 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
29955 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
29957 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
29959 * config/arm/cortex-a53.md: Add bypasses for
29960 cortex_a53_r2f_cvt.
29961 (cortex_a53_r2f): Only use for transfers.
29962 (cortex_a53_f2r): Likewise.
29963 (cortex_a53_r2f_cvt): Add reservation for conversions.
29964 (cortex_a53_f2r_cvt): Likewise.
29966 2017-01-11 Tamar Christina <tamar.christina@arm.com>
29968 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
29969 to all inlined functions, change static to extern.
29971 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
29974 * config/arm/arm.c (legitimize_pic_address): Handle reference to
29976 (arm_assemble_integer): Likewise.
29978 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
29980 * config.gcc: Use new awk script to check CPU, FPU and architecture
29981 parameters for --with-... options.
29982 * config/arm/parsecpu.awk: New file
29983 * config/arm/arm-cpus.in: New file.
29984 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
29986 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
29988 * config/arm/t-arm: Update dependency rules.
29989 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
29990 of processing .def files.
29991 * config/arm/genopt.sh: Deleted.
29992 * config/arm/gentune.sh: Deleted.
29993 * config/arm/arm-cores.def: Deleted.
29994 * config/arm/arm-arches.def: Deleted.
29995 * config/arm/arm-fpus.def: Deleted.
29996 * config/arm/arm-tune.md: Regenerated.
29997 * config/arm/arm-tables.opt: Regenerated.
29998 * config/arm/arm-cpu.h: New generated file.
29999 * config/arm/arm-cpu-data.h: New generated file.
30000 * config/arm/arm-cpu-cdata.h: New generated file.
30002 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
30005 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
30007 (input_varpool_node): Unpack dynamically_initialized bit.
30009 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
30011 PR rtl-optimization/79032
30012 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
30013 the alignment of the adjusted memory reference against that of MODE,
30014 instead of the alignment of the original memory reference.
30016 2017-01-11 Martin Jambor <mjambor@suse.cz>
30018 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
30020 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
30021 decorated functions.
30023 2017-01-11 Richard Biener <rguenther@suse.de>
30025 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
30026 set range/nonnull info for PHI results. Do not set it on
30027 stmts marked for removal.
30029 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
30031 * expr.c (store_field): In the bitfield case, fetch the return value
30032 from the registers before applying a single big-endian adjustment.
30033 Always do a final load for a BLKmode value not larger than a word.
30035 2017-01-10 David Malcolm <dmalcolm@redhat.com>
30038 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30039 that we correctly handle column numbers greater than
30040 LINE_MAP_MAX_COLUMN_NUMBER.
30042 2017-01-10 Martin Sebor <msebor@redhat.com>
30044 PR middle-end/78245
30045 * gimple-ssa-sprintf.c (get_destination_size): Call
30046 {init,fini}object_sizes.
30047 * tree-object-size.c (addr_object_size): Adjust.
30048 (pass_through_call): Adjust.
30049 (pass_object_sizes::execute): Adjust.
30050 * tree-object-size.h (fini_object_sizes): Declare.
30052 2017-01-10 Martin Sebor <msebor@redhat.com>
30054 PR tree-optimization/78775
30055 * builtins.c (get_size_range): Move...
30056 * calls.c: ...to here.
30057 (alloc_max_size): Accept zero argument.
30058 (operand_signed_p): Remove.
30059 (maybe_warn_alloc_args_overflow): Call get_size_range.
30060 * calls.h (get_size_range): Declare.
30062 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
30064 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
30065 from TI's devices.csv file as of September 2016.
30066 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
30068 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
30070 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
30071 * doc/invoke.texi: Likewise.
30072 * doc/md.texi: Likewise.
30073 * doc/objc.texi: Likewise.
30075 2017-01-10 Joshua Conner <joshconner@google.com>
30077 * config/arm/fuchsia-elf.h: New file.
30078 * config/fuchsia.h: New file.
30079 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
30080 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
30082 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
30084 2016-01-10 Richard Biener <rguenther@suse.de>
30086 PR tree-optimization/79034
30087 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30088 Propagate out degenerate PHIs in the joiner.
30090 2017-01-10 Martin Liska <mliska@suse.cz>
30092 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
30093 (sort_congruence_classes_by_decl_uid): Likewise.
30094 (sort_congruence_class_groups_by_decl_uid): Likewise.
30095 (sem_item_optimizer::merge_classes): Sort class, groups in these
30096 classes and members in the groups by DECL_UID of declarations.
30097 This would make merge operations stable.
30099 2017-01-10 Martin Liska <mliska@suse.cz>
30101 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30102 usage of m_classes_vec.
30103 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30104 (sem_item_optimizer::get_group_by_hash): Likewise.
30105 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30106 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30107 (sem_item_optimizer::verify_classes): Likewise.
30108 (sem_item_optimizer::process_cong_reduction): Likewise.
30109 (sem_item_optimizer::dump_cong_classes): Likewise.
30110 (sem_item_optimizer::merge_classes): Likewise.
30111 * ipa-icf.h (congruence_class_hash): Rename from
30112 congruence_class_group_hash. Remove declaration of m_classes_vec.
30114 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30116 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30117 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30118 * config.gcc: Add avx512vpopcntdqintrin.h.
30119 * config/i386/avx512vpopcntdqintrin.h: New.
30120 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30121 * config/i386/i386-builtin-types.def: Add new types.
30122 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30123 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30124 __builtin_ia32_vpopcountq_v8di_mask): New.
30125 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30126 __AVX512VPOPCNTDQ__.
30127 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30128 (PTA_AVX512VPOPCNTDQ): Define.
30129 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30130 TARGET_AVX512VPOPCNTDQ_P): Define.
30131 * config/i386/i386.opt: Add mavx512vpopcntdq.
30132 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30133 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30135 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30137 PR middle-end/77484
30138 * predict.def (PRED_CALL): Set to 67.
30140 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30142 * expr.c (store_field): In the bitfield case, if the value comes from
30143 a function call and is of an aggregate type returned in registers, do
30144 not modify the field mode; extract the value in all cases if the mode
30145 is BLKmode and the size is not larger than a word.
30147 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30150 * config/i386/cpuid.h: Fix undefined behavior.
30152 2017-01-04 Jeff Law <law@redhat.com>
30154 PR tree-optimization/79007
30155 PR tree-optimization/67955
30156 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30157 conservative for pt.null when flag_non_call_exceptions is on.
30159 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30161 PR translation/79019
30162 PR translation/79020
30163 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30164 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30165 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30167 * config/avr/avr.opt (maccumulate-args): Likewise.
30168 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30169 * common.opt (freport-bug): Likewise.
30170 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30171 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30172 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30173 translatable string.
30174 * config/i386/i386.c (function_value_32): Likewise.
30175 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
30176 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
30178 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
30179 * common/config/msp430/msp430-common.c (msp430_handle_option):
30181 * symtab.c (symtab_node::verify_base): Likewise.
30182 * opts.c (set_debug_level): Likewise.
30183 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
30184 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
30185 missing whitespace to translatable strings.
30186 * config/avr/avr.md (bswapsi2): Fix typo in comment.
30187 * config/sh/superh.h: Likewise.
30188 * config/i386/xopintrin.h: Likewise.
30189 * config/i386/znver1.md: Likewise.
30190 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
30191 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
30192 * double-int.h (struct double_int): Likewise.
30193 * double-int.c (div_and_round_double): Likewise.
30194 * wide-int.cc: Likewise.
30195 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
30196 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
30197 * cfgcleanup.c (crossjumps_occured): Renamed to ...
30198 (crossjumps_occurred): ... this.
30199 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
30202 PR tree-optimization/78899
30203 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
30204 returning bool return struct loop *, NULL for failure and the new
30206 (versionable_outer_loop_p): Don't version outer loop if it has
30207 dont_vectorized bit set.
30208 (tree_if_conversion): When versioning outer loop, ensure
30209 tree_if_conversion is performed also on the inner loop of the
30210 non-vectorizable outer loop copy.
30211 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
30212 LOOP_VECTORIZED in inner loop of the scalar outer loop and
30213 prevent vectorization of it.
30214 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
30215 the outer loop vectorization of the non-scalar version is attempted
30216 before vectorization of the inner loop in scalar version. If
30217 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
30218 vectorization of its inner loop.
30219 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
30220 has 2 inner loops, rename also on edges from bb whose single pred
30221 is outer_loop->header. Fix typo in function comment.
30223 2017-01-09 Martin Sebor <msebor@redhat.com>
30226 * asan.c (asan_emit_stack_protection): Increase local buffer size
30227 to avoid snprintf truncation warning.
30229 2017-01-09 Andrew Pinski <apinski@cavium.com>
30231 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
30232 to reference thunderx2t99 for the tuning structure
30233 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
30235 (thunderx2t99_extra_costs): This.
30236 * config/aarch64/aarch64-tune.md: Regenerate.
30237 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
30238 (vulcan_addrcost_table): This.
30239 (vulcan_regmove_cost): Rename to ...
30240 (thunderx2t99_regmove_cost): This.
30241 (vulcan_vector_cost): Rename to ...
30242 (thunderx2t99_vector_cost): this.
30243 (vulcan_branch_cost): Rename to ...
30244 (thunderx2t99_branch_cost): This.
30245 (vulcan_tunings): Rename to ...
30246 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
30247 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
30249 2017-01-09 Martin Jambor <mjambor@suse.cz>
30253 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
30254 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
30255 (propagate_vr_accross_jump_function): Use the above function for all
30256 value range computations for pass-through jump functions and type
30257 converasion from explicit value range values.
30258 (ipcp_propagate_stage): Do not attempt to deduce types of formal
30259 parameters from TYPE_ARG_TYPES.
30260 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
30261 (ipa_write_node_info): Stream type of the actual argument.
30262 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
30264 2017-01-09 Martin Liska <mliska@suse.cz>
30267 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
30268 (lookup_compiler): Do not show error message with have_E.
30270 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30272 PR tree-optimization/78938
30273 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
30274 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
30275 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
30276 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
30279 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30281 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
30284 2017-01-09 Richard Biener <rguenther@suse.de>
30286 PR tree-optimization/78997
30287 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
30288 name condition properly.
30290 2017-01-09 Richard Biener <rguenther@suse.de>
30293 * dwarf2out.c (is_cxx): New overload with context.
30294 (is_naming_typedef_decl): Use it.
30296 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30298 * invoke.texi (Option Summary): Correct spacing in option lists
30299 and add line breaks to fix over-long lines.
30301 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30303 PR middle-end/17660
30305 * extend.texi (Common Variable Attributes): Add xref to GCC
30306 Internals manual to explain mode attribute keywords.
30308 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30311 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
30312 and Preprocessor Options.
30313 (Options for Linking): Document -pthread here....
30314 (RS/6000 and PowerPC Options): ...not here.
30315 (Solaris 2 Options): ...or here.
30316 * doc/cppopts.texi: Document -pthread.
30318 2017-01-08 Martin Sebor <msebor@redhat.com>
30320 PR middle-end/77708
30321 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
30322 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
30323 New member functions.
30324 (format_directive): Used them.
30326 (pass_sprintf_length::handle_gimple_call): Same.
30327 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
30328 to avoid truncation for any argument.
30329 (extract_affine_mul): Same.
30330 * tree.c (get_file_function_name): Same.
30332 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30334 PR middle-end/77484
30335 * predict.def (PRED_INDIR_CALL): Set to 86.
30337 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30339 PR preprocessor/54124
30340 * doc/cppopts.texi: Reformat -d subtable to list the full name
30341 of the options. Add cross-reference to the docs for the general
30342 compiler -d options.
30343 * doc/invoke.texi (Developer Options): Add cross-reference to the
30344 preprocessor-specific -d option documentation.
30346 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30348 PR preprocessor/13498
30349 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
30350 redudant material, and reflect new command-line options.
30351 (System Headers): Likewise.
30353 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30355 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
30356 -isystem, and -idirafter. Copy-edit.
30357 * doc/cppopts.texi: Copy-edit. Remove contradiction about
30358 default for -ftrack-macro-expansion. Delete obsolete and
30359 badly-formatted implementation details about -fdebug-cpp output.
30360 * doc/cppwarnopts.texi: Copy-edit.
30362 2017-01-07 David Malcolm <dmalcolm@redhat.com>
30365 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30366 that the transition from a max line width >= 1<<10 to narrower
30367 lines works correctly.
30369 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
30371 * doc/options.texi (PerFunction): New.
30372 * opt-functions.awk (switch_flags): Map both Optimization and
30373 PerFunction to CL_OPTIMIZATION.
30374 * opth-gen.awk: Test for PerFunction flag along with
30376 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
30377 it only when the latter is present. Skip those that don't in
30378 the hash function generator.
30379 * common.opt (fvar-tracking): Mark as PerFunction instead of
30381 (fvar-tracking-assignments): Likewise.
30382 (fvar-tracking-assignments-toggle): Likewise.
30383 (fvar-tracking-uninit): Likewise.
30385 2017-01-07 Jakub Jelinek <jakub@redhat.com>
30387 PR translation/79018
30388 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
30391 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
30394 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
30395 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
30396 TARGET_LONG_JUMP_TABLE_OFFSETS.
30397 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
30398 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
30399 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
30400 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
30401 * config/m68k/m68k.md (tablejump expander): Likewise.
30402 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
30403 TARGET_LONG_JUMP_TABLE_OFFSETS.
30404 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
30405 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
30407 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
30408 David Holsgrove <david.holsgrove@xilinx.com>
30410 * common/config/microblaze/microblaze-common.c
30411 (TARGET_EXCEPT_UNWIND_INFO): Remove.
30412 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
30414 * config/microblaze/microblaze.c (microblaze_must_save_register)
30415 (microblaze_expand_epilogue, microblaze_return_addr): Handle
30417 (microblaze_eh_return): New function.
30418 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
30419 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
30420 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
30421 * config/microblaze/microblaze.md (eh_return): New pattern.
30423 2017-01-06 Jakub Jelinek <jakub@redhat.com>
30425 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
30426 GCC_DIAGNOSTIC_STRINGIFY): Define.
30428 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
30430 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30432 * config/arm/arm.md (<mcrr>): New.
30434 * config/arm/arm.c (arm_arch5te): New.
30435 (arm_option_override): Set arm_arch5te.
30436 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
30438 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
30439 (arm_mcrr_qualifiers): ... this. New.
30440 (MRRC_QUALIFIERS): Define to...
30441 (arm_mrrc_qualifiers): ... this. New.
30442 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
30444 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
30445 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
30446 (MRRCI, mrrc, MRRC): New.
30447 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
30448 VUNSPEC_MRRC2): New.
30450 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30452 * config/arm/arm.md (<mcr>): New.
30454 * config/arm/arm.c (arm_coproc_builtin_available): Add
30455 support for mcr, mrc, mcr2 and mrc2.
30456 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
30457 (arm_mcr_qualifiers): ... this. New.
30458 (MRC_QUALIFIERS): Define to ...
30459 (arm_mrc_qualifiers): ... this. New.
30460 (MCR_QUALIFIERS): Define to ...
30461 (arm_mcr_qualifiers): ... this. New.
30462 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
30464 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
30465 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
30466 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
30467 VUNSPEC_MRC2): New.
30469 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30471 * config/arm/arm.md (*ldc): New.
30475 * config/arm/arm.c (arm_coproc_builtin_available): Add
30476 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
30477 (arm_coproc_ldc_stc_legitimate_address): New.
30478 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
30479 'qualifier_const_pointer'.
30480 (LDC_QUALIFIERS): Define to...
30481 (arm_ldc_qualifiers): ... this. New.
30482 (STC_QUALIFIERS): Define to...
30483 (arm_stc_qualifiers): ... this. New.
30484 * config/arm/arm-protos.h
30485 (arm_coproc_ldc_stc_legitimate_address): New.
30486 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
30487 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
30488 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
30489 stc2, stcl, stc2l): New.
30490 * config/arm/constraints.md (Uz): New.
30491 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
30492 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
30493 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
30494 VUNSPEC_STC2L): New.
30496 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30498 * config/arm/arm.md (<cdp>): New.
30499 * config/arm/arm.c (neon_const_bounds): Rename this ...
30500 (arm_const_bounds): ... this.
30501 (arm_coproc_builtin_available): New.
30502 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
30503 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
30504 (CDP_QUALIFIERS): Define to...
30505 (arm_cdp_qualifiers): ... this. New.
30507 (arm_expand_builtin_args): Add case for 6 arguments.
30508 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
30509 (arm_const_bounds): ... this.
30510 (arm_coproc_builtin_available): New.
30511 * config/arm/arm_acle.h (__arm_cdp): New.
30513 * config/arm/arm_acle_builtins.def (cdp): New.
30515 * config/arm/iterators.md (CDPI,CDP,cdp): New.
30516 * config/arm/neon.md: Rename all 'neon_const_bounds' to
30517 'arm_const_bounds'.
30518 * config/arm/types.md (coproc): New.
30519 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
30520 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
30521 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
30522 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
30524 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30526 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
30527 (UBINOP_QUALIFIERS): New.
30529 (acle_builtin_data): New. Change comment.
30530 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
30531 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
30532 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
30533 arm_acle_builtins.def.
30534 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
30535 (arm_init_acle_builtins): New.
30536 (CRC32_BUILTIN): Remove.
30537 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
30538 crc32cb, crc32ch and crc32cw.
30539 (arm_init_crc32_builtins): Remove.
30540 (arm_init_builtins): Use arm_init_acle_builtins rather
30541 than arm_init_crc32_builtins.
30542 (arm_expand_acle_builtin): New.
30543 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
30544 * config/arm/arm_acle_builtins.def: New.
30546 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30548 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
30549 (arm_builtin_datum): ... this.
30550 (arm_init_neon_builtin): Rename to ...
30551 (arm_init_builtin): ... this. Add a new parameters PREFIX
30552 and USE_SIG_IN_NAME.
30553 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
30554 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
30555 'arm_builtin_datum'.
30556 (arm_init_vfp_builtins): Likewise.
30557 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
30558 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
30559 (arm_expand_neon_args): Rename to ...
30560 (arm_expand_builtin_args): ... this. Rename builtin_arg
30561 enum values and differentiate between ARG_BUILTIN_MEMORY
30562 and ARG_BUILTIN_NEON_MEMORY.
30563 (arm_expand_neon_builtin_1): Rename to ...
30564 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
30565 values, arm_expand_builtin_args and add bool parameter NEON.
30566 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
30567 (arm_expand_vfp_builtin): Likewise.
30568 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
30570 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30572 PR middle-end/77484
30573 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
30574 * predict.c (tree_estimate_probability_bb): Reverse direction of
30575 polymorphic call predictor.
30577 2017-01-06 David Malcolm <dmalcolm@redhat.com>
30579 * passes.c (execute_one_pass): Split out pass-skipping logic into...
30580 (determine_pass_name_match): ...this new function and...
30581 (should_skip_pass_p): ...this new function.
30583 2017-01-06 Nathan Sidwell <nathan@acm.org>
30585 * ipa-visibility.c (function_and_variable_visibility): Reformat
30586 comments and long lines. Remove extrneous if.
30587 * symtab.c (symtab_node::make_decl_local): Fix code format.
30588 (symtab_node::set_section_for_node): Fix comment typo.
30590 2017-01-06 Martin Liska <mliska@suse.cz>
30593 * lra-constraints.c: Rename invariant to lra_invariant.
30594 * predict.c (set_even_probabilities): Initialize e to NULL.
30596 2017-01-05 Martin Sebor <msebor@redhat.com>
30598 PR tree-optimization/78910
30599 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
30600 (format_integer): Correct off-by-one error in the handling
30601 of precision with negative numbers in signed conversions..
30603 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
30605 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
30607 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30609 PR tree-optimization/71016
30610 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
30611 factor_out_conditional_conversion. Formatting fix.
30612 (factor_out_conditional_conversion): Add cond_stmt argument.
30613 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
30614 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
30617 2017-01-05 David Malcolm <dmalcolm@redhat.com>
30619 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
30620 read-rtl-function.o, and selftest-rtl.o.
30621 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
30622 (selftest::aarch64_test_loading_full_dump): New function.
30623 (selftest::aarch64_run_selftests): New function.
30624 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
30625 selftest::aarch64_run_selftests.
30626 * config/i386/i386.c
30627 (selftest::ix86_test_loading_dump_fragment_1): New function.
30628 (selftest::ix86_test_loading_call_insn): New function.
30629 (selftest::ix86_test_loading_full_dump): New function.
30630 (selftest::ix86_test_loading_unspec): New function.
30631 (selftest::ix86_run_selftests): Call the new functions.
30632 * emit-rtl.c (maybe_set_max_label_num): New function.
30633 * emit-rtl.h (maybe_set_max_label_num): New decl.
30634 * function.c (instantiate_decls): Guard call to
30635 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
30636 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
30638 * gensupport.c (gen_reader::gen_reader): Pass "false"
30639 for new "compact" param of rtx_reader.
30640 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
30641 rather than an empty string for NULL strings.
30642 * read-md.c: Potentially include config.h rather than bconfig.h.
30643 Wrap include of errors.h with #ifdef GENERATOR_FILE.
30644 (have_error): New global, copied from errors.c.
30645 (md_reader::read_name): Rename to...
30646 (md_reader::read_name_1): ...this, adding "out_loc" param,
30647 and converting "missing name or number" to returning false, rather
30649 (md_reader::read_name): Reimplement in terms of read_name_1.
30650 (md_reader::read_name_or_nil): New function.
30651 (md_reader::read_string): Handle "(nil)" by returning NULL.
30652 (md_reader::md_reader): Add new param "compact".
30653 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
30654 (md_reader::read_file): New method.
30655 * read-md.h (md_reader::md_reader): Add new param "compact".
30656 (md_reader::read_file): New method.
30657 (md_reader::is_compact): New accessor.
30658 (md_reader::read_name): Convert return type from void to file_location.
30659 (md_reader::read_name_or_nil): New decl.
30660 (md_reader::read_name_1): New decl.
30661 (md_reader::m_compact): New field.
30662 (noop_reader::noop_reader): Pass "false" for new "compact" param
30664 (rtx_reader::rtx_reader): Add new "compact" param.
30665 (rtx_reader::read_rtx_operand): Make virtual and convert return
30666 type from void to rtx.
30667 (rtx_reader::read_until): New decl.
30668 (rtx_reader::handle_any_trailing_information): New virtual function.
30669 (rtx_reader::postprocess): New virtual function.
30670 (rtx_reader::finalize_string): New virtual function.
30671 (rtx_reader::m_in_call_function_usage): New field.
30672 (rtx_reader::m_reuse_rtx_by_id): New field.
30673 * read-rtl-function.c: New file.
30674 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
30675 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
30676 (selftest::verify_three_block_rtl_cfg): New decl.
30677 * read-rtl-function.h: New file.
30678 * read-rtl.c: Potentially include config.h rather than bconfig.h.
30679 For host, include function.h, memmodel.h, and emit-rtl.h.
30680 (one_time_initialization): New function.
30681 (struct compact_insn_name): New struct.
30682 (compact_insn_names): New array.
30683 (find_code): Handle insn codes in compact dumps.
30684 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
30685 (bind_subst_iter_and_attr): Likewise.
30686 (add_condition_to_string): Likewise.
30687 (add_condition_to_rtx): Likewise.
30688 (apply_attribute_uses): Likewise.
30689 (add_current_iterators): Likewise.
30690 (apply_iterators): Likewise.
30691 (initialize_iterators): Guard usage of apply_subst_iterator with
30692 #ifdef GENERATOR_FILE.
30693 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
30694 (md_reader::read_mapping): Likewise.
30695 (add_define_attr_for_define_subst): Likewise.
30696 (add_define_subst_attr): Likewise.
30697 (read_subst_mapping): Likewise.
30698 (check_code_iterator): Likewise.
30699 (rtx_reader::read_rtx): Likewise. Move one-time initialization
30701 (one_time_initialization): New function.
30702 (rtx_reader::read_until): New method.
30703 (read_flags): New function.
30704 (parse_reg_note_name): New function.
30705 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
30706 Handle reuse_rtx ids.
30707 Wrap iterator lookup within #ifdef GENERATOR_FILE.
30708 Add parsing support for RTL dumps, mirroring the special-cases in
30709 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
30710 values, and calling handle_any_trailing_information.
30711 (rtx_reader::read_rtx_operand): Convert return type from void
30712 to rtx, returning return_rtx. Handle case 'e'. Call
30713 finalize_string on XSTR and XTMPL fields.
30714 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
30715 "(nil)" values were omitted. Call the postprocess vfunc on the
30717 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
30718 class ctor. Initialize m_in_call_function_usage. Call
30719 one_time_initialization.
30720 * rtl-tests.c (selftest::test_uncond_jump): Call
30721 set_new_first_and_last_insn.
30722 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
30723 * selftest-rtl.c: New file.
30724 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
30725 (selftest::get_insn_by_uid): New decl.
30726 * selftest-run-tests.c (selftest::run_tests): Call
30727 read_rtl_function_c_tests.
30728 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
30729 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
30732 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
30734 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
30735 operands in a special way. Assert that pos+len <= mode precision.
30737 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30739 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
30740 3 argument Alias with unlimited for the negative form.
30741 (fno-vect-cost-model): Removed.
30743 2017-01-05 Martin Liska <mliska@suse.cz>
30745 * hsa-gen.c (gen_hsa_divmod): New function.
30746 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
30748 2017-01-05 Martin Liska <mliska@suse.cz>
30751 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
30754 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30756 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
30757 small constant length operands.
30759 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30761 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
30762 between loop iterations.
30764 2017-01-05 Martin Liska <mliska@suse.cz>
30767 * gimplify.c (gimplify_decl_expr): Compare to
30768 asan_poisoned_variables instread of checking flags.
30769 (gimplify_target_expr): Likewise.
30770 (gimplify_expr): Likewise.
30771 (gimplify_function_tree): Conditionally initialize
30772 asan_poisoned_variables.
30774 2017-01-04 Jeff Law <law@redhat.com>
30776 PR tree-optimizatin/78812
30777 * rtl.h (contains_mem_rtx_p): Prototype.
30778 * ifcvt.c (containts_mem_rtx_p): Move from here to...
30779 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
30780 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
30781 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
30782 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
30784 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30786 * input.c (assert_char_at_range): Default-initialize actual_range.
30788 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30790 * df-scan.c (df_ref_create_structure): Make regno unsigned,
30791 to match the caller.
30793 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30795 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
30796 insns after final jump in test to emit dummy move.
30798 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30800 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
30801 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
30803 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30805 * multiple_target.c (create_dispatcher_calls): Init e_next.
30806 * tree-ssa-loop-split.c (split_loop): Init border.
30807 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
30810 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
30815 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
30816 (altivec_register_operand): Do not return true if the operand
30817 contains a SUBREG mixing SImode and SFmode.
30818 (vsx_register_operand): Likewise.
30819 (vsx_reg_sfsubreg_ok): New predicate.
30820 (vfloat_operand): Do not return true if the operand contains a
30821 SUBREG mixing SImode and SFmode.
30822 (vint_operand): Likewise.
30823 (vlogical_operand): Likewise.
30824 (gpc_reg_operand): Likewise.
30825 (int_reg_operand): Likewise.
30826 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
30827 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
30828 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
30830 (rs6000_emit_move_si_sf_subreg): New helper function.
30831 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
30832 fixup SUBREGs involving SImode and SFmode.
30833 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
30834 numbers for the new peephole2 optimization.
30835 (peephole2 for SFmode unions): New peephole2 to optimize cases in
30836 the GLIBC math library that do AND/IOR/XOR operations on single
30837 precision floating point.
30838 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
30839 target macros to say whether we need to avoid SUBREGs mixing
30841 (TARGET_ALLOW_SF_SUBREG): Likewise.
30842 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
30843 (UNSPEC_SI_FROM_SF): Likewise.
30844 (iorxor): Change spacing.
30845 (and_ior_xor): New iterator for AND, IOR, and XOR.
30846 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
30847 (movdi_from_sf_zero_ext): Likewise.
30848 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
30849 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
30850 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
30851 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
30852 (fms<mode>4): Likewise.
30853 (fnma<mode>4): Likewise.
30854 (fnms<mode>4): Likewise.
30855 (nfma<mode>4): Likewise.
30856 (nfms<mode>4): Likewise.
30858 2017-01-04 Marek Polacek <polacek@redhat.com>
30861 * doc/invoke.texi: Document -Wpointer-compare.
30863 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30865 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
30868 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
30869 descriptions for -gdwarf-5 and emit them as uleb128 instead of
30872 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
30875 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
30876 documentation of the powerpc_popcntb_ok attribute.
30877 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30878 code to issue warning messages if a requested CPU configuration is
30879 not supported by the binary (assembler and loader) toolchain.
30880 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
30881 made to define a built-in function that has been disabled.
30882 (paired_init_builtins): Add assertion to prevent ICE if attempt is
30883 made to define a built-in function that has been disabled.
30884 (altivec_init_builtins): Add comment explaining why definition
30885 of the DST built-in functions is not preceded by an assertion
30886 check. Add assertions to prevent ICE if attempts are made to
30887 define an altivec predicate or an abs* built-in function that has
30889 (htm_init_builtins): Add comment explaining why definition of the
30890 htm built-in functions is not preceded by an assertion check.
30892 2017-01-04 Jeff Law <law@redhat.com>
30894 PR tree-optimizatin/67955
30895 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
30896 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
30897 the points-to solution does not include pt_null. Use DECL_PT_UID
30900 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
30902 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
30903 Use gen_int_mode instead of gen_lopwart for const_int operands.
30905 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30907 PR tree-optimization/71563
30908 * match.pd: Simplify X << Y into X if Y is known to be 0 or
30909 out of range value - has low bits known to be zero.
30911 2017-01-04 Alan Modra <amodra@gmail.com>
30913 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
30914 * configure: Regenerate.
30915 * config.in: Regenerate.
30917 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30920 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
30921 a substring of the message, but strcmp with the whole message. Ifdef
30922 ENABLE_NLS, translate the message first using dgettext.
30924 2017-01-03 Jeff Law <law@redhat.com>
30926 PR tree-optimizatin/78856
30927 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
30928 (mark_threaded_blocks): Remove code to truncate thread paths that
30929 cross multiple loop headers. Instead invalidate the cached loop
30930 iteration information and handle case of a thread path walking
30931 into an irreducible region.
30933 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
30936 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
30937 assertions. Add support for doing the signbit if the IEEE 128-bit
30938 floating point value is in a GPR.
30939 * config/rs6000/rs6000.md (Fsignbit): Delete.
30940 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
30941 Update the length attribute if the value is in a GPR.
30942 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
30943 the sign or zero extension instruction, since the value is always 0/1.
30944 (signbit<mode>2_dm2): Delete using <Fsignbit>.
30947 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
30948 extracting SImode to a GPR register so that we can generate a
30949 store, limit the vector to be in a traditional Altivec register
30950 for the vextuwrx instruction.
30952 2017-01-03 Ian Lance Taylor <iant@google.com>
30954 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
30956 2017-01-03 Martin Sebor <msebor@redhat.com>
30958 PR tree-optimization/78696
30959 * gimple-ssa-sprintf.c (format_floating): Correct handling of
30960 precision. Use MPFR for %f for greater fidelity. Correct handling
30962 (pass_sprintf_length::compute_format_length): Set width and precision
30963 specified by asrerisk to void_node for vararg functions.
30964 (try_substitute_return_value): Adjust dump output.
30966 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
30968 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
30970 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
30972 * doc/invoke.texi (SPARC options): Document -mlra as the default.
30973 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
30974 -mlra/-mno-lra was passed to the compiler.
30976 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
30978 PR rtl-optimization/65618
30979 * emit-rtl.c (try_split): Move initialization of "before" and
30980 "after" to just before the call to emit_insn_after_setloc.
30982 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
30984 * doc/md.texi (Standard Names): Remove reference to Java frontend.
30986 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
30988 * dwarf2out.c (gen_enumeration_type_die): When
30989 -gno-strict-dwarf, add a DW_AT_encoding attribute.
30991 2017-01-03 Jakub Jelinek <jakub@redhat.com>
30993 PR tree-optimization/78965
30994 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
30995 Change first argument from const call_info & to call_info &. For %n
30996 set info.nowrite to false.
30998 PR middle-end/78901
30999 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
31000 possibly throwing calls.
31002 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
31003 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
31004 and fns handling, rather than in a separate case SSA_NAME.
31006 2017-01-02 Jeff Law <law@redhat.com>
31008 * config/darwin-driver.c (darwin_driver_init): Const-correctness
31009 fixes for first_period and second_period variables.
31011 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
31014 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
31015 (*insvqi_1): New insn pattern.
31016 (*insvqi_1_mem_rex64): Ditto.
31017 (*insvqi_2): Ditto.
31018 (*insvqi_3): Rename from *insvqi.
31020 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
31022 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
31024 * doc/cfg.texi (Edges): Remove reference to Java.
31025 (Maintaining the CFG): Ditto.
31027 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31029 PR middle-end/77674
31030 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
31031 transparent aliases.
31033 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31035 PR middle-end/77484
31036 * predict.def (PRED_CALL): Update hitrate.
31037 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
31038 * predict.c (tree_estimate_probability_bb): Split CALL predictor
31039 into direct/indirect/polymorphic variants.
31041 2017-01-01 Jakub Jelinek <jakub@redhat.com>
31043 Update copyright years.
31045 * gcc.c (process_command): Update copyright notice dates.
31046 * gcov-dump.c (print_version): Ditto.
31047 * gcov.c (print_version): Ditto.
31048 * gcov-tool.c (print_version): Ditto.
31049 * gengtype.c (create_file): Ditto.
31050 * doc/cpp.texi: Bump @copying's copyright year.
31051 * doc/cppinternals.texi: Ditto.
31052 * doc/gcc.texi: Ditto.
31053 * doc/gccint.texi: Ditto.
31054 * doc/gcov.texi: Ditto.
31055 * doc/install.texi: Ditto.
31056 * doc/invoke.texi: Ditto.
31058 Copyright (C) 2017 Free Software Foundation, Inc.
31060 Copying and distribution of this file, with or without modification,
31061 are permitted in any medium without royalty provided the copyright
31062 notice and this notice are preserved.